This action might not be possible to undo. Are you sure you want to continue?
and
Forcing:
How to prove theorems about Borel sets
the hard way.
Arnold W. Miller
Department of Mathematics
480 Lincoln Dr.
Van Vleck Hall
University of Wisconsin
Madison, WI. 53706
miller@math.wisc.edu
http://www.math.wisc.edu/∼miller
This page left blank.
Note to the readers
Departing from the usual author’s statementI would like to say that I
am not responsible for any of the mistakes in this document. Any mistakes
here are the responsibility of the reader. If anybody wants to point out a
mistake to me, I promise to respond by saying “but you know what I meant
to say, don’t you?”
These are lecture notes from a course I gave at the University of Wis
consin during the Spring semester of 1993. Some knowledge of forcing is
assumed as well as a modicum of elementary Mathematical Logic, for exam
ple, the LowenheimSkolem Theorem. The students in my class had a one
semester course, introduction to mathematical logic covering the complete
ness theorem and incompleteness theorem, a set theory course using Kunen
[56], and a model theory course using Chang and Keisler [17]. Another good
reference for set theory is Jech [44]. Oxtoby [90] is a good reference for the
basic material concerning measure and category on the real line. Kuratowski
[59] and Kuratowski and Mostowski [60] are excellent references for classical
descriptive set theory. Moschovakis [89] and Kechris [54] are more modern
treatments of descriptive set theory.
The ﬁrst part is devoted to the general area of Borel hierarchies, a subject
which has always interested me. The results in section 14 and 15 are new and
answer questions from my thesis. I have also included (without permission)
an unpublished result of Fremlin (Theorem 13.4).
Part II is devoted to results concerning the low projective hierarchy. It
ends with a theorem of Harrington from his thesis that is consistent to have
Π
1
2
¯
sets of arbitrary size.
The general aim of part III and IV is to get to Louveau’s theorem. Along
the way many of the classical theorems of descriptive set theory are presented
“justintime” for when they are needed. This technology allows the reader
to keep from overﬁlling his or her memory storage device. I think the proof
given of Louveau’s Theorem 33.1 is also a little diﬀerent.
1
Questions like “Who proved what?” always interest me, so I have included
my best guess here. Hopefully, I have managed to oﬀend a large number of
1
In a randomly inﬁnite Universe, any event occurring here and now with ﬁnite probabil
ity must be occurring simultaneously at an inﬁnite number of other sites in the Universe.
It is hard to evaluate this idea any further, but one thing is certain: if it is true then it is
certainly not original!– The Anthropic Cosmological Principle, by John Barrow and Frank
Tipler.
mathematicians.
AWM April 1995
Added April 2001: Several brave readers ignored my silly joke in the ﬁrst
paragraph and sent me corrections and comments. Since no kind act should
go unpunished, let me say that any mistakes introduced into the text are
their fault.
Contents
1 What are the reals, anyway? 1
I On the length of Borel hierarchies 4
2 Borel Hierarchy 4
3 Abstract Borel hierarchies 10
4 Characteristic function of a sequence 13
5 Martin’s Axiom 16
6 Generic G
δ
18
7 αforcing 22
8 Boolean algebras 28
9 Borel order of a ﬁeld of sets 33
10 CH and orders of separable metric spaces 35
11 MartinSolovay Theorem 38
12 Boolean algebra of order ω
1
43
13 Luzin sets 48
14 Cohen real model 52
15 The random real model 65
16 Covering number of an ideal 73
II Analytic sets 78
17 Analytic sets 78
18 Constructible wellorderings 82
19 Hereditarily countable sets 84
20 Shoenﬁeld Absoluteness 86
21 MansﬁeldSolovay Theorem 88
22 Uniformity and Scales 90
23 Martin’s axiom and Constructibility 95
24 Σ
1
2
wellorderings 97
25 Large Π
1
2
sets 98
III Classical Separation Theorems 102
26 SouslinLuzin Separation Theorem 102
27 Kleene Separation Theorem 104
28 Π
1
1
Reduction 107
29 ∆
1
1
codes 109
IV Gandy Forcing 113
30 Π
1
1
equivalence relations 113
31 Borel metric spaces and lines in the plane 119
32 Σ
1
1
equivalence relations 125
33 Louveau’s Theorem 130
34 Proof of Louveau’s Theorem 137
References 142
Index 152
1 WHAT ARE THE REALS, ANYWAY? 1
1 What are the reals, anyway?
Let ω = ¦0, 1, . . .¦ and let ω
ω
(Baire space) be the set of functions from
ω to ω. Let ω
<ω
be the set of all ﬁnite sequences of elements of ω. [s[ is
the length of s, ¸) is the empty sequence, and for s ∈ ω
<ω
and n ∈ ω let
sˆn denote the sequence which starts out with s and has one more element
n concatenated onto the end. The basic open sets of ω
ω
are the sets of the
form:
[s] = ¦x ∈ ω
ω
: s ⊆ x¦
for s ∈ ω
<ω
. A subset of ω
ω
is open iﬀ it is the union of basic open subsets.
It is separable (has a countable dense subset) since it is second countable
(has a countable basis). The following deﬁnes a complete metric on ω
ω
:
d(x, y) =
_
0 if x = y
1
n+1
if x n = y n and x(n) ,= y(n)
Cantor space 2
ω
is the subspace of ω
ω
consisting of all functions from ω to
2 = ¦0, 1¦. It is compact.
Theorem 1.1 (Baire [4]) ω
ω
is homeomorphic to the irrationals P.
Proof:
First replace ω by the integers Z. We will construct a mapping from Z
ω
to P. Enumerate the rationals Q = ¦q
n
: n ∈ ω¦. Inductively construct a
sequence of open intervals ¸I
s
: s ∈ Z
<ω
) satisfying the following:
1. I
= R, and for s ,= ¸) each I
s
is a nontrivial open interval in R with
rational endpoints,
2. for every s ∈ Z
<ω
and n ∈ Z I
sˆn
⊆ I
s
,
3. the right end point of I
sˆn
is the left end point of I
sˆn+1
,
4. ¦I
sˆn
: n ∈ Z¦ covers all of I
s
except for their endpoints,
5. the length of I
s
is less than
1
s
for s ,= ¸), and
6. the n
th
rational q
n
is an endpoint of I
t
for some [t[ ≤ n + 1.
Descriptive Set Theory and Forcing 2
Deﬁne the function f : Z
ω
→ P as follows. Given x ∈ Z
ω
the set
n∈ω
I
xn
must consist of a singleton irrational. It is nonempty because
closure(I
xn+1
) ⊆ I
xn
.
It is a singleton because their diameters shrink to zero.
So we can deﬁne f by
¦f(x)¦ =
n∈ω
I
xn
.
The function f is onetoone because if s and t are incomparable then I
s
and
I
t
are disjoint. It is onto since for every u ∈ P and n ∈ ω there is a unique s
of length n with u ∈ I
s
. It is a homeomorphism because
f([s]) = I
s
∩ P
and the sets of the form I
s
∩ P form a basis for P.
Note that the map given is also an order isomorphism from Z
ω
with the
lexicographical order to P with it’s usual order.
We can identify 2
ω
with P(ω), the set of all subsets of ω, by identifying
a subset with its characteristic function. Let F = ¦x ∈ 2
ω
: ∀
∞
n x(n) = 0¦
(the quantiﬁer ∀
∞
stands for “for all but ﬁnitely many n”). F corresponds
to the ﬁnite sets and so 2
ω
¸ F corresponds to the inﬁnite subsets of ω which
we write as [ω]
ω
.
Theorem 1.2 ω
ω
is homeomorphic to [ω]
ω
.
Proof:
Let f ∈ ω
ω
and deﬁne F(f) ∈ 2
ω
to be the sequence of 0’s and 1’s
determined by:
F(f) = 0
f(0)
ˆ1ˆ0
f(1)
ˆ1ˆ0
f(2)
ˆ1ˆ
where 0
f(n)
refers to a string of length f(n) of zeros. The function F is a
onetoone onto map from ω
ω
to 2
ω
¸ F. It is a homeomorphism because
F([s]) = [t] where t = 0
s(0)
ˆ1ˆ0
s(1)
ˆ1ˆ0
s(2)
ˆ1ˆ ˆ0
s(n)
ˆ1 where [s[ = n +1.
Descriptive Set Theory and Forcing 3
Note that sets of the form [t] where t is a ﬁnite sequence ending in a one
form a basis for 2
ω
¸ F.
I wonder why ω
ω
is called Baire space? The earliest mention of this I have
seen is in Sierpi´ nski [99] where he refers to ω
ω
as the 0dimensional space of
Baire. Sierpi´ nski also says that Frechet was the ﬁrst to describe the metric
d given above. Unfortunately, Sierpi´ nski [99] gives very few references.
2
The classical proof of Theorem 1.1 is to use “continued fractions” to
get the correspondence. Euler [19] proved that every rational number gives
rise to a ﬁnite continued fraction and every irrational number gives rise to
an inﬁnite continued fraction. Brezinski [13] has more on the history of
continued fractions.
My proof of Theorem 1.1 allows me to remain blissfully ignorant
3
of even
the elementary theory of continued fractions.
Cantor space, 2
ω
, is clearly named so because it is homeomorphic to
Cantor’s middle two thirds set.
2
I am indebted to John C. Morgan II for supplying the following reference and comment.
“Baire introduced his space in Baire [3]. Just as coeﬃcients of linear equations evolved into
matrices the sequences of natural numbers in continued fraction developments of irrational
numbers were liberated by Baire’s mind to live in their own world.”
3
It is impossible for a man to learn what he thinks he already knows.Epictetus
4
Part I
On the length of Borel
hierarchies
2 Borel Hierarchy
Deﬁnitions. For X a topological space deﬁne Σ
0
1
¯
to be the open subsets of
X. For α > 1 deﬁne A ∈ Σ
0
α
¯
iﬀ there exists a sequence ¸B
n
: n ∈ ω)
with each B
n
∈ Σ
0
β
n
¯
for some β
n
< α such that
A =
_
n∈ω
∼ B
n
where ∼ B is the complement of B in X, i.e., ∼ B = X ¸ B. Deﬁne
Π
0
α
¯
= ¦∼ B : B ∈ Σ
0
α
¯
¦
and
∆
0
α
¯
= Σ
0
α
¯
∩ Π
0
α
¯
.
The Borel subsets of X are deﬁned by Borel(X) =
α<ω
1
Σ
0
α
¯
(X). It is clearly
the smallest family of sets containing the open subsets of X and closed under
countable unions and complementation.
Theorem 2.1 Σ
0
α
¯
is closed under countable unions, Π
0
α
¯
is closed under
countable intersections, and ∆
0
α
¯
is closed under complements. For any α,
Π
0
α
¯
(X) ⊆ Σ
0
α+1
¯
(X) and Σ
0
α
¯
(X) ⊆ Π
0
α+1
¯
(X).
Proof:
That Σ
0
α
¯
is closed under countable unions is clear from its deﬁnition.
It follows from DeMorgan’s laws by taking complements that Π
0
α
¯
is closed
under countable intersections.
Theorem 2.2 If f : X → Y is continuous and A ∈ Σ
0
α
¯
(Y ), then f
−1
(A) is
in Σ
0
α
¯
(X).
2 BOREL HIERARCHY 5
This is an easy induction since it is true for open sets (Σ
0
1
¯
) and f
−1
passes
over complements and unions.
Theorem 2.2 is also, of course, true for Π
0
α
¯
or ∆
0
α
¯
in place of Σ
0
α
¯
.
Theorem 2.3 Suppose X is a subspace of Y , then
Σ
0
α
¯
(X) = ¦A ∩ X : A ∈ Σ
0
α
¯
(Y )¦.
Proof:
For Σ
0
1
¯
it follows from the deﬁnition of subspace. For α > 1 it is an easy
induction.
The class of sets Σ
0
2
¯
is also referred to as F
σ
and the class Π
0
2
¯
as G
δ
.
Theorem 2.3 is true for Π
0
α
¯
in place of Σ
0
α
¯
, but not in general for ∆
0
α
¯
.
For example, let X be the rationals in [0, 1] and Y be [0,1]. Then since X is
countable every subset of X is Σ
0
2
¯
in X and hence ∆
0
2
¯
in X. If Z contained
in X is dense and codense then Z is ∆
0
2
¯
in X (every subset of X is), but
there is no ∆
0
2
¯
set Q in Y = [0, 1] whose intersection with X is Z. (If Q
is G
δ
and F
σ
and contains Z then its comeager, but a comeager F
σ
in [0, 1]
contains an interval.)
Theorem 2.4 For X a topological space and Π
0
1
¯
(X) ⊆ Π
0
2
¯
(X) (i.e., closed
sets are G
δ
), then
1. Π
0
α
¯
(X) ⊆ Π
0
α+1
¯
(X),
2. Σ
0
α
¯
(X) ⊆ Σ
0
α+1
¯
(X), and hence
3. Π
0
α
¯
(X) ∪ Σ
0
α
¯
(X) ⊆ ∆
0
α+1
¯
(X)
4. Σ
0
α
¯
is closed under ﬁnite intersections,
5. Π
0
α
¯
is closed under ﬁnite unions, and
6. ∆
0
α
¯
is closed under ﬁnite intersections, ﬁnite unions, and complements.
Proof:
2 BOREL HIERARCHY 6
Induction on α. For example, to see that Σ
0
α
¯
is closed under ﬁnite inter
sections, use that
(
_
n∈ω
P
n
) ∩ (
_
n∈ω
Q
n
) =
_
n,m∈ω
(P
n
∩ Q
m
)
It follows by DeMorgan’s laws that Π
0
α
¯
is closed under ﬁnite unions. ∆
0
α
¯
is
closed under ﬁnite intersections, ﬁnite unions, and complements since it is
the intersection of the two classes.
In metric spaces closed sets are G
δ
, since
C =
n∈ω
¦x : ∃y ∈ C d(x, y) <
1
n + 1
¦
for C a closed set.
The assumption that closed sets are G
δ
is necessary since if
X = ω
1
+ 1
with the order topology, then the closed set consisting of the singleton point
¦ω
1
¦ is not G
δ
; in fact, it is not in the σδlattice generated by the open
sets (the smallest family containing the open sets and closed under countable
intersections and countable unions).
Williard [112] gives an example which is a second countable Hausdorﬀ
space. Let X ⊆ 2
ω
be any nonBorel set. Let 2
ω
∗
be the space 2
ω
with the
smallest topology containing the usual topology and X as an open set. The
family of all sets of the form (B ∩ X) ∪ C where B, C are (ordinary) Borel
subsets of 2
ω
is the σδlattice generated by the open subsets of 2
ω
∗
, because:
n
(B
n
∩ X) ∪ C
n
= ( (
n
B
n
∪ C
n
) ∩ X) ∪
n
C
n
_
n
(B
n
∩ X) ∪ C
n
= ((
_
n
B
n
) ∩ X) ∪
_
n
C
n
.
Note that ∼ X is not in this σδlattice.
M.Laczkovich has pointed out to me that the class Π
0
3
¯
(X) where for the
ordered space X = ω
1
+ 1 is not closed under ﬁnite unions:
The elements of Π
0
3
¯
are of the form ∩
∞
n=1
A
n
where each A
n
is either open
or F
σ
. This implies that Π
0
3
¯
contains the open sets and the closed sets.
2 BOREL HIERARCHY 7
However, the union of an open set and a closed set is not necessarily in Π
0
3
¯
.
Let A be the set of isolated points of ω
1
+ 1 and let B = ¦ω
1
¦. Then A is
open and B is closed. But A ∪ B / ∈ Π
0
3
¯
. Suppose A ∪ B = ∩
∞
n=1
A
n
, where
each A
n
is either open or F
σ
. If A
n
is open then ω
1
∈ A
n
implies that A
n
contains an unbounded closed subset of ω
1
. If A
n
is F
σ
then A ⊆ A
n
implies
the same. Therefore ∩
n
A
n
also contains an unbounded closed subset of ω
1
.
Thus A ∩ B contains a countable limit point, which is impossible.
Theorem 2.5 (Lebesgue [63]) For every α with 1 ≤ α < ω
1
Σ
0
α
¯
(2
ω
) ,= Π
0
α
¯
(2
ω
).
The proof of this is a diagonalization argument applied to a universal
set. We will need the following lemma.
Lemma 2.6 Suppose X is second countable (i.e. has a countable base), then
for every α with 1 ≤ α < ω
1
there exists a universal Σ
0
α
¯
set U ⊆ 2
ω
X,
i.e., a set U which is Σ
0
α
¯
(2
ω
X) such that for every A ∈ Σ
0
α
¯
(X) there exists
x ∈ 2
ω
such that A = U
x
where U
x
= ¦y ∈ X : (x, y) ∈ U¦.
Proof:
The proof is by induction on α. Let ¦B
n
: n ∈ ω¦ be a countable base for
X. For α = 1 let
U = ¦(x, y) : ∃n (x(n) = 1 ∧y ∈ B
n
)¦ =
_
n
(¦x : x(n) = 1¦ B
n
).
For α > 1 let β
n
be a sequence which sups up to α if α a limit, or equals
α −1 if α is a successor. Let U
n
be a universal Σ
0
β
n
¯
set. Let
¸n, m) = 2
n
(2m+ 1) −1
be the usual pairing function which gives a recursive bijection between ω
2
and ω. For any n the map g
n
: 2
ω
X → 2
ω
X is deﬁned by (x, y) →
(x
n
, y) where x
n
(m) = x(¸n, m)). This map is continuous so if we deﬁne
U
∗
n
= g
−1
n
(U
n
), then U
∗
n
is Σ
0
β
n
¯
, and because the map x → x
n
is onto it is also
a universal Σ
0
β
n
¯
set. Now deﬁne U by:
U =
_
n
∼ U
∗
n
.
2 BOREL HIERARCHY 8
U is universal for Σ
0
α
¯
because given any sequence B
n
∈ Σ
0
β
n
¯
for n ∈ ω there
exists x ∈ 2
ω
such that for every n ∈ ω we have that B
n
= (U
∗
n
)
x
= (U
n
)
x
n
(this is because the map x → ¸x
n
: n < ω) takes 2
ω
onto (2
ω
)
ω
.) But then
U
x
= (
_
n
∼ U
∗
n
)
x
=
_
n
∼ (U
∗
n
)
x
=
_
n
∼ (B
n
).
Proof of Theorem 2.5:
Let U ⊆ 2
ω
2
ω
be a universal Σ
0
α
¯
set. Let
D = ¦x : ¸x, x) ∈ U¦.
D is the continuous preimage of U under the map x → ¸x, x), so it is Σ
0
α
¯
,
but it cannot be Π
0
α
¯
because if it were, then there would be x ∈ 2
ω
with
∼ D = U
x
and then x ∈ D iﬀ ¸x, x) ∈ U iﬀ x ∈ U
x
iﬀ x ∈∼ D.
Deﬁne ord(X) to be the least α such that Borel(X) = Σ
0
α
¯
(X). Lebesgue’s
theorem says that ord(X) = ω
1
. Note that ord(X) = 1 if X is a discrete
space and that ord(Q) = 2.
Corollary 2.7 For any space X which contains a homeomorphic copy of 2
ω
(i.e., a perfect set) we have that ord(X) = ω
1
, consequently ω
ω
, R, and any
uncountable complete separable metric space have ord = ω
1
.
Proof:
If the Borel hierarchy on X collapses, then by Theorem 2.3 it also collapses
on all subspaces of X. Every uncountable complete separable metric space
contains a perfect set (homeomorphic copy of 2
ω
). To see this suppose X
is an uncountable complete separable metric space. Construct a family of
open sets ¸U
s
: s ∈ 2
<ω
) such that
1. U
s
is uncountable,
2. cl(U
sˆ0
) ∩ cl(U
sˆ1
) = ∅,
3. cl(U
sˆi
) ⊆ U
s
for i=0,1, and
4. diameter of U
s
less than 1/[s[
2 BOREL HIERARCHY 9
Then the map f : 2
ω
→ X deﬁned so that
¦f(x)¦ =
n∈ω
U
xn
gives an embedding of 2
ω
into X.
Lebesgue [63] used universal functions instead of sets, but the proof is
much the same. Corollary 33.5 of Louveau’s Theorem shows that there can be
no Borel set which is universal for all ∆
0
α
¯
sets. Miller [82] contains examples
from model theory of Borel sets of arbitrary high rank.
The notation Σ
0
α
, Π
0
β
was ﬁrst popularized by Addison [1]. I don’t know
if the “bold face” and “light face” notation is such a good idea, some copy
machines wipe it out. Consequently, I use
Σ
0
α
¯
which is blackboard boldface.
3 ABSTRACT BOREL HIERARCHIES 10
3 Abstract Borel hierarchies
Suppose F ⊆ P(X) is a family of sets. Most of the time we would like to
think of F as a countable ﬁeld of sets (i.e. closed under complements and
ﬁnite intersections) and so analogous to the family of clopen subsets of some
space.
We deﬁne the classes Π
0
α
¯
(F) analogously. Let Π
0
0
¯
(F) = F and for every
α > 0 deﬁne A ∈ Π
0
α
¯
(F) iﬀ there exists B
n
∈ Π
0
β
n
¯
for some β
n
< α such that
A =
n∈ω
∼ B
n
.
Deﬁne
• Σ
0
α
¯
(F) = ¦∼ B : B ∈ Π
0
α
¯
(F)¦,
• ∆
0
α
¯
(F) = Π
0
α
¯
(F) ∩ Σ
0
α
¯
(F),
• Borel(F) =
α<ω
1
Σ
0
α
¯
(F), and
• let ord(F) be the least α such that Borel(F) = Σ
0
α
¯
(F).
Theorem 3.1 (Bing, Bledsoe, Mauldin [12]) Suppose F ⊆ P(2
ω
) is a count
able family such that Borel(2
ω
) ⊆ Borel(F). Then ord(F) = ω
1
.
Corollary 3.2 Suppose X is any space containing a perfect set and F is a
countable family of subsets of X with Borel(X) ⊆ Borel(F). Then ord(F) =
ω
1
.
Proof:
Suppose 2
ω
⊆ X and let
ˆ
F = ¦A∩2
ω
: A ∈ F¦. By Theorem 2.3 we have
that Borel(2
ω
) ⊆ Borel(
ˆ
F) and so by Theorem 3.1 we know ord(
ˆ
F) = ω
1
.
But this implies ord(F) = ω
1
.
The proof of Theorem 3.1 is a generalization of Lebesgue’s universal set
argument. We need to prove the following two lemmas.
Lemma 3.3 (Universal sets) Suppose H ⊆ P(X) is countable and deﬁne
R = ¦A B : A ⊆ 2
ω
is clopen and B ∈ H¦.
Then for every α with 1 ≤ α < ω
1
there exists U ⊆ 2
ω
X with U ∈ Π
0
α
¯
(R)
such that for every A ∈ Π
0
α
¯
(H) there exists x ∈ 2
ω
with A = U
x
.
Descriptive Set Theory and Forcing 11
Proof:
This is proved exactly as Theorem 2.6, replacing the basis for X with H.
Note that when we replace U
n
by U
∗
n
it is necessary to prove by induction on
β that for every set A ∈ Π
0
β
¯
(R) and n ∈ ω that the set
A
∗
= ¦(x, y) : (x
n
, y) ∈ A¦
is also in Π
0
β
¯
(R).
Lemma 3.4 Suppose H ⊆ P(2
ω
), R is deﬁned as in Lemma 3.3, and
Borel(2
ω
) ⊆ Borel(H).
Then for every set A ∈ Borel(R) the set D = ¦x : (x, x) ∈ A¦ is in Borel(H).
Proof:
If A = B C where B is clopen and C ∈ H, then D = B ∩ C which is
in Borel(H) by assumption. Note that
¦x : (x, x) ∈
n
A
n
¦ =
n
¦x : (x, x) ∈ A
n
¦
and
¦x : (x, x) ∈∼ A¦ =∼ ¦x : (x, x) ∈ A¦,
so the result follows by induction.
Proof of Theorem 3.1:
Suppose Borel(H) = Π
0
α
¯
(H) and let U ⊆ 2
ω
2
ω
be universal for Π
0
α
¯
(H)
given by Lemma 3.3. By Lemma 3.4 the set D = ¦x : (x, x) ∈ U¦ is in
Borel(H) and hence its complement is in Borel(H) = Π
0
α
¯
(H). Hence we get
the same old contradiction: if U
x
=∼ D, then x ∈ D iﬀ x / ∈ D.
Theorem 3.5 (Reclaw) If X is a second countable space and X can be
mapped continuously onto the unit interval, [0, 1], then ord(X) = ω
1
.
Proof:
Let f : X → [0, 1] be continuous and onto. Let B be a countable base for
X and let H = ¦f(B) : B ∈ B¦. Since the preimage of an open subset of [0, 1]
Descriptive Set Theory and Forcing 12
is open in X it is clear that Borel([0, 1]) ⊆ Borel(H). So by Corollary 3.2
it follows that ord(H) = ω
1
. But f maps the Borel hierarchy of X directly
over to the hierarchy generated by H, so ord(X) = ω
1
.
Note that if X is a discrete space of cardinality the continuum then there
is a continuous map of X onto [0, 1] but ord(X) = 1.
The Cantor space 2
ω
can be mapped continuously onto [0, 1] via the map
x →
∞
n=0
x(n)
2
n+1
.
This map is even onetoone except at countably many points where it is
twotoone. It is also easy to see that R can be mapped continuously onto
[0, 1] and ω
ω
can be mapped onto 2
ω
. It follows that in Theorem 3.5 we may
replace [0, 1] by 2
ω
, ω
ω
, or R.
Myrna Dzamonja points out that any completely regular space Y which
contains a perfect set can be mapped onto [0, 1]. This is true because if
P ⊆ Y is perfect, then there is a continuous map f from P onto [0, 1]. But
since Y is completely regular this map extends to Y .
Reclaw did not publish his result, but I did, see Miller [86] and [87].
4 CHARACTERISTIC FUNCTION OF A SEQUENCE 13
4 Characteristic function of a sequence
The idea of a characteristic function of a sequence of sets is due to Ku
ratowski and generalized the notion of a characteristic function of a set in
troduced by de le Vall´eePoussin. The general notion was introduced by
Szpilrajn [106]. He also exploited it in Szpilrajn [107]. Szpilrajn changed his
name to Marczewski soon after the outbreak of World War II most likely to
hide from the Nazis. He kept the name Marczewski for the rest of his life.
Suppose F ⊆ P(X) is a countable ﬁeld of sets (i.e. F is a family of
sets which is closed under complements in X and ﬁnite intersections). Let
F = ¦A
n
: n ∈ ω¦. Deﬁne c : X → 2
ω
by
c(x)(n) =
_
1 if x ∈ A
n
0 if x / ∈ A
n
Let Y = c(X), then there is a direct correspondence between F and
¦C ∩ Y : C ⊆ 2
ω
clopen ¦.
In general, c maps X into 2
F
.
Theorem 4.1 (Szpilrajn [107]) If F ⊆ P(X) is a countable ﬁeld of sets,
then there exists a subspace Y ⊆ 2
ω
such that ord(F) = ord(Y ).
Proof:
If we deﬁne x ≈ y iﬀ ∀n (x ∈ A
n
iﬀ y ∈ A
n
), then we see that members
of Borel(F) respect ≈. The preimages of points of Y under c are exactly the
equivalence classes of ≈. The map c induces a bijection between X/ ≈ and
Y which takes the family F exactly to a clopen basis for the topology on Y .
Hence ord(F) = ord(Y ).
The following theorem says that bounded Borel hierarchies must have a
top.
Theorem 4.2 (Miller [75]) Suppose F ⊆ P(X) is a ﬁeld of sets and
ord(F) = λ
where λ is a countable limit ordinal. Then there exists B ∈ Borel(F) which
is not in Π
0
α
¯
(F) for any α < λ.
Descriptive Set Theory and Forcing 14
Proof:
By the characteristic function of a sequence of sets argument we may
assume without loss of generality that
F = ¦C ∩ Y : C ⊆ 2
κ
clopen ¦.
A set C ⊆ 2
κ
is clopen iﬀ it is a ﬁnite union of sets of the form
[s] = ¦x ∈ 2
κ
: s ⊆ x¦
where s : D → 2 is a map with D ∈ [κ]
<ω
(i.e. D is a ﬁnite subset of κ). Note
that by induction for every A ∈ Borel(F) there exists an S ∈ [κ]
ω
(called a
support of A) with the property that for every x, y ∈ 2
κ
if x S = y S
then (x ∈ A iﬀ y ∈ A). That is to say, membership in A is determined by
restrictions to S.
Lemma 4.3 There exists a countable S ⊆ κ with the properties that α < λ
and s : D → 2 with D ∈ [S]
<ω
if ord(Y ∩ [s]) > α then there exists A in
Σ
0
α
¯
(F) but not in ∆
0
α
¯
(F) such that A ⊆ [s] and A is supported by S.
Proof:
This is proved by a LowenheimSkolem kind of an argument.
By permuting κ around we may assume without loss of generality that
S = ω. Deﬁne
T = ¦s ∈ ω
<ω
: ord(Y ∩ [s]) = λ¦.
Note that T is a tree, i.e., s ⊆ t ∈ T implies s ∈ T. Also for any s ∈ T either
sˆ0 ∈ T or sˆ1 ∈ T, because
[s] = [sˆ0] ∪ [sˆ1].
Since ¸) ∈ T it must be that T has an inﬁnite branch. Let x : ω → 2 be such
that x n ∈ T for all n < ω. For each n deﬁne
t
n
= (x n)ˆ(1 −x(n))
and note that
2
κ
= [x] ∪
_
n∈ω
[t
n
]
Descriptive Set Theory and Forcing 15
is a partition of 2
κ
into clopen sets and one closed set [x].
Claim: For every α < λ and n ∈ ω there exists m > n with
ord(Y ∩ [t
m
]) > α.
Proof:
Suppose not and let α and n witness this. Note that
[x n] = [x] ∪
_
n≤m<ω
[t
m
].
Since ord([x n] ∩ Y ) = λ we know there exists A ∈ Σ
0
α+1
¯
(F) ¸ ∆
0
α+1
¯
(F)
such that A ⊆ [x n] and A is supported by S = ω. Since A is supported by
ω either [x] ⊆ A or A is disjoint from [x]. But if ord([t
m
] ∩ Y ) ≤ α for each
m > n, then
A
0
=
_
n≤m<ω
(A ∩ [t
m
])
is Σ
0
α
¯
(F) and A = A
0
or A = A
0
∪ [x] either of which is Σ
0
α
¯
(F) (as long as
α > 1). This proves the Claim.
The claim allows us to construct a set which is not at a level below λ as
follows. Let α
n
< λ be a sequence unbounded in λ and let k
n
be a distinct
sequence with ord([t
k
n
] ∩ Y ) ≥ α
n
. Let A
n
⊆ [t
k
n
] be in Borel(F) ¸ ∆
0
α
n
¯
(F).
Then ∪
n
A
n
is not at any level bounded below λ.
Question 4.4 Suppose R ⊆ P(X) is a ring of sets, i.e., closed under ﬁnite
unions and ﬁnite intersections. Let R
∞
be the σring generated by R, i.e., the
smallest family containing R and closed under countable unions and countable
intersections. For n ∈ ω deﬁne R
n
as follows. R
0
= R and let R
n+1
be the
family of countable unions (if n even) or family of countable intersections (if
n odd) of sets from R
n
. If R
∞
=
n<ω
R
n
, then must there be n < ω such
that R
∞
= R
n
?
5 MARTIN’S AXIOM 16
5 Martin’s Axiom
The following result is due to Rothberger [94] and Solovay [45][74]. The
forcing we use is due to Silver. However, it is probably just another view of
Solovay’s ‘almost disjoint sets forcing’.
Theorem 5.1 Assuming Martin’s Axiom if X is any second countable Haus
dorﬀ space of cardinality less than the continuum, then ord(X) ≤ 2 and, in
fact, every subset of X is G
δ
.
Proof:
Let A ⊆ X be arbitrary and let B be a countable base for the topology on
X. The partial order P is deﬁned as follows. p ∈ P iﬀ p is a ﬁnite consistent
set of sentences of the form
1. “x / ∈
◦
Un
” where x ∈ X ¸ A or
2. “B ⊆
◦
Un
” where B ∈ B and n ∈ ω.
Consistent means that there is not a pair of sentences “x / ∈
◦
Un
”, “B ⊆
◦
Un
”
in p where x ∈ B. The ordering on P is reverse containment, i.e. p is stronger
than q, p ≤ q iﬀ p ⊇ q. The circle in the notation
◦
Un
’s means that it is the
name for the set U
n
which will be determined by the generic ﬁlter. For an
element x of the ground model we should use ˇ x to denote the canonical name
of x, however to make it more readable we often just write x. For standard
references on forcing see Kunen [56] or Jech [44].
We call this forcing Silver forcing.
Claim: P satisﬁes the ccc.
Proof:
Note that since B is countable there are only countably many sentences
of the type “B ⊆
◦
Un
”. Also if p and q have exactly the same sentences of
this type then p ∪ q ∈ P and hence p and q are compatible. It follows that
P is the countable union of ﬁlters and hence we cannot ﬁnd an uncountable
set of pairwise incompatible conditions.
For x ∈ X ¸ A deﬁne
D
x
= ¦p ∈ P : ∃n “x / ∈
◦
Un
” ∈ p¦.
Descriptive Set Theory and Forcing 17
For x ∈ A and n ∈ ω deﬁne
E
n
x
= ¦p ∈ P : ∃B ∈ B x ∈ B and “B ⊆
◦
Un
” ∈ p¦.
Claim: D
x
is dense for each x ∈ X ¸ A and E
n
x
is dense for each x ∈ A and
n ∈ ω.
Proof:
To see that D
x
is dense let p ∈ P be arbitrary. Choose n large enough so
that
◦
U
n
is not mentioned in p, then (p ∪ ¦“x / ∈
◦
Un
”¦) ∈ P.
To see that E
n
x
is dense let p be arbitrary and let Y ⊆ X ¸ A be the set
of elements of X ¸ A mentioned by p. Since x ∈ A and X is Hausdorﬀ there
exists B ∈ B with B ∩ Y = ∅ and x ∈ B. Then q = (p ∪ ¦“B ⊆
◦
Un
”¦) ∈ P
and q ∈ E
n
x
.
Since the cardinality of X is less than the continuum we can ﬁnd a Pﬁlter
G with the property that G meets each D
x
for x ∈ X ¸ A and each E
n
x
for
x ∈ A and n ∈ ω. Now deﬁne
U
n
=
_
¦B : “B ⊆
◦
Un
” ∈ G¦.
Note that A =
n∈ω
U
n
and so A is G
δ
in X.
Spaces X in which every subset is G
δ
are called Qsets.
The following question was raised during an email correspondence with
Zhou.
Question 5.2 Suppose every set of reals of cardinality ℵ
1
is a Qset. Then
is p > ω
1
, i.e., is it true that for every family T ⊆ [ω]
ω
of size ω
1
with the
ﬁnite intersection property there exists an X ∈ [ω]
ω
with X ⊆
∗
Y for all
Y ∈ T?
It is a theorem of Bell [11] that p is the ﬁrst cardinal for which MA for
σcentered forcing fails. Another result along this line due to Alan Taylor is
that p is the cardinality of the smallest set of reals which is not a γset, see
Galvin and Miller [30].
Fleissner and Miller [23] show it is consistent to have a Qset whose union
with the rationals is not a Qset.
For more information on Martin’s Axiom see Fremlin [27]. For more on
Qsets, see Fleissner [24] [25], Miller [83] [87], Przymusinski [92], Judah and
Shelah [46] [47], and Balogh [5].
6 GENERIC G
δ
18
6 Generic G
δ
It is natural
4
to ask
“What are the possibly lengths of Borel hierarchies?”
In this section we present a way of forcing a generic G
δ
.
Let X be a Hausdorﬀ space with a countable base B. Consider the fol
lowing forcing notion.
p ∈ P iﬀ it is a ﬁnite consistent set of sentences of the form:
1. “B ⊆
◦
Un
” where B ∈ B and n ∈ ω, or
2. “x / ∈
◦
Un
” where x ∈ X and n ∈ ω, or
3. “x ∈
n<ω
◦
Un
” where x ∈ X.
Consistency means that we cannot say that both “B ⊆
◦
Un
” and “x / ∈
◦
Un
” if it
happens that x ∈ B and we cannot say both “x / ∈
◦
Un
” and “x ∈
n<ω
◦
Un
”.
The ordering is reverse inclusion. A P ﬁlter G determines a G
δ
set U as
follows: Let
U
n
=
_
¦B ∈ B : “B ⊆
◦
Un
” ∈ G¦.
Let U =
n
U
n
. If G is Pgeneric over V , a density argument shows that for
every x ∈ X we have that
x ∈ U iﬀ “x ∈
n<ω
◦
Un
” ∈ G.
Note that U is not in V (as long as X is inﬁnite). For suppose p ∈ P and
A ⊆ X is in V is such that
p [¬
◦
U
=
ˇ
A.
Since X is inﬁnite there exist x ∈ X which is not mentioned in p. Note that
p
0
= p ∪ ¦“x ∈
n<ω
◦
Un
”¦ is consistent and also p
1
= p ∪ ¦“x / ∈
◦
Un
”¦ is
consistent for all suﬃciently large n (i.e. certainly for U
n
not mentioned in
p.) But p
0
[¬ x ∈
◦
U
and p
1
[¬ x / ∈
◦
U
, and since x is either in A or not in A
we arrive at a contradiction.
4
‘Gentlemen, the great thing about this, like most of the demonstrations of the higher
mathematics, is that it can be of no earthly use to anybody.’ Baron Kelvin
Descriptive Set Theory and Forcing 19
In fact, U is not F
σ
in the extension (assuming X is uncountable). To
see this we will ﬁrst need to prove that P has ccc.
Lemma 6.1 P has ccc.
Proof:
Note that p and q are compatible iﬀ (p ∪ q) ∈ P iﬀ (p ∪ q) is a consistent
set of sentences. Recall that there are three types of sentences:
1. B ⊆
◦
Un
2. x / ∈
◦
Un
3. x ∈
n<ω
◦
Un
where B ∈ B, n ∈ ω, and x ∈ X. Now if for contradiction A were an
uncountable antichain, then since there are only countably many sentences
of type 1 above we may assume that all p ∈ A have the same set of type
1 sentences. Consequently for each distinct pair p, q ∈ A there must be an
x ∈ X and n such that either “x / ∈
◦
Un
” ∈ p and “x ∈
n<ω
◦
Un
” ∈ q
or viceversa. For each p ∈ A let D
p
be the ﬁnitely many elements of X
mentioned by p and let s
p
: D
p
→ ω be deﬁned by
s
p
(x) =
_
0 if “x ∈
n<ω
◦
Un
” ∈ p
n + 1 if “x / ∈
◦
Un
” ∈ p
But now ¦s
p
: p ∈ A¦ is an uncountable family of pairwise incompatible ﬁnite
partial functions from X into ω which is impossible. (FIN(X, ω) has the ccc,
see Kunen [56].)
If V [G] is a generic extension of a model V which contains a topological
space X, then we let X also refer to the space in V [G] whose topology is
generated by the open subsets of X which are in V .
Theorem 6.2 (Miller [75]) Suppose X in V is an uncountable Hausdorﬀ
space with countable base B and G is Pgeneric over V . Then in V [G] the
G
δ
set U is not F
σ
.
Descriptive Set Theory and Forcing 20
Proof:
We call this argument the old switcheroo. Suppose for contradiction
p [¬
n∈ω
◦
Un
=
_
n∈ω
◦
Cn
where
◦
Cn
are closed in X .
For Y ⊆ X let P(Y ) be the elements of P which only mention y ∈ Y in type
2 or 3 statements. Let Y ⊆ X be countable such that
1. p ∈ P(Y ) and
2. for every n and B ∈ B there exists a maximal antichain A ⊆ P(Y )
which decides the statement “B∩
◦
Cn
= ∅”.
Since X is uncountable there exists x ∈ X ¸ Y . Let
q = p ∪ ¦“x ∈
n∈ω
◦
Un
”¦.
Since q extends p, clearly
q [¬ x ∈
_
n∈ω
◦
Cn
so there exists r ≤ q and n ∈ ω so that
r [¬ x ∈
◦
Cn
.
Let
r = r
0
∪ ¦“x ∈
n∈ω
◦
Un
”¦
where r
0
does not mention x. Now we do the switch. Let
t = r
0
∪ ¦“x / ∈
◦
Um
”¦
where m is chosen suﬃciently large so that t is a consistent condition. Since
t [¬ x / ∈
n∈ω
◦
Un
we know that
t [¬ x / ∈
◦
Cn
.
Consequently there exist s ∈ P(Y ) and B ∈ B such that
Descriptive Set Theory and Forcing 21
1. s and t are compatible,
2. s [¬ B∩
◦
Cn
= ∅, and
3. x ∈ B.
But s and r are compatible, because s does not mention x. This is a contra
diction since s ∪ r [¬ x ∈
◦
Cn
and s ∪ r [¬ x / ∈
◦
Cn
.
7 αFORCING 22
7 αforcing
In this section we generalize the forcing which produced a generic G
δ
to
arbitrarily high levels of the Borel hierarchy. Before doing so we must prove
some elementary facts about wellfounded trees.
Let OR denote the class of all ordinals. Deﬁne T ⊆ Q
<ω
to be a tree iﬀ
s ⊆ t ∈ T implies s ∈ T. Deﬁne the rank function r : T → OR ∪ ¦∞¦ of
T as follows:
1. r(s) ≥ 0 iﬀ s ∈ T,
2. r(s) ≥ α + 1 iﬀ ∃q ∈ Q r(sˆq) ≥ α,
3. r(s) ≥ λ (for λ a limit ordinal) iﬀ r(s) ≥ α for every α < λ.
Now deﬁne r(s) = α iﬀ r(s) ≥ α but not r(s) ≥ α + 1 and r(s) = ∞ iﬀ
r(s) ≥ α for every ordinal α.
Deﬁne [T] = ¦x ∈ Q
ω
: ∀n x n ∈ T¦. We say that T is wellfounded
iﬀ [T] = ∅.
Theorem 7.1 T is wellfounded iﬀ r(¸)) ∈ OR.
Proof:
It follows easily from the deﬁnition that if r(s) is an ordinal, then
r(s) = sup¦r(sˆq) + 1 : q ∈ Q¦.
Hence, if r(¸)) = α ∈ OR and x ∈ [T], then
r(x (n + 1)) < r(x n)
is a descending sequence of ordinals.
On the other hand, if r(s) = ∞ then for some q ∈ Q we must have
r(sˆq) = ∞. So if r(¸)) = ∞ we can construct (using the axiom of choice) a
sequence s
n
∈ T with r(s
n
) = ∞ and s
n+1
= s
n
ˆx(n). Hence x ∈ [T].
Deﬁnition. T is a nice αtree iﬀ
1. T ⊆ ω
<ω
is a tree,
2. r : T → (α + 1) is its rank function (r(¸)) = α),
Descriptive Set Theory and Forcing 23
3. if r(s) > 0, then for every n ∈ ω sˆn ∈ T,
4. if r(s) = β is a successor ordinal, then for every n ∈ ω r(sˆn) = β −1,
and
5. if r(s) = λ is a limit ordinal, then r(sˆ0) ≥ 2 and r(sˆn) increases to
λ as n → ∞.
It is easy to see that for every α < ω
1
nice αtrees exist. For X a Hausdorﬀ
space with countable base, B, and T a nice αtree (α ≥ 2), deﬁne the partial
order P = P(X, B, T) which we call αforcing as follows:
p ∈ P iﬀ p = (t, F) where
1. t : D → B where D ⊆ T
0
= ¦s ∈ T : r(s) = 0¦ is ﬁnite,
2. F ⊆ T
>0
X is ﬁnite where
T
>0
= T ¸ T
0
= ¦s ∈ T : r(s) > 0¦,
3. if (s, x), (sˆn, y) ∈ F, then x ,= y, and
4. if (s, x) ∈ F and t(sˆn) = B, then x / ∈ B.
The ordering on P is given by p ≤ q iﬀ t
p
⊇ t
q
and F
p
⊇ F
q
.
Lemma 7.2 P has ccc.
Proof:
Suppose A is uncountable antichain. Since there are only countably many
diﬀerent t
p
without loss we may assume that there exists t such that t
p
= t
for all p ∈ A. Consequently for p, q ∈ A the only thing that can keep p ∪ q
from being a condition is that there must be an x ∈ X and an s, sˆn ∈ T
>0
such that
(s, x), (sˆn, x) ∈ (F
p
∪ F
q
).
But now for each p ∈ A let H
p
: X → [T
>0
]
<ω
be the ﬁnite partial function
deﬁned by
H
p
(x) = ¦s ∈ T
>0
: (s, x) ∈ F
p
¦
where domain H
p
is ¦x : ∃s ∈ T
>0
(s, x) ∈ F
p
¦. Then ¦H
p
: p ∈ A¦ is
an uncountable antichain in the order of ﬁnite partial functions from X to
[T
>0
]
<ω
, a countable set.
Deﬁne for G a Pﬁlter the set U
s
⊆ X for s ∈ T as follows:
Descriptive Set Theory and Forcing 24
1. for s ∈ T
0
let U
s
= B iﬀ ∃p ∈ G such that t
p
(s) = B and
2. for s ∈ T
>0
let U
s
=
n∈ω
∼ U
sˆn
Note that U
s
is a Π
0
β
¯
(X)set where r(s) = β.
Lemma 7.3 If G is Pgeneric over V then in V [G] we have that for every
x ∈ X and s ∈ T
>0
x ∈ U
s
⇐⇒ ∃p ∈ G (s, x) ∈ F
p
.
Proof:
First suppose that r(s) = 1 and note that the following set is dense:
D = ¦p ∈ P : (s, x) ∈ F
p
or ∃n∃B ∈ B x ∈ B and t
p
(sˆn) = B¦.
To see this let p ∈ P be arbitrary. If (s, x) ∈ F
p
then p ∈ D and we are
already done. If (s, x) / ∈ F
p
then let
Y = ¦y : (s, y) ∈ F
p
¦.
Choose B ∈ B with x ∈ B and Y disjoint from B. Choose sˆn not in the
domain of t
p
, and let q = (t
q
, F
p
) be deﬁned by t
q
= t
p
∪ (sˆn, B). So q ≤ p
and q ∈ D. Hence D is dense.
Now by deﬁnition x ∈ U
s
iﬀ x ∈
n∈ω
∼ U
sˆn
. So let G be a generic
ﬁlter and p ∈ G ∩ D. If (s, x) ∈ F
p
then we know that for every q ∈ G and
for every n, if t
q
(sˆn) = B then x / ∈ B. Consequently, x ∈ U
s
. On the other
hand if t
p
(sˆn) = B where x ∈ B, then x / ∈ U
s
and for every q ∈ G it must
be that (s, x) / ∈ F
q
(since otherwise p and q would be incompatible).
Now suppose r(s) > 1. In this case note that the following set is dense:
E = ¦p ∈ P : (s, x) ∈ F
p
or ∃n (sˆn, x) ∈ F
p
¦.
To see this let p ∈ P be arbitrary. Then either (s, x) ∈ F
p
and already
p ∈ E or by choosing n large enough q = (t
p
, F
p
∪ ¦(sˆn, x)¦) ∈ E. (Note
r(sˆn) > 0.)
Now assume the result is true for all U
sˆn
. Let p ∈ G∩ E. If (s, x) ∈ F
p
then for every q ∈ G and n we have (sˆn, x) / ∈ F
q
and so by induction
x / ∈ U
sˆn
and so x ∈ U
s
. On the other hand if (sˆn, x) ∈ F
p
, then by
Descriptive Set Theory and Forcing 25
induction x ∈ U
sˆn
and so x / ∈ U
s
, and so again for every q ∈ G we have
(s, x) / ∈ F
q
.
The following lemma is the heart of the old switcheroo argument used
in Theorem 6.2. Given any Q ⊂ X deﬁne the rank(p, Q) as follows:
rank(p, Q) = max¦r(s) : (s, x) ∈ F
p
for some x ∈ X ¸ Q¦.
Lemma 7.4 (Rank Lemma). For any β ≥ 1 and p ∈ P there exists ˆ p
compatible with p such that
1. rank(ˆ p, Q) < β + 1 and
2. for any q ∈ P if rank(q, Q) < β, then
ˆ p and q compatible implies p and q compatible.
Proof:
Let p
0
≤ p be any extension which satisﬁes: for any (s, x) ∈ F
p
and n ∈ ω,
if r(s) = λ > β is a limit ordinal and r(sˆn) < β +1, then there exist m ∈ ω
such that (sˆnˆm, x) ∈ F
p
0
. Note that since r(sˆn) is increasing to λ there
are only ﬁnitely many (s, x) and sˆn to worry about. Also r(sˆnˆm) > 0 so
this is possible to do.
Now let ˆ p be deﬁned as follows:
t
ˆ p
= t
p
and
F
ˆ p
= ¦(s, x) ∈ F
p
0
: x ∈ Q or r(s) < β + 1¦.
Suppose for contradiction that there exists q such that rank(q, Q) < β, ˆ p
and q compatible, but p and q incompatible. Since p and q are incompatible
either
1. there exists (s, x) ∈ F
q
and t
p
(sˆn) = B with x ∈ B, or
2. there exists (s, x) ∈ F
p
and t
q
(sˆn) = B with x ∈ B, or
3. there exists (s, x) ∈ F
p
and (sˆn, x) ∈ F
q
, or
4. there exists (s, x) ∈ F
q
and (sˆn, x) ∈ F
p
.
Descriptive Set Theory and Forcing 26
(1) cannot happen since t
ˆ p
= t
p
and so ˆ p, q would be incompatible. (2) cannot
happen since r(s) = 1 and β ≥ 1 means that (s, x) ∈ F
ˆ p
and so again ˆ p and
q are incompatible. If (3) or (4) happens for x ∈ Q then again (in case 3)
(s, x) ∈ F
ˆ p
or (in case 4) (sˆn, x) ∈ F
ˆ p
and so ˆ p, q incompatible.
So assume x / ∈ Q. In case (3) by the deﬁnition of rank(q, Q) < β we know
that r(sˆn) < β. Now since T is a nice tree we know that either r(s) ≤ β
and so (s, x) ∈ F
ˆ p
or r(s) = λ a limit ordinal. Now if λ ≤ β then (s, x) ∈ F
ˆ p
.
If λ > β then by our construction of p
0
there exist m with (sˆnˆm, x) ∈ F
ˆ p
and so ˆ p, q are incompatible. Finally in case (4) since x / ∈ Q and so r(s) < β
we have that r(sˆn) < β and so (sˆn, x) ∈ F
ˆ p
and so ˆ p, q are incompatible.
Intuitively, it should be that statements of small rank are forced by con
ditions of small rank. The next lemma will make this more precise. Let
L
∞
(P
α
: α < κ) be the inﬁnitary propositional logic with ¦P
α
: α < κ¦ as
the atomic sentences. Let Π
0
sentences be the atomic ones, ¦P
α
: α < κ¦.
For any β > 0 let θ be a Π
β
sentence iﬀ there exists Γ ⊆
δ<β
Π
δ
sentences
and
θ =
ψ∈Γ
ψ.
Models for this propositional language can naturally be regarded as subsets
Y of κ where we deﬁne
1. Y [= P
α
iﬀ α ∈ Y ,
2. Y [= θ iﬀ not Y [= θ, and
3. Y [=
__
Γ iﬀ Y [= θ for every θ ∈ Γ.
Lemma 7.5 (Rank and Forcing Lemma) Suppose rank : P → OR is any
function on a poset P which satisﬁes the Rank Lemma 7.4. Suppose [¬
P
◦
Y
⊂ κ
and for every p ∈ P and α < κ if
p [¬ α ∈
◦
Y
then there exist ˆ p compatible with p such that rank(ˆ p) = 0 and
ˆ p [¬ α ∈
◦
Y
.
Descriptive Set Theory and Forcing 27
Then for every Π
β
sentence θ (in the ground model) and every p ∈ P, if
p [¬ “
◦
Y
[= θ”
then there exists ˆ p compatible with p such that rank(ˆ p) ≤ β and
ˆ p [¬ “
◦
Y
[= θ”.
Proof:
This is one of those lemmas whose statement is longer than its proof. The
proof is induction on β and for β = 0 the conclusion is true by assumption.
So suppose β > 0 and θ =
__
ψ∈Γ
ψ where Γ ⊆
δ<β
Π
δ
sentences. By the
rank lemma there exists ˆ p compatible with p such that rank(ˆ p) ≤ β and for
every q ∈ P with rank(q) < β if ˆ p, q compatible then p, q compatible. We
claim that
ˆ p [¬ “
◦
Y
[= θ”.
Suppose not. Then there exists r ≤ ˆ p and ψ ∈ Γ such that
r [¬ “
◦
Y
[= ψ”.
By inductive assumption there exists ˆ r compatible with r such that
rank(ˆ r) < β
such that
ˆ r [¬ “
◦
Y
[= ψ”.
But ˆ r, ˆ p compatible implies ˆ r, p compatible, which is a contradiction because
θ implies ψ and so
p [¬ “
◦
Y
[= ψ”.
Some earlier uses of rank in forcing arguments occur in Steel’s forcing,
see Steel [108], Friedman [29], and Harrington [36]. It also occurs in Silver’s
analysis of the collapsing algebra, see Silver [101].
In Miller [77] αforcing for all α is used to construct generic Souslin sets.
8 BOOLEAN ALGEBRAS 28
8 Boolean algebras
In this section we consider the length of Borel hierarchies generated by a
subset of a complete boolean algebra. We ﬁnd that the generators of the
complete boolean algebra associated with αforcing generate it in exactly
α + 1 steps. We start by presenting some background information.
Let B be a cBa, i.e, complete boolean algebra. This means that in
addition to being a boolean algebra, inﬁnite sums and products, also exist;
i.e., for any C ⊆ B there exists b (denoted
C) such that
1. c ≤ b for every c ∈ C and
2. for every d ∈ B if c ≤ d for every c ∈ C, then b ≤ d.
Similarly we deﬁne
C = −
c∈C
−c where −c denotes the complement of
c in B.
A partial order P is separative iﬀ for any p, q ∈ P we have
p ≤ q iﬀ ∀r ∈ P(r ≤ p implies q, r compatible).
Theorem 8.1 (Scott, Solovay see [44]) A partial order P is separative iﬀ
there exists a cBa B such that P ⊆ B is dense in B, i.e. for every b ∈ B if
b > 0 then there exists p ∈ P with p ≤ b.
It is easy to check that the αforcing P is separative (as long as B is
inﬁnite): If p ,≤ q then either
1. t
p
does not extend t
q
, so there exists s such that t
q
(s) = B and either
s not in the domain of t
p
or t
p
(s) = C where C ,= B and so in either
case we can ﬁnd r ≤ p with r, q incompatible, or
2. F
p
does not contain F
q
, so there exists (s, x) ∈ (F
q
¸ F
p
) and we can
either add (sˆn, x) for suﬃciently large n or add t
r
(sˆn) = B for some
suﬃciently large n and some B ∈ B with x ∈ B and get r ≤ p which is
incompatible with q.
The elegant (but as far as I am concerned mysterious) approach to forcing
using complete boolean algebras contains the following facts:
1. for any sentence θ in the forcing language
[[ θ [] =
¦b ∈ B : b [¬ θ¦ =
¦p ∈ P : p [¬ θ¦
where P is any dense subset of B,
Descriptive Set Theory and Forcing 29
2. p [¬ θ iﬀ p ≤ [[ θ [],
3. [[ θ [] = −[[ θ [],
4. [[ θ ∧ψ [] = [[ θ [] ∧[[ ψ [],
5. [[ θ ∨ψ [] = [[ θ [] ∨[[ ψ [],
6. for any set X in the ground model,
[[ ∀x ∈
ˇ
X θ(x) [] =
x∈X
[[ θ(ˇ x [].
Deﬁnitions. For B a cBa and C ⊆ B deﬁne:
Π
0
0
¯
(C) = C and
Π
0
α
¯
(C) = ¦
Γ : Γ ⊆ ¦−c : c ∈
β<α
Π
0
β
¯
(C)¦¦ for α > 0.
ord(B) = min¦α : ∃C ⊆ B countable with Π
0
α
¯
(C) = B¦.
Theorem 8.2 (Miller [75]) For every α ≤ ω
1
there exists a countably gen
erated ccc cBa B with ord(B) = α.
Proof:
Let P be αforcing and B be the cBa given by the ScottSolovay Theo
rem 8.1. We will show that ord(B) = α + 1.
Let
C = ¦p ∈ P : F
p
= ∅¦.
C is countable and we claim that P ⊆ Π
0
α
¯
(C). Since B = Σ
0
1
¯
(P) this will
imply that B = Σ
0
α+1
¯
(C) and so ord(B) ≤ α + 1.
First note that for any s ∈ T with r(s) = 0 and x ∈ X,
[[ x ∈ U
s
[] =
¦p ∈ C : ∃B ∈ B t
p
(s) = B and x ∈ B¦.
By Lemma 7.3 we know for generic ﬁlters G that for every x ∈ X and
s ∈ T
>0
x ∈ U
s
⇐⇒ ∃p ∈ G (s, x) ∈ F
p
.
Hence [[ x ∈ U
s
[] = ¸∅, ¦(s, x)¦) since if they are not equal, then
b = [[ x ∈ U
s
[] ∆ ¸∅, ¦(s, x)¦) > 0,
Descriptive Set Theory and Forcing 30
but letting G be a generic ultraﬁlter with b in it would lead to a contradiction.
We get that for r(s) > 0:
¸∅, ¦(s, x)¦) = [[ x ∈ U
s
[] = [[ x ∈
n∈ω
∼ U
sˆn
[] =
n∈ω
−[[ x ∈ U
sˆn
[].
Remembering that for r(sˆn) = 0 we have [[ x ∈ U
sˆn
[] ∈ Σ
0
1
¯
(C), we see by
induction that for every s ∈ T
>0
if r(s) = β then
¸∅, ¦(s, x)¦) ∈ Π
0
β
¯
(C).
For any p ∈ P
p = ¸t
p
, ∅) ∧
(s,x)∈F
p
¸∅, ¦(s, x)¦).
So we have that p ∈ Π
0
α
¯
(C).
Now we will see that ord(B) > α. We use the following Lemmas.
B
+
are the nonzero elements of B.
Lemma 8.3 If r : P → OR is a rank function, i.e. it satisﬁes the Rank
Lemma 7.4 and in addition p ≤ q implies r(p) ≤ r(q), then if P is dense in
the cBa B then r extends to r
∗
on B
+
:
r
∗
(b) = min¦β ∈ OR : ∃C ⊆ P : b =
C and ∀p ∈ C r(p) ≤ β¦
and still satisﬁes the Rank Lemma.
Proof:
Easy induction.
Lemma 8.4 If r : B
+
→ ord is a rank function and E ⊆ B is a countable
collection of rank zero elements, then for any a ∈ Π
0
γ
¯
(E) and a ,= 0 there
exists b ≤ a with r(b) ≤ γ.
Proof:
To see this let E = ¦e
n
: n ∈ ω¦ and let
◦
Y
be a name for the set in the
generic extension
Y = ¦n ∈ ω : e
n
∈ G¦.
Descriptive Set Theory and Forcing 31
Note that e
n
= [[ n ∈
◦
Y
[]. For elements b of B in the complete subalgebra
generated by E let us associate sentences θ
b
of the inﬁnitary propositional
logic L
∞
(P
n
: n ∈ ω) as follows:
θ
e
n
= P
n
θ
−b
= θ
b
θ
R
=
r∈R
θ
r
Note that [[ Y [= θ
b
[] = b and if b ∈ Π
0
γ
¯
(E) then θ
b
is a Π
γ
sentence. The
Rank and Forcing Lemma 7.5 gives us, by translating
p [¬ Y [= θ
b
into p ≤ [[ Y [= θ
b
[] = b) that:
For any γ ≥ 1 and p ≤ b ∈ Π
0
γ
¯
(E) there exists a ˆ p compatible
with p such that ˆ p ≤ b and r(ˆ p) ≤ γ.
Now we use the lemmas to see that ord(B) > α.
Given any countable E ⊆ B, let Q ⊆ X be countable so that for any
e ∈ E there exists H ⊆ P countable so that e =
H and for every p ∈ H
we have rank(p, Q) = 0. Let x ∈ X ¸ Q be arbitrary; then we claim:
[[ x ∈ U
[] / ∈ Σ
0
α
¯
(E).
We have chosen Q so that r(p) = rank(p, Q) = 0 for any p ∈ E so the
hypothesis of Lemma 8.4 is satisﬁed. Suppose for contradiction that
[[ x ∈ U
[] = b ∈ Σ
0
α
¯
(E).
Let b =
n∈ω
b
n
where each b
n
is Π
0
γ
n
¯
(C) for some γ
n
< α. For some n and
p ∈ P we would have p ≤ b
n
. By Lemma 8.4 we have that there exists ˆ p
with ˆ p ≤ b
n
≤ b = [[ x ∈ U
[] and rank(ˆ p, Q) ≤ γ
n
. But by the deﬁnition of
rank(ˆ p, Q) the pair (¸), x) is not in F
ˆ p
, but this contradicts
ˆ p ≤ b
n
≤ b = [[ x ∈ U
[] = ¸∅, ¦(¸), x)¦).
This takes care of all countable successor ordinals. (We leave the case
of α = 0, 1 for the reader to contemplate.) For λ a limit ordinal take α
n
Descriptive Set Theory and Forcing 32
increasing to λ and let P =
n<ω
P
α
n
be the direct sum, where P
α
n
is α
n

forcing. Another way to describe essentially the same thing is as follows: Let
P
λ
be λforcing. Then take P to be the subposet of P
λ
such that ¸) doesn’t
occur, i.e.,
P = ¦p ∈ P
λ
: ∃x ∈ X (¸), x) ∈ F
p
¦.
Now if P is dense in the cBa B, then ord(B) = λ. This is easy to see,
because for each p ∈ P there exists β < λ with p ∈ Π
0
β
¯
(C). Consequently,
P ⊆
β<λ
Π
0
β
¯
(C) and so since B = Σ
0
1
¯
(P) we get B = Σ
0
λ
¯
(C). Similarly to the
other argument we see that for any countable E we can choose a countable
Q ⊆ X such for any s ∈ T with 2 ≤ r(s) = β < λ (so s ,= ¸)) we have that
[[ x ∈ U
s
[] is not Σ
0
β
¯
(E). Hence ord(B) = λ.
For ord(B) = ω
1
we postpone until section 12.
9 BOREL ORDER OF A FIELD OF SETS 33
9 Borel order of a ﬁeld of sets
In this section we use the SikorskiLoomis representation theorem to transfer
the abstract Borel hierarchy on a complete boolean algebra into a ﬁeld of
sets.
A family F ⊆ P(X) is a σﬁeld iﬀ it contains the empty set and is closed
under countable unions and complements in X. I ⊆ F is a σideal in F iﬀ
1. I contains the empty set,
2. I is closed under countable unions,
3. A ⊆ B ∈ I and A ∈ F implies A ∈ I, and
4. X / ∈ I.
F/I is the countably complete boolean algebra formed by taking F and
modding out by I, i.e. A ≈ B iﬀ A∆B ∈ I. For A ∈ F we use [A] or [A]
I
to
denote the equivalence class of A modulo I.
Theorem 9.1 (Sikorski,Loomis, see [100] section 29) For any countably
complete boolean algebra B there exists a σﬁeld F and a σideal I such
that B is isomorphic to F/I.
Proof:
Recall that the Stone space of B, stone(B), is the space of ultraﬁlters u
on B with the topology generated by the clopen sets of the form:
[b] = ¦u ∈ stone(B) : b ∈ u¦.
This space is a compact Hausdorﬀ space in which the ﬁeld of clopen sets
exactly corresponds to B. B is countably complete means that for any se
quence
¦b
n
: n < ω¦ in B
there exists b ∈ B such that b =
n∈ω
b
n
. This translates to the fact that
given any countable family of clopen sets ¦C
n
: n ∈ ω¦ in stone(B) there
exists a clopen set C such that
n∈ω
C
n
⊆ C and the closed set C ¸
n∈ω
C
n
cannot contain a clopen set, hence it has no interior, so it is nowhere dense.
Let F be the σﬁeld generated by the clopen subsets of stone(B). Let I be
the σideal generated by the closed nowhere dense subsets of F (i.e. the ideal
Descriptive Set Theory and Forcing 34
of meager sets). The Baire category theorem implies that no nonempty open
subset of a compact Hausdorﬀ space is meager, so st(B) / ∈ I and the same
holds for any nonempty clopen subset of stone(B). Since the countable union
of clopen sets is equivalent to a clopen set modulo I it follows that the map
C → [C] is an isomorphism taking the clopen algebra of stone(B) onto F/I.
Shortly after I gave a talk about my boolean algebra result (Theorem
8.2), Kunen pointed out the following result.
Theorem 9.2 (Kunen see [75]) For every α ≤ ω
1
there exists a ﬁeld of sets
H such that ord(H) = α.
Proof:
Clearly we only have to worry about α with 2 < α < ω
1
. Let B be the
complete boolean algebra given by Theorem 8.2. Let B · F/I where F is a
σﬁeld of sets and I a σideal. Let C ⊆ F/I be a countable set of generators.
Deﬁne
H = ¦A ∈ F : [A]
I
∈ C¦.
By induction on β it is easy to prove that for any Q ∈ F:
Q ∈ Σ
0
β
¯
(H) iﬀ [Q]
I
∈ Σ
0
β
¯
(C).
From which it follows that ord(H) = α.
Note that there is no claim that the family H is countable. In fact, it
is consistent (Miller [75]) that either ord(H) ≤ 2 or ord(H) = ω
1
for every
countable H.
10 CH AND ORDERS OF SEPARABLE METRIC SPACES 35
10 CH and orders of separable metric spaces
In this section we prove that assuming CH that there exists countable ﬁeld
of sets of all possible Borel orders, which we know is equivalent to existence
of separable metric spaces of all possible orders. We will need a sharper form
of the representation theorem.
Theorem 10.1 (Sikorski, see [100] section 31) B is a countably generated
ccc cBa iﬀ there exists a ccc σideal I in Borel(2
ω
) such that B · Borel(2
ω
)/I.
Furthermore if B is generated by the countable set C ⊆ B, then this isomor
phism can be taken so as to map the clopen sets mod I onto C.
Proof:
To see that Borel(2
ω
)/I is countably generated is trivial since the clopen
sets modulo I generate it. A general theorem of Tarski is that any κcomplete
κcc boolean algebra is complete.
For the other direction, we may assume by using the SikorskiLoomis
Theorem, that B is F/J where F is a σﬁeld and J a σideal in F. Since B is
countably generated there exists C
n
∈ F for n ∈ ω such that ¦[C
n
] : n ∈ ω¦
generates F/J where [C] denotes the equivalence class of C modulo J. Now
let h : X → 2
ω
be deﬁned by
h(x)(n) =
_
1 if x ∈ C
n
0 if x / ∈ C
n
and deﬁne φ : Borel(2
ω
) → F by
φ(A) = h
−1
(A).
Deﬁne I = ¦A ∈ Borel(2
ω
) : φ(A) ∈ J¦. Finally, we claim that
ˆ
φ : Borel(2
ω
)/I → F/I deﬁned by
ˆ
φ([A]
I
) = [φ(A)]
J
is an isomorphism of the two boolean algebras.
For I a σideal in Borel(2
ω
) we say that X ⊆ 2
ω
is an ILuzin set
iﬀ for every A ∈ I we have that X ∩ A is countable. We say that X is
superILuzin iﬀ X is ILuzin and for every B ∈ Borel(2
ω
) ¸I we have that
B ∩X ,= ∅. The following Theorem was ﬁrst proved by Mahlo [70] and later
by Luzin [69] for the ideal of meager subsets of the real line. Apparently,
Mahlo’s paper was overlooked and hence these kinds of sets have always been
referred to as Luzin sets.
Descriptive Set Theory and Forcing 36
Theorem 10.2 (Mahlo [70]) CH. Suppose I is a σideal in Borel(2
ω
) con
taining all the singletons. Then there exists a superILuzin set.
Proof:
Let
I = ¦A
α
: α < ω
1
¦
and let
Borel(2
ω
) ¸ I = ¦B
α
: α < ω
1
¦.
Inductively choose x
α
∈ 2
ω
so that
x
α
∈ B
α
¸ (¦x
β
: β < α¦ ∪
_
β<α
A
α
).
Then X = ¦x
α
: α < ω
1
¦ is a superILuzin set.
Theorem 10.3 (Kunen see [75]) Suppose B = Borel(2
ω
)/I is a cBa, C ⊆ B
are the clopen mod I sets, ord(C) = α > 2, and X is superILuzin. Then
ord(X) = α.
Proof:
Note that the ord(X) is the minimum α such that for every B ∈ Borel(2
ω
)
there exists A ∈ Π
0
α
¯
(2
ω
) with A ∩ X = B ∩ X.
Since ord(C) = α we know that given any Borel set B there exists a Π
0
α
¯
set A such that A∆B ∈ I. Since X is Luzin we know that X ∩ (A∆B) is
countable. Hence there exist countable sets F
0
, F
1
such that
X ∩ B = X ∩ ((A ¸ F
0
) ∪ F
1
).
But since α > 2 we have that ((A¸F
0
)∪F
1
) is also Π
0
α
¯
and hence ord(X) ≤ α.
On the other hand for any β < α we know there exists a Borel set B such
that for every Π
0
β
¯
set A we have B∆A / ∈ I (since ord(C) > β). But since X
is superILuzin we have that for every Π
0
β
¯
set A that X ∩ (B∆A) ,= ∅ and
hence X ∩ B ,= X ∩ A. Consequently, ord(X) > β.
Corollary 10.4 (CH) For every α ≤ ω
1
there exists a separable metric space
X such that ord(X) = α.
Descriptive Set Theory and Forcing 37
While a graduate student at Berkeley I had obtained the result that it
was consistent with any cardinal arithmetic to assume that for every α ≤ ω
1
there exists a separable metric space X such that ord(X) = α. It never
occurred to me at the time to ask what CH implied. In fact, my way of
thinking at the time was that proving something from CH is practically the
same as just showing it is consistent. I found out in the real world (outside
of Berkeley) that they are considered very diﬀerently.
In Miller [75] it is shown that for every α < ω
1
it is consistent there exists
a separable metric space of order β iﬀ α < β ≤ ω
1
. But the general question
is open.
Question 10.5 For what C ⊆ ω
1
is it consistent that
C = ¦ord(X) : X separable metric ¦?
11 MARTINSOLOVAY THEOREM 38
11 MartinSolovay Theorem
In this section we the theorem below. The technique of proof will be used in
the next section to produce a boolean algebra of order ω
1
.
Theorem 11.1 (MartinSolovay [74]) The following are equivalent for an
inﬁnite cardinal κ:
1. MA
κ
, i.e., for any poset P which is ccc and family T of dense subsets
of P with [T[ < κ there exists a Pﬁlter G with G∩D ,= ∅ for all D ∈ T
2. For any ccc σideal I in Borel(2
ω
) and 1 ⊂ I with [1[ < κ we have that
2
ω
¸
_
1 ,= ∅.
Lemma 11.2 Let B = Borel(2
ω
)/I for some ccc σideal I and let P = B ¸
¦0¦. The following are equivalent for an inﬁnite cardinal κ:
1. for any family T of dense subsets of P with [T[ < κ there exists a
Pﬁlter G with G∩ D ,= ∅ for all D ∈ T
2. for any family T ⊆ B
ω
with [T[ < κ there exists an ultraﬁlter U on B
which is Tcomplete, i.e., for every ¸b
n
: n ∈ ω) ∈ T
n∈ω
b
n
∈ U iﬀ ∃n b
n
∈ U
3. for any 1 ⊂ I with [1[ < κ
2
ω
¸
_
1 ,= ∅
Proof:
To see that (1) implies (2) note that for any ¸b
n
: n ∈ ω) ∈ B
ω
the set
D = ¦p ∈ P : p ≤ −
n
b
n
or ∃n p ≤ b
n
¦
is dense. Note also that any ﬁlter extends to an ultraﬁlter.
Descriptive Set Theory and Forcing 39
To see that (2) implies (3) do as follows. Let H
γ
stand for the family of
sets whose transitive closure has cardinality less than the regular cardinal γ,
i.e. they are hereditarily of cardinality less than γ. The set H
γ
is a natural
model of all the axioms of set theory except possibly the power set axiom,
see Kunen [56]. Let M be an elementary substructure of H
γ
for suﬃciently
large γ with [M[ < κ, I ∈ M, 1 ⊆ M.
Let T be all the ωsequences of Borel sets which are in M. Since [T[ < κ
we know there exists U an Tcomplete ultraﬁlter on B. Deﬁne x ∈ 2
ω
by the
rule:
x(n) = i iﬀ [¦y ∈ 2
ω
: y(n) = i¦] ∈ U.
Claim: For every Borel set B ∈ M:
x ∈ B iﬀ [B] ∈ U.
Proof:
This is true for subbasic clopen sets by deﬁnition. Inductive steps just
use that U is an Mcomplete ultraﬁlter.
To see that (3) implies (1), let M be an elementary substructure of H
γ
for suﬃciently large γ with [M[ < κ, I ∈ M, T ⊆ M. Let
1 = M ∩ I.
By (3) there exists
x ∈ 2
ω
¸
_
1.
Let B
M
= B ∩ M. Then deﬁne
G = ¦[B] ∈ B
M
: x ∈ B¦.
Check G is a P ﬁlter which meets every D ∈ T.
This proves Lemma 11.2.
To prove the theorem it necessary to do a two step iteration. Let P be
a poset and
◦
Q∈ V
P
be the Pname of a poset, i.e.,
[¬
P
Q is a poset.
Descriptive Set Theory and Forcing 40
Then we form the poset
P∗
◦
Q= ¦(p,
◦
q
) : p [¬
◦
q
∈
◦
Q¦
ordered by (ˆ p, ˆ q) ≤ (p, q) iﬀ ˆ p ≤ p and ˆ p [¬ ˆ q ≤ q. In general there are
two problems with this. First, P∗
◦
Q is a class. Second, it does not satisfy
antisymmetry: x ≤ y and y ≤ x implies x = y. These can be solved by
cutting down to a suﬃciently large set of nice names and modding out by
the appropriate equivalence relation. Three of the main theorems are:
Theorem 11.3 If G is Pgeneric over V and H is Q
G
generic over V [G],
then
G∗ H = ¦(p, q) ∈ P∗
◦
Q: p ∈ G, q
G
∈ H¦.
is a P∗
◦
Q ﬁlter generic over V .
Theorem 11.4 If K is a P∗
◦
Qﬁlter generic over V , then
G = ¦p : ∃q (p, q) ∈ K¦
is Pgeneric over V and
H = ¦q
G
: ∃p (p, q) ∈ K¦
is Q
G
generic over V [G].
Theorem 11.5 (SolovayTennenbaum [104]) If P is ccc and [¬
P
“
◦
Q is ccc”,
then P∗
◦
Q is ccc.
For proofs of these results, see Kunen [56] or Jech [44].
Finally we prove Theorem 11.1. (1) implies (2) follows immediately from
Lemma 11.2. To see (2) implies (1) proceed as follows.
Note that κ ≤ c, since (1) fails for FIN(c
+
, 2). We may also assume
that the ccc poset P has cardinality less than κ. Use a LowenheimSkolem
argument to obtain a set Q ⊆ P with the properties that [Q[ < κ, D ∩ Q is
dense in Q for every D ∈ T, and for every p, q ∈ Q if p and q are compatible
(in P) then there exists r ∈ Q with r ≤ p and r ≤ q. Now replace P by Q.
The last condition on Q guarantees that Q has the ccc.
Descriptive Set Theory and Forcing 41
Choose X = ¦x
p
: p ∈ P¦ ⊆ 2
ω
distinct elements of 2
ω
. If G is Pﬁlter
generic over V let Q be Silver’s forcing for forcing a G
δ
set,
n∈ω
U
n
, in X
such that
G = ¦p ∈ P : x
p
∈
n∈ω
U
n
¦.
Let B ∈ V be a countable base for X. A simple description of P∗
◦
Q can be
given by:
(p, q) ∈ P∗
◦
Q
iﬀ p ∈ P and q ∈ V is a ﬁnite set of consistent sentences of the form:
1. “x / ∈
◦
Un
” where x ∈ X or
2. “B ⊆
◦
Un
” where B ∈ B and n ∈ ω.
with the additional requirement that whenever the sentence “x / ∈
◦
Un
” is in
q and x = x
r
, then p and r are incompatible (so p [¬ r / ∈ G).
Note that if D ⊆ P is dense in P, then D is predense in P∗
◦
Q, i.e., every
r ∈ P∗
◦
Q is compatible with an element of D. Consequently, it is enough
to ﬁnd suﬃciently generic ﬁlters for P∗
◦
Q. By Lemma 11.2 and Sikorski’s
Theorem 10.1 it is enough to see that if P∗
◦
Q⊆ B is dense in the ccc cBa
algebra B, then B is countably generated. Let
C = ¦[[ B ⊆ U
n
[] : B ∈ B, n ∈ ω¦.
We claim that C generates B. To see this, note that for each p ∈ P
[[ x
p
∈ ∩
n
U
n
[] =
n∈ω
[[ x
p
∈ U
n
[]
[[ x
p
∈ U
n
[] =
B∈B,x
p
∈B
[[ B ⊆ U
n
[]
furthermore
(p, ∅) = [[ x
p
∈ ∩
n
U
n
[]
Descriptive Set Theory and Forcing 42
and so it follows that every element of P∗
◦
Q is in the boolean algebra gen
erated by C and so since P∗
◦
Q is dense in B it follows that C generates B.
Deﬁne X ⊆ 2
ω
to be a generalized ILuzin set for an ideal I in the Borel
sets iﬀ [X[ = c and [X ∩ A[ < c for every A ∈ I. It follows from the
MartinSolovay Theorem 11.1 that (assuming that the continuum is regular)
MA is equivalent to
for every ccc ideal I in the Borel subsets of 2
ω
there exists a generalized
ILuzin set.
Miller and Prikry [84] show that it is necessary to assume the continuum
is regular in the above observation.
12 BOOLEAN ALGEBRA OF ORDER ω
1
43
12 Boolean algebra of order ω
1
Now we use the MartinSolovay technique to produce a countably generated
ccc cBa with order ω
1
. Before doing so we introduce a countable version of
αforcing which will be useful for other results also. It is similar to one used
in Miller [76] to give a simple proof about generating sets in the category
algebra.
Let T be a nice tree of rank α (2 ≤ α < ω
1
). Deﬁne
P
α
= ¦p : D → ω : D ∈ [ω]
<ω
, ∀s, sˆn ∈ D p(s) ,= p(sˆn)¦.
This is ordered by p ≤ q iﬀ p ⊇ q. For p ∈ P
α
deﬁne
rank(p) = max¦r
T
(s) : s ∈ domain(p)¦
where r
T
is the rank function on T.
Lemma 12.1 rank : P
α
→ α+1 satisﬁes the Rank Lemma 7.4, i.e, for every
p ∈ P
α
and β ≥ 1 there exists ˆ p ∈ P
α
such that
1. ˆ p is compatible with p,
2. rank(ˆ p) ≤ β, and
3. for any q ∈ P
α
if rank(q) < β and ˆ p and q are compatible, then p and
q are compatible.
Proof:
First let p
0
≤ p be such that for every s ∈ domain(p) and n ∈ ω if
r
T
(sˆn) < β < λ = r
T
(s)
then there exists m ∈ ω with p
0
(sˆnˆm) = p(s). Note that
r
T
(sˆn) < β < λ = r
T
(s)
can happen only when λ is a limit ordinal and for any such s there can be
at most ﬁnitely many n (because T is a nice tree).
Now let
E = ¦s ∈ domain(p
0
) : r
T
(s) ≤ β¦
and deﬁne ˆ p = p
0
E. It is compatible with p since p
0
is stronger than both.
From its deﬁnition it has rank ≤ β. So let q ∈ P
α
have rank(q) < β and be
incompatible with p. We need to show it is incompatible with ˆ p. There are
only three ways for q and p to be incompatible:
Descriptive Set Theory and Forcing 44
1. ∃s ∈ domain(p) ∩ domain(q) p(s) ,= q(s),
2. ∃s ∈ domain(q) ∃sˆn ∈ domain(p) q(s) = p(s
ˆ
n), or
3. ∃s ∈ domain(p) ∃sˆn ∈ domain(q) p(s) = q(s
ˆ
n).
For (1) since rank(q) < β we know r
T
(s) < β and hence by construction s is
in the domain of ˆ p and so q and ˆ p are incompatible. For (2) since
r
T
(sˆn) < r
T
(s) < β
we get the same conclusion. For (3) since sˆn ∈ domain(q) we know
r
T
(sˆn) < β.
If r
T
(s) = β, then s ∈ domain(ˆ p) and so q and ˆ p are incompatible. Otherwise
since T is a nice tree,
r
T
(sˆn) < β < r
T
(s) = λ
a limit ordinal. In this case we have arranged ˆ p so that there exists m with
p(s) = ˆ p(sˆnˆm) and so again q and ˆ p are incompatible.
Lemma 12.2 There exists a countable family T of dense subsets of P
α
such
that for every G a P
α
ﬁlter which meets each dense set in T the ﬁlter G
determines a map x : T → ω by p ∈ G iﬀ p ⊆ x. This map has the property
that for every s ∈ T
>0
the value of x(s) is the unique element of ω not in
¦x(sˆn) : n ∈ ω¦.
Proof:
For each s ∈ T the set
D
s
= ¦p : s ∈ domain(p)¦
is dense. Also for each s ∈ T
>0
and k ∈ ω the set
E
k
s
= ¦p : p(s) = k or ∃n p(sˆn) = k¦
is dense.
The poset P
α
is separative, since if p ,≤ q then either p and q are incom
patible or there exists s ∈ domain(q) ¸ domain(p) in which case we can ﬁnd
ˆ p ≤ p with ˆ p(s) ,= q(s).
Descriptive Set Theory and Forcing 45
Now if P
α
⊆ B is dense in the cBa B, it follows that for each p ∈ P
α
p = [[ p ⊆ x []
and for any s ∈ T
>0
and k
[[ x(s) = k [] =
m∈ω
[[ x(s
ˆ
m) ,= k [].
Consequently if
C = ¦p ∈ P
α
: domain(p) ⊆ T
0
¦
then C ⊆ B has the property that ord(C) = α + 1.
Now let
α<ω
1
P
α
be the direct sum, i.e., p = ¸p
α
: α < ω
1
) with p
α
∈ P
α
and p
α
= 1
α
= ∅ for all but ﬁnitely many α. This forcing is equivalent to
adding ω
1
Cohen reals, so the usual deltalemma argument shows that it is
ccc. Let
X = ¦x
α,s,n
∈ 2
ω
: α < ω
1
, s ∈ T
0
α
, n ∈ ω¦
be distinct elements of 2
ω
. For G = ¸G
α
: α < ω
1
) which is
α<ω
1
P
α
generic
over V , use X and Silver forcing to code the rank zero parts of each G
α
, i.e.,
deﬁne (
α<ω
1
P
α
)∗
◦
Q by (p, q) ∈ (
α<ω
1
P
α
)∗
◦
Q
iﬀ
p ∈
α<ω
1
P
α
and q is a ﬁnite set of consistent sentences of the form:
1. “x / ∈
◦
Un
” where x ∈ X or
2. “B ⊆
◦
Un
” where B is clopen and n ∈ ω.
with the additional proviso that whenever “x
α,s,n
/ ∈
◦
Un
” ∈ q then s is in the
domain of p
α
and p
α
(s) ,= n. This is a little stronger than saying p [¬ ˇ q ∈ Q,
but would be true for a dense set of conditions.
The rank function
rank : (
α<ω
1
P
α
)∗
◦
Q:→ ω
1
is deﬁned by
rank(¸p
α
: α < ω
1
), q) = max¦rank(p
α
) : α < ω
1
¦
which means we ignore q entirely.
Descriptive Set Theory and Forcing 46
Lemma 12.3 For every p ∈ (
α<ω
1
P
α
)∗
◦
Q and β ≥ 1 there exists ˆ p in the
poset (
α<ω
1
P
α
)∗
◦
Q such that
1. ˆ p is compatible with p,
2. rank(ˆ p) ≤ β, and
3. for any q ∈ (
α<ω
1
P
α
)∗
◦
Q if rank(q) < β and ˆ p and q are compatible,
then p and q are compatible.
Proof:
Apply Lemma 12.1 to each p
α
to obtain ˆ p
α
and then let
ˆ p = (¸ ˆ p
α
: α < ω
1
), q).
This is still a condition because ˆ p
α
retains all the rank zero part of p
α
which
is needed to force q ∈
◦
Q.
Let (
α<ω
1
P
α
)∗
◦
Q⊆ B be a dense subset of the ccc cBa B. We show
that B is countably generated and ord(B) = ω
1
. A strange thing about ω
1
is that if one countable set of generators has order ω
1
, then all countable
sets of generators have order ω
1
. This is because any countable set will be
generated by a countable stage.
One set of generators for B is
C = ¦[[
ˇ
B ⊆
◦
U
n
[] : B clopen , n ∈ ω¦.
Note that
[[ x ∈ ∩
n∈ω
U
n
[] =
n∈ω
[[ x ∈ U
n
[] =
n∈ω
¦[[
ˇ
B ⊆
◦
U
n
[] : x ∈ B¦
and also each P
α
is generated by
¦p ∈ P
α
: domain(p) ⊆ T
0
α
¦.
We know that for each α < ω
1
, s ∈ T
0
α
and n ∈ ω if p = (¸p
α
: α < ω
1
), q) is
the condition for which p
α
is the function with domain ¦s¦, and p
α
(s) = n,
and the rest of p is the trivial condition, then
p = [[ ˇ x
α,s,n
∈
n∈ω
◦
U
n
[].
Descriptive Set Theory and Forcing 47
From these facts it follows that C generates B.
It follows from Lemma 8.4 that the order of C is ω
1
. For any β < ω
1
let
b = (¸p
α
: α < ω
1
), q) be the condition all of whose components are trivial
except for p
β
, and p
β
any the function with domain ¸). Then b / ∈ Σ
0
β
¯
(C).
Otherwise by Lemma 8.4, there would be some a ≤ b with rank(a, C) < β,
but then p
a
β
would not have ¸) in its domain.
This proves the ω
1
case of Theorem 8.2.
13 LUZIN SETS 48
13 Luzin sets
In this section we use Luzin sets and generalized ILuzin sets to construct
separable metric spaces of various Borel orders. Before doing so we review
some standard material on the property of Baire.
Given a topological space X the σideal of meager sets is deﬁned as fol
lows. Y ⊆ X is nowhere dense iﬀ the interior of the closure of Y is empty,
i.e, int(cl(Y )) = ∅. A subset of X is meager iﬀ it is the countable union
of nowhere dense sets. The Baire category Theorem is the statement that
nonempty open subsets of compact Hausdorﬀ spaces or completely metriz
able spaces are not meager. A subset B of X has the Baire property iﬀ there
exists U open such that B∆U is meager.
Theorem 13.1 (Baire) The family of sets with the Baire property forms a
σﬁeld.
Proof:
If B∆U is meager where U is open, then
B∆cl(U) = (B ¸ cl(U)) ∪ (cl(U) ¸ B)
and (B ¸ cl(U)) ⊆ B ¸ U is meager and (cl(U) ¸ B) ⊆ (U ¸ B) ∪ (cl(U) ¸ U)
is meager because cl(U) ¸ U is nowhere dense. Therefore,
∼ B ∆ ∼ cl(U) = B∆cl(U)
is meager.
If B
n
∆U
n
is meager for each n, then
(
_
n∈ω
B
n
)∆(
_
n∈ω
U
n
) ⊆
_
n∈ω
B
n
∆U
n
is meager.
Hence every Borel set has the property of Baire.
Theorem 13.2 Suppose that every nonempty open subset of X is nonmea
ger, then B = Borel(X)/meager(X) is a cBa.
Descriptive Set Theory and Forcing 49
Proof:
It is enough to show that it is complete. Suppose Γ ⊆ B is arbitrary. Let
 be a family of open sets such that
Γ = ¦[U]
meager(X)
: U ∈ ¦.
Let V =
 and we claim that [V ] is the minimal upper bound of Γ in B.
Clearly it is an upper bound. Suppose [W] is any upper bound for Γ with W
open. So U ¸ W is meager for every U ∈ . We need to show that V ¸ W is
meager (so [V ] ≤ [W]). V ¸ W ⊆ cl(V ) ¸ W and if the latter is not nowhere
dense, then there exists P a nonempty open set with P ⊆ cl(V ) ¸ W. Since
V =
 we may assume that there exists U ∈  with P ⊆ U. But P is
a nonempty open set and [P] ≤ [W] so it is impossible for P to be disjoint
from W.
We say that X ⊆ 2
ω
is a super Luzin set iﬀ for every Borel set B the
set X ∩ B is countable iﬀ B is meager. (This is equivalent to superILuzin
where I is the ideal of meager sets.) It is easy to see that if X is an ordinary
Luzin set, then in some basic clopen set C it is a super Luzin set relative to
C. Also since ω
ω
can be obtained by deleting countably many points from
2
ω
it is clear that having a Luzin set for one is equivalent to having it for the
other. With a little more work it can be seen that it is equivalent to having
one for any completely metrizable separable metric space without isolated
points.
The generic set of Cohen reals in the Cohen real model is a Luzin set.
Let FIN(κ, 2) be the partial order of ﬁnite partial functions from κ into
2. If G is FIN(κ, 2)generic over V and for each α < κ we deﬁne x
α
by
x
α
(n) = G(ω ∗ α +n), then X = ¦x
α
: α < κ¦ is a Luzin set in V [G].
Theorem 13.3 (Miller [75]) If there exists a Luzin set in ω
ω
, then for every
α with 3 ≤ α < ω
1
there exists Y ⊆ ω
ω
with ord(Y ) = α.
Proof:
Let T
α
be the nice αtree used in the deﬁnition of
P
α
= ¦p : p : D → ω, D ∈ [T
α
]
<ω
, ∀s, sˆn ∈ D p(s) ,= p(sˆn)¦.
Let Q
α
be the closed subspace of ω
T
α
Q
α
= ¦x ∈ ω
T
α
: ∀s, sˆn ∈ T
α
x(s) ,= x(sˆn).
Descriptive Set Theory and Forcing 50
For each p ∈ P
α
we get a basic clopen set
[p] = ¦x ∈ Q
α
: p ⊆ x¦.
It easy to check that Q
α
is homeomorphic to ω
ω
. Hence there exists a super
Luzin set X ⊆ Q
α
. Consider the map r : Q
α
→ ω
T
0
α
deﬁned by restriction,
i.e., r(x) = x T
0
α
. Note that by Lemma 12.2 there exists a countable
sequence of dense open subsets of Q
α
, ¸D
n
: n ∈ ω), such that r is onetoone
on
n∈ω
D
n
. Since
n∈ω
D
n
is a comeager set in Q
α
and X is Luzin we may
assume that
X ⊆
n∈ω
D
n
.
Y is just the image of X under r. (So, in fact, Y is the onetoone
continuous image of a Luzin set.) An equivalent way to view Y is just to
imagine X with the topology given by
B = ¦[p] : p ∈ P
α
, domain(p) ⊆ T
0
α
¦.
We know by Lemma 8.4 that
ord¦[B] : B ∈ B¦ = α + 1
as a subset of Borel(Q
α
)/meager(Q
α
) which means that:
α+1 is minimal such that for every B ∈ Borel(Q
α
) there exists a Σ
0
α+1
¯
(B)
set A such that such that B∆A is meager in Q
α
.
This translates (since X is superLuzin) to:
α+1 is minimal such that for every B ∈ Borel(Q
α
) there exists a Σ
0
α+1
¯
(B)
set A such that such that (B∆A) ∩ X is countable.
Which means for Y that:
α+1 is minimal such that for every B ∈ Borel(Y ) there exists a Σ
0
α+1
¯
(Y )
set A such that such that B∆A is countable.
But since countable subsets of Y are Σ
0
2
¯
and α > 2, this means ord(Y ) =
α + 1.
To get Y of order λ for a limit λ < ω
1
just take a clopen separated union
of sets whose order increases to λ.
Now we clean up a loose end from Miller [75]. In that paper we had shown
that assuming MA for every α < ω
1
there exists a separable metric space X
with α ≤ ord(X) ≤ α + 2 or something silly like that. Shortly afterwards,
Fremlin supplied the missing arguments to show the following.
Descriptive Set Theory and Forcing 51
Theorem 13.4 (Fremlin [26]) MA implies that for every α with 2 ≤ α ≤ ω
1
there exists a second countable Hausdorﬀ space X with ord(X) = α.
Proof:
Since the union of less than continuum many meager sets is meager, the
Mahlo construction 10.2 gives us a set X ⊆ Q
α
of cardinality c such that
for every Borel set B ∈ Borel(Q
α
) we have that B is meager iﬀ B ∩ X has
cardinality less than c.
Letting B be deﬁned as in the proof of Theorem 13.3 we see that:
α+1 is minimal such that for every B ∈ Borel(Q
α
) there exists a Σ
0
α+1
¯
(B)
set A such that such that (B∆A) ∩ X has cardinality less than c.
What we need to see to complete the proof is that:
for every Z ⊆ X of cardinality less than c there exists a Σ
0
2
¯
(B) set F such
that F ∩ X = Z.
Lemma 13.5 (MA) For any Z ⊆ Q
α
of cardinality less than c, there exists
¸D
n
: n ∈ ω) such that:
1. D
n
is predense in P
α
,
2. p ∈ D
n
implies domain(p) ⊆ T
0
α
, and
3. Z ∩
n∈ω
s∈D
n
[s] = ∅.
Proof:
Force with the following poset
P = ¦(F, ¸p
n
: n < N)) : F ∈ [Z]
<ω
, N < ω, domain(p) ∈ [T
0
α
]
<ω
¦
where (F, ¸p
n
: n < N)) ≤ (H, ¸q
n
: n < M)) iﬀ F ⊇ H, N ≥ M, p
n
= q
n
for n < M, and for each x ∈ H and M ≤ n < N we have x / ∈ [p
n
]. Since
this forcing is ccc we can apply MA with the appropriate choice of family of
dense sets to get D
n
= ¦p
m
: m > n¦ to do the job.
By applying the Lemma we get that for every Z ⊆ X of cardinality less
than c there exists a Σ
0
2
¯
(B) set F which is meager in Q
α
and such that
Z ⊆ F ∩X. But since F is meager we know F ∩X has cardinality less than
c. By Theorem 5.1 every subset of r(F ∩ X) is a relative Σ
0
2
¯
in Y , so there
exists an F
0
a Σ
0
2
¯
(B) set such that Z = (F ∩ X) ∩ F
0
. This proves Theorem
13.4.
14 COHEN REAL MODEL 52
14 Cohen real model
I have long wondered if there exists an uncountable separable metric space
of order 2 in the Cohen real model. I thought there weren’t any. We already
know from Theorem 13.3 that since there is an uncountable Luzin set in
Cohen real model that for every α with 3 ≤ α ≤ ω
1
there is an uncountable
separable metric space X with ord(X) = α.
Theorem 14.1 Suppose G is FIN(κ, 2)generic over V where κ ≥ ω
1
. Then
in V [G] there is a separable metric space X of cardinality ω
1
with ord(X) = 2.
Proof:
We may assume that κ = ω
1
. This is because FIN(κ, 2) FIN(ω
1
, 2) is
isomorphic to FIN(κ, 2) and so by the product lemma we may replace V by
V [H] where (H, G) is FIN(κ, 2) FIN(ω
1
, 2)generic over V .
We are going to use the fact that forcing with FIN(ω
1
, 2) is equivalent
to any ﬁnite support ω
1
iteration of countable posets. The main idea of the
proof is to construct an Aronszajn tree of perfect sets, a technique ﬁrst used
by Todorcevic (see Galvin and Miller [30]). We construct an Aronszajn
tree (A, ¢) and a family of perfect sets ([T
s
] : s ∈ A) such that ⊇ is the
same order as ¢. We will then show that if X = ¦x
s
: s ∈ A¦ is such that
x
s
∈ [T
s
], then the order of X is 2.
In order to insure the construction can keep going at limit ordinals we
will need to use a fusion argument. Recall that a perfect set corresponds to
the inﬁnite branches [T] of a perfect tree T ⊆ 2
<ω
, i.e., a tree with the
property that for every s ∈ T there exist a t ∈ T such that both tˆ0 ∈ T
and tˆ1 ∈ T. Such a T is called a splitting node of T. There is a natural
correspondence of the splitting nodes of a perfect tree T and 2
<ω
.
Given two perfect trees T and T
and n ∈ ω deﬁne T ≤
n
T
iﬀ T ⊆ T
and the ﬁrst 2
<n
splitting nodes of T remain in T
.
Lemma 14.2 (Fusion) Suppose (T
n
: n ∈ ω) is a sequence of perfect sets
such that T
n+1
≤
n
T
n
for every n ∈ ω. Then T =
n∈ω
T
n
is a perfect tree
and T ≤
n
T
n
for every n ∈ ω.
Proof:
Descriptive Set Theory and Forcing 53
If T =
n<ω
T
n
, then T is a perfect tree because the ﬁrst 2
<n
splitting
nodes of T
n
are in T
m
for every m > n and thus in T.
By identifying FIN(ω
1
, 2) with
α<ω
1
FIN(ω, 2) we may assume that
G = ¸G
α
: α < ω
1
)
where G
β
is FIN(ω, 2)generic over V [G
α
: α < β] for each β < ω
1
.
Given an Aronszajn tree A we let A
α
be the nodes of A at level α, i.e.
A
α
= ¦s ∈ A : ¦t ∈ A : t ¡s¦ has order type α¦
and
A
<α
=
_
β<α
A
β
.
We use ¸G
α
: α < ω
1
) to construct an Aronszajn tree (A, ¢) and a family of
perfect sets ([T
s
] : s ∈ A) such that
1. s ¢t implies T
s
⊇ T
t
,
2. if s and t are distinct elements of A
α
, then [T
s
] and [T
t
] are disjoint,
3. every s ∈ A
α
has inﬁnitely many distinct extensions in A
α+1
,
4. for each s ∈ A
<α
and n < ω there exists t ∈ A
α
such that T
t
≤
n
T
s
,
5. for each s ∈ A
α
and t ∈ A
α+1
with s ¡t, we have that [T
t
] is a generic
perfect subset of [T
s
] obtained by using G
α
(explained below in Case
2), and
6. ¦T
s
: s ∈ A
<α
¦ ∈ V [G
β
: β < α].
The ﬁrst three items simply say that ¦[T
s
] : s ∈ A¦ and its ordering by
⊆ determines (A, _), so what we really have here is an Aronszajn tree of
perfect sets. Item (4) is there in order to allow the construction to proceed
at limits levels.
Item (5) is what we do a successor levels and guarantees the set we are
building has order 2. Item (6) is a consequence of the construction and would
be true for a closed unbounded set of ordinals no matter what we did anyway.
Here are the details of our construction.
Descriptive Set Theory and Forcing 54
Case 1. α a limit ordinal.
The construction is done uniformly enough so that we already have that
¦T
s
: s ∈ A
<α
¦ ∈ V [G
β
: β < α].
Working in V [G
β
: β < α] choose a sequence α
n
for n ∈ ω which strictly
increases to α. Given any s
n
∈ A
α
n
we can choose by inductive hypothesis a
sequence s
m
∈ A
α
m
for m ≥ n such that
T
s
m+1
≤
m
T
s
m
.
If T =
m>n
T
s
m
, then by Lemma 14.2 we have that T ≤
n
T
s
n
. Now let
¦T
t
: t ∈ A
α
¦ be a countable collection of perfect trees so that for every n
and s ∈ A
α
n
there exists t ∈ A
α
with T
t
≤
n
T
s
. This implies item (4) because
for any s ∈ A
<α
and n < ω there exists some m ≥ n with s ∈ A
<α
m
hence by
inductive hypothesis there exists ˆ s ∈ A
α
m
with T
ˆ s
≤
n
T
s
and by construction
there exists t ∈ A
α
with T
t
≤
m
T
ˆ s
and so T
t
≤
n
T
s
as desired.
Case 2. Successor stages.
Suppose we already have constructed
¦T
s
: s ∈ A
<α+1
¦ ∈ V [G
β
: β < α + 1].
Given a perfect tree T ⊆ 2
<ω
deﬁne the countable partial order P(T) as
follows. p ∈ P(T) iﬀ p is a ﬁnite subtree of T and p ≤ q iﬀ p ⊇ q and p is an
end extension of q, i.e., every new node of p extends a terminal node of q. It
is easy to see that if G is P(T)generic over a model M, then
T
G
=
_
¦p : p ∈ G¦
is a perfect subtree of T. Furthermore, for any D ⊆ [T] dense open in [T]
and coded in M, [T
G
] ⊆ D. i.e., the branches of T
G
are Cohen reals (relative
to T) over M. This means that for any Borel set B ⊆ [T] coded in M, there
exists an clopen set C ∈ M such that
C ∩ [T
G
] = B ∩ [T
G
].
To see why this is true let p ∈ P(T) and B Borel. Since B has the Baire
property relative to [T] by extending each terminal node of p, if necessary,
we can obtain q ≥ p such that for every terminal node s of q either [s] ∩B is
Descriptive Set Theory and Forcing 55
meager in [T] or [s] ∩ B is comeager in [T] ∩ [s]. If we let C be union of all
[s] for s a terminal node of q such that [s] ∩ B is comeager in [T] ∩ [s], then
q [¬ B ∩ T
G
= C ∩ T
G
.
To get T
G
≤
n
T we could instead force with
P(T, n) = ¦p ∈ P(T) : p end extends the ﬁrst 2
<n
splitting nodes of T¦.
Finally to determine A
α+1
consider
¦P(T
s
, m) : s ∈ A
α
, m ∈ ω¦.
This poset is countable and hence G
α+1
determines a sequence
¸T
s,m
: s ∈ A
α
, m ∈ ω)
of generic perfect trees such that T
s,m
≤
m
T
s
. Note that genericity also
guarantees that corresponding perfect sets will be disjoint. We deﬁne A
α+1
to be this set of generic trees.
This ends the construction.
By taking generic perfect sets at successor steps we have guaranteed the
following. For any Borel set B coded in V [G
β
: β < α+1] and T
t
for t ∈ A
α+1
there exists a clopen set C
t
such that
C
t
∩ [T
t
] = B ∩ [T
t
].
Suppose X = ¦x
s
: s ∈ A¦ is such that x
s
∈ [T
s
] for every s ∈ A. Then X
has order 2. To verify this, let B ⊆ 2
ω
be any Borel set. By ccc there exists
a countable α such that B is coded in V [G
β
: β < α + 1]. Hence,
B ∩
_
t∈A
α+1
[T
t
] =
_
t∈A
α+1
(C
t
∩ [T
t
]).
Hence B ∩ X is equal to a Σ
0
2
¯
set intersected X:
X ∩
_
t∈A
α+1
(C
t
∩ [T
t
])
Descriptive Set Theory and Forcing 56
union a countable set:
(B ∩ X) ¸
_
t∈A
α+1
[T
t
]
and therefore B ∩ X is Σ
0
2
¯
in X.
Another way to get a space of order 2 is to use the following argument. If
the ground model satisﬁes CH, then there exists a Sierpi´ nski set. Such a set
has order 2 (see Theorem 15.1) in V and therefore by the next theorem it has
order 2 in V [G]. It also follows from the next theorem that if X = 2
ω
∩ V ,
then X has order ω
1
in V [G]. Consequently, in what I think of as “the Cohen
real model”, i.e. the model obtained by adding ω
2
Cohen reals to a model of
CH, there are separable metric spaces of cardinality ω
1
and order α for every
α with 2 ≤ α ≤ ω
1
.
Theorem 14.3 Suppose G is FIN(κ, 2)generic over V and
V [= ord(X) = α
Then
V [G] [= ord(X) = α
By the usual ccc arguments it is clearly enough to prove the Theorem for
FIN(ω, 2). To prove it we will need the following lemma.
Lemma 14.4 (Kunen, see [57]) Suppose p ∈ FIN(ω, 2), X is a second count
able Hausdorﬀ space in V , and
◦
B
is a name such that
p [¬
◦
B
⊆
ˇ
X is a Π
0
α
¯
set.
Then the set
¦x ∈ X : p [¬ ˇ x ∈
◦
B
¦
is a Π
0
α
¯
set in X.
Proof:
This is proved by induction on α.
For α = 1 let B ∈ V be a countable base for the closed subsets of X,
i.e., every closed set is the intersection of elements of B. Suppose p [¬“
◦
B
is
Descriptive Set Theory and Forcing 57
a closed set in
ˇ
X”. Then for every x ∈ X p [¬“ˇ x ∈
◦
B
” iﬀ for every q ≤ p and
for every C ∈ B if q [¬“
◦
B
⊆
ˇ
C”, then x ∈ C. But
¦x ∈ X : ∀q ≤ p ∀B ∈ B (q [¬ “
◦
B
⊆
ˇ
C” → x ∈ C)¦
is closed in X.
Now suppose α > 1 and p [¬
◦
B
∈ Π
0
α
¯
(X). Let β
n
be a sequence which
is either constantly α − 1 if α is a successor or which is unbounded in α
if α is a limit. By the usual forcing facts there exists a sequence of names
¸B
n
: n ∈ ω) such that for each n,
p [¬ B
n
∈ Π
0
β
n
¯
,
and
p [¬ B =
n<ω
∼ B
n
.
Then for every x ∈ X
p [¬ ˇ x ∈ B
iﬀ
∀n ∈ ω p [¬ x ∈∼ B
n
iﬀ
∀n ∈ ω ∀q ≤ p q ,[¬ x ∈ B
n
.
Consequently,
¦x ∈ X : p [¬ x ∈
◦
B
¦ =
n∈ω
q≤p
∼ ¦x : q [¬ ˇ x ∈
◦
B
m
¦.
Now let us prove the Theorem. Suppose V [=“ord(X) = α”. Then in
V [G] for any Borel set B ∈ Borel(X)
B =
_
p∈G
¦x ∈ X : p [¬ ˇ x ∈
◦
B
¦.
By the lemma, each of the sets ¦x ∈ X : p [¬ ˇ x ∈
◦
B
¦ is a Borel set in V , and
since ord(X) = α, it is a Σ
0
α
¯
set. Hence, it follows that B is a Σ
0
α
¯
set. So,
Descriptive Set Theory and Forcing 58
V [G] [= ord(X) ≤ α. To see that ord(X) ≥ α let β < α and suppose in V
the set A ⊆ X is Σ
0
β
¯
but not Π
0
β
¯
. This must remain true in V [G] otherwise
there exists a p ∈ G such that
p [¬ “
ˇ
A is Π
0
β
¯
”
but by the lemma
¦x ∈ X : p [¬ ˇ x ∈
ˇ
A¦ = A
is Π
0
β
¯
which is a contradiction.
Part of this argument is similar to one used by Judah and Shelah [46]
who showed that it is consistent to have a Qset which does not have strong
measure zero.
It is natural to ask if there are spaces of order 2 of higher cardinality.
Theorem 14.5 Suppose G is FIN(κ, 2)generic over V where V is a model
of CH and κ ≥ ω
2
. Then in V [G] for every separable metric space X with
[X[ > ω
1
, we have ord(X) ≥ 3.
Proof:
This will follow easily from the next lemma.
Lemma 14.6 (Miller [81]) Suppose G is FIN(κ, 2)generic over V where V
is a model of CH and κ ≥ ω
2
. Then V [G] models that for every X ⊆ 2
ω
with [X[ = ω
2
there exists a Luzin set Y ⊆ 2
ω
and a onetoone continuous
function f : Y → X.
Proof:
Let ¸τ
α
: α < ω
2
) be a sequence of names for distinct elements of X. For
each α and n choose a maximal antichain A
α
n
∪ B
α
n
such that
p [¬ τ
α
(n) = 0 for each p ∈ A
α
n
and
p [¬ τ
α
(n) = 1 for each p ∈ B
α
n
.
Let X
α
⊆ κ be union of domains of elements from
n∈ω
A
α
n
∪B
α
n
. Since each
X
α
is countable we may as well assume that the X
α
’s form a ∆system, i.e.
there exists R such that X
α
∩X
β
= R for every α ,= β. We can assume that
R is the empty set. The reason is we can just replace A
α
n
by
ˆ
A
α
n
= ¦p (∼ R) : p ∈ A
α
n
and p R ∈ G¦
Descriptive Set Theory and Forcing 59
and similarly for B
α
n
. Then let V [G R] be the new ground model.
Let
¸j
α
: X
α
→ ω : α < ω
2
)
be a sequence of bijections in the ground model. Each j
α
extends to an
order preserving map from FIN(X
α
, 2) to FIN(ω, 2). By CH, we may as well
assume that there exists a single sequence, ¸(A
n
, B
n
) : n ∈ ω), such that
every j
α
maps ¸A
α
n
, B
α
n
: n ∈ ω) to ¸(A
n
, B
n
) : n ∈ ω).
The Luzin set is Y = ¦y
α
: α < ω
2
¦ where y
α
(n) = G(j
−1
α
(n)). The
continuous function, f, is the map determined by ¸(A
n
, B
n
) : n ∈ ω):
f(x)(n) = 0 iﬀ ∃m x m ∈ A
n
.
This proves the Lemma.
If f : Y → X is onetoone continuous function from a Luzin set Y ,
then ord(X) ≥ 3. To see this assume that Y is dense and let D ⊆ Y be
a countable dense subset of Y . Then D is not G
δ
in Y . This is because
any G
δ
set containing D is comeager and therefore must meet Y in an un
countable set. But note that f(D) is a countable set which cannot be G
δ
in
X, because f
−1
(f(D)) would be G
δ
in Y and since f is onetoone we have
D = f
−1
(f(D)). This proves the Theorem.
It is natural to ask about the cardinalities of sets of various orders in
this model. But note that there is a trivial way to get a large set of order
β. Take a clopen separated union of a large Luzin set (which has order 3)
and a set of size ω
1
with order β. One possible way to strengthen the notion
of order is to say that a space X of cardinality κ has essential order β iﬀ
every nonempty open subset of X has order β and cardinality κ. But this is
also open to a simple trick of combining a small set of order β with a large
set of small order. For example, let X ⊆ 2
ω
be a clopen separated union of
a Luzin set of cardinality κ and set of cardinality ω
1
of order β ≥ 3. Let
¸P
n
: n ∈ ω) be a sequence of disjoint nowhere dense perfect subsets of 2
ω
with the property that for every s ∈ 2
<ω
there exists n with P
n
⊆ [s]. Let
X
n
⊆ P
n
be a homeomorphic copy of X for each n < ω. Then
n∈ω
X
n
is a
set of cardinality κ which has essential order β.
With this cheat in mind let us deﬁne a stronger notion of order. A
separable metric space X has hereditary order β iﬀ every uncountable
Y ⊆ X has order β. We begin with a stronger version of Theorem 13.3.
Descriptive Set Theory and Forcing 60
Theorem 14.7 If there exists a Luzin set X of cardinality κ, then for every
α with 2 < α < ω
1
there exists a separable metric space Y of cardinality κ
which is hereditarily of order α.
Proof:
This is a slight modiﬁcation of the proof of Theorem 13.3. Let Q
α
be the
following partial order. Let ¸α
n
: n ∈ ω) be a sequence such that if α is a
limit ordinal, then α
n
is a coﬁnal increasing sequence in α and if α = β + 1
then α
n
= β for every n.
The rest of the proof is same except we use Q
α+1
instead of P
α
for suc
cessors and Q
α
for limit α instead of taking a clopen separated union. By
using the direct sum (even in the successor case) we get a stronger property
for the order. Let
ˆ
Q
α
=
Q
α
n
be the closed subspace of
n∈ω
ω
T
α
n
and let B be the collection of clopen subsets of Q
α
which are given by rank
zero conditions of Q(α), i.e., all rectangles of the form
n∈ω
[p
n
] such that
p
n
∈ Q
α
n
with domain(p) ⊆ T
0
α
and p
n
the trivial condition for all but ﬁnitely
many n.
As in the proof of Theorem 13.3 we get that the order of ¦[B] : B ∈ B¦
as a subset of Borel(
ˆ
Q
α
)/meager(
ˆ
Q
α
) is α. Because we took the direct sum
we get the stronger property that for any nonempty clopen set C in
ˆ
Q
α
the
order of ¦[B ∩ C] : B ∈ B¦ is α.
But know given X a Luzin set in
ˆ
Q
α
we know that for any uncountable
Y ⊆ X there is a nonempty clopen set C ⊆
ˆ
Q
α
such that Y ∩ C is a super
Luzin set relative to C. (The accumulation points of Y , the set of all
points all of whose neighborhoods contain uncountably many points of Y , is
closed and uncountable, therefore must have nonempty interior.) If C is a
nonempty clopen set in the interior of the accumulation points of Y , then
since ¦[B∩C] : B ∈ B¦ is α, we have by the proof of Theorem 13.3, that the
order of Y is α.
Theorem 14.8 Suppose that in V there is a separable metric space, X, with
hereditary order β for some β ≤ ω
1
. Let G be FIN(κ, 2)generic over V for
any κ ≥ ω. Then in V [G] the space X has hereditary order β.
Descriptive Set Theory and Forcing 61
Proof:
In V [G] let Y ⊆ X be uncountable. For contradiction, suppose that
p [¬ ord(
◦
Y
) ≤ α and [
◦
Y
[ = ω
1
for some p ∈ FIN(κ, 2) and α < β. Working in V by the usual ∆system
argument we can get q ≤ p and
¸p
x
: x ∈ A)
for some A ∈ [X]
ω
1
such that and p
x
≤ q and
p
x
[¬ ˇ x ∈
◦
Y
for each x ∈ A and
dom(p
x
) ∩ dom(p
y
) = dom(q)
for distinct x and y in A. Since A is uncountable we know that in V the
order of A is ω
1
. Consequently, there exists R ⊆ A which is Σ
0
α
¯
(A) but not
Π
0
α
¯
(A). We claim that in V [G] the set R ∩ Y is not Π
0
α
¯
(Y ). If not, there
exists r ≤ q and
◦
S
such that
r [¬ “
◦
Y
∩R =
◦
Y
∩
◦
S
and
◦
S
∈ Π
0
α
¯
(A)”.
Since Borel sets are coded by reals there exists Γ ∈ [κ]
ω
∩ V such that for
any x ∈ A the statement “ˇ x ∈
◦
S
” is decided by conditions in FIN(Γ, 2) and
also let Γ be large enough to contain the domain of r.
Deﬁne
T = ¦x ∈ A : q [¬ ˇ x ∈
◦
S
¦.
According to Lemma 14.4 the set T is Π
0
α
¯
(A). Consequently, (assuming
α ≥ 3) there are uncountably many x ∈ A with x ∈ R∆T. Choose such an
x which also has the property that dom(p
x
) ¸dom(q) is disjoint from Γ. This
can be done since the p
x
form a ∆ system. But now, if x ∈ T ¸ R, then
r ∪ p
x
[¬ “ˇ x ∈
◦
Y
∩
◦
S
and x / ∈
◦
Y
∩
ˇ
R”.
On the other hand, if x ∈ R ¸ T, then there exists ˆ r ≤ r in FIN(Γ, 2) such
that
ˆ r [¬ ˇ x / ∈
◦
S
Descriptive Set Theory and Forcing 62
and consequently,
ˆ r ∪ p
x
[¬ “ˇ x / ∈
◦
Y
∩
◦
S
and x ∈
◦
Y
∩
ˇ
R”.
Either way we get a contradiction and the result is proved.
Theorem 14.9 (CH) There exists X ⊆ 2
ω
such that X has hereditary order
ω
1
.
Proof:
By Theorem 8.2 there exists a countably generated ccc cBa B which has
order ω
1
. For any b ∈ B with b ,= 0 let ord(b) be the order of the boolean
algebra you get by looking only at ¦c ∈ B : c ≤ b¦. Note that in fact B has
the property that for every b ∈ B we have ord(b) = ω
1
. Alternatively, it easy
to show that any ccc cBa of order ω
1
would have to contain an element b
such that every c ≤ b has order ω
1
.
By the proof of the SikorskiLoomis Theorem 9.1 we know that B is
isomorphic to Borel(Q)/meager(Q) where Q is a ccc compact Hausdorﬀ space
with a basis of cardinality continuum.
Since Q has ccc, every open dense set contains an open dense set which
is a countable union of basic open sets. Consequently, by using CH, there
exists a family T of meager sets with [T[ = ω
1
such that every meager set
is a subset of one in T. Also note that for any nonmeager Borel set B in Q
there exists a basic open set C and F ∈ T with C¸F ⊆ B. Hence by Mahlo’s
construction (Theorem 10.2) there exists a set X ⊆ Q with the property that
for any Borel subset B of Q
[B ∩ X[ ≤ ω iﬀ B meager.
Let B be a countable ﬁeld of clopen subsets of Q such that
¦[B]
meager(Q)
: B ∈ B¦
generates B. Let
R = ¦X ∩ B : B ∈ B¦.
If
˜
X ⊆ 2
ω
is the image of X under the characteristic function of the sequence
B (see Theorem 4.1), then
˜
X has hereditary order ω
1
. Of course
˜
X is really
just the same as X but retopologized using B as a family of basic open sets.
Descriptive Set Theory and Forcing 63
Let Y ∈ [X]
ω
1
. Since ord(p) = ω
1
for any basic clopen set the following claim
shows that the order of Y (or rather the image of Y under the characteristic
function of the sequence B) is ω
1
.
Claim: There exists a basic clopen p in Q such that for every Borel B ⊆ p,
[B ∩ Y [ ≤ ω iﬀ B meager.
Proof:
Let p and q stand for nonempty basic clopen sets. Obviously if B is
meager then B ∩ Y is countable, since B ∩ X is countable. To prove the
other direction, suppose for contradiction that for every p there exists q ≤ p
and Borel B
q
⊆ q such that B
q
is comeager in q and B
q
∩ Y is countable.
By using ccc there exists a countable dense family Σ and B
q
for q ∈ Σ with
B
q
⊆ q Borel and comeager in q such that B
q
∩ Y is countable. But
B =
_
¦B
q
: q ∈ Σ¦
is a comeager Borel set which meets Y in a countable set. This implies that
Y is countable since X is contained in B except for countable many points.
Theorem 14.10 Suppose G is FIN(κ, 2)generic over V where V is a model
of CH and κ ≥ ω. Then in V [G] there exists a separable metric space X with
[X[ = ω
1
and hereditarily of order ω
1
.
Proof:
Immediate from Theorem 14.8 and 14.9.
Finally, we show that there are no large spaces of hereditary order ω
1
in
the Cohen real model.
Theorem 14.11 Suppose G is FIN(κ, 2)generic over V where V is a model
of CH and κ ≥ ω
2
. Then in V [G] for every separable metric space X with
[X[ = ω
2
there exists Y ∈ [X]
ω
2
with ord(Y ) < ω
1
.
Proof:
By the argument used in the proof of Lemma 14.6 we can ﬁnd
¸G
α
: α < ω
2
) ∈ V [G]
Descriptive Set Theory and Forcing 64
which is
α<ω
2
FIN(ω, 2)generic over V and a FIN(ω, 2)name τ for an
element of 2
ω
such that Y = ¦τ
G
α
: α < ω
2
¦ is subset of X. We claim that
ord(Y ) < ω
1
. Let
T = ¦[[ τ ∈ C [] : C ⊆ 2
ω
clopen ¦
where boolean values are in the unique complete boolean algebra B in which
FIN(ω, 2) is dense. Let F be the complete subalgebra of B which is generated
by T. First note that the order of T in F is less than ω
1
. This is because F
contains a countable dense set:
D = ¦
¦c ∈ F : p ≤ c¦ : p ∈ FIN(ω, 2)¦.
Since D is countable and Σ
0
1
¯
(D) = F, it follows that the order of T is
countable.
I claim that the order of Y is essentially less than or equal to the order
of T in F.
Lemma 14.12 Let B be a cBa, τ a Bname for an element of 2
ω
, and
T = ¦[[ τ ∈ C [] : C ⊆ 2
ω
clopen ¦.
Then for each B ⊆ 2
ω
a Π
0
α
¯
set coded in V the boolean value [[ τ ∈
ˇ
B [] is
Π
0
α
¯
(T) and conversely, for every c ∈ Π
0
α
¯
(T) there exists a B ⊆ 2
ω
a Π
0
α
¯
set
coded in V such that c = [[ τ ∈
ˇ
B [].
Proof:
Both directions of the lemma are simple inductions.
Now suppose the order of T in F is α. Let B ⊆ 2
ω
be any Borel set coded
in V [G]. By ccc there exists H = G Σ where Σ ⊆ κ is countable set in V
such that B is coded in V [H]. Consequently, since we could replace V with
V [H] and delete countably many elements of Y we may as well assume that
B is coded in the ground model. Since the order of T is α we have by the
lemma that there exists a Π
0
α
¯
set A such that
[[ τ ∈
ˇ
A [] = [[ τ ∈
ˇ
B [].
It follows that
Y ∩ A = Y ∩ B
and hence order of Y is less than or equal to α (or three since we neglected
countably many elements of Y ).
15 THE RANDOM REAL MODEL 65
15 The random real model
In this section we consider the question of Borel orders in the random real
model. We conclude with a few remarks about perfect set forcing.
A set X ⊆ 2
ω
is a Sierpi´ nski set iﬀ X is uncountable and for every
measure zero set Z we have X∩Z countable. Note that by Mahlo’s Theorem
10.2 we know that under CH Sierpi´ nski sets exists. Also it is easy to see that
in the random real model, the set of reals given by the generic ﬁlter is a
Sierpi´ nski set.
Theorem 15.1 (Poprougenko [91]) If X is Sierpi´ nski, then ord(X) = 2.
Proof:
For any Borel set B ⊆ 2
ω
there exists an F
σ
set with F ⊆ B and B ¸ F
measure zero. Since X is Sierpi´ nski (B ¸ F) ∩ X = F
0
is countable, hence
F
σ
. So
B ∩ X = (F ∪ F
0
) ∩ X.
I had been rather hoping that every uncountable separable metric space
in the random real model has order either 2 or ω
1
. The following result shows
that this is deﬁnitely not the case.
Theorem 15.2 Suppose X ∈ V is a subspace of 2
ω
of order α and G is
measure algebra 2
κ
generic over V , i.e. adjoin κ many random reals.
Then V [G] [= α ≤ ord(X) ≤ α + 1.
The result will easily follow from the next two lemmas.
Presumably, ord(X) = α in V [G], but I haven’t been able to prove this.
Fremlin’s proof (Theorem 13.4) having ﬁlled up one such missing gap, leaving
this gap here restores a certain cosmic balance of ignorance.
5
Clearly, by the usual ccc arguments, we may assume that κ = ω and
G is just a random real. In the following lemmas boolean values [[ θ [] will
be computed in the measure algebra B on 2
ω
. Let µ be the usual product
measure on 2
ω
.
5
All things I thought I knew; but now confess, the more I know I know, I know the
less. John Owen (15601622)
Descriptive Set Theory and Forcing 66
Lemma 15.3 Suppose a real, b ∈ B, and
◦
U
the name of a Π
0
α
¯
subset of 2
ω
in V [G]. Then the set
¦x ∈ 2
ω
: µ(b ∧[[ ˇ x ∈
◦
U
[]) ≥ ¦
is Π
0
α
¯
in V .
Proof:
The proof is by induction on α.
Case α = 1.
If
◦
U
is a name for a closed set, then
[[ ˇ x ∈
◦
U
[] =
n∈ω
[[ [x n]∩
◦
U
,= ∅ [].
Consequently,
µ(b ∧[[ ˇ x ∈
◦
U
[]) ≥
iﬀ
∀n ∈ ω µ(b ∧[[ [x n]∩
◦
U
,= ∅ []) ≥
and the set is closed.
Case α > 1.
Suppose
◦
U
=
n∈ω
∼
◦
U
n
where each
◦
U
n
is a name for a Π
0
α
n
¯
set for some
α
n
< α. We can assume that the sequence ∼ U
n
is descending. Consequently,
µ(b ∧[[ ˇ x ∈
ˇ
U []) ≥
iﬀ
µ(b ∧[[ ˇ x ∈
n∈ω
∼
◦
U
n
[]) ≥
iﬀ
∀n ∈ ω µ(b ∧[[ ˇ x ∈∼
◦
U
n
[]) ≥
iﬀ
∀n ∈ ω not µ(b ∧[[ ˇ x ∈
◦
U
n
[]) > µ(b) −.
iﬀ
∀n ∈ ω not ∃m ∈ ω µ(b ∧[[ ˇ x ∈
◦
U
n
[]) ≥ µ(b) − + 1/m
Descriptive Set Theory and Forcing 67
By induction, each of the sets
¦x ∈ 2
ω
: µ(b ∧[[ ˇ x ∈
◦
U
n
[]) ≥ µ(b) − + 1/m¦
is Π
0
α
n
¯
and so the result is proved.
It follows from this lemma that if X ⊆ 2
ω
and V [=“ord(X) > α”, then
V [G] [=“ord(X) > α”. For suppose F ⊆ 2
ω
is Σ
0
α
¯
such that for every H ⊆ 2
ω
which is Π
0
α
¯
we have F ∩ X ,= H ∩ X. Suppose for contradiction that
b [¬ “
◦
U
∩
ˇ
X =
ˇ
F ∩
ˇ
X and
◦
U
is Π
0
α
¯
”.
But then
¦x ∈ 2
ω
: µ(b ∧[[ ˇ x ∈
ˇ
U []) = µ(b)¦
is a Π
0
α
¯
set which must be equal to F on X, which is a contradiction.
To prove the other direction of the inequality we will use the following
lemma.
Lemma 15.4 Let G be Bgeneric (where B is the measure algebra on 2
ω
)
and r ∈ 2
ω
is the associated random real. Then for any b ∈ B
b ∈ G iﬀ ∀
∞
n µ([r n] ∧b) ≥
3
4
µ([r n]).
Proof:
Since G is an ultraﬁlter it is enough to show that b ∈ G implies
∀
∞
n µ([r n] ∧b) ≥
3
4
µ([r n]).
Let B
+
be the nonzero elements of B. To prove this it suﬃces to show:
Claim: For any b ∈ B
+
and for every d ≤ b in B
+
there exists a tree T ⊆ 2
<ω
with [T] of positive measure, [T] ≤ d, and
µ([s] ∩ b) ≥
3
4
µ([s])
for all but ﬁnitely many s ∈ T.
Proof:
Descriptive Set Theory and Forcing 68
Without loss we may assume that d is a closed set and let T
d
be a tree
such that d = [T
d
]. Let t
0
∈ T
d
be such that
µ([t
0
] ∩ [T
d
]) ≥
9
10
µ([t
0
]).
Deﬁne a subtree T ⊆ T
d
by r ∈ T iﬀ r ⊆ t
0
or t
0
⊆ r and
∀t (t
0
⊆ t ⊆ r implies µ([t] ∩ b) ≥
3
4
µ([t]) ).
So we only need to see that [T] has positive measure. So suppose for contra
diction that µ([T]) = 0. Then for some suﬃciently large N
µ(
_
s∈T∩2
N
[s]) ≤
1
10
µ([t
0
]).
For every s ∈ T
d
∩2
N
with t
0
⊆ s, if s / ∈ T then there exists t with t
0
⊆ t ⊆ s
and µ([t] ∩ b) <
3
4
µ([t]). Let Σ be a maximal antichain of t like this. But
note that
[t
0
] ∩ [T
d
] ⊆
_
s∈2
N
∩T
[s] ∪
_
t∈Σ
([t] ∩ b).
By choice of Σ
µ(
_
s∈Σ
[s] ∩ b) ≤
3
4
µ([t
0
])
and by choice of N
µ(
_
s∈2
N
∩T
[s]) ≤
1
10
µ([t
0
])
which contradicts the choice of t
0
:
µ([t
0
] ∩ [T
d
]) ≤ (
1
10
+
3
4
)µ([t
0
]) =
6
17
20
µ([t
0
]) <
9
10
µ([t
0
]).
This proves the claim and the lemma.
In eﬀect, what we have done in Lemma 15.4 is reprove the Lebesgue
density theorem, see Oxtoby [90].
6
Trust me on this, I have been teaching a lot of Math 99 “College Fractions”.
Descriptive Set Theory and Forcing 69
So now suppose that the order of X in V is ≤ α. We show that it is ≤ α+1
in V [G]. Let
◦
U
be any name for a Borel subset of X in the extension. Then
we know that x ∈ U
G
iﬀ [[ ˇ x ∈
◦
U
[] ∈ G. By Lemma 15.3 we know that for
any s ∈ 2
<ω
the set
B
s
= ¦x ∈ X : µ([s] ∩ [[ ˇ x ∈
◦
U
[]) ≥
3
4
µ([s])¦
is a Borel subset of X in the ground model and hence is Π
0
α
¯
(X). By Lemma
15.4 we have that for any x ∈ X
x ∈ U iﬀ ∀
∞
n x ∈ B
rn
and so U is Σ
0
α+1
¯
(X) in V[G].
This concludes the proof of Theorem 15.2.
Note that this result does allow us to get sets of order λ for any countable
limit ordinal λ by taking a clopen separated union of a sequence of sets whose
order goes up λ.
Also a Luzin set X from the ground model has order 3 in the random
real extension. Since (ord(X) = 3)
V
we know that (3 ≤ ord(X) ≤ 4)
V [G]
.
To see that (ord(X) ≤ 3)
V [G]
suppose that B ⊆ X is Borel in V [G]. The
above proof shows that there exists Borel sets B
n
each coded in V (but the
sequence may not be in V ) such that
x ∈ B iﬀ ∀
∞
n x ∈ B
n
.
For each B
n
there exists an open set U
n
⊆ X such that B
n
∆U
n
is countable.
If we let
C =
_
n∈ω
m>n
U
m
then C is Σ
0
3
¯
(X) and B∆C is countable. Since subtracting and adding a
countable set from a Σ
0
3
¯
(X) is still Σ
0
3
¯
(X) we have that B is Σ
0
3
¯
(X) and so
the order of X is ≤ 3 in V [G].
Theorem 15.5 Suppose V models CH and G is measure algebra on 2
κ

generic over V for some κ ≥ ω
2
. Then in V [G] for every X ⊆ 2
ω
of
cardinality ω
2
there exists Y ∈ [X]
ω
2
with ord(Y ) = 2.
Descriptive Set Theory and Forcing 70
Proof:
Using the same argument as in the proof of Theorem 14.11 we can get a
Sierpi´ nski set S ⊆ 2
ω
of cardinality ω
2
and a term τ for any element of 2
ω
such that Y = ¦τ
r
: r ∈ S¦ is a set of distinct elements of X. This Sierpi´ nski
set has two additional properties: every element of it is random over the
ground model and it meets every set of positive measure, i.e. it is a super
Sierpi´ nski set.
We will show that the order of Y is 2.
Lemma 15.6 Let T ⊆ B be any subset of a measure algebra B closed under
ﬁnite conjunctions. Then Π
0
2
¯
(T) = Σ
0
2
¯
(T), i.e. T has order ≤ 2.
Proof:
Let µ be the measure on B.
(1) For any b ∈ Π
0
1
¯
(T) and real > 0 there exists a ∈ T with b ≤ a and
µ(a −b) < .
pf:
7
b =
n∈ω
a
n
. Let a =
n<N
a
n
for some suﬃciently large N.
(2) For any b ∈ Σ
0
2
¯
(T) and real > 0 there exists a ∈ Σ
0
1
¯
(T) with b ≤ a and
µ(a −b) < .
pf: b =
n<ω
b
n
. Applying (1) we get a
n
∈ T with b
n
≤ a
n
and
µ(a
n
−b
n
) <
2
n
.
Then let a =
n∈ω
a
n
.
Now suppose b ∈ Σ
0
2
¯
(T). Then by applying (2) there exists a
n
∈ Σ
0
1
¯
(T)
with b ≤ a
n
and µ(a
n
−b) < 1/n. Consequently, if a =
n∈ω
a
n
, then b ≤ a
and µ(a −b) = 0 and so a = b.
Let
T = ¦[[ τ ∈ C [] : C ⊆ 2
ω
clopen ¦
where boolean values are in the measure algebra B on 2
ω
. Let F be the
complete subalgebra of F which is generated by T.
Since the order of T is 2, by Lemma 14.12 we have that for any Borel
set B ⊆ Y there exists a Σ
0
2
¯
(Y ) set F such that y ∈ B iﬀ y ∈ F for all but
7
Pronounced ‘puﬀ’.
Descriptive Set Theory and Forcing 71
countably many y ∈ Y . Thus we see that the order of Y is ≤ 3. To get it
down to 2 we use the following lemma. If B = (F ¸ F
0
) ∪ F
1
where F
0
and
F
1
are countable and F is Σ
0
2
¯
, then by the lemma F
0
would be Π
0
2
¯
and thus
B would be Σ
0
2
¯
.
Lemma 15.7 Every countable subset of Y is Π
0
2
¯
(Y ).
Proof:
It suﬃces to show that every countable subset of Y can be covered by a
countable Π
0
2
¯
(Y ) since one can always subtract a countable set from a Π
0
2
¯
(Y )
and remain Π
0
2
¯
(Y ).
For any s ∈ 2
<ω
deﬁne
b
s
= [[ s ⊆ τ [].
Working in the ground model let B
s
be a Borel set with [B
s
]
B
= b
s
. Since
the Sierpi´ nski set consists only of reals random over the ground model we
know that for every r ∈ S
r ∈ B
s
iﬀ s ⊆ τ
r
.
Also since the Sierpi´ nski set meets every Borel set of positive measure we
know that for any z ∈ Y the set
n<ω
B
zn
has measure zero. Now let Z =
¦z
n
: n < ω¦ ⊆ Y be arbitrary but listed with inﬁnitely many repetitions.
For each n choose m so that if s
n
= z
n
m, then µ(B
s
n
) < 1/2
n
. Now for
every r ∈ S we have that
r ∈
n<ω
_
m>n
B
s
m
iﬀ τ
r
∈
n<ω
_
m>n
[s
m
].
The set H =
n<ω
m>n
[s
m
] covers Z and is Π
0
2
¯
. It has countable intersec
tion with Y because the set
n<ω
m>n
B
s
m
has measure zero.
This proves the Lemma and Theorem 15.5.
Perfect Set Forcing
Descriptive Set Theory and Forcing 72
In the iterated Sack’s real model the continuum is ω
2
and every set
X ⊆ 2
ω
of cardinality ω
2
can be mapped continuously onto 2
ω
(Miller [81]).
It follows from Reclaw’s Theorem 3.5 that in this model every separable
metric space of cardinality ω
2
has order ω
1
. On the other hand this forcing
(and any other with the Sack’s property) has the property that every meager
set in the extension is covered by a meager set in the ground model and every
measure set in the extension is covered by a measure zero set in the ground
model (see Miller [78]). Consequently, in this model there are Sierpi´ nski sets
and Luzin sets of cardinality ω
1
. Therefore in the iterated Sacks real model
there are separable metric spaces of cardinality ω
1
of every order α with
2 ≤ α < ω
1
. I do not know if there is an uncountable separable metric space
which is hereditarily of order ω
1
in this model.
Another way to obtain the same orders is to use the construction of
Theorem 22 of Miller [75]. What was done there implies the following:
For any model V there exists a ccc extension V [G] in which every
uncountable separable metric space has order ω
1
.
If we apply this result ω
1
times with a ﬁnite support extension,
we get a model, V [G
α
: α < ω
1
], where there are separable metric spaces
of all orders of cardinality ω
1
, but every separable metric space of cardinality
ω
2
has order ω
1
.
To see the ﬁrst fact note that ω
1
length ﬁnite support iteration always
adds a Luzin set. Consequently, by Theorem 14.7, for each α with 2 < α < ω
1
there exists a separable metric space of cardinality ω
1
which is hereditarily of
order α. Also there is such an X of order 2 by the argument used in Theorem
14.1.
On the other hand if X has cardinality ω
2
in V [G
α
: α < ω
1
], then for
some β < ω
1
there exists and uncountable Y ∈ V [G
α
: α < β] with Y ⊆ X.
Hence Y will have order ω
1
in V [G
α
: α < β +1] and by examining the proof
it is easily seen that it remains of order ω
1
in V [G
α
: α < ω
1
].
16 COVERING NUMBER OF AN IDEAL 73
16 Covering number of an ideal
This section is a small diversion.
8
It is motivated by Theorem 11.1 of Martin
and Solovay.
Deﬁne for any ideal I in Borel(2
ω
)
cov(I) = min¦[1[ : 1 ⊆ I,
_
1 = 2
ω
¦.
The following theorem is wellknown.
Theorem 16.1 For any cardinal κ the following are equivalent:
1. MA
κ
(ctbl), i.e. for any countable poset, P, and family T of dense
subsets of P with [T[ < κ there exists a Pﬁlter G with G ∩ D ,= ∅ for
every D ∈ T, and
2. cov(meager(2
ω
)) ≥ κ.
Proof:
MA
κ
(ctbl) implies cov(meager(2
ω
)) ≥ κ, is easy because if U ⊆ 2
ω
is a
dense open set, then
D = ¦s ∈ 2
<ω
: [s] ⊆ U¦
is dense in 2
<ω
.
cov(meager(2
ω
)) ≥ κ implies MA
κ
(ctbl) follows from the fact that any
countable poset, P, either contains a dense copy of 2
<ω
or contains a p such
that every two extensions of p are compatible.
Theorem 16.2 (Miller [79]) cof(cov(meager(2
ω
))) > ω, e.g., it is impossi
ble to have cov(meager(2
ω
)) = ℵ
ω
.
Proof:
Suppose for contradiction that κ = cov(meager(2
ω
)) has countable coﬁ
nality and let κ
n
for n ∈ ω be a coﬁnal sequence in κ. Let ¸C
α
: α < κ) be
a family of closed nowhere dense sets which cover 2
ω
. We will construct a
sequence P
n
⊆ 2
ω
of perfect sets with the properties that
1. P
n+1
⊆ P
n
,
8
All men’s gains are the fruit of venturing. Herodotus BC 484425.
Descriptive Set Theory and Forcing 74
2. P
n
∩
¦C
α
: α < κ
n
¦ = ∅, and
3. ∀α < κ C
α
∩ P
n
is nowhere dense in P
n
.
This easily gives a contradiction, since
n<ω
P
n
is nonempty and disjoint
from all C
α
, contradicting the fact that the C
α
’s cover 2
ω
.
We show how to obtain P
0
, since the argument easily relativizes to show
how to obtain P
n+1
given P
n
. Since cov(meager(2
ω
)) > κ
n
there exists a
countable sequence
D = ¦x
n
: n ∈ ω¦ ⊆ 2
ω
such that D is dense and for every n
x
n
/ ∈
_
α<κ
n
C
α
.
Consider the following forcing notion P.
P = ¦(H, n) : n ∈ ω and H ∈ [D]
<ω
¦
This is ordered by (H, n) ≤ (K, m) iﬀ
1. H ⊇ K,
2. n ≥ m, and
3. for every x ∈ H there exists y ∈ K with x m = y m.
Note that P is countable.
For each n ∈ ω deﬁne E
n
⊆ P by (H, m) ∈ E
n
iﬀ
1. m > n and
2. ∀x ∈ H∃y ∈ H x n = y n but x m ,= y m.
and for each α < κ
0
let
F
α
= ¦(H, m) ∈ P : ∀x ∈ H [x m] ∩ C
α
= ∅¦.
For G a Pﬁlter, deﬁne X ⊆ D by
X =
_
¦H : ∃n (H, n) ∈ G¦
Descriptive Set Theory and Forcing 75
and let P = cl(X). It easy to check that the E
n
’s are dense and if G meets
each one of them, then P is perfect (i.e. has no isolated points). The F
α
for
α < κ
0
are dense in P. This is because D∩C
α
= ∅ so given (H, n) ∈ P there
exists m ≥ n such that for every x ∈ H we have [x m] ∩ C
α
= ∅ and thus
(H, m) ∈ F
α
. Note that if G ∩ F
α
,= ∅, then P ∩ C
α
= ∅. Consequently, by
Theorem 16.1, there exists a Pﬁlter G such that G meets each E
n
and all
F
α
for α < κ
0
. Hence P = cl(X) is a perfect set which is disjoint from each
C
α
for α < κ
0
. Note also that for every α < κ we have that C
α
∩ D is ﬁnite
and hence C
α
∩X is ﬁnite and therefore C
α
∩P is nowhere dense in P. This
ends the construction of P = P
0
and since the P
n
can be obtained with a
similar argument, this proves the Theorem.
Question 16.3 (Fremlin) Is the same true for the measure zero ideal in
place of the ideal of meager sets?
Some partial results are known (see Bartoszynski, Judah, Shelah [7][8][9]).
Theorem 16.4 (Miller [79]) It is consistent that cov(meager(2
ω
1
)) = ℵ
ω
.
Proof:
In fact, this holds in the model obtained by forcing with FIN(ℵ
ω
, 2) over
a model of GCH.
cov(meager(2
ω
1
)) ≥ ℵ
ω
: Suppose for contradiction that
¦C
α
: α < ω
n
¦ ∈ V [G]
is a family of closed nowhere dense sets covering 2
ω
1
. Deﬁne
E
α
= ¦s ∈ FIN(ω
1
, 2) : [s] ∩ C
α
= ∅¦.
Using ccc, there exists Σ ∈ [ℵ
ω
]
ω
n
in V with
¦E
α
: α < ω
n
¦ ∈ V [G Σ].
Let X ⊆ ℵ
ω
be a set in V of cardinality ω
1
which is disjoint from Σ. By the
product lemma G X is FIN(X, 2)generic over V [G Σ]. Consequently, if
H : ω
1
→ 2 corresponds to G via an isomorphism of X and ω
1
, then H / ∈ C
α
for every α < ω
n
.
Descriptive Set Theory and Forcing 76
cov(meager(2
ω
1
)) ≤ ℵ
ω
: Note that for every uncountable X ⊆ ω
1
such
that X ∈ V [G] there exists
n ∈ ω and Z ∈ [ω
1
]
ω
1
∩ V [G ω
n
]
with Z ⊆ X. To see this, note that for every α ∈ X there exists p ∈ G such
that p [¬ α ∈ X and p ∈ FIN(ω
n
, 2) for some n ∈ ω. Consequently, by ccc,
some n works for uncountably many α.
Consider the family of all closed nowhere dense sets C ⊆ 2
ω
1
which are
coded in some V [G ω
n
] for some n. We claim that these cover 2
ω
1
. This
follows from above, because for any Z ⊆ ω
1
which is inﬁnite the set
C = ¦x ∈ 2
ω
1
: ∀α ∈ Z x(α) = 1¦
is nowhere dense.
Theorem 16.5 (Miller [79]) It is consistent that there exists a ccc σideal
I in Borel(2
ω
) such that cov(I) = ℵ
ω
.
Proof:
Let P = FIN(ω
1
, 2)∗
◦
Q where
◦
Q is a name for the Silver forcing which
codes up generic ﬁlter for FIN(ω
1
, 2) just like in the proof of Theorem 11.1.
Let
α<ℵ
ω
P be the direct sum (i.e. ﬁnite support product) of ℵ
ω
copies of
P. Forcing with the direct sum adds a ﬁlter G = ¸G
α
: α < ℵ
ω
) where each
G
α
is Pgeneric. In general, a direct sum is ccc iﬀ every ﬁnite subproduct is
ccc. This follows by a deltasystem argument. Every ﬁnite product of P has
ccc, because P is σcentered, i.e., it is the countable union of centered sets.
Let V be a model of GCH and G = ¸G
α
: α < ℵ
ω
) be
α<ℵ
ω
P generic over
V . We claim that in V [G] if I is the σideal given by Sikorski’s Theorem 9.1
such that
α<ℵ
ω
P is densely embedded into Borel(2
ω
)/I then cov(I) = ℵ
ω
.
First deﬁne, m
P
, to be the cardinality of the minimal failure of MA for
P, i.e., the least κ such that there exists a family [T[ = κ of dense subsets of
P such that there is no Pﬁlter meeting all the D ∈ T.
Lemma 16.6 In V [¸G
α
: α < ℵ
ω
)] we have that m
P
= ℵ
ω
.
Proof:
Note that for any set D ⊂ P there exists a set Σ ∈ [ℵ
ω
]
ω
1
in V with
D ∈ V [¸G
α
: α ∈ Σ)]. So if [T[ = ω
n
then there exists Σ ∈ [ℵ
ω
]
ω
n
in V with
Descriptive Set Theory and Forcing 77
T ∈ V [¸G
α
: α ∈ Σ)]. Letting α ∈ ℵ
ω
¸ Σ we get G
α
a Pﬁlter meeting every
D ∈ T. Hence m
P
≥ ℵ
ω
.
On the other hand:
Claim: For every X ∈ [ω
1
]
ω
1
∩ V [¸G
α
: α < ℵ
ω
)] there exists n ∈ ω and
Y ∈ [ω
1
]
ω
1
∩ V [¸G
α
: α < ℵ
n
)] with Y ⊆ X.
Proof:
For every α ∈ X there exist p ∈ G and n < ω such that p [¬ ˇ α ∈
◦
X
and domain(p) ⊆ ℵ
n
. Since X is uncountable there is one n which works for
uncountably many α ∈ X.
It follows from the Claim that there is no H which is FIN(ω
1
, 2) generic
over all the models V [¸G
α
: α < ℵ
n
)], but forcing with P would add such an
H and so m
P
≤ ℵ
ω
and the Lemma is proved.
Lemma 16.7 If P is ccc and dense in the cBa Borel(2
ω
)/I, then m
P
=
cov(I).
Proof:
This is the same as Lemma 11.2 equivalence of (1) and (3), except you
have to check that m is the same for both P and Borel(2
ω
)/I.
Kunen [58] showed that least cardinal for which MA fails can be a singular
cardinal of coﬁnality ω
1
, although it is impossible for it to have coﬁnality ω
(see Fremlin [27]). It is still open whether it can be a singular cardinal
of coﬁnality greater than ω
1
(see Landver [61]). Landver [62] generalizes
Theorem 16.2 to the space 2
κ
with basic clopen sets of the form [s] for
s ∈ 2
<κ
. He uses a generalization of a characterization of cov(meager(2
ω
))
due to Bartoszynski [6] and Miller [80].
78
Part II
Analytic sets
17 Analytic sets
Analytic sets were discovered by Souslin when he encountered a mistake of
Lebesgue. Lebesgue had erroneously proved that the Borel sets were closed
under projection. I think the mistake he made was to think that the count
able intersection commuted with projection. A good reference is the volume
devoted to analytic sets edited by Rogers [93]. For the more classical view
point of operationA, see Kuratowski [59]. For the whole area of descriptive
set theory and its history, see Moschovakis [89].
Deﬁnition. A set A ⊆ ω
ω
is Σ
1
1
iﬀ there exists a recursive
R ⊆
_
n∈ω
(ω
n
ω
n
)
such that for all x ∈ ω
ω
x ∈ A iﬀ ∃y ∈ ω
ω
∀n ∈ ω R(x n, y n).
A similar deﬁnition applies for A ⊆ ω and also for A ⊆ ω ω
ω
and so forth.
For example, A ⊆ ω is Σ
1
1
iﬀ there exists a recursive R ⊆ ω ω
<ω
such that
for all m ∈ ω
m ∈ A iﬀ ∃y ∈ ω
ω
∀n ∈ ω R(m, y n).
A set C ⊆ ω
ω
ω
ω
is Π
0
1
iﬀ there exists a recursive predicate
R ⊆
_
n∈ω
(ω
n
ω
n
)
such that
C = ¦(x, y) : ∀n R(x n, y n)¦.
That means basically that C is a recursive closed set.
The Π classes are the complements of the Σ’s and ∆ is the class of sets
which are both Π and Σ. The relativized classes, e.g. Σ
1
1
(x) are obtained by
allowing R to be recursive in x, i.e., R ≤
T
x. The boldface classes, e.g., Σ
1
1
¯
,
Π
1
1
¯
, are obtained by taking arbitrary R’s.
17 ANALYTIC SETS 79
Lemma 17.1 A ⊆ ω
ω
is Σ
1
1
iﬀ there exists set C ⊆ ω
ω
ω
ω
which is Π
0
1
and
A = ¦x ∈ ω
ω
: ∃y ∈ ω
ω
(x, y) ∈ C¦.
Lemma 17.2 The following are all true:
1. For every s ∈ ω
<ω
the basic clopen set [s] = ¦x ∈ ω
ω
: s ⊆ x¦ is Σ
1
1
,
2. if A ⊆ ω
ω
ω
ω
is Σ
1
1
, then so is
B = ¦x ∈ ω
ω
: ∃y ∈ ω (x, y) ∈ A¦,
3. if A ⊆ ω ω
ω
is Σ
1
1
, then so is
B = ¦x ∈ ω
ω
: ∃n ∈ ω (n, x) ∈ A¦,
4. if A ⊆ ω ω
ω
is Σ
1
1
, then so is
B = ¦x ∈ ω
ω
: ∀n ∈ ω (n, x) ∈ A¦,
5. if ¸A
n
: n ∈ ω) is sequence of Σ
1
1
sets given by the recursive predicates
R
n
and ¸R
n
: n ∈ ω) is (uniformly) recursive, then both
_
n∈ω
A
n
and
n∈ω
A
n
are Σ
1
1
.
6. if the graph of f : ω
ω
→ ω
ω
is Σ
1
1
and A ⊆ ω
ω
is Σ
1
1
, then f
−1
(A) is
Σ
1
1
.
Of course, the above lemma is true with ω or ω ω
ω
, etc., in place of ω
ω
.
It also relativizes to any class Σ
1
1
(x). It follows from the Lemma that every
Borel subset of ω
ω
is Σ
1
1
¯
and that the continuous preimage of Σ
1
1
¯
set is Σ
1
1
¯
.
Theorem 17.3 There exists a Σ
1
1
set U ⊆ ω
ω
ω
ω
which is universal for
all Σ
1
1
¯
sets, i.e., for every Σ
1
1
¯
set A ⊆ ω
ω
there exists x ∈ ω
ω
with
A = ¦y : (x, y) ∈ U¦.
17 ANALYTIC SETS 80
Proof:
There exists C ⊆ ω
ω
ω
ω
ω
ω
a Π
0
1
set which is universal for Π
0
1
¯
subsets
of ω
ω
ω
ω
. Let U be the projection of C on its second coordinate.
Similarly we can get Σ
1
1
sets contained in ω ω (or ω ω
ω
) which are
universal for Σ
1
1
subsets of ω (or ω
ω
).
The usual diagonal argument shows that there are Σ
1
1
subsets of ω
ω
which
are not Π
1
1
¯
and Σ
1
1
subsets of ω which are not Π
1
1
.
Theorem 17.4 (Normal form) A set A ⊆ ω
ω
is Σ
1
1
iﬀ there exists a recursive
map
ω
ω
→ 2
ω
<ω
x → T
x
such that T
x
⊆ ω
<ω
is a tree for every x ∈ ω
ω
, and x ∈ A iﬀ T
x
is illfounded.
By recursive map we mean that there is a Turing machine ¦e¦ such that for
x ∈ ω
ω
the machine e computing with an oracle for x, ¦e¦
x
computes the
characteristic function of T
x
.
Proof:
Suppose
x ∈ A iﬀ ∃y ∈ ω
ω
∀n ∈ ω R(x n, y n).
Deﬁne
T
x
= ¦s ∈ ω
<ω
: ∀i ≤ [s[ R(x i, s i)¦.
A similar thing is true for A ⊆ ω, i.e., A is Σ
1
1
iﬀ there is a uniformly
recursive list of recursive trees ¸T
n
: n < ω) such that n ∈ A iﬀ T
n
is ill
founded.
The connection between Σ
1
1
and wellfounded trees, gives us the following:
Theorem 17.5 (Mostowski’s Absoluteness) Suppose M ⊆ N are two tran
sitive models of ZFC
∗
and θ is Σ
1
1
¯
sentence with parameters in M. Then
M [= θ iﬀ N [= θ.
Proof:
17 ANALYTIC SETS 81
ZFC
∗
is a nice enough ﬁnite fragment of ZFC to know that trees are well
founded iﬀ they have rank functions (Theorem 7.1). θ is Σ
1
1
¯
sentence with
parameters in M means there exists R in M such that
θ = ∃x ∈ ω
ω
∀n R(x n).
This means that for some tree T ⊆ ω
<ω
in M θ is equivalent to “T has an
inﬁnite branch”. So if M [= θ then N [= θ since a branch T exists in M. On
the other hand if M [= θ, then
M [= ∃r : T → OR a rank function”
and then for this same r ∈ M
N [= r : T → OR is a rank function”
and so N [= θ.
18 CONSTRUCTIBLE WELLORDERINGS 82
18 Constructible wellorderings
G¨odel proved the axiom of choice relatively consistent with ZF by producing a
deﬁnable wellorder of the constructible universe. He announced in G¨odel [32]
that if V=L, then there exists an uncountable Π
1
1
set without perfect subsets.
Kuratowski wrote down a proof of the theorem below but the manuscript was
lost during World War II (see Addison [2]).
A set is Σ
1
2
iﬀ it is the projection of a Π
1
1
set.
Theorem 18.1 [V=L] There exists a ∆
1
2
wellordering of ω
ω
.
Proof:
Recall the deﬁnition of G¨odel’s Constructible sets L. L
0
= ∅, L
λ
=
α<λ
L
α
for λ a limit ordinal, and L
α+1
is the deﬁnable subsets of L
α
. De
ﬁnable means with parameters from L
α
. L =
α∈OR
L
α
.
The set x is constructed before y, (x <
c
y) iﬀ the least α such that x ∈ L
α
is less than the least β such that y ∈ L
β
, or α = β and the “least” deﬁning
formula for x is less than the one for y. Here “least” basically boils down
to lexicographical order. Whatever the exact formulation of x <
c
y is it
satisﬁes:
x <
c
y iﬀ L
α
[= x <
c
y
where x, y ∈ L
α
and L
α
[=ZFC
∗
where ZFC
∗
is a suﬃciently large ﬁnite
fragment of ZFC. (Actually, it is probably enough for α to be a limit ordinal.)
Assuming V = L, for x, y ∈ ω
ω
we have that x <
c
y iﬀ there exists E ⊆ ωω
and
◦
x,
◦
y
∈ ω such that letting M = (ω, E) then
1. E is extensional and wellfounded,
2. M [=ZFC
∗
+ V=L
3. M [=
◦
x<
c
◦
y
,
4. for all n, m ∈ ω (x(n) = m iﬀ M [=
◦
x (
◦
n) =
◦
m), and
5. for all n, m ∈ ω (y(n) = m iﬀ M [=
◦
y
(
◦
n) =
◦
m).
The ﬁrst clause guarantees (by the Mostowski collapsing lemma) that M is
isomorphic to a transitive set. The second, that this transitive set will be
Descriptive Set Theory and Forcing 83
of the form L
α
. The last two clauses guarantee that the image under the
collapse of
◦
x is x and
◦
y
is y.
Wellfoundedness of E is Π
1
1
. The remaining clauses are all Π
0
n
for some
n ∈ ω. Hence, we have given a Σ
1
2
deﬁnition of <
c
. But a total ordering <
which is Σ
1
n
is ∆
1
n
, since x ,< y iﬀ y = x or y < x. It follows that <
c
is also
Π
1
2
and hence ∆
1
2
.
19 HEREDITARILY COUNTABLE SETS 84
19 Hereditarily countable sets
HC is the set consisting of all hereditarily countable sets. There is a
close connection between the projective hierarchy above level 2 and a natural
hierarchy on the subsets of HC. A formula of set theory is ∆
0
iﬀ it is in
the smallest family of formulas containing the atomic formulas of the form
“x ∈ y” or “x = y”, and closed under conjunction, θ ∧φ, negation, θ, and
bounded quantiﬁcation, ∀x ∈ y or ∃x ∈ y. A formula θ of set theory is Σ
1
iﬀ it of the form ∃u
1
, . . . , u
n
ψ where ψ is ∆
0
.
Theorem 19.1 A set A ⊆ ω
ω
is Σ
1
2
iﬀ there exists a Σ
1
formula θ() of set
theory such that
A = ¦x ∈ ω
ω
: HC [= θ(x)¦.
Proof:
We note that ∆
0
formulas are absolute between transitive sets, i.e., if
ψ( ) is ∆
0
formula, M a transitive set and y a ﬁnite sequence of elements
of M, then M [= ψ(y) iﬀ V [= ψ(y). Suppose that θ() is a Σ
1
formula of
set theory. Then for every x ∈ ω
ω
we have that HC [= θ(x) iﬀ there exists a
countable transitive set M ∈ HC with x ∈ M such that M [= θ(x). Hence,
HC [= θ(x) iﬀ there exists E ⊆ ωω and
◦
x∈ ω such that letting M = (ω, E)
then
1. E is extensional and wellfounded,
2. M [=ZFC
∗
, (or just that ω exists)
3. M [= θ(
◦
x),
4. for all n, m ∈ ω x(n) = m iﬀ M [=
◦
x (
◦
n) =
◦
m.
Therefore, ¦x ∈ ω
ω
: HC [= θ(x)¦ is a Σ
1
2
set. On the other hand given a Σ
1
2
set A there exists a Π
1
1
formula θ(x, y) such that A = ¦x : ∃yθ(x, y)¦. But
then by Mostowski absoluteness (Theorem 17.5) we have that x ∈ A iﬀ there
exists a countable transitive set M with x ∈ M and there exists y ∈ M such
that M [=ZFC
∗
and M [= θ(x, y). But this is a Σ
1
formula for HC.
The theorem says that Σ
1
2
= Σ
HC
1
. Similarly, Σ
1
n+1
= Σ
HC
n
. Let us
illustrate this with an example construction.
Descriptive Set Theory and Forcing 85
Theorem 19.2 If V=L, then there exists a ∆
1
2
Luzin set X ⊆ ω
ω
.
Proof:
Let ¦T
α
: α < ω
1
¦ (ordered by <
c
) be all subtrees T of ω
<ω
whose
branches [T] are a closed nowhere dense subset of ω
ω
. Deﬁne x
α
to be the
least constructed (<
c
) element of ω
ω
which is not in
_
β<α
[T
β
] ∪ ¦x
β
: β < α¦.
Deﬁne X = ¦x
α
: α < ω
1
¦. So X is a Luzin set.
To see that X is Σ
HC
1
note that x ∈ X iﬀ there exists a transitive count
able M which models ZFC
∗
+V=L such that M [=“x ∈ X”(i.e. M models
the ﬁrst paragraph of this proof).
To see that X is Π
HC
1
note that x ∈ X iﬀ for all M if M is a transitive
countable model of ZFC
∗
+V=L with x ∈ M and M [=“∃y ∈ X x <
c
y”,
then M [=“x ∈ X”. This is true because the nature of the construction is
such that if you put a real into X which is constructed after x, then x will
never get put into X after this. So x will be in X iﬀ it is already in X.
20 SHOENFIELD ABSOLUTENESS 86
20 Shoenﬁeld Absoluteness
For a tree T ⊆
n<ω
κ
n
ω
n
deﬁne
p[T] = ¦y ∈ ω
ω
: ∃x ∈ κ
ω
∀n(x n, y n) ∈ T¦.
A set deﬁned this way is called κSouslin. Thus Σ
1
1
¯
sets are precisely the
ωSouslin sets. Note that if A ⊆ ω
ω
ω
ω
and A = p[T] then the projection
of A, ¦y : ∃x ∈ ω
ω
(x, y) ∈ A¦ is κSouslin. To see this let <, >: κ ω → κ
be a pairing function. For s ∈ κ
n
let s
0
∈ κ
n
and s
1
∈ ω
n
be deﬁned by
s(i) =< s
0
(i), s
1
(i) >. Let T
∗
be the tree deﬁned by
T
∗
=
_
n∈ω
¦(s, t) ∈ κ
n
ω
n
: (s
0
, s
1
, t) ∈ T¦.
Then p[T
∗
] = ¦y : ∃x ∈ ω
ω
(x, y) ∈ A¦.
Theorem 20.1 (Shoenﬁeld [98]) If A is a Σ
1
2
set, then A is ω
1
Souslin set
coded in L, i.e. A = p[T] where T ∈ L.
Proof:
From the construction of T
∗
it is clear that is enough to see this for A
which is Π
1
1
.
We know that a countable tree is wellfounded iﬀ there exists a rank
function r : T → ω
1
. Suppose
x ∈ A iﬀ ∀y∃n (x n, y n) / ∈ T
where T is a recursive tree. So deﬁning T
x
= ¦t : (x [t[, t) ∈ T¦ we have
that x ∈ A iﬀ T
x
is wellfounded (Theorem 17.4).
The ω
1
tree
ˆ
T is just the tree of partial rank functions. Let ¦s
n
: n ∈ ω¦
be a recursive listing of ω
<ω
with [s
n
[ ≤ n. Then for every N < ω, and
(r, s) ∈ ω
N
1
ω
N
we have (r, t) ∈
ˆ
T iﬀ
∀n, m < N [(t, s
n
), (t, s
m
) ∈ T and s
n
⊂ s
m
] implies r(n) > r(m).
Then A = p[
ˆ
T]. To see this, note that if x ∈ A, then T
x
is wellfounded and
so it has a rank function and therefore there exists r with (x, r) ∈ [
ˆ
T] and
so x ∈ p[
ˆ
T]. On the other hand if (x, r) ∈ [
ˆ
T], then r determines a rank
function on T
x
and so T
x
is wellfounded and hence x ∈ A.
Descriptive Set Theory and Forcing 87
Theorem 20.2 (Shoenﬁeld Absoluteness [98]) If M ⊆ N are transitive mod
els of ZFC
∗
and ω
1
N
⊆ M, then for any Σ
1
2
(x) sentence θ with parameter
x ∈ M
M [= θ iﬀ N [= θ.
Proof:
If M [= θ, then N [= θ, because Π
1
1
¯
sentences are absolute. On the other
hand suppose N [= θ. Working in N using the proof of Theorem 20.1 we get
a tree T ⊆ ω
<ω
1
with T ∈ L[x] such that T is illfounded, i.e., there exists
r ∈ [T]. Note that r codes a witness to a Π
1
1
(x) predicate and a rank function
showing the tree corresponding to this predicate is wellfounded. Since for
some α < ω
1
, r ∈ α
ω
we see that
T
α
= T ∩ α
<ω
is illfounded. But T
α
∈ M (since by assumption (ω
1
)
N
⊆ M) and so by the
absoluteness of wellfounded trees, M thinks that T
α
is illfounded. But a
branch thru [T] gives a witness and a rank function showing that θ is true,
and consequently, M [= θ.
21 MANSFIELDSOLOVAY THEOREM 88
21 MansﬁeldSolovay Theorem
Theorem 21.1 (Mansﬁeld [72], Solovay [103]) If A ⊆ ω
ω
is a Σ
1
2
¯
set with
constructible parameter which contains a nonconstructible element of ω
ω
,
then A contains a perfect set which is coded in L.
Proof:
By Shoenﬁeld’s Theorem 20.1, we may assume A = p[T] where T ∈ L and
T ⊆
n<ω
ω
n
1
ω
n
. Working in L deﬁne the following decreasing sequence
of subtrees as follows.
T
0
= T,
T
λ
=
β<λ
T
β
, if λ a limit ordinal, and
T
α+1
= ¦(r, s) ∈ T
α
: ∃(r
0
, s
0
), (r
1
, s
1
) ∈ T
α
such that (r
0
, s
0
), (r
1
, s
1
)
extend (r, s), and s
0
and s
1
are incompatible¦.
Each T
α
is tree, and for α < β we have T
β
⊆ T
α
. Thus there exists some
α
0
such that T
α
0
+1
= T
α
0
.
Claim: [T
α
0
] is nonempty.
Proof:
Let (x, y) ∈ [T] be any pair with y not constructible. Since A = p[T] and
A is not a subset of L, such a pair must exist. Prove by induction on α that
(x, y) ∈ [T
α
]. This is easy for α a limit ordinal. So suppose (x, y) ∈ [T
α
] but
(x, y) / ∈ [T
α+1
]. By the deﬁnition it must be that there exists n < ω such
that (x n, y n) = (r, s) / ∈ T
α+1
. But in L we can deﬁne the tree:
T
(r,s)
α
= ¦(ˆ r, ˆ s) ∈ T
α
: (ˆ r, ˆ s) ⊆ (r, s) or (r, s) ⊆ (ˆ r, ˆ s)¦
which has the property that p[T
(r,s)
α
] = ¦y¦. But by absoluteness of well
founded trees, it must be that there exists (u, y
0
) ∈ [T
(r,s)
α
] with (u, y
0
) ∈ L.
But then y
0
= y ∈ L which is a contradiction. This proves the claim.
Since T
α
0
+1
= T
α
0
, it follows that for every (r, s) ∈ T
α
0
there exist
(r
0
, s
0
), (r
1
, s
1
) ∈ T
α
0
such that (r
0
, s
0
), (r
1
, s
1
) extend (r, s) and s
0
and s
1
are incompatible. This
allows us to build by induction (working in L):
¸(r
σ
, s
σ
) : σ ∈ 2
<ω
)
Descriptive Set Theory and Forcing 89
with (r
σ
, s
σ
) ∈ T
α
0
and for each σ ∈ 2
<ω
(r
σ
0
, s
σ
0
), (r
σ
1
, s
σ
1
) extend (r
σ
, s
σ
)
and s
σ
0
and s
σ
1
are incompatible. For any q ∈ 2
ω
deﬁne
x
q
=
_
n<ω
r
qn
and y
q
=
_
n<ω
s
qn
.
Then we have that (x
q
, y
q
) ∈ [T
α
0
] and therefore P = ¦y
q
: q ∈ 2
ω
¦ is a
perfect set such that
P ⊆ p[T
α
0
] ⊆ p[T] = A
and P is coded in L.
This proof is due to Mansﬁeld. Solovay’s proof used forcing. Thus we
have departed
9
from our theme of giving forcing proofs.
9
“Consistency is the hobgoblin of little minds. With consistency a great soul has simply
nothing to do.” Ralph Waldo Emerson.
22 UNIFORMITY AND SCALES 90
22 Uniformity and Scales
Given R ⊆ X Y we say that S ⊆ X Y uniformizes R iﬀ
1. S ⊆ R,
2. for all x ∈ X if there exists y ∈ Y such that R(x, y), then there exists
y ∈ Y such that S(x, y), and
3. for all x ∈ X and y, z ∈ Y if S(x, y) and S(x, z), then y = z.
Another way to say the same thing is that S is a subset of R which is the
graph of a function whose domain is the same as R’s.
The Π
1
1
sets have the uniformization property.
Theorem 22.1 (Kondo [49]) Every Π
1
1
set R can be uniformized by a Π
1
1
set S.
Here, X and Y can be taken to be either ω or ω
ω
or even a singleton ¦0¦.
In this last case, this amounts to saying for any nonempty Π
1
1
set A ⊆ ω
ω
there exists a Π
1
1
set B ⊆ A such that B is a singleton, i.e., [B[ = 1. The
proof of this Theorem is to use a property which has become known as the
scale property.
Lemma 22.2 (scale property) For any Π
1
1
set A there exists ¸φ
i
: i < ω)
such that
1. each φ
i
: A → OR,
2. for all i and x, y ∈ A if φ
i+1
(x) ≤ φ
i+1
(y), then φ
i
(x) ≤ φ
i
(y),
3. for every x, y ∈ A if ∀i φ
i
(x) = φ
i
(y), then x = y,
4. for all ¸x
n
: n < ω) ∈ A
ω
and ¸α
i
: i < ω) ∈ OR
ω
if for every i and for
all but ﬁnitely many n φ
i
(x
n
) = α
i
, then there exists x ∈ A such that
lim
n→∞
x
n
= x and for each i φ
i
(x) ≤ α
i
,
5. there exists P a Π
1
1
set such that for all x, y ∈ A and i
P(i, x, y) iﬀ φ
i
(x) ≤ φ
i
(y)
and for all x ∈ A, y / ∈ A, i ∈ ω P(i, x, y), and
Descriptive Set Theory and Forcing 91
6. there exists S a Σ
1
1
set such that for all x, y ∈ A and i
S(i, x, y) iﬀ φ
i
(x) ≤ φ
i
(y)
and for all y ∈ A, x, i ∈ ω if S(i, x, y), then x ∈ A.
Another way to view a scale is from the point of view of the relations on
A deﬁned by x ≤
i
y iﬀ φ
i
(x) ≤ φ
i
(y). These are called prewellorderings.
They are well orderings if we mod out by x ≡
i
y which is deﬁned by
x ≡
i
y iﬀ x ≤
i
y and y ≤
i
x.
The second item says that these relations get ﬁner and ﬁner as i increases.
The third item says that in the “limit” we get a linear order. The fourth
item is some sort of continuity condition. And the last two items are the
deﬁnability properties of the scale.
Before proving the lemma, let us deduce uniformity from it. We do not
use the last item in the lemma. First let us show that for any nonempty Π
1
1
set A ⊆ ω
ω
there exists a Π
1
1
singleton B ⊆ A. Deﬁne
x ∈ B iﬀ x ∈ A and ∀n∀y P(n, x, y).
Since P is Π
1
1
the set B is Π
1
1
. Clearly B ⊆ A, and also by item (2) of the
lemma, B can have at most one element. So it remains to show that B is
nonempty. Deﬁne α
i
= min¦φ
i
(x) : x ∈ A¦. For each i choose x
i
∈ A such
that φ
i
(x
i
) = α
i
.
Claim: If n > i then φ
i
(x
n
) = α
i
.
Proof:
By choice of x
n
for every y ∈ A we have φ
n
(x
n
) ≤ φ
n
(y). By item
(2) in the lemma, for every y ∈ A we have that φ
i
(x
n
) ≤ φ
i
(y) and hence
φ
i
(x
n
) = α
i
.
By item (4) there exists x ∈ A such that lim
n→∞
x
n
= x and φ
i
(x) ≤ α
i
all i. By the minimality of α
i
it must be that φ
i
(x) = α
i
. So x ∈ B and we
are done.
Now to prove a more general case of uniformity suppose that R ⊆ ω
ω
ω
ω
is Π
1
1
. Let φ
i
: R → OR be scale given by the lemma and
P ⊆ ω (ω
ω
ω
ω
) (ω
ω
ω
ω
)
Descriptive Set Theory and Forcing 92
be the Π
1
1
predicate given by item (5). Then deﬁne the Π
1
1
set S ⊆ ω
ω
ω
ω
by
(x, y) ∈ S iﬀ (x, y) ∈ R and ∀z∀n P(n, (x, y), (x, z)).
The same proof shows that S uniformizes R.
The proof of the Lemma will need the following two elementary facts
about wellfounded trees. For T,
ˆ
T subtrees of Q
<ω
we say that σ : T →
ˆ
T
is a tree embedding iﬀ for all s, t ∈ T if s ⊂ t then σ(s) ⊂ σ(t). Note
that s ⊂ t means that s is a proper initial segment of t. Also note that tree
embeddings need not be onetoone. We write T _
ˆ
T iﬀ there exists a tree
embedding from T into
ˆ
T. We write T ≺
ˆ
T iﬀ there is a tree embedding which
takes the root node of T to a nonroot node of
ˆ
T. Recall that r : T → OR is a
rank function iﬀ for all s, t ∈ T if s ⊂ t then r(s) > r(t). Also the rank of T
is the minimal ordinal α such that there exists a rank function r : T → α+1.
Lemma 22.3 Suppose T _
ˆ
T and
ˆ
T is wellfounded, i.e., [
ˆ
T] = ∅, then T
is wellfounded and rank of T is less than or equal to rank of
ˆ
T.
Proof:
Let σ : T →
ˆ
T be a tree embedding and r :
ˆ
T → OR a rank function.
Then r ◦ σ is a rank function on T.
Lemma 22.4 Suppose T and
ˆ
T are well founded trees and rank of T is less
than or equal rank of
ˆ
T, then T _
ˆ
T.
Let r
T
and r
ˆ
T
be the canonical rank functions on T and
ˆ
T (see Theorem 7.1).
Inductively deﬁne σ : T ∩ Q
n
→
ˆ
T ∩ Q
n
, so as to satisfy r
T
(s) ≤ r
ˆ
T
(σ(s)).
Now we a ready to prove the existence of scales (Lemma 22.2). Let
ω
−
1
= ¦−1¦ ∪ ω
1
be wellordered in the obvious way. Given a wellfounded tree T ⊆ ω
<ω
with
rank function r
T
extend r
T
to all of ω
<ω
by deﬁning r
T
(s) = −1 if s / ∈ T.
Now suppose A ⊆ ω
ω
is Π
1
1
and x ∈ A iﬀ T
x
is wellfounded (see Theorem
17.4). Let ¦s
n
: n < ω¦ be a recursive listing of ω
<ω
with s
0
= ¸). For each
n < ω deﬁne ψ
n
: A → ω
−
1
ω ω
−
1
ω by
ψ
n
(x) = ¸r
T
x
(s
0
), x(0), r
T
x
(s
1
), x(1), . . . , r
T
x
(s
n
), x(n)).
Descriptive Set Theory and Forcing 93
The set ω
−
1
ω ω
−
1
ω is wellordered by the lexicographical order. The
scale φ
i
is just obtained by mapping the range of ψ
i
order isomorphically
to the ordinals. (Remark: by choosing s
0
= ¸), we guarantee that the ﬁrst
coordinate is always the largest coordinate, and so the range of ψ
i
is less than
or equal to ω
1
.) Now we verify the properties.
For item (2): if ψ
i+1
(x) ≤
lex
ψ
i+1
(y), then ψ
i
(x) ≤
lex
ψ
i
(y). This is true
because we are just taking the lexicographical order of a longer sequence.
For item (3): if ∀i ψ
i
(x) = ψ
i
(y), then x = y. This is true, because
ψ
i
(x) = ψ
i
(y) implies x i = y i.
For item (4): Suppose ¸x
n
: n < ω) ∈ A
ω
and for every i and for all but
ﬁnitely many n ψ
i
(x
n
) = t
i
. Then since ψ
i
(x
n
) contains x
n
i there must
be x ∈ ω
ω
such that lim
n→∞
x
n
= x. Note that since ¦s
n
: n ∈ ω¦ lists every
element of ω
<ω
, we have that for every s ∈ ω
<ω
there exists r(s) ∈ OR such
that r
T
x
n
(s) = r(s) for all but ﬁnitely many n. Using this and
lim
n→∞
T
x
n
= T
x
it follows that r is a rank function on T
x
. Consequently x ∈ A. Now since
r
T
x
(s) ≤ r(s), it follows that ψ
i
(x) ≤
lex
t
i
.
For item (5),(6): The following set is Σ
1
1
:
¦(T,
ˆ
T) : T,
ˆ
T are subtrees of ω
<ω
, T _
ˆ
T¦.
Consequently, assuming that T,
ˆ
T are wellfounded, to say that r
T
(s) ≤ r
ˆ
T
(s)
is equivalent to saying there exists a tree embedding which takes s to s. Note
that this is Σ
1
1
. This shows that it is possible to deﬁne a Σ
1
1
set S ⊆ ωω
ω
ω
ω
such that for every x, y ∈ A we have (n, x, y) ∈ S iﬀ
¸r
T
x
(s
0
), x(0), r
T
x
(s
1
), x(1), . . . , r
T
x
(s
n
), x(n))
is lexicographically less than or equal to
¸r
T
y
(s
0
), y(0), r
T
y
(s
1
), y(1), . . . , r
T
y
(s
n
), y(n)).
Note that if (n, x, y) ∈ S and y ∈ A then x ∈ A, since T
y
is a wellfounded
tree and S implies T
x
_ T
y
, so T
x
is wellfounded and so x ∈ A.
To get the Π
1
1
relation P (item (5)), instead of saying T can be embedded
into
ˆ
T we say that
ˆ
T cannot be embedded properly into T,i.e.,
ˆ
T ,≺ T or
Descriptive Set Theory and Forcing 94
in other words, there does not exists a tree embedding σ :
ˆ
T → T such
that σ(¸)) ,= ¸). This is a Π
1
1
statement. For T and
ˆ
T wellfounded trees
saying that rank of T is less than or equal to
ˆ
T is equivalent to saying rank
of
ˆ
T is not strictly smaller than the rank of T. But by Lemma 22.4 this is
equivalent to the nonexistence of such an embedding. Note also that if x ∈ A
and y / ∈ A, then we will have P(n, x, y) for every n. This is because T
y
is not
wellfounded and so cannot be embedded into the wellfounded tree T
x
.
This ﬁnishes the proof of the Scale Lemma 22.2.
The scale property was invented by Moschovakis [88] to show how deter
minacy could be used to get uniformity properties
10
in the higher projective
classes. He was building on earlier ideas of Blackwell, Addison, and Martin.
The 500 page book by Kuratowski and Mostowski [60] ends with a proof of
the uniformization theorem.
10
I have yet to see any problem, however complicated, which, when you looked at it in
the right way, did not become still more complicated. Poul Anderson
23 MARTIN’S AXIOM AND CONSTRUCTIBILITY 95
23 Martin’s axiom and Constructibility
Theorem 23.1 (G¨odel see Solovay [103]) If V=L, there exists uncountable
Π
1
1
set A ⊆ ω
ω
which contains no perfect subsets.
Proof:
Let X be any uncountable Σ
1
2
set containing no perfect subsets. For
example, a ∆
1
2
Luzin set would do (Theorem 18.1). Let R ⊂ ω
ω
ω
ω
be Π
1
1
such that x ∈ X iﬀ ∃y R(x, y). Use Π
1
1
uniformization (Theorem 22.1) to
get S ⊆ R with the property that X is the onetoone image of S via the
projection map π(x, y) = x. Then S is an uncountable Π
1
1
set which contains
no perfect subset. This is because if P ⊆ S is perfect, then π(P) is a perfect
subset of X.
Note that it is suﬃcient to assume that ω
1
= (ω
1
)
L
. Suppose A ∈ L is
deﬁned by the Π
1
1
formula θ. Then let B be the set which is deﬁned by θ in
V . So by Π
1
1
absoluteness A = B ∩ L. The set B cannot contain a perfect
set since the sentence:
∃T T is a perfect tree and ∀x (x ∈ [T] implies θ(x))
is a Σ
1
2
and false in L and so by Shoenﬁeld absoluteness (Theorem 20.2) must
be false in V . It follows then by the MansﬁeldSolovay Theorem 21.1 that B
cannot contain a nonconstructible real and so A = B.
Actually, by tracing thru the actual deﬁnition of X one can see that
the elements of the uniformizing set S (which is what A is) consist of pairs
(x, y) where y is isomorphic to some L
α
and x ∈ L
α
. These pairs are reals
which witness their own constructibility, so one can avoid using the Solovay
Mansﬁeld Theorem.
Corollary 23.2 If ω
1
= ω
L
1
, then there exists a Π
1
1
set of constructible reals
which contains no perfect set.
Theorem 23.3 (MartinSolovay [74]) Suppose MA + CH + ω
1
= (ω
1
)
L
.
Then every A ⊆ 2
ω
of cardinality ω
1
is Π
1
1
¯
.
Proof:
Let A ⊆ 2
ω
be a uncountable Π
1
1
set of constructible reals and let B be an
arbitrary subset of 2
ω
of cardinality ω
1
. Arbitrarily wellorder the two sets,
A = ¦a
α
: α < ω
1
¦ and B = ¦b
α
: α < ω
1
¦.
Descriptive Set Theory and Forcing 96
By Theorem 5.1 there exists two sequences of G
δ
sets ¸U
n
: n < ω) and
¸V
n
: n < ω) such that for every α < ω
1
for every n < ω
a
α
(n) = 1 iﬀ b
α
∈ U
n
and
b
α
(n) = 1 iﬀ a
α
∈ V
n
.
This is because the set ¦a
α
: b
α
(n) = 1¦, although it is an arbitrary subset
of A, is relatively G
δ
by Theorem 5.1.
But note that b ∈ B iﬀ ∀a ∈ 2
ω
[∀n (a(n) = 1 iﬀ b ∈ U
n
)] implies [a ∈ A and ∀n (b(n) = 1 iﬀ a ∈ U
n
)].
Since A is Π
1
1
this deﬁnition of B has the form:
∀a([Π
0
3
¯
] implies [Π
1
1
and Π
0
3
¯
])
So B is Π
1
1
¯
.
Note that if every set of reals of size ω
1
is Π
1
1
¯
then every ω
1
union of Borel
sets is Σ
1
2
¯
. To see this let ¸B
α
: α < ω
1
) be any sequence of Borel sets. Let
U be a universal Π
1
1
¯
set and let ¸x
α
: α < ω
1
) be a sequence such that
B
α
= ¦y : (x
α
, y) ∈ U ¦.
Then
y ∈
_
α<ω
1
B
α
iﬀ ∃x x ∈ ¦x
α
: α < ω
1
¦ ∧(x, y) ∈ U.
But ¦x
α
: α < ω
1
¦ is Π
1
1
¯
and so the union is Σ
1
2
¯
.
24 Σ
1
2
WELLORDERINGS 97
24 Σ
1
2
wellorderings
Theorem 24.1 (Mansﬁeld [71]) If (F, ¡) is a Σ
1
2
wellordering, i.e.,
F ⊆ ω
ω
and ¡ ⊆ F
2
are both Σ
1
2
, then F is a subset of L.
Proof:
We will use the following:
Lemma 24.2 Assume there exists z ∈ 2
ω
such that z / ∈ L. Suppose f : P →
F is a 11 continuous function from the perfect set P and both f and P are
coded in L, then there exists Q ⊆ P perfect and g : Q → F 11 continuous so
that both g and Q are coded in L and for every x ∈ Q we have g(x) ¡f(x).
Proof:
(Kechris [52]) First note that there exists σ : P → P an autohomeomor
phism coded in L such that for every x ∈ P we have σ(x) ,= x but σ
2
(x) = x.
To get this let c : 2
ω
→ 2
ω
be the complement function, i.e., c(x)(n) = 1−x(n)
which just switches 0 and 1. Then c(x) ,= x but c
2
(x) = x. Now if h : P → 2
ω
is a homeomorphism coded in L, then σ = h
−1
◦ c ◦ h works.
Now let A = ¦x ∈ P : f(σ(x)) ¡f(x)¦. The set A is a Σ
1
2
¯
set with code
in L. Now since P is coded in L there must be a z ∈ P such that z / ∈ L.
Note that σ(z) / ∈ L also. But either
f(σ(z)) ¡f(z) or f(z) = f(σ
2
(z)) ¡f(σ(z))
and so either z ∈ A or σ(z) ∈ A. In either case A has a nonconstructible
member and so by the MansﬁeldSolovay Theorem 21.1 the set A contains a
perfect set Q coded in L. Let g = f ◦ σ.
Assume there exists z ∈ F such that z / ∈ L. By the MansﬁeldSolovay
Theorem there exists a perfect set P coded in L such that P ⊆ F. Let
P
0
= P and f
0
be the identity function. Repeatedly apply the Lemma to
obtain f
n
: P
n
→ F so that for every n and P
n+1
⊆ P
n
, for every x ∈ P
n+1
f
n+1
(x) ¡ f
n
(x). But then if x ∈
n<ω
the sequence ¸f
n
(x) : n < ω) is a
descending ¡ sequence with contradicts the fact that ¡ is a wellordering.
Friedman [28] proved the weaker result that if there is a Σ
1
2
¯
wellordering
of the real line, then ω
ω
⊆ L[g] for some g ∈ ω
ω
.
25 LARGE Π
1
2
SETS 98
25 Large Π
1
2
sets
A set is Π
1
2
¯
iﬀ it is the complement of a Σ
1
2
¯
set. Unlike Σ
1
2
¯
sets which cannot
have size strictly in between ω
1
and the continuum (Theorem 21.1), Π
1
2
¯
sets
can be practically anything.
11
Theorem 25.1 (Harrington [35]) Suppose V is a model of set theory which
satisﬁes ω
1
= ω
L
1
and B is arbitrary subset of ω
ω
in V . Then there exists a
ccc extension of V , V [G], in which B is a Π
1
2
¯
set.
Proof:
Let P
B
be the following poset. p ∈ P
B
iﬀ p is a ﬁnite consistent set of
sentences of the form:
1. “[s]∩
◦
Cn
= ∅”, or
2. “x ∈
◦
Cn
, where x ∈ B.
This partial order is isomorphic to Silver’s view of almost disjoint sets forcing
(Theorem 5.1). So forcing with P
B
creates an F
σ
set
n∈ω
C
n
so that
∀x ∈ ω
ω
∩ V (x ∈ B iﬀ x ∈
_
n<ω
C
n
).
Forcing with the direct sum of ω
1
copies of P
B
,
α<ω
1
P
B
, we have that
∀x ∈ ω
ω
∩ V [¸G
α
: α < ω
1
)](x ∈ B iﬀ x ∈
α<ω
1
∪
n<ω
C
α
n
).
One way to see this is as follows. Note that in any case
B ⊆
α<ω
1
∪
n<ω
C
α
n
.
So it is the other implication which needs to be proved. By ccc, for any
x ∈ V [¸G
α
: α < ω
1
)] there exists β < ω
1
with x ∈ V [¸G
α
: α < β)]. But
considering V [¸G
α
: α < β)] as the new ground model, then G
β
would be P
B

generic over V [¸G
α
: α < β)] and hence if x / ∈ B we would have x / ∈ ∪
n<ω
C
β
n
.
Another argument will be given in the proof of the next lemma.
11
It’s life Jim, but not as we know it. Spock of Vulcan
Descriptive Set Theory and Forcing 99
Lemma 25.2 Suppose ¸c
α
: α < ω
1
be a sequence in V of elements of ω
ω
and ¸a
α
: α < ω
1
) is a sequence in V [¸G
α
: α < ω
1
)] of elements of 2
ω
. Using
Silver’s forcing add a sequence of Π
0
2
¯
sets ¸U
n
: n < ω) such that
∀n ∈ ω∀α < ω
1
(a
α
(n) = 1 iﬀ c
α
∈ U
n
).
Then
V [¸G
α
: α < ω
1
)][¸U
n
: n < ω)] [= ∀x ∈ ω
ω
(x ∈ B iﬀ x ∈
α<ω
1
∪
n<ω
C
α
n
).
Proof:
The lemma is not completely trivial, since adding the ¸U
n
: n < ω) adds
new elements of ω
ω
which may somehow sneak into the ω
1
intersection.
Working in V deﬁne p ∈ Q iﬀ p is a ﬁnite set of consistent sentences of
the form:
1. “[s] ⊆ U
n,m
” where s ∈ ω
<ω
, or
2. “c
α
∈ U
n,m
”.
Here we intend that U
n
= ∩
m∈ω
U
n,m
. Since the c’s are in V it is clear that
the partial order Q is too. Deﬁne
P = ¦(p, q) ∈ (
α<ω
1
P
B
) Q : if “c
α
∈ U
n,m
”∈ q, then p [¬ a
α
(n) = 1¦.
Note that P is a semilowerlattice, i.e., if (p
0
, q
0
) and (p
1
, q
1
) are compatible
elements of P, then (p
0
∪ p
1
, q
0
∪ q
1
) is their greatest lower bound. This is
another way to view the iteration, i.e, P is dense in the usual iteration. Not
every iteration has this property, one which Harrington calls “innocuous”.
Now to prove the lemma, suppose for contradiction that
(p, q) [¬
◦
x∈
α<ω
1
∪
n<ω
C
α
n
and
◦
x/ ∈ B.
To simplify the notation, assume (p, q) = (∅, ∅). Since P has the ccc a se
quence of Working in V let )A
n
: n ∈ ω¸ be a sequence of maximal antichains
of P which decide
◦
x, i.e. for (p, q) ∈ A
n
there exists s ∈ ω
n
such that
(p, q) [¬
◦
x n = ˇ s.
Descriptive Set Theory and Forcing 100
Since P has the ccc, the A
n
are countable and we can ﬁnd an α < ω which
does not occur in the support of any p for any (p, q) in
n∈ω
A
n
. Since x is
forced to be in ∪
n<ω
C
α
n
there exists (p, q) and n ∈ ω such that
(p, q) [¬
◦
x∈ C
α
n
.
Let “x
i
∈ C
α
n
” for i < N be all the sentences of this type which occur in
p(α). Since we are assuming x is being forced not in B it must be diﬀerent
than all the x
i
, so there must be an m, (ˆ p, ˆ q) ∈ A
m
, and s ∈ ω
m
, such that
1. (ˆ p, ˆ q) and (p, q) are compatible,
2. (ˆ p, ˆ q) [¬
◦
x m = ˇ s, and
3. x
i
m ,= s for every i < N.
((To get (ˆ p, ˆ q) and s let G be a generic ﬁlter containing (p, q), then since
x
G
,= x
i
for every i < N there must be m < ω and s ∈ ω
m
such that
x
G
m = s and s ,= x
i
m for every i < N. Let (ˆ p, ˆ q) ∈ G∩ A
m
.))
Now consider (p ∪ ˆ p, q ∪ ˆ q) ∈ P. Since α was not in the support of ˆ p,
(p ∪ ˆ p)(α) = p(α).
Since s was chosen so that x
i
/ ∈ [s] for every i < N,
p(α) ∪ ¦[s] ∩ C
α
n
= ∅¦
is a consistent set of sentences, hence an element of P
B
. This is a contradic
tion, the condition
(p ∪ ˆ p ∪ ¦[s] ∩ C
α
n
= ∅¦, q ∪ ˆ q)
forces x ∈ C
α
n
and also x / ∈ C
α
n
.
Let F be a universal Σ
0
2
set coded in V and let ¸a
α
∈ 2
ω
: α < ω
1
) be
such that
F
a
α
=
_
n∈ω
C
α
n
.
Let C = ¸c
α
: α < ω
1
) be a Π
1
1
set in V . Such a set exists since ω
1
= ω
L
1
.
Lemma 25.3 In V [¸G
α
: α < ω
1
)][¸U
n
: n < ω)] the set B is Π
1
2
¯
.
Descriptive Set Theory and Forcing 101
Proof:
x ∈ B iﬀ x ∈
α<ω
1
n∈ω
C
α
n
iﬀ x ∈
α<ω
1
F
a
α
iﬀ
∀a, c if c ∈ C and ∀n (a(n) = 1 iﬀ c ∈ U
n
), then (a, x) ∈ F, i.e. (x ∈ F
a
).
Note that
• “c ∈ C” is Π
1
1
,
• “∀n (a(n) = 1 iﬀ c ∈ U
n
)” is Borel, and
• “(a, x) ∈ F” is Borel,
and so this ﬁnal deﬁnition for B has the form:
∀((Π
1
1
∧Borel)) → Borel)
Therefore B is Π
1
2
¯
.
Harrington [35] also shows how to choose B so that the generic extension
has a ∆
1
3
¯
wellordering of ω
ω
. He also shows how to take a further innocuous
extensions to make B a ∆
1
3
set and to get a ∆
1
3
wellordering.
102
Part III
Classical Separation Theorems
26 SouslinLuzin Separation Theorem
Deﬁne A ⊆ ω
ω
to be κSouslin iﬀ there exists a tree T ⊆
n<ω
(κ
n
ω
n
)
such that
y ∈ A iﬀ ∃x ∈ κ
ω
∀n < ω (x n, y n) ∈ T.
In this case we write A = p[T], the projection of the inﬁnite branches of the
tree T. Note that ωSouslin is the same as Σ
1
1
¯
.
Deﬁne the κBorel sets to be the smallest family of subsets of ω
ω
con
taining the usual Borel sets and closed under intersections or unions of size
κ and complements.
Theorem 26.1 Suppose A and B are disjoint κSouslin subsets of ω
ω
. Then
there exists a κBorel set C which separates A and B, i.e., A ⊆ C and
C ∩ B = ∅.
Proof:
Let A = p[T
A
] and B = p[T
B
]. Given a tree T ⊆
n<ω
(κ
n
ω
n
), and
s ∈ κ
<ω
, t ∈ ω
<ω
(possibly of diﬀerent lengths), deﬁne
T
s,t
= ¦(ˆ s,
ˆ
t) ∈ T : (s ⊆ ˆ s or ˆ s ⊆ s) and (t ⊆
ˆ
t or
ˆ
t ⊆ t)¦.
Lemma 26.2 Suppose p[T
s,t
A
] cannot be separated from p[T
r,t
B
] by a κBorel
set. Then for some α < κ the set
p[T
sˆα,t
A
] cannot be separated from p[T
r,t
B
] by a κBorel set.
Proof:
Note that p[T
s,t
A
] =
α<κ
p[T
sˆα,t
A
]. If there were no such α, then for every
α we would have a κBorel set C
α
with
p[T
sˆα,t
A
] ⊆ C
α
and C
α
∩ p[T
r,t
B
] = ∅.
But then
α<κ
C
α
is a κBorel set separating p[T
s,t
A
] and p[T
r,t
B
].
26 SOUSLINLUZIN SEPARATION THEOREM 103
Lemma 26.3 Suppose p[T
s,t
A
] cannot be separated from p[T
r,t
B
] by a κBorel
set. Then for some β < κ
p[T
s,t
A
] cannot be separated from p[T
rˆβ,t
B
] by a κBorel set.
Proof:
Since p[T
r,t
B
] =
β<κ
p[T
rˆβ,t
B
], if there were no such β then for every β we
would have κBorel set C
β
with
p[T
s,t
A
] ⊆ C
β
and C
β
∩ p[T
rˆβ,t
B
] = ∅.
But then
β<κ
C
β
is a κBorel set separating p[T
s,t
A
] and p[T
r,t
B
].
Lemma 26.4 Suppose p[T
s,t
A
] cannot be separated from p[T
r,t
B
] by a κBorel
set. Then for some n < ω
p[T
s,tˆn
A
] cannot be separated from p[T
r,tˆn
B
] by a κBorel set.
Proof:
Note that
p[T
s,tˆn
A
] = p[T
s,t
A
] ∩ [tˆn]
and
p[T
r,tˆn
B
] = p[T
r,t
B
] ∩ [tˆn].
Thus if C
n
⊆ [tˆn] were to separate p[T
s,tˆn
A
] and p[T
r,tˆn
B
] for each n, then
n<ω
C
n
would separate p[T
s,t
A
] from p[T
r,t
B
].
To prove the separation theorem apply the lemmas iteratively in rotation
to obtain, u, v ∈ κ
ω
and x ∈ ω
ω
so that for every n, p[T
un,xn
A
] cannot be
separated from p[T
vn,xn
B
]. But necessarily, for every n
(u n, x n) ∈ T
A
and (v n, x n) ∈ T
B
otherwise either p[T
un,xn
A
] = ∅ or p[T
vn,xn
B
] = ∅ and they could be separated.
But this means that x ∈ p[T
A
] = A and x ∈ p[T
B
] = B contradicting the fact
that they are disjoint.
27 KLEENE SEPARATION THEOREM 104
27 Kleene Separation Theorem
We begin by deﬁning the hyperarithmetic subsets of ω
ω
. We continue
with our view of Borel sets as wellfounded trees with little dohickey’s (basic
clopen sets) attached to its terminal nodes.
A code for a hyperarithmetic set is a triple (T, p, q) where T is a
recursive wellfounded subtree of ω
<ω
, p : T
>0
→ 2 is recursive, and q :
T
0
→ B is a recursive map, where B is the set of basic clopen subsets of ω
ω
including the empty set. Given a code (T, p, q) we deﬁne ¸C
s
: s ∈ T) as
follows.
• if s is a terminal node of T, then
C
s
= q(s)
• if s is a not a terminal node and p(s) = 0, then
C
s
=
_
¦C
sˆn
: sˆn ∈ T¦,
and
• if s is a not a terminal node and p(s) = 1, then
C
s
=
¦C
sˆn
: sˆn ∈ T¦.
Here we are being a little more ﬂexible by allowing unions and intersections
at various nodes.
Finally, the set C coded by (T, p, q) is the set C
. A set C ⊆ ω
ω
is
hyperarithmetic iﬀ it is coded by some recursive (T, p, q).
Theorem 27.1 (Kleene [55]) Suppose A and B are disjoint Σ
1
1
subsets of
ω
ω
. Then there exists a hyperarithmetic set C which separates them, i.e.,
A ⊆ C and C ∩ B = ∅.
Proof:
This amounts basically to a constructive proof of the classical Separation
Theorem 26.1.
Let A = p[T
A
] and B = p[T
B
] where T
A
and T
B
are recursive subtrees of
n∈ω
(ω
n
ω
n
), and
p[T
A
] = ¦y : ∃x∀n (x n, y n) ∈ T
A
¦
Descriptive Set Theory and Forcing 105
and similarly for p[T
B
]. Now deﬁne the tree
T = ¦(u, v, t) : (u, t) ∈ T
A
and (v, t) ∈ T
B
¦.
Notice that T is recursive tree which is wellfounded. Any inﬁnite branch thru
T would give a point in the intersection of A and B which would contradict
the fact that they are disjoint.
Let T
+
be the tree of all nodes which are either “in” or “just out” of T,
i.e., (u, v, t) ∈ T
+
iﬀ (u n, v n, t n) ∈ T where [u[ = [v[ = [t[ = n + 1.
Now we deﬁne the family of sets
¸C
(u,v,t)
: (u, v, t) ∈ T
+
)
as follows.
Suppose (u, v, t) ∈ T
+
is a terminal node of T
+
. Then since (u, v, t) / ∈ T
either (u, t) / ∈ T
A
in which case we deﬁne C
(u,v,t)
= ∅ or (u, t) ∈ T
A
and
(v, t) / ∈ T
B
in which case we deﬁne C
(u,v,t)
= [t]. Note that in either case
C
(u,v,t)
⊆ [t] separates p[T
u,t
A
] from p[T
v,t
B
].
Lemma 27.2 Suppose ¸A
n
: n < ω), ¸B
m
: m < ω) ¸C
nm
: n, m < ω)
are such that for every n and m C
nm
separates A
n
from B
m
. Then both
n<ω
m<ω
C
nm
and
m<ω
n<ω
C
nm
separate
n<ω
A
n
from
m<ω
B
m
.
Proof:
Left to reader.
It follows from the Lemma that if we let
C
(u,v,t)
=
_
k<ω
m<ω
_
n<ω
C
(uˆn,vˆm,tˆk)
(or any other permutation
12
of
and
), then by induction on rank of
(u, v, t) in T
+
that C
(u,v,t)
⊆ [t] separates p[T
u,t
A
] from p[T
v,t
B
]. Hence, C =
C
(,,)
separates A = p[T
A
] from B = p[T
B
].
To get a hyperarithmetic code use the tree consisting of all subsequences
of sequences of the form,
¸t(0), v(0), u(0), . . . , t(n), v(n), u(n))
12
Algebraic symbols are used when you do not know what you are talking about
(Philippe Schnoebelen).
Descriptive Set Theory and Forcing 106
where (u, v, t) ∈ T
+
. Details are left to the reader.
The theorem also holds for A and B disjoint Σ
1
1
subsets of ω. One way
to see this is to identify ω with the constant functions in ω
ω
. The deﬁnition
of hyperarithmetic code (T, p, q) is changed only by letting q map into the
ﬁnite subsets of ω.
Theorem 27.3 If C is a hyperarithmetic set, then C is ∆
1
1
.
Proof:
This is true whether C is a subset of ω
ω
or ω. We just do the case C ⊆ ω
ω
.
Let (T, p, q) be a hyperarithmetic code for C. Then x ∈ C iﬀ
there exists a function in : T → ¦0, 1¦ such that
1. if s a terminal node of T, then in(s) = 1 iﬀ x ∈ q(s),
2. if s ∈ T and not terminal and p(s) = 0, then in(s) = 1 iﬀ there exists
n with s
ˆ
n ∈ T and in(s
ˆ
n) = 1,
3. if s ∈ T and not terminal and p(s) = 1, then
in(s) = 1 iﬀ for all n with s
ˆ
n ∈ T we have in(s
ˆ
n) = 1, and ﬁnally,
4. in(¸)) = 1.
Note that (1) thru (4) are all ∆
1
1
(being a terminal node in a recursive tree
is Π
0
1
, etc). It is clear that in is just coding up whether or not x ∈ C
s
for
s ∈ T. Consequently, C is Σ
1
1
. To see that ∼ C is Σ
1
1
note that x / ∈ C iﬀ
there exists in : T → ¦0, 1¦ such that (1), (2), (3), and (4)
where
4
in(¸)) = 0.
Corollary 27.4 A set is ∆
1
1
iﬀ it is hyperarithmetic.
Corollary 27.5 If A and B are disjoint Σ
1
1
sets, then there exists a ∆
1
1
set
which separates them.
For more on the eﬀective Borel hierarchy, see Hinman [40]. See Barwise
[10] for a model theoretic or admissible sets approach to the hyperarithmetic
hierarchy.
28 Π
1
1
REDUCTION 107
28 Π
1
1
Reduction
We say that A
0
,B
0
reduce A,B iﬀ
1. A
0
⊆ A and B
0
⊆ B,
2. A
0
∪ B
0
= A ∪ B, and
3. A
0
∩ B
0
= ∅.
Π
1
1
reduction is the property that every pair of Π
1
1
sets can be reduced by
a pair of Π
1
1
sets. The sets can be either subsets of ω or of ω
ω
.
Theorem 28.1 Π
1
1
uniformity implies Π
1
1
reduction.
Proof:
Suppose A, B ⊆ X are Π
1
1
where X = ω or X = ω
ω
. Let
P = (A ¦0¦) ∪ (B ¦1¦).
Then P is a Π
1
1
subset of X ω
ω
and so by Π
1
1
uniformity (Theorem 22.1)
there exists Q ⊆ P which is Π
1
1
and for every x ∈ X, if there exists i ∈ ¦0, 1¦
such that (x, i) ∈ P, then there exists a unique i ∈ ¦0, 1¦ such that (x, i) ∈ Q.
Hence, letting
A
0
= ¦x ∈ X : (x, 0) ∈ Q¦
and
B
0
= ¦x ∈ X : (x, 1) ∈ Q¦
gives a pair of Π
1
1
sets which reduce A and B.
There is also a proof of reduction using the prewellordering property,
which is a weakening of the scale property used in the proof of Π
1
1
uniformity.
So, for example, suppose A and B are Π
1
1
subsets of ω
ω
. Then we know there
are maps from ω
ω
to trees,
x → T
a
x
and y → T
b
y
which are “recursive” and
x ∈ A iﬀ T
a
x
is wellfounded and
y ∈ B iﬀ T
b
y
is wellfounded.
Now deﬁne
Descriptive Set Theory and Forcing 108
1. x ∈ A
0
iﬀ x ∈ A and not (T
b
x
≺ T
a
x
), and
2. x ∈ B
0
iﬀ x ∈ B and not (T
a
x
_ T
b
x
).
Since ≺ and _ are both Σ
1
1
it is clear, that A
0
and B
0
are Π
1
1
subsets of A
and B respectively. If x ∈ A and x / ∈ B, then T
a
x
is wellfounded and T
b
x
is
illfounded and so not (T
b
x
≺ T
a
x
) and a ∈ A
0
. Similarly, if x ∈ B and x / ∈ A,
then x ∈ B
0
. If x ∈ A∩B, then both T
b
x
and T
a
x
are wellfounded and either
T
a
x
_ T
b
x
, in which case x ∈ A
0
and x / ∈ B
0
, or T
b
x
≺ T
a
x
, in which case x ∈ B
0
and x / ∈ A
0
.
Theorem 28.2 Π
1
1
reduction implies Σ
1
1
separation, i.e., for any two dis
joint Σ
1
1
sets A and B there exists a ∆
1
1
set C which separates them. i.e.,
A ⊆ C and C ∩ B = ∅.
Proof:
Note that ∼ A ∪ ∼ B = X. If A
0
and B
0
are Π
1
1
sets reducing ∼ A and
∼ B, then ∼ A
0
= B
0
, so they are both ∆
1
1
. If we set C = B
0
, then
C = B
0
=∼ A
0
⊆∼ A
so C ⊆∼ A and therefore A ⊆ C. On the other hand C = B
0
⊆∼ B implies
C ∩ B = ∅.
29 ∆
1
1
CODES 109
29 ∆
1
1
codes
Using Π
1
1
reduction and universal sets it is possible to get codes for ∆
1
1
subsets
of ω and ω
ω
.
Here is what we mean by ∆
1
1
codes for subsets of X where X = ω or
X = ω
ω
.
There exists a Π
1
1
sets C ⊆ ω ω
ω
and P ⊆ ω ω
ω
X and a Σ
1
1
set
S ⊆ ω ω
ω
X such that
• for any (e, u) ∈ C
¦x ∈ X : (e, u, x) ∈ P¦ = ¦x ∈ X : (e, u, x) ∈ S¦
• for any u ∈ ω
ω
and ∆
1
1
(u) set D ⊆ X there exists a (e, u) ∈ C such
that
D = ¦x ∈ X : (e, u, x) ∈ P¦ = ¦x ∈ X : (e, u, x) ∈ S¦.
From now on we will write
“e is a ∆
1
1
(u)code for a subset of X”
to mean (e, u) ∈ C and remember that it is a Π
1
1
predicate.
We also write “D is the ∆
1
1
(u) set coded by e” if “e is a ∆
1
1
(u)code for a
subset of X” and
D = ¦x ∈ X : (e, x) ∈ P¦ = ¦x ∈ X : (e, x) ∈ S¦.
Note that x ∈ D can be said in either a Σ
1
1
(u) way or Π
1
1
(u) way, using either
S or P.
Theorem 29.1 (SpectorGandy Theorem [105],[31] ) ∆
1
1
codes exist.
Proof:
Let U ⊆ ω ω
ω
X be a Π
1
1
set which is universal for all Π
1
1
(u) sets,
i.e., for every u ∈ ω
ω
and A ∈ Π
1
1
(u) with A ⊆ X there exists e ∈ ω such
that A = ¦x ∈ X : (e, u, x) ∈ U¦. For example, to get such a U proceed
as follows. Let ¦e¦
u
be the partial function you get by using the e
th
Turing
machine with oracle u. Then deﬁne (e, u, x) ∈ U iﬀ ¦e¦
u
is the characteristic
function of a tree T ⊆
n<ω
(ω
n
ω
n
) and T
x
= ¦s : (s, x [s[) ∈ T¦ is
wellfounded.
Descriptive Set Theory and Forcing 110
Now get a doubly universal pair. Let e → (e
0
, e
1
) be the usual recursive
unpairing function from ω to ω ω and deﬁne
U
0
= ¦(e, u, x) : (e
0
, u, x) ∈ U¦
and
U
1
= ¦(e, u, x) : (e
1
, u, x) ∈ U¦.
The pair of sets U
0
and U
1
are Π
1
1
and doubly universal, i.e., for any u ∈ ω
ω
and A and B which are Π
1
1
(u) subsets of X there exists e ∈ ω such that
A = ¦x : (e, u, x) ∈ U
0
¦
and
B = ¦x : (e, u, x) ∈ U
1
¦.
Now apply reduction to obtain P
0
⊆ U
0
and P
1
⊆ U
1
which are Π
1
1
sets.
Note that the by the nature of taking cross sections, P
0
e,u
and P
1
e,u
reduce U
0
e,u
and U
1
e,u
. Now we deﬁne
• “e is a ∆
1
1
(u) code” iﬀ ∀x ∈ X(x ∈ P
0
e,u
or x ∈ P
1
e,u
), and
• P = P
0
and S =∼ P
1
.
Note that e is a ∆
1
1
(u) code is a Π
1
1
statement in (e, u). Also if e is a ∆
1
1
(u)
code, then P
(e,u)
= S
e,u
and so its a ∆
1
1
(u) set. Furthermore if D ⊆ X is a
∆
1
1
(u) set, then since U
0
and U
1
were a doubly universal pair, there exists e
such that U
0
e,u
= D and U
1
e,u
=∼ D. For this e it must be that U
0
e,u
= P
0
e,u
and U
1
e,u
= P
1
e,u
since the P’s reduce the U’s. So this e is a ∆
1
1
(u) code which
codes the set D.
Corollary 29.2 ¦(x, u) ∈ P(ω) ω
ω
: x ∈ ∆
1
1
(u)¦ is Π
1
1
.
Proof:
x ∈ ∆
1
1
(u) iﬀ ∃e ∈ ω such that
1. e is a ∆
1
1
(u) code,
2. ∀n if n ∈ x, then n is in the ∆
1
1
(u)set coded by e, and
3. ∀n if n is the ∆
1
1
(u)set coded by e, then n ∈ x.
Descriptive Set Theory and Forcing 111
Note that clause (1) is Π
1
1
. Clause (2) is Π
1
1
if we use that (e, u, n) ∈ P is
equivalent to “n is in the ∆
1
1
(u)set coded by e”. While clause (3) is Π
1
1
if we
use that (e, u, n) ∈ S is equivalent to “n is in the ∆
1
1
(u)set coded by e”.
We say that y ∈ ω
ω
is ∆
1
1
(u) iﬀ its graph ¦(n, m) : y(n) = m¦ is ∆
1
1
(u).
Since being the graph a function is a Π
0
2
property it is easy to see how to
obtain ∆
1
1
(u) codes for functions y ∈ ω
ω
.
Corollary 29.3 Suppose θ(x, y, z) is a Π
1
1
formula, then
ψ(y, z) = ∃x ∈ ∆
1
1
(y) θ(x, y, z)
is a Π
1
1
formula.
Proof:
ψ(y, z) iﬀ
∃e ∈ ω such that
1. e is a ∆
1
1
(y) code, and
2. ∀x if x is the set coded by (e, y), then θ(x, y, z).
This will be Π
1
1
just in case the clause “x is the set coded by (e, y)” is Σ
1
1
.
But this is ∆
1
1
provided that e is a ∆
1
1
(y) code, e.g., for x ⊆ ω we just say:
∀n ∈ ω
1. if n ∈ x then (e, y, n) ∈ S and
2. if (e, y, n) ∈ P, then n ∈ x.
Both of these clauses are Σ
1
1
since S is Σ
1
1
and P is Π
1
1
. A similar argument
works for x ∈ ω
ω
.
The method of this corollary also works for the quantiﬁer
∃D ⊆ ω
ω
such that D ∈ ∆(y) θ(D, y, z).
It is equivalent to say ∃e ∈ ω such that e is a ∆
1
1
(y) code for a subset of
ω
ω
and θ(. . . , y, z) where occurrences of the “q ∈ D” in the formula θ have
been replaced by either (e, y, q) ∈ P or (e, y, q) ∈ S, whichever is necessary
to makes θ come out Π
1
1
.
Descriptive Set Theory and Forcing 112
Corollary 29.4 Suppose f : ω
ω
→ ω
ω
is Borel, B ⊆ ω
ω
is Borel, and f is
onetoone on B. Then the image of B under f, f(B), is Borel.
Proof:
By relativizing the following argument to an arbitrary parameter we may
assume that the graph of f and the set B are ∆
1
1
. Deﬁne
R = ¦(x, y) : f(x) = y and x ∈ B¦.
Then for any y the set
¦x : R(x, y)¦
is a ∆
1
1
(y) singleton (or empty). Consequently, its unique element is ∆
1
1
in
y. It follows that
y ∈ f(B) iﬀ ∃x R(x, y) iﬀ ∃x ∈ ∆
1
1
(y) R(x, y)
and so f(B) is both Σ
1
1
and Π
1
1
.
Many applications of the GandySpector Theorem exist. For example, it
is shown (assuming V=L in all three cases) that
1. there exists an uncountable Π
1
1
set which is concentrated on the ratio
nals (Erdos, Kunen, and Mauldin [21]),
2. there exists a Π
1
1
Hamel basis (Miller [85]), and
3. there exists a topologically rigid Π
1
1
set (Van Engelen, Miller, and Steel
[18]).
113
Part IV
Gandy Forcing
30 Π
1
1
equivalence relations
Theorem 30.1 (Silver [101]) Suppose (X, E) is a Π
1
1
¯
equivalence relation,
i.e. X is a Borel set and E ⊆ X
2
is a Π
1
1
¯
equivalence relation on X. Then
either E has countably many equivalence classes or there exists a perfect set
of pairwise inequivalent elements.
Before giving the proof consider the following example. Let WO be the
set of all characteristic functions of wellorderings of ω. This is a Π
1
1
subset
of 2
ω×ω
. Now deﬁne x · y iﬀ there exists an isomorphism taking x to y or
x, y / ∈ WO. Note that (2
ω×ω
, ·) is a Σ
1
1
equivalence relation with exactly ω
1
equivalence classes. Furthermore, if we restrict · to WO, then (WO, ·) is
a Π
1
1
equivalence relation (since wellorderings are isomorphic iﬀ neither is
isomorphic to an initial segment of the other). Consequently, Silver’s theorem
is the best possible.
The proof we are going to give is due to Harrington [33], see also Kechris
and Martin [53], Mansﬁeld and Weitkamp [73] and Louveau [64]. A model
theoretic proof is given in Harrington and Shelah [38].
We can assume that X is ∆
1
1
and E is Π
1
1
, since the proof readily relativizes
to an arbitrary parameter. Also, without loss, we may assume that X = ω
ω
since we just make the complement of X into one more equivalence class.
Let P be the partial order of nonempty Σ
1
1
subsets of ω
ω
ordered by
inclusion. This is known as Gandy forcing. Note that there are many
trivial generic ﬁlters corresponding to Σ
1
1
singletons.
Lemma 30.2 If G is Pgeneric over V , then there exists a ∈ ω
ω
such that
G = ¦p ∈ P : a ∈ p¦ and ¦a¦ =
G.
Proof:
For every n an easy density argument shows that there exists a unique
s ∈ ω
n
such that [s] ∈ G where [s] = ¦x ∈ ω
ω
: s ⊆ x¦. Deﬁne a ∈ ω
ω
by
[a n] ∈ G for each n. Clearly,
G ⊆ ¦a¦.
Now suppose B ∈ G, we need to show a ∈ B. Let B = p[T].
30 Π
1
1
EQUIVALENCE RELATIONS 114
Claim: There exists x ∈ ω
ω
such that p[T
xn,an
] ∈ G for every n ∈ ω.
Proof:
This is by induction on n. Suppose p[T
xn,an
] ∈ G. Then
p[T
xn,an+1
] ∈ G
since
p[T
xn,an+1
] = [a n + 1] ∩ p[T
xn,an
]
and both of these are in G. But note that
p[T
xn,an+1
] =
_
m∈ω
p[T
xnˆm,an+1
]
and so by a density argument there exists m = x(n) such that
p[T
xnˆm,an+1
] ∈ G.
This proves the Claim.
By the Claim we have that (x, a) ∈ [T] (since elements of P are nonempty)
and so a ∈ p[T] = B. Consequently,
G = ¦a¦. Now suppose that a ∈ p ∈ P
and p / ∈ G. Then since
¦q ∈ P : q ≤ p or q ∩ p = ∅¦
is dense there must be q ∈ G with q ∩p = ∅. But this is impossible, because
a ∈ q ∩ p, but q ∩ p = ∅ is a Π
1
1
sentence and hence absolute.
We say that a ∈ ω
ω
is Pgeneric over V iﬀ G = ¦p ∈ P : a ∈ p¦ is
Pgeneric over V .
Lemma 30.3 If a is Pgeneric over V and a = ¸a
0
, a
1
) (where ¸, ) is the
standard pairing function), then a
0
and a
1
are both Pgeneric over V .
Proof:
The proof is symmetric so we just do it for a
0
. Note that we are not
claiming that they are product generic only that each is separately generic.
Suppose D ⊆ P is dense open. Let
E = ¦p ∈ P : ¦x
0
: x ∈ p¦ ∈ D¦.
30 Π
1
1
EQUIVALENCE RELATIONS 115
To see that E is dense let q ∈ P be arbitrary. Deﬁne
q
0
= ¦x
0
: x ∈ q¦.
Since q
0
is a nonempty Σ
1
1
set and D is dense, there exists r
0
≤ q
0
with
r
0
∈ D. Let
r = ¦x ∈ q : x
0
∈ r
0
¦.
Then r ∈ E and r ≤ q.
Since E is dense we have that there exists p ∈ E with a ∈ p and conse
quently,
a
0
∈ p
0
= ¦x
0
: x ∈ p¦ ∈ D.
Lemma 30.4 Suppose B ⊆ ω
ω
is Π
1
1
and for every x, y ∈ B we have that
xEy. Then there exists a ∆
1
1
set D with B ⊆ D ⊆ ω
ω
and such that for
every x, y ∈ D we have that xEy.
Proof:
Let A = ¦x ∈ ω
ω
: ∀y y ∈ B → xEy¦. Then A is a Π
1
1
set which
contains the Σ
1
1
set B, consequently by the Separation Theorem 27.5 or 28.2
there exists a ∆
1
1
set D with B ⊆ D ⊆ A. Since all elements of B are
equivalent, so are all elements of A and hence D is as required.
Now we come to the heart of Harrington’s proof. Let B be the union of
all ∆
1
1
subsets of ω
ω
which meet only one equivalence class of E, i.e.
B =
_
¦D ⊆ ω
ω
: D ∈ ∆
1
1
and ∀x, y ∈ D xEy¦.
Since E is Π
1
1
we know that by using ∆
1
1
codes that this union is Π
1
1
, i.e.,
z ∈ B iﬀ ∃e ∈ ω such that
1. e is a ∆
1
1
code for a subset of ω
ω
,
2. ∀x, y in the set coded by e we have xEy, and
3. z is in the set coded by e.
30 Π
1
1
EQUIVALENCE RELATIONS 116
Note that item (1) is Π
1
1
and (2) and (3) are both ∆
1
1
(see Theorem 29.1).
If B = ω
ω
, then since there are only countably many ∆
1
1
sets, there would
only be countably many E equivalence classes and we are done. So assume
A =∼ B is a nonempty Σ
1
1
set and in this case we will prove that there is a
perfect set of Einequivalent reals.
Lemma 30.5 Suppose c ∈ ω
ω
∩ V . Then
A [¬
P
ˇ c , E
◦
a
where
◦
a is a name for the generic real (Lemma 30.2).
Proof:
Suppose not, and let C ⊆ A be a nonempty Σ
1
1
set such that C [¬ cEa.
We know that there must exists c
0
, c
1
∈ C with c
0
,Ec
1
. Otherwise there
would exists a ∆
1
1
superset of C which meets only one equivalence class
(Lemma 30.4). But we these are all disjoint from A. Let
Q = ¦c : c
0
∈ C, c
1
∈ C, and c
0
, Ec
1
¦.
Note that Q is a nonempty Σ
1
1
set. Let a ∈ Q be Pgeneric over V . Then by
Lemma 30.3 we have that both a
0
and a
1
are Pgeneric over V and a
0
∈ C,
a
1
∈ C, and a
0
, Ea
1
. But a
i
∈ C and C [¬ a
i
Ec means that
a
0
Ec, a
1
Ec, and a
0
, Ea
1
.
This contradicts the fact that E is an equivalence relation.
Note that “E is an equivalence relation” is a Π
1
1
statement hence it is
absolute. Note also that we don’t need to assume that there are a which are
Pgeneric over V . To see this replace V by a countable transitive model M
of ZFC
∗
(a suﬃciently large fragment of ZFC) and use absoluteness.
Note that the lemma implies that if (a
0
, a
1
) is P Pgeneric over V and
a
1
∈ A, then a
0
,Ea
1
. This is because a
1
is Pgeneric over V [a
0
] and so a
0
can be regarded as an element of the ground model.
Lemma 30.6 Suppose M is a countable transitive model of ZFC
∗
and P is
a partially ordered set in M. Then there exists ¦G
x
: x ∈ 2
ω
¦, a “perfect” set
of Pﬁlters, such that for every x ,= y we have that (G
x
, G
y
) is PPgeneric
over M.
30 Π
1
1
EQUIVALENCE RELATIONS 117
Proof:
Let D
n
for n < ω list all dense open subsets of P P which are in M.
Construct ¸p
s
: s ∈ 2
<ω
) by induction on the length of s so that
1. s ⊆ t implies p
t
≤ p
s
and
2. if [s[ = [t[ = n + 1 and s and t are distinct, then (p
s
, p
t
) ∈ D
n
.
Now deﬁne for any x ∈ 2
ω
G
x
= ¦p ∈ P : ∃n p
xn
≤ p¦.
Finally to prove Theorem 30.1 let M be a countable transitive set iso
morphic to an elementary substructure of (V
κ
, ∈) for some suﬃciently large
κ. Let
¦G
x
: x ∈ 2
ω
¦
be given by Lemma 30.6 with A ∈ G
x
for all x and let
P = ¦a
x
: x ∈ 2
ω
¦
be the corresponding generic reals. By Lemma 30.5 we know that for every
x ,= y ∈ 2
ω
we have that a
x
,Ea
y
. Note also that P is perfect because the
map x → a
x
is continuous. This is because for any n ∈ ω there exists m < ω
such that every p
s
with s ∈ 2
m
decides a n.
Corollary 30.7 Every Σ
1
1
set which contains a real which is not ∆
1
1
contains
a perfect subset.
Proof:
Let A ⊆ ω
ω
be a Σ
1
1
set. Deﬁne xEy iﬀ x, y / ∈ A or x=y. Then is E is a
Π
1
1
equivalence relation. A ∆
1
1
singleton is a ∆
1
1
real, hence Harrington’s set
B in the above proof must be nonempty. Any perfect set of Einequivalent
elements can contain at most one element of ∼ A.
Corollary 30.8 Every uncountable Σ
1
1
¯
set contains a perfect subset.
30 Π
1
1
EQUIVALENCE RELATIONS 118
Perhaps this is not such a farfetched way of proving this result, since one
of the usual proofs looks like a combination of Lemma 30.2 and 30.6.
V.Kanovei has pointed out to me (email, see also Kanovei [48]) that there
is a shorter proof of
If (a, b) is a PPgeneric over V pair of reals a, b ∈ A then a , Eb.
which avoids Lemma 30.5 and absoluteness:
1. Assume not. Then there exist conditions X, Y ⊆ A such that XY [¬
aEb.
2. Thus if (a, b) ∈ X Y is P P generic then aEb.
3. It follows that aEa
for any two Pgeneric a, a
∈ X. Indeed take b ∈ Y
which is Pgeneric over V [a, a
]. [Or over V [a] ∪ V [a
] if you see diﬃculties
with V [a, a
] when the pair (a, a
) is not generic over V .] Then both (a, b)
and (a
, b) are P Pgeneric, and use item 2.
4. Similarly bEb
for any pair of Pgeneric b, b
∈ Y .
5. Therefore aEb for any pair of Pgeneric a ∈ X and b ∈ Y .
6. Finally aEb for all a ∈ X and b ∈ Y . Indeed otherwise the nonempty
set
Q = ¦(a, b) ∈ X Y : a , Eb¦
produces (by Lemma 30.3) a pair (a, b) ∈ Q such that a ∈ X and b ∈ Y are
Pgeneric. Contradiction with item 5.
31 BOREL METRIC SPACES AND LINES IN THE PLANE 119
31 Borel metric spaces and lines in the plane
We give two applications of Harrington’s technique of using Gandy forcing.
First let us begin by isolating a principal which we call overﬂow. It is an
easy consequence of the Separation Theorem.
Lemma 31.1 (Overﬂow) Suppose θ(x
1
, x
2
, . . . , x
n
) is a Π
1
1
formula and A
is a Σ
1
1
set such that
∀x
1
, . . . , x
n
∈ A θ(x
1
, . . . , x
n
).
Then there exists a ∆
1
1
set D ⊇ A such that
∀x
1
, . . . , x
n
∈ D θ(x
1
, . . . , x
n
).
Proof:
For n = 1 this is just the Separation Theorem 27.5.
For n = 2 deﬁne
B = ¦x : ∀y(y ∈ A → θ(x, y))¦.
Then B is Π
1
1
set which contains A. Hence by separation there exists a ∆
1
1
set E with A ⊆ E ⊆ B. Now deﬁne
C = ¦x : ∀y(y ∈ E → θ(x, y)¦.
Then C is a Π
1
1
set which also contains A. By applying separation again we
get a ∆
1
1
set F with A ⊆ F ⊆ C. Letting D = E ∩ F does the job. The
proof for n > 2 is similar.
We say that (B, δ) is a Borel metric space iﬀ B is Borel, δ is a metric
on B, and for every ∈ Q the set
¦(x, y) ∈ B
2
: δ(x, y) ≤ ¦
is Borel.
Theorem 31.2 (Harrington [39]) If (B, δ) is a Borel metric space, then
either (B, δ) is separable (i.e., contains a countable dense set) or for some
> 0 there exists a perfect set P ⊆ B such that δ(x, y) > for every distinct
x, y ∈ P.
Descriptive Set Theory and Forcing 120
Proof:
By relativizing the proof to an arbitrary parameter we may assume that
B and the sets ¦(x, y) ∈ B
2
: δ(x, y) ≤ ¦ are ∆
1
1
.
Lemma 31.3 For any ∈ Q
+
if A ⊆ B is Σ
1
1
and the diameter of A is less
than , then there exists a ∆
1
1
set D with diameter less than and A ⊆ D ⊆ B.
Proof:
This follows from Lemma 31.1, since
θ(x, y) iﬀ δ(x, y) < and x, y ∈ B
is a Π
1
1
formula.
For any ∈ Q
+
look at
Q
=
_
¦D ∈ ∆
1
1
: D ⊆ B and diam(D) < ¦.
Note that Q is a Π
1
1
set. If for every ∈ Q
+
Q
= B, then since there are
only countably many ∆
1
1
sets, (B, δ) is separable and we are done. On the
other hand suppose for some ∈ Q
+
we have that
P
= B ¸ Q
,= ∅.
Lemma 31.4 For every c ∈ V ∩ B
P
[¬ δ(
◦
a, ˇ c) > /3
where [¬ is Gandy forcing and
◦
a is a name for the generic real (see Lemma
30.2).
Proof:
Suppose not. Then there exists P ≤ P
such that
P [¬ δ(a, c) ≤ /3.
Since P is disjoint from Q
by Lemma 31.3 we know that the diameter of P
is ≥ . Let
R = ¦(a
0
, a
1
) : a
0
, a
1
∈ P and δ(a
0
, a
1
) > (2/3)¦.
Descriptive Set Theory and Forcing 121
Then R is in P and by Lemma 30.3, if a is Pgeneric over V with a ∈ R, then
a
0
and a
1
are each separately Pgeneric over V . But a
0
∈ R and a
1
∈ R means
that δ(a
0
, c) ≤ /3 and δ(a
1
, c) ≤ /3. But by absoluteness δ(a
0
, a
1
) > (2/3).
This contradicts the fact that δ must remain a metric by absoluteness.
Using this lemma and Lemma 30.6 is now easy to get a perfect set P ⊆ B
such that δ(x, y) > /3 for each distinct x, y ∈ P. This proves Theorem 31.2.
Theorem 31.5 (van Engelen, Kunen, Miller [20]) For any Σ
1
1
¯
set A in the
plane, either A can be covered by countably many lines or there exists a perfect
set P ⊆ A such that no three points of P are collinear.
Proof:
This existence of this proof was pointed out to me by Dougherty, Jackson,
and Kechris. The proof in [20] is more elementary.
By relativizing the proof we may as well assume that A is Σ
1
1
.
Lemma 31.6 Suppose B is a Σ
1
1
set lying on a line in the plane. Then there
exists a ∆
1
1
set D with B ⊆ D such that all points of D are collinear.
Proof:
This follows from Lemma 31.1 since
θ(x, y, z) iﬀ x, y, and z are collinear
is Π
1
1
(even Π
0
1
).
Deﬁne
∼ P =
_
¦D ⊆ R
2
: D ∈ ∆
1
1
and all points of D are collinear¦.
It is clear that ∼ P is Π
1
1
and therefore P is Σ
1
1
. If P ∩ A = ∅, then A can
be covered by countably many lines.
So assume that
Q = P ∩ A ,= ∅.
For any two distinct points in the plane, p and q, let line(p, q) be the unique
line on which they lie.
Descriptive Set Theory and Forcing 122
Lemma 31.7 For any two distinct points in the plane, p and q, with p, q ∈ V
Q [¬
◦
a/ ∈ line(ˇ p, ˇ q).
Proof:
Suppose for contradiction that there exists R ≤ Q such that
R [¬
◦
a∈ line(ˇ p, ˇ q).
Since R is disjoint from
_
¦D ⊆ R
2
: D ∈ ∆
1
1
and all points of D are collinear¦
it follows from Lemma 31.6 that not all triples of points from R are collinear.
Deﬁne the nonempty Σ
1
1
set
S = ¦a : a
0
, a
1
, a
2
∈ R and a
0
, a
1
, a
2
are not collinear¦
where a = (a
0
, a
1
, a
2
) via some standard tripling function. Then S ∈ P and
by the obvious generalization of Lemma 30.3 each of the a
i
is Pgeneric if
a is. But this is a contradiction since all a
i
∈ line(p, q) which makes them
collinear.
The following Lemma is an easy generalization of Lemma 30.6 so we leave
the proof to the reader.
Lemma 31.8 Suppose M is a countable transitive model of ZFC
∗
and P is
a partially ordered set in M. Then there exists ¦G
x
: x ∈ 2
ω
¦, a “perfect”
set of Pﬁlters, such that for every x, y, z distinct, we have that (G
x
, G
y
, G
z
)
is P P Pgeneric over M.
Using Lemma 31.7 and 31.8 it is easy to get (just as in the proof of
Theorem 30.1) a perfect set of triply generic points in the plane, hence no
three of which are collinear. This proves Theorem 31.5.
Obvious generalizations of Theorem 31.5 are:
1. Any Σ
1
1
¯
subset of R
n
which cannot be covered by countably many lines
contains a perfect set all of whose points are collinear.
Descriptive Set Theory and Forcing 123
2. Any Σ
1
1
¯
subset of R
2
which cannot be covered by countably many circles
contains a perfect set which does not contain four points on the same
circle.
3. Any Σ
1
1
¯
subset of R
2
which cannot be covered by countably many
parabolas contains a perfect set which does not contain four points on
the same parabola.
4. For any n any Σ
1
1
¯
subset of R
2
which cannot be covered by countably
many polynomials of degree < n contains a perfect set which does not
contain n + 1 points on the same polynomial of degree < n.
5. Higher dimensional version of the above involving spheres or other sur
faces.
A very general statement of this type is due to Solecki [102]. Given any
Polish space X, family of closed sets Q in X, and analytic A ⊆ X; either
A can be covered by countably many elements of Q or there exists a G
δ
set
B ⊆ A such that B cannot be covered by countably many elements of Q.
Solecki deduces Theorem 31.5 from this.
Another result of this type is known as the BorelDilworth Theorem.
It is due to Harrington [39]. It says that if P is a Borel partially ordered set,
then either P is the union of countably many chains or there exist a perfect
set P of pairwise incomparable elements. One of the early Lemmas from [39]
is the following:
Lemma 31.9 Suppose A is a Σ
1
1
chain in a ∆
1
1
poset P. Then there exists
a ∆
1
1
superset D ⊇ A which is a chain.
Proof:
Suppose P = (P, ≤) where P and ≤ are ∆
1
1
. Then
θ(x, y) iﬀ x, y ∈ P and (x ≤ y or y ≤ x)
is Π
1
1
and so the result follows by Lemma 31.1.
For more on Borel linear orders, see Louveau [67]. Louveau [68] is a survey
paper on Borel equivalence relations, linear orders, and partial orders.
Q.Feng [22] has shown that given an open partition of the two element
subsets of ω
ω
, that either ω
ω
is the union of countably many 0−homogenous
Descriptive Set Theory and Forcing 124
sets or there exists a perfect 1−homogeneous set. Todorcevic [111] has given
an example showing that this is false for Borel partitions (even replacing
open by closed).
32 Σ
1
1
EQUIVALENCE RELATIONS 125
32 Σ
1
1
equivalence relations
Theorem 32.1 (Burgess [14]) Suppose E is a Σ
1
1
¯
equivalence relation. Then
either E has ≤ ω
1
equivalence classes or there exists a perfect set of pairwise
Einequivalent reals.
Proof:
We will need to prove the boundedness theorem for this result. Deﬁne
WF = ¦T ⊆ ω
<ω
: Tis a wellfounded tree¦.
For α < ω
1
deﬁne WF
<α
to the subset of WF of all wellfounded trees of
rank < α. WF is a complete Π
1
1
set, i.e., for every B ⊆ ω
ω
which is Π
1
1
¯
there
exists a continuous map f such that f
−1
(WF) = B (see Theorem 17.4).
Consequently, WF is not Borel. On the other hand each of the WF
<α
are
Borel.
Lemma 32.2 For each α < ω
1
the set WF
<α
is Borel.
Proof:
Deﬁne for s ∈ ω
<ω
and α < ω
1
WF
s
<α
= ¦T ⊆ ω
<ω
: T is a tree, s ∈ T, r
T
(s) < α¦.
The fact that WF
s
<α
is Borel is proved by induction on α. The set of trees
is Π
0
1
. For λ a limit
WF
s
<λ
=
_
α<λ
WF
s
<α
.
For a successor α + 1
T ∈ WF
s
<α+1
iﬀ s ∈ T and ∀n (s
ˆ
n ∈ T → T ∈ WF
sˆn
<α
).
Another way to prove this is take a tree T of rank α and note that
WF
<α
= ¦
ˆ
T :
ˆ
T ≺ T ¦
and this set is ∆
1
1
¯
and hence Borel by Theorem 26.1.
Descriptive Set Theory and Forcing 126
Lemma 32.3 Boundedness Theorem If A ⊆ WF is Σ
1
1
¯
, then there exists
α < ω
1
such that A ⊆ WF
α
.
Proof:
Suppose no such α exists. Then
T ∈ WF iﬀ there exists
ˆ
T ∈ A such that T _
ˆ
T.
But this would give a Σ
1
1
¯
deﬁnition of WF, contradiction.
There is also a lightface version of the boundedness theorem, i.e., if A is
a Σ
1
1
subset of WF, then there exists a recursive ordinal α < ω
CK
1
such that
A ⊆ WF
<α
. Otherwise,
¦e ∈ ω : e is the code of a recursive wellfounded tree ¦
would be Σ
1
1
.
Now suppose that E is a Σ
1
1
¯
equivalence relation. By the Normal Form
Theorem 17.4 we know there exists a continuous mapping (x, y) → T
xy
such
that T
xy
is always a tree and
xEy iﬀ T
xy
/ ∈ WF.
Deﬁne
xE
α
y iﬀ T
xy
/ ∈ WF
<α
.
By Lemma 32.2 we know that the binary relation E
α
is Borel. Note that E
α
reﬁnes E
β
for α > β. Clearly,
E =
α<ω
1
E
α
and for any limit ordinal λ
E
λ
=
α<λ
E
α
.
While there is no reason to expect that any of the E
α
are equivalence
relations, we use the boundedness theorem to show that many are.
Lemma 32.4 For unboundedly many α < ω
1
the binary relation E
α
is an
equivalence relation.
Descriptive Set Theory and Forcing 127
Proof:
Note that every E
α
must be reﬂexive, since E is reﬂexive and E =
α<ω
1
E
α
.
The following claim will allow us to handle symmetry.
Claim: For every α < ω
1
there exists β < ω
1
such that for every x, y
if xE
α
y and y , E
α
x, then x , E
β
y.
Proof:
Let
A = ¦T
xy
: xE
α
y and y , E
α
x¦.
Then A is a Borel set. Since y , E
α
x implies y , Ex and hence x , Ey, it follows
that A ⊆ WF. By the Boundedness Theorem 32.3 there exists β < ω
1
such
that A ⊆ WF
<β
.
The next claim is to take care of transitivity.
Claim: For every α < ω
1
there exists β < ω
1
such that for every x, y, z
if xE
α
y and yE
α
z, and x , E
α
z, then either x , E
β
y or y , E
β
z.
Proof:
Let
B = ¦T
xy
⊕T
yz
: xE
α
y, yE
α
z, and x , E
α
z¦.
The operation ⊕ on a pair of trees T
0
and T
1
is deﬁned by
T
0
⊕T
1
= ¦(s, t) : s ∈ T
0
, t ∈ T
1
, and [s[ = [t[¦.
Note that the rank of T
0
⊕T
1
is the minimum of the rank of T
0
and the rank
of T
1
. (Deﬁne the rank function on T
0
⊕ T
1
by taking the minimum of the
rank functions on the two trees.)
The set B is Borel because the relation E
α
is. Note also that since
x ,E
α
z implies x ,Ez and E is an equivalence relation, then either x ,Ey
or y , Ez. It follows that either T
xy
∈ WF or T
yz
∈ WF and so in either case
T
xy
⊕ T
yz
∈ WF and so B ⊆ WF. Again, by the Boundedness Theorem
there is a β < ω
1
such that B ⊆ WF
<β
and this proves the Claim.
Now we use the Claims to prove the Lemma. Using the usual Lowenheim
Skolem argument we can ﬁnd arbitrarily large countable ordinals λ such that
Descriptive Set Theory and Forcing 128
for every α < λ there is a β < λ which satisﬁes both Claims for α. But this
means that E
λ
is an equivalence relation. For suppose xE
λ
y and y , E
λ
x. Then
since E
λ
=
α<λ
E
α
there must be α < λ such that xE
α
y and y , E
α
x. But by
the Claim there exist β < λ such that x , E
β
y and hence x , E
λ
y, a contradiction.
A similar argument using the second Claim works for transitivity.
Let G be any generic ﬁlter over V with the property that it collapses ω
1
but not ω
2
. For example, Levy forcing with ﬁnite partial functions from ω
to ω
1
(see Kunen [56] or Jech [44]). Then ω
V [G]
1
= ω
V
2
. By absoluteness, E is
still an equivalence relation and for any α if E
α
was an equivalence relation
in V , then it still is one in V [G]. Since
E
ω
V
1
=
α<ω
V
1
E
α
and the intersection of equivalence relations is an equivalence relation, it
follows that the Borel relation E
ω
V
1
is an equivalence relation. So now suppose
that E had more than ω
2
equivalence classes in V . Let Q be a set of size ω
2
in V of pairwise Einequivalent reals. Then Q has cardinality ω
1
in V [G] and
for every x ,= y ∈ Q there exists α < ω
V
1
with x , E
α
y. Hence it must be that
the elements of Q are in diﬀerent E
ω
V
1
equivalence classes. Consequently,
by Silver’s Theorem 30.1 there exists a perfect set P of E
ω
V
1
inequivalent
reals. Since in V [G] the equivalence relation E reﬁnes E
ω
V
1
, it must be that
the elements of P are pairwise Einequivalent also. The following is a Σ
1
2
¯
statement:
V [G] [= ∃P perfect ∀x∀y (x, y ∈ P and x ,= y) → x , Ey.
Hence, by Shoenﬁeld Absoluteness 20.2, V must think that there is a perfect
set of Einequivalent reals.
A way to avoid taking a generic extension of the universe is to suppose
Burgess’s Theorem is false. Then let M be the transitive collapse of an
elementary substructure of some suﬃciently large V
κ
(at least large enough
to know about absoluteness and Silver’s Theorem). Let M[G] be obtained as
in the above proof by Levy collapsing ω
M
1
. Then we can conclude as above
that M thinks E has a perfect set of inequivalent elements, which contradicts
the assumption that M thought Burgess’s Theorem was false.
By Harrington’s Theorem 25.1 it is consistent to have Π
1
2
¯
sets of arbitrary
cardinality, e.g it is possible to have c = ω
23
and there exists a Π
1
2
¯
set B with
Descriptive Set Theory and Forcing 129
[B[ = ω
17
. Hence, if we deﬁne
xEy iﬀ x = y or x, y / ∈ B
then we get Σ
1
2
¯
equivalence relation with exactly ω
17
equivalence classes, but
since the continuum is ω
23
there is no perfect set of Einequivalent reals.
See Burgess [15] [16] and Hjorth [41] for more results on analytic equiva
lence relations. For further results concerning projective equivalence relations
see Harrington and Sami [37], Sami [96], Stern [109] [110], Kechris [51], Har
rington and Shelah [38], Shelah [97], and Harrington, Marker, and Shelah
[39].
33 LOUVEAU’S THEOREM 130
33 Louveau’s Theorem
Let us deﬁne codes for Borel sets in our usual way of thinking of them as
trees with basic clopen sets attached to the terminal nodes.
Deﬁnitions
1. Deﬁne (T, q) is an αcode iﬀ T ⊆ ω
<ω
is a tree of rank ≤ α and
q : T
0
→ B is a map from the terminal nodes, T
0
, of T (i.e. rank zero
nodes) to a nice base, B, for the clopen sets of ω
ω
, say all sets of the
form [s] for s ∈ ω
<ω
plus the empty set.
2. Deﬁne S
s
(T, q) and P
s
(T, q) for s ∈ T by induction on the rank of s as
follows. For s ∈ T
0
deﬁne
P
s
(T, q) = q(s) and S
s
(T, q) =∼ q(s).
For s ∈ T
>0
deﬁne
P
s
(T, q) =
_
¦S(T, q)
sˆm
: sˆm ∈ T¦ and S
s
(T, q) =∼ P
s
(T, q).
3. Deﬁne
P(T, q) = P
(T, q) and S(T, q) = S
(T, q)
the Π
0
α
set and the Σ
0
α
set coded by (T, q), respectively. (S is short for
Sigma and P is short for Pi.)
4. Deﬁne C ⊆ ω
ω
is Π
0
α
(hyp) iﬀ it has an αcode which is hyperarithmetic.
5. ω
CK
1
is the ﬁrst nonrecursive ordinal.
Theorem 33.1 (Louveau [65]) If A, B ⊆ ω
ω
are Σ
1
1
sets, α < ω
CK
1
, and
A and B can be separated by Π
0
α
¯
set, then A and B can be separated by a
Π
0
α
(hyp)set.
Corollary 33.2 ∆
1
1
∩ Π
0
α
¯
= Π
0
α
(hyp)
Corollary 33.3 (Section Problem) If B ⊆ ω
ω
ω
ω
is Borel and α < ω
1
is such that B
x
∈ Σ
0
α
¯
for every x ∈ ω
ω
, then
B ∈ Σ
0
α
¯
(¦D C : D ∈ Borel(ω
ω
) and Cis clopen¦).
Descriptive Set Theory and Forcing 131
Note that the converse is trivial.
This result was proved by Dellecherie for α = 1 who conjectured it in gen
eral. SaintRaymond proved it for α = 2 and Louveau and SaintRaymond
independently proved it for α = 3 and then Louveau proved it in general. In
their paper [66] Louveau and SaintRaymond give a diﬀerent proof of it. We
will need the following lemma.
Lemma 33.4 For α < ω
CK
1
the following sets are ∆
1
1
:
¦y : y is a βcode for some β < α¦,
¦(x, y) : y is a βcode for some β < α and x ∈ P(T, q)¦ , and
¦(x, y) : y is a βcode for some β < α and x ∈ S(T, q)¦.
Proof:
For the ﬁrst set it is enough to see that WF
<α
the set of trees of rank < α
is ∆
1
1
. Let
ˆ
T be a recursive tree of rank α. Then T ∈ WF
<α
iﬀ T ≺
ˆ
T shows
that WF
<α
is Σ
1
1
. But since
ˆ
T is wellfounded T ≺
ˆ
T iﬀ (
ˆ
T _ T) and so
it is Π
1
1
. For the second set just use an argument similar to Theorem 27.3.
The third set is just the complement of the second one.
Now we prove Corollary 33.3 by induction on α. By relativizing the proof
to a parameter we may assume α < ω
CK
1
and that B is ∆
1
1
. By taking
complements we may assume that the result holds for Π
0
β
for all β < α.
Deﬁne
R(x, (T, q)) iﬀ (T, q) ∈ ∆
1
1
(x), (T, q) is an αcode, and P(T, q) = B
x
.
where P(T, q) is the Π
0
α
set coded by (T, q). Note that by the relativized
version of Louveau’s Theorem for every x there exists a (T, q) such that
R(x, (T, q)). By Π
1
1
uniformization (Theorem 22.1) there exist a Π
1
1
set
ˆ
R ⊆ R
such that for every x there exists a unique (T, q) such that
ˆ
R(x, (T, q)). Fix
β < α and n < ω and deﬁne
B
β,n
(x, z) iﬀ there exists (T, q) ∈ ∆
1
1
(x) such that
1.
ˆ
R(x, (T, q)),
2. rank
T
(¸n)) = β and
3. z ∈ P
n
(T, q).
Descriptive Set Theory and Forcing 132
Since quantiﬁcation over ∆
1
1
(x) preserves Π
1
1
(Theorem 29.3),
ˆ
R is Π
1
1
, and the
rest is ∆
1
1
by Lemma 33.4, we see that B
β,n
is Π
1
1
. But note that B
β,n
(x, z)
iﬀ there exists (T, q) ∈ ∆
1
1
(x) such that
1.
ˆ
R(x, (T, q)),
2. rank
T
(¸n)) ,= β, or
3. z ∈ S
n
(T, q).
and consequently, ∼ B
β,n
is Π
1
1
and therefore B
β,n
is ∆
1
1
. Note that every
cross section of B
β,n
is a Π
0
β
¯
set and so by induction (in case α > 1)
B
β,n
∈ Π
0
α
¯
(¦D C : D ∈ Borel(ω
ω
) and Cis clopen¦).
But then
B =
_
n<ω,β<α
B
β,n
and so
B ∈ Σ
0
α
¯
(¦D C : D ∈ Borel(ω
ω
) and Cis clopen¦).
Now to do the case for α = 1, deﬁne for every n ∈ ω and s ∈ ω
<ω
B
s,n
(x, z) iﬀ there exists (T, q) ∈ ∆
1
1
(x) such that
1.
ˆ
R(x, (T, q)),
2. rank
T
(¸n)) = 0,
3. q(¸n)) = s, and
4. z ∈ [s].
As in the other case B
s,n
is ∆
1
1
. Let z
0
∈ [s] be arbitrary, then deﬁne the
Borel set C
s,n
= ¦x : (x, z
0
) ∈ B
s,n
¦. Then B
s,n
= C
s,n
[s] where But now
B =
_
n<ω,s∈ω
<ω
B
s,n
and so
B ∈ Σ
0
1
¯
(¦D C : D ∈ Borel(ω
ω
) and C clopen¦).
Descriptive Set Theory and Forcing 133
Note that for every α < ω
1
there exists a Π
1
1
¯
set U which is universal for
all ∆
0
α
¯
sets, i.e., every cross section of U is ∆
0
α
¯
and every ∆
0
α
¯
set occurs as a
cross section of U. To see this, let V be a Π
0
α
set which is universal for Π
0
α
¯
sets. Now put
(x, y) ∈ U iﬀ y ∈ V
x
0
and ∀z(z ∈ V
x
0
iﬀ z / ∈ V
x
1
)
where x = (x
0
, x
1
) is some standard pairing function. Note also that the
complement of U is also universal for all ∆
0
α
¯
sets, so there is a Σ
1
1
¯
which is
universal for all ∆
0
α
¯
sets. Louveau’s Theorem implies that there can be no
Borel set universal for all ∆
0
α
¯
sets.
Corollary 33.5 There can be no Borel set universal for all ∆
0
α
¯
sets.
In order to prove this corollary we will need the following lemmas. A
space is Polish iﬀ it is a separable complete metric space.
Lemma 33.6 If X is a 0dimensional Polish space, then there exists a closed
set Y ⊆ ω
ω
such that X and Y are homeomorphic.
Proof:
Build a tree ¸C
s
: s ∈ T) of nonempty clopen sets indexed by a tree
T ⊆ ω
<ω
such that
1. C
= X,
2. the diameter of C
s
is less that 1/[s[ for s ,= ¸), and
3. for each s ∈ T the clopen set C
s
is the disjoint union of the clopen sets
¦C
sˆn
: sˆn ∈ T¦.
If Y = [T] (the inﬁnite branch of T), then X and Y are homeomorphic.
I am not sure who proved this ﬁrst. I think the argument for the next
lemma comes from a theorem about Hausdorﬀ that lifts the diﬀerence hier
archy on the ∆
0
2
¯
sets to the ∆
0
α
¯
sets. This presentation is taken from Kechris
[54] mutatis mutandis.
13
13
Latin for plagiarized.
Descriptive Set Theory and Forcing 134
Lemma 33.7 For any sequence ¸B
n
: n ∈ ω) of Borel subsets of ω
ω
there
exists 0dimensional Polish topology, τ, which contains the standard topology
and each B
n
is a clopen set in τ.
Proof:
This will follow easily from the next two claims.
Claim: Suppose (X, τ) is a 0dimensional Polish space and F ⊆ X is closed,
then there exists a 0dimensional Polish topology σ ⊇ τ such that F is clopen
in (X, σ). (In fact, τ ∪ ¦F¦ is a subbase for σ.)
Proof:
Let X
0
be F with the subspace topology given by τ and X
1
be ∼ F with
the subspace topology. Since X
0
is closed in X the complete metric on X is
complete when restricted to X
0
. Since ∼ F is open there is another metric
which is complete on X
1
. This is a special case of Alexandroﬀ’s Theorem
which says that a G
δ
set in a completely metrizable space is completely
metrizable in the subspace topology. In this case the complete metric
ˆ
d on
∼ F would be deﬁned by
ˆ
d(x, y) = d(x, y) +
¸
¸
¸
¸
1
d(x, F)
−
1
d(y, F)
¸
¸
¸
¸
where d is a complete metric on X and d(x, F) is the distance from x to the
closed set F.
Let
(X, σ) = X
0
⊕X
1
be the discrete topological sum, i.e., U is open iﬀ U = U
0
∪U
1
where U
0
⊆ X
0
is open in X
0
and U
1
⊆ X
1
is open in X
1
.
Claim: If (X, τ) is a Hausdorﬀ space and (X, τ
n
) for n ∈ ω are 0dimensional
Polish topologies extending τ, then there exists a 0dimensional Polish topol
ogy (X, σ) such that τ
n
⊆ σ for every n. (In fact
n<ω
τ
n
is a subbase for
σ.)
Proof:
Consider the 0dimensional Polish space
n∈ω
(X, τ
n
).
Descriptive Set Theory and Forcing 135
Let f : X →
n∈ω
(X, τ
n
) be the embedding which takes each x ∈ X to the
constant sequence x (i.e., f(x) = ¸x
n
: n ∈ ω) where x
n
= x for every n).
Let D ⊆
n∈ω
(X, τ
n
) be the range of f, the set of constant sequences. Note
that f : (X, τ) → (D, τ) is a homeomorphism. Let σ be the topology on X
deﬁned by
U ∈ σ iﬀ there exists V open in
n∈ω
(X, τ
n
) with U = f
−1
(V ).
Since each τ
n
extends τ we get that D is a closed subset of
n∈ω
(X, τ
n
).
Consequently, D with the subspace topology inherited from
n∈ω
(X, τ
n
) is
Polish. It follows that σ is a Polish topology on X. To see that τ
n
⊆ σ for
every n let U ∈ τ
N
and deﬁne
V =
n<N
X U
n>N
X.
Then f
−1
(V ) = U and so U ∈ σ.
We prove Lemma 33.7 by induction on the rank of the Borel sets. Note
that by the second Claim it is enough to prove it for one Borel set at a time.
So suppose B is a Σ
0
α
¯
subset of (X, τ). Let B =
n∈ω
B
n
where each B
n
is
Π
0
β
¯
for some β < α. By induction on α there exists a 0dimensional Polish
topology τ
n
extending τ in which each B
n
is clopen. Applying the second
Claim gives us a 0dimensional topology σ extending τ in which each B
n
is
clopen and therefore B is open. Apply the ﬁrst Claim to get a 0dimensional
Polish topology in which B is clopen.
Proof:
(of Corollary 33.5). The idea of this proof is to reduce it to the case of a
∆
0
α
¯
set universal for ∆
0
α
¯
 sets, which is easily seen to be impossible by the
standard diagonal argument.
Suppose B is a Borel set which is universal for all ∆
0
α
¯
sets. Then by the
Corollary 33.3
B ∈ ∆
0
α
(¦D C : D ∈ Borel(ω
ω
) and C is clopen¦).
By Lemma 33.7 there exists a 0dimensional Polish topology τ such that if
X = (ω
ω
, τ)
Descriptive Set Theory and Forcing 136
then B is ∆
0
α
¯
(Xω
ω
). Now by Lemma 33.6 there exists a closed set Y ⊆ ω
ω
and a homeomorphism h : X → Y . Consider
C = ¦(x, y) ∈ X X : (x, h(y)) ∈ B¦.
The set C is ∆
0
α
¯
in XX because it is the continuous preimage of the set B
under the map (x, y) → (x, h(y)). The set C is also universal for ∆
0
α
¯
subsets
of X because the set Y is closed. To see this for α > 1 if H ∈ ∆
0
α
¯
(Y ), then
H ∈ ∆
0
α
¯
(ω
ω
), consequently there exists x ∈ X with B
x
= H. For α = 1 just
use that disjoint closed subsets of ω
ω
can be separated by clopen sets.
Finally, the set C gives a contradiction by the usual diagonal argument:
D = ¦(x, x) : x / ∈ C¦
would be ∆
0
α
¯
in X but cannot be a cross section of C.
Question 33.8 (Mauldin) Does there exists a Π
1
1
set which is universal for
all Π
1
1
¯
sets which are not Borel?
14
We could also ask for the complexity of a set which is universal for Σ
0
α
¯
¸∆
0
α
¯
sets.
14
This was answered by Greg Hjorth [42], who showed it is independent.
34 PROOF OF LOUVEAU’S THEOREM 137
34 Proof of Louveau’s Theorem
Finally, we arrive at our last section. The following summarizes how I feel
now.
You are walking down the street minding your own business and
someone stops you and asks directions. Where’s xxx hall? You
don’t know and you say you don’t know. Then they point at the
next street and say: Is that xxx street? Well by this time you feel
kind of stupid so you say, yea yea that’s xxx street, even though
you haven’t got the slightest idea whether it is or not. After all,
who wants to admit they don’t know where they are going or
where they are.
For α < ω
CK
1
deﬁne D ⊆ ω
ω
is Σ
0
α
(semihyp) iﬀ there exists S a Π
1
1
set
of hyperarithmetic reals such that every element of S is a βcode for some
β < α and
D =
_
¦P(T, q) : (T, q) ∈ S¦.
A set is Π
0
α
(semihyp) iﬀ it is the complement of a Σ
0
α
(semihyp) set. The
Π
0
0
(semihyp) sets are just the usual clopen basis ([s] for s ∈ ω
<ω
together
with the empty set) and Σ
0
0
(semihyp) sets are their complements.
Lemma 34.1 Σ
0
α
(semihyp) sets are Π
1
1
and consequently Π
0
α
(semihyp) sets
are Σ
1
1
.
Proof:
x ∈
¦P(T, q) : (T, q) ∈ S¦ iﬀ there exists (T, q) ∈ ∆
1
1
such that (T, q) ∈
S and x ∈ P(T, q). Quantiﬁcation over ∆
1
1
preserves Π
1
1
( see Corollary 29.3
) and Lemma 33.4 implies that “x ∈ P(T, q)” is ∆
1
1
.
We will need the following reﬂection principle in order to prove the Main
Lemma 34.3.
A predicate Φ ⊆ P(ω) is called Π
1
1
on Π
1
1
iﬀ for any Π
1
1
set N ⊆ ω ω
the set ¦e : Φ(N
e
)¦ is Π
1
1
(where N
e
= ¦n : (e, n) ∈ N¦).
Lemma 34.2 (Harrington [39] Kechris [50]) Π
1
1
Reﬂection. Suppose Φ(X)
is Π
1
1
on Π
1
1
and Q is a Π
1
1
set.
If Φ(Q), then there exists a ∆
1
1
set D ⊆ Q such that Φ(D).
Descriptive Set Theory and Forcing 138
Proof:
By the normal form theorem 17.4 there is a recursive mapping e → T
e
such that e ∈ Q iﬀ T
e
is wellfounded. Deﬁne for e ∈ ω
N
0
e
= ¦ˆ e : T
ˆ e
_ T
e
¦
N
1
e
= ¦ˆ e : (T
e
≺ T
ˆ e
)¦
then N
0
is Σ
1
1
and N
1
is Π
1
1
. For e ∈ Q we have N
0
e
= N
1
e
= D
e
⊆ Q is ∆
1
1
;
and for e / ∈ Q we have that N
1
e
= Q. If we assume for contradiction that
Φ(N
1
e
) for all e ∈ Q, then
e / ∈ Q iﬀ φ(N
1
e
).
But this would mean that Q is ∆
1
1
and this proves the Lemma.
Note that a Π
1
1
predicate need not be Π
1
1
on Π
1
1
since the predicate
Φ(X) = “0 / ∈ X”
is ∆
0
0
but not Π
1
1
on Π
1
1
. Some examples of Π
1
1
on Π
1
1
predicates Φ(X) are
Φ(X) iﬀ ∀x / ∈ X θ(x)
or
Φ(X) iﬀ ∀x, y / ∈ X θ(x, y)
where θ is a Π
1
1
sentence.
Lemma 34.3 Suppose A is Σ
1
1
and A ⊆ B ∈ Σ
0
α
(semihyp), then there exists
C ∈ Σ
0
α
(hyp) with A ⊆ C ⊆ B.
Proof:
Let B =
¦P(T, q) : (T, q) ∈ S¦ where S is a Π
1
1
set of hyperarithmetic
< αcodes. Let
ˆ
S ⊆ ω be the Π
1
1
set of ∆
1
1
codes for elements of S, i.e.
e ∈
ˆ
S iﬀ e is a ∆
1
1
code for (T
e
, q
e
) and (T
e
, q
e
) ∈ S.
Now deﬁne the predicate Φ(X) for X ⊆ ω as follows:
Φ(X) iﬀ X ⊆
ˆ
S and A ⊆
e∈X
P(T
e
, q
e
).
Descriptive Set Theory and Forcing 139
The predicate Φ(X) is Π
1
1
on Π
1
1
and Φ(
ˆ
S). Therefore by reﬂection
(Lemma 34.2) there exists a ∆
1
1
set D ⊆
ˆ
S such that Φ(D). Deﬁne (T, q) by
T = ¦eˆs : e ∈ D and s ∈ T
e
¦ q(eˆs) = q
e
(s) for e ∈ D and s ∈ T
0
e
.
Since D is ∆
1
1
it is easy to check that (T, q) is ∆
1
1
and hence hyperarithmetic.
Since Φ(D) holds it follows that C = S(T, q) the Σ
0
α
(hyp) set coded by (T, q)
has the property that A ⊆ C and since D ⊆
ˆ
S it follows that C ⊆ B.
Deﬁne for α < ω
CK
1
the αtopology by taking for basic open sets the
family
_
¦Π
0
β
(semihyp) : β < α¦.
As usual, cl
α
(A) denotes the closure of the set A in the αtopology.
The 1topology is just the standard topology on ω
ω
. The αtopology has
its basis certain special Σ
1
1
sets so it is intermediate between the standard
topology and the Gandy topology corresponding to Gandy forcing.
Lemma 34.4 If A is Σ
1
1
, then cl
α
(A) is Π
0
α
(semihyp).
Proof:
Since the Σ
0
β
(semihyp) sets for β < α form a basis for the αclosed sets,
cl
α
(A) =
¦X ⊇ A : ∃β < α X ∈ Σ
0
β
(semihyp)¦.
By Lemma 34.3 this same intersection can be written:
cl
α
(A) =
¦X ⊇ A : ∃β < α X ∈ Σ
0
β
(hyp)¦.
But now deﬁne (T, q) ∈ Q iﬀ (T, q) ∈ ∆
1
1
, (T, q) is a βcode for some β < α,
and A ⊆ S(T, q). Note that Q is a Π
1
1
set and consequently, cl
α
(A) is a
Π
0
α
(semihyp) set, as desired.
Note that it follows from the Lemmas that for A a Σ
1
1
set, cl
α
(A) is a Σ
1
1
set which is a basic open set in the βtopology for any β > α.
Let P be Gandy forcing, i.e., the partial order of all nonempty Σ
1
1
subsets
of ω
ω
and let
◦
a be a name for the real obtained by forcing with P, so that by
Lemma 30.2, for any G which is Pgeneric, we have that p ∈ G iﬀ a
G
∈ p.
Descriptive Set Theory and Forcing 140
Lemma 34.5 For any α < ω
CK
1
, p ∈ P, and C ∈ Π
0
α
¯
(coded in V ) if
p [¬
◦
a∈ C,
then
cl
α
(p) [¬
◦
a∈ C.
Proof:
This is proved by induction on α.
For α = 1 recall that the αtopology is the standard topology and C
is a standard closed set. If p [¬
◦
a∈ C, then it better be that p ⊆ C, else
there exists s ∈ ω
<ω
with q = p ∩ [s] nonempty and [s] ∩ C = ∅. But then
q ≤ p and q [¬
◦
a/ ∈ C. Hence p ⊆ C and since C is closed, cl(p) ⊆ C. Since
cl(p) [¬ a ∈ cl(p), it follows that cl(p) [¬ a ∈ C.
For α > 1 let
C =
n<ω
∼ C
n
where each C
n
is Π
0
β
¯
for some β < α. Suppose for contradiction that
cl
α
(p) ,[¬
◦
a∈ C
Then for some n < ω and r ≤ cl
α
(p) it must be that
r [¬
◦
a∈ C
n
.
Suppose that C
n
is Π
0
β
¯
for some β < α. Then by induction
cl
β
(r) [¬
◦
a∈ C
n
.
But cl
β
(r) is a Π
0
β
(semihyp) set by Lemma 34.4 and hence a basic open
set in the αtopology. Note that since they force contradictory information
(cl
β
(r) [¬
◦
a/ ∈ C and p [¬
◦
a∈ C) it must be that cl
β
(r) ∩ p = ∅, (otherwise the
two conditions would be compatible in P). But since cl
β
(r) is αopen this
means that
cl
β
(r) ∩ cl
α
(p) = ∅
which contradicts the fact that r ≤ cl
α
(p).
Now we are ready to prove Louveau’s Theorem 33.1. Suppose A and B
are Σ
1
1
sets and C is a Π
0
α
¯
set with A ⊆ C and C ∩ B = ∅. Since A ⊆ C it
follows that
A [¬
◦
a∈ C.
Descriptive Set Theory and Forcing 141
By Lemma 34.5 it follows that
cl
α
(A) [¬
◦
a∈ C.
Now it must be that cl
α
(A) ∩B = ∅, otherwise letting p = cl
α
(A) ∩B would
be a condition of P such that
p [¬
◦
a∈ C
and
p [¬
◦
a∈ B
which would imply that B ∩ C ,= ∅ in the generic extension. But by ab
soluteness B and C must remain disjoint. So cl
α
(A) is a Π
α
(semihyp)set
(Lemma 34.4) which is disjoint from the set B and thus by applying Lemma
34.3 to its complement there exists a Π
0
α
(hyp)set C with cl
α
(A) ⊆ C and
C ∩ B = ∅.
The argument presented here is partially from Harrington [34], but con
tains even more simpliﬁcation brought about by using forcing and abso
luteness. Louveau’s Theorem is also proved in Sacks [95], Mansﬁeld and
Weitkamp [73] and Kanovei [48]. For a generalization to higher levels of the
projective hierarchy using determinacy, see Hjorth [43].
Elephant Sandwiches
A man walks by a restaurant. Splashed all over are signs saying “Order
any sandwich”, “Just ask us, we have it”, and “All kinds of sandwiches”.
Intrigued, he walks in and says to the proprietor, “I would like an elephant
sandwich.”
The proprietor responds “Sorry, but you can’t have an elephant sand
wich.”
“What do you mean?” says the man, “All your signs say to order any
sandwich. And here the ﬁrst thing I ask for, you don’t have.”
Says the proprietor “Oh we have elephant. Its just that here it is 5pm
already and I just don’t want to start another elephant.”
REFERENCES 142
References
[1] J.W.Addison, Separation principles in the hierarchies of classical and
eﬀective descriptive set theory, Fundamenta Mathematicae, 46(1959),
123135. Cited on page: 9
[2] J.W.Addison, Some consequences of the axiom of constructibility, Fun
damenta Mathematicae, 46(1957), 337357. Cited on page: 82
[3] R.Baire, Sur la th´eorie des ensembles, Comptes Rendus Academie Sci
ence Paris, 129(1899), 946949. Cited on page: 3
[4] R.Baire, Sur la repr´esentation des fonctions discontinues (2me partie),
Acta Mathematica, 32(1909), 97176. Cited on page: 1
[5] Z.Balogh, There is a Qset space in ZFC, Proceedings of the American
Mathematical Society, 113(1991), 557561. Cited on page: 17
[6] T.Bartoszynski, Combinatorial aspects of measure and category, Fun
damenta Mathematicae, 127(1987), 225239. Cited on page: 77
[7] T.Bartoszynski, On covering of real line by null sets, Paciﬁc Journal of
Mathematics, 131(1988),112. Cited on page: 75
[8] T.Bartoszynski, H.Judah, S.Shelah, The coﬁnality of cardinal invariants
related to measure and category, Journal of Symbolic Logic, 54(1989),
719726. Cited on page: 75
[9] T.Bartoszynski, H.Judah, On the coﬁnality of the smallest covering of
the real line by meager sets, Journal of Symbolic Logic, 54(1989), 828
832. Cited on page: 75
[10] J.Barwise, Admissible sets and structures, SpringerVerlag, 1975.
Cited on page: 106
[11] M.Bell, On the combinatorial principal P(c), Fundamenta Mathemati
cae, 114(1981), 149157. Cited on page: 17
[12] R.H.Bing, W.W.Bledsoe, R.D.Mauldin, Sets generated by rectangles,
Paciﬁc J. Math., 51(1974, 2736. Cited on page: 10
REFERENCES 143
[13] C.Brezinski, History of Continued Fractions and Pad´e Approxi
mants, SpringerVerlag, 1991. Cited on page: 3
[14] J.Burgess, Inﬁnitary languages and descriptive set theory, PhD Univer
sity of California, Berkeley, 1974. Cited on page: 125
[15] J.Burgess, Equivalences generated by families of Borel sets, Proceed
ings of the American Mathematical Society, 69(1978), 323326. Cited on
page: 129
[16] J.Burgess, Eﬀective enumeration of classes in a Σ
1
1
equivalence relation,
Indiana University Mathematics Journal, 28(1979), 353364. Cited on
page: 129
[17] C.C.Chang, H.J.Keisler, Model Theory, North Holland, 1973. Cited
on page: 0
[18] F.van Engelen, A.Miller, J.Steel, Rigid Borel sets and better quasiorder
theory, Contemporary Mathematics (American Mathematical Society),
65(1987), 199222. Cited on page: 112
[19] L.Euler, De fractionibus continuis dissertatio, Comm. Acad. Sc. Imp.
St. P´etersbourg, 9(1737), 98137. Cited on page: 3
[20] F.van Engelen, K.Kunen, A.Miller, Two remarks about analytic sets,
Lecture Notes in Mathematics, SpringerVerlag, 1401(1989), 6872.
Cited on page: 121
[21] P.Erdos, K.Kunen, R.Mauldin, Some additive properties of sets of real
numbers, Fundamenta Mathematicae, 113(1981), 187199. Cited on
page: 112
[22] Q.Feng, Homogeneity for open partitions of pairs of reals, Transactions
of the American Mathematical Society, 339(1993), 659684. Cited on
page: 123
[23] W.Fleissner, A.Miller, On Qsets, Proceedings of the American Mathe
matical Society, 78(1980), 280284. Cited on page: 17
[24] W.Fleissner, Current research on Q sets, in Topology, Vol. I (Proc.
Fourth Colloq., Budapest, 1978), Colloq. Math. Soc. Janos Bolyai,
23(1980), 413431. Cited on page: 17
REFERENCES 144
[25] W.Fleissner, Squares of Q sets, Fundamenta Mathematicae, 118(1983),
223231. Cited on page: 17
[26] D.H.Fremlin, On the Baire order problem, Note of 29.4.82. Cited on
page: 51
[27] D.H.Fremlin, Consequences of Martin’s Axiom, Cambridge Univer
sity Press, 1984. Cited on page: 17 77
[28] H.Friedman, PCA wellorderings of the line, Journal of Symbolic Logic,
39(1974), 7980. Cited on page: 97
[29] S.Friedman, Steel Forcing and Barwise compactness, Annals of Mathe
matical Logic, 22(1982), 3146. Cited on page: 27
[30] F.Galvin, A.Miller, γsets and other singular sets of real numbers, Topol
ogy and Its Applications, 17(1984), 145155. Cited on page: 17 52
[31] R.O.Gandy, Proof of Mostowski’s conjecture, Bulletin de L’Academie
Polonaise des Science, 8 (1960), 571575. Cited on page: 109
[32] K.G¨odel, The consistency of the axiom of choice and of the generalized
continuum hypothesis, Proceedings of the National Academy of Science,
24(1938), 556557. Cited on page: 82
[33] L.Harrington, A powerless proof of a Theorem of Silver, handwritten
note dated 1176. Cited on page: 113
[34] L.Harrington, Theorem (A. Louveau), handwritten note undated, circa
1976. Cited on page: 141
[35] L.Harrington, Long projective wellorderings, Annals of Mathematical
Logic, 12(1977), 124. Cited on page: 98 101
[36] L.Harrington, Analytic determinacy and 0
#
, Journal of Symbolic Logic,
43(1978), 685693. Cited on page: 27
[37] L.Harrington, R.Sami, Equivalence relations, projective and beyond,
Logic Colloquium 78, ed Boﬀa et al, North Holland 1979, 247263. Cited
on page: 129
REFERENCES 145
[38] L.Harrington and S.Shelah, Counting equivalence classes for coκ
Souslin equivalence relations, in Logic Colloquium 80, North Holland,
1982, 147152. Cited on page: 113 129
[39] L.Harrington, D.Marker, S.Shelah, Borel orderings, Transactions of the
American Mathematical Society, 310(1988), 293302. Cited on page: 119
123 129 137
[40] P.Hinman, RecursionTheoretic Hierarchies, SpringerVerlag, 1978.
Cited on page: 106
[41] G.Hjorth, Thin equivalence relations and eﬀective decompositions, Jour
nal of Symbolic Logic, 58(1993), 11531164. Cited on page: 129
[42] G.Hjorth, Universal coanalytic sets, Proceedings of the American Math
ematical Society, 124(1996), 38673873. Cited on page: 136
[43] G.Hjorth, Two applications of inner model theory to the study of Σ
1
2
sets, Bulletin of Symbolic Logic, 2(1996), 94107. Cited on page: 141
[44] T.Jech, Set Theory, Academic Press, 1978. Cited on page: 0 16 28 40
128
[45] R.B.Jensen, R.M.Solovay, Some applications of almost disjoint sets, in
Mathematical Logic and Foundations of Set Theory, ed by Y.Bar
Hillel, 1970, North Holland, 84104. Cited on page: 16
[46] H.Judah, S.Shelah, Qsets do not necessarily have strong measure zero,
Proceedings of the American Mathematical Society, 102 (1988), 681683.
Cited on page: 17 58
[47] H.Judah, S.Shelah, Qsets, Sierpi´ nski sets, and rapid ﬁlters, Proceedings
of the American Mathematical Society, 111(1991), 821832. Cited on
page: 17
[48] V.G.Kanove˘ı, Topologies generated by eﬀectively Suslin sets and their
applications in descriptive set theory, Russian Mathematical Surveys,
51(1996), 385417. Cited on page: 118 141
[49] M.Kondˇo, Sur l’uniformisation des compl´ementaires analytiques et les
ensembles projectifs de la seconde classe, Japan J. Math. 15(1939), 197
230. Cited on page: 90
REFERENCES 146
[50] A. Kechris, Spector second order classes and reﬂection, in Generalized
recursion theory, II , North Holland 1978, Studies Logic Foundations
Math., 94(1978), 147183. Cited on page: 137
[51] A.Kechris, On transﬁnite sequences of projective sets with an appli
cation to Σ
1
2
equivalence relations, in Logic Colloquium ’77 ed by
Macintyre et al, North Holland, 1978, 155160. Cited on page: 129
[52] A. Kechris, The perfect set theorem and deﬁnable wellorderings of the
continuum, Journal of Symbolic Logic, 43(1978), 630634. Cited on page:
97
[53] A.S.Kechris, D.A.Martin, Inﬁnite games and eﬀective descriptive set
theory, in Analytic sets, ed by C.A.Rogers, Academic Press, 1980.
Cited on page: 113
[54] A.S.Kechris, Classical descriptive set theory, Graduate Texts in
Mathematics, 156, SpringerVerlag, New York, 1995. Cited on page: 0
133
[55] S.Kleene, Hierarchies of numbertheoretic predicates, Bulletin of the
American Mathematical Society, 63(1955), 193213. Cited on page: 104
[56] K.Kunen, Set Theory , North Holland, 1980. Cited on page: 0 16 19
39 40 128
[57] K. Kunen, A.Miller, Borel and projective sets from the point of view of
compact sets, Mathematical Proceedings of the Cambridge Philosophical
Society, 94(1983), 399409. Cited on page: 56
[58] K.Kunen, Where MA ﬁrst fails, Journal of Symbolic Logic, 53(1988),
429433. Cited on page: 77
[59] K.Kuratowski, Topology, vol 1, Academic Press, 1966. Cited on page:
0 78
[60] K.Kuratowski, A.Mostowski, Set theory, with an introduction to
descriptive set theory, North Holland, 1976. Cited on page: 0 94
[61] A.Landver, Singular σdense trees, Journal of Symbolic Logic, 57(1992),
14031416. Cited on page: 77
REFERENCES 147
[62] A.Landver, Baire numbers, uncountable Cohen sets and perfectset forc
ing, Journal of Symbolic Logic, 57(1992), 10861107. Cited on page: 77
[63] H.Lebesgue, Sur les fonctions repr´esentables analytiquement, Journal de
Math´ematiques Pures et Appliqu´es, 1(1905), 139216. Cited on page: 7
9
[64] A. Louveau, Relations d’equivalence dans les espaces polonais, Collec
tion: General logic seminar, (Paris, 198283), 113122, Publ. Math.
Univ. Paris VII. Cited on page: 113
[65] A.Louveau, A separation theorem for Σ
1
1
sets, Transactions of the Amer
ican Mathematical Society, 260(1980), 363378. Cited on page: 130
[66] A.Louveau, J.Saint Raymond, Borel classes and closed games: Wadge
type and Hurewicztype results, Transactions of the American Mathe
matical Society, 304(1987), 431467. Cited on page: 131
[67] A. Louveau, Two results on Borel orders, Journal of Symbolic Logic,
54(1989), 865874. Cited on page: 123
[68] A.Louveau, Classifying Borel Structures, in Set Theory of the Con
tinuum, ed by Judah et al., SpringerVerlag, 1992, 103112. Cited on
page: 123
[69] N.Luzin, Sur un probl`eme de M. Baire, Comptes Rendus Hebdomad
daires Seances Academie Science Paris, 158(1914), 12581261. Cited on
page: 35
[70] P.Mahlo,
¨
Uber Teilmengen des Kontinuums von dessen Machtigkeit,
Sitzungsberichte der Sachsischen Akademie der Wissenschaften zu Leip
siz, Mathematisch Naturwissenschaftliche Klasse 65, (1913), 283315.
Cited on page: 35 36
[71] R.Mansﬁeld, The nonexistence of Σ
1
2
wellordering of the Baire space,
Fundamenta Mathematicae, 86(1975), 279282. Cited on page: 97
[72] R.Mansﬁeld, Perfect subsets of deﬁnable sets of real numbers, Paciﬁc
Journal of Mathematics, 35(1970), 451457. Cited on page: 88
[73] R.Mansﬁeld, G.Weitkamp, Recursive Aspects of Descriptive Set
Theory, Oxford University Press (1985). Cited on page: 113 141
REFERENCES 148
[74] D.A.Martin, R.M.Solovay, Internal Cohen extensions, Annals of Math
ematical Logic, 2(1970), 143178. Cited on page: 16 38 95
[75] A.Miller, On the length of Borel hierarchies, Annals of Mathematical
Logic, 16(1979), 233267. Cited on page: 13 19 29 34 36 37 49 50 72
[76] A.Miller, On generating the category algebra and the Baire order prob
lem, Bulletin de L’Academie Polonaise des Science, 27(1979), 751755.
Cited on page: 43
[77] A.Miller, Generic Souslin sets, Paciﬁc Journal of Mathematics, 97(1981),
171181. Cited on page: 27
[78] A.Miller, Some properties of measure and category, Transactions of the
American Mathematical Society, 266(1981), 93114. Corrections and ad
ditions to some properties of measure and category, Transactions of the
American Mathematical Society, 271(1982), 347348. Cited on page: 72
[79] A.Miller, The Baire category theorem and cardinals of countable coﬁ
nality, Journal of Symbolic Logic, 47(1982), 275288. Cited on page: 73
75 76
[80] A.Miller, A characterization of the least cardinal for which the Baire cat
egory theorem fails, Proceedings of the American Mathematical Society,
86(1982), 498502. Cited on page: 77
[81] A.Miller, Mapping a set of reals onto the reals, Journal of Symbolic
Logic, 48(1983), 575584. Cited on page: 58 72
[82] A.Miller, The Borel classiﬁcation of the isomorphism class of a countable
model, Notre Dame Journal of Formal Logic, 24(1983), 2234. Cited on
page: 9
[83] A.Miller, Special subsets of the real line, in Handbook of set
theoretic topology, North Holland, 1984, 201233. Cited on page:
17
[84] A.Miller, K. Prikry, When the continuum has coﬁnality ω
1
, Paciﬁc Jour
nal of Mathematics, 115(1984), 399407. Cited on page: 42
[85] A.Miller, Inﬁnite combinatorics and deﬁnability, Annals of Pure and
Applied Mathematical Logic, 41(1989), 179203. Cited on page: 112
REFERENCES 149
[86] A.Miller, Projective subsets of separable metric spaces, Annals of Pure
and Applied Logic, 50(1990), 5369. Cited on page: 12
[87] A.Miller, Special sets of reals, in Set Theory of the Reals, ed Haim
Judah, Israel Mathematical Conference Proceedings, vol 6(1993), 415
432, American Math Society. Cited on page: 12 17
[88] Y.Moschovakis, Uniformization in a playful universe, Bulletin of the
American Mathematical Society, 77(1971), 731736. Cited on page: 94
[89] Y.Moschovakis, Descriptive set theory, North Holland 1980. Cited
on page: 0 78
[90] J.C.Oxtoby, Measure and category, SpringerVerlag, 1971. Cited on
page: 0 68
[91] G.Poprougenko, Sur un probl/‘eme de M. Mazurkiewicz, Fundamenta
Mathematicae, 15(1930), 284286. Cited on page: 65
[92] T.Przymusinski, The existence of Qsets is equivalent to the existence
of strong Qsets, Proceedings of the American Mathematical Society,
79(1980), 626628. Cited on page: 17
[93] C.A.Rogers et al, editors, Analytic Sets, Academic Press, 1980. Cited
on page: 78
[94] F.Rothberger, On some problems of Hausdorﬀ and of Sierpi´ nski, Fun
damenta Mathematicae, 35(1948), 2946. Cited on page: 16
[95] G.Sacks, Higher Recursion Theory, SpringerVerlag, 1990. Cited on
page: 141
[96] R.Sami, On Σ
1
1
equivalence relations with Borel classes of bounded rank,
Journal of Symbolic Logic, 49(1984), 12731283. Cited on page: 129
[97] S. Shelah, On coκSouslin relations, Israel Journal of Mathematics,
47(1984), 139153. Cited on page: 129
[98] J.Shoenﬁeld, The problem of predicativity, Essays on the foundations of
mathematics, Magnes press, Hebrew University, Jerusalem, 1961, 132
139. Cited on page: 86 87
REFERENCES 150
[99] W.Sierpi´ nski, Introduction to general topology, University of
Toronto Press, 1934, translated from original Polish version of 1928.
Cited on page: 3
[100] R. Sikorski, Boolean Algebras, SpringerVerlag New York, 1969, 3rd
edition. Cited on page: 33 35
[101] J.Silver, Counting the number of equivalence classes of Borel and co
analytic equivalence relations, Annals of Mathematical Logic, 18(1980),
128. Cited on page: 27 113
[102] S.Solecki, Covering analytic sets by families of closed sets, Covering an
alytic sets by families of closed sets, Journal of Symbolic Logic, 59(1994),
10221031. Cited on page: 123
[103] R.M.Solovay, On the cardinality of Σ
1
2
sets of reals, in Foundations of
Mathematics, Biulof et al eds., SpringerVerlag (1969), 5873. Cited
on page: 88 95
[104] R.M.Solovay, S.Tennenbaum, Iterated Cohen extensions and Souslin’s
problem, Annals of Mathematics, 94(1971), 201245. Cited on page: 40
[105] C.Spector, Hyperarithmetical quantiﬁers, Fundamenta Mathematicae,
48(1960), 313320. Cited on page: 109
[106] E.Szpilrajn, Sur l’´equivence des suites d’ensembles et l’´equivence des
fonctions, Fundamenta Mathematicae, 26(1936), 302326. Cited on page:
13
[107] E.Szpilrajn, The characteristic function of a sequence of sets and some
of its applications, Fundamenta Mathematicae, 31(1938), 207223. Cited
on page: 13
[108] J.R.Steel, Forcing with tagged trees, Annals of Mathematical Logic,
15(1978), 5574. Cited on page: 27
[109] J.Stern, Perfect set theorems for analytic and coanalytic equivalence
relations, in Logic Colloquium ’77 ed by Macintyre et al, North Hol
land, 1978, 277284. Cited on page: 129
REFERENCES 151
[110] J. Stern, Eﬀective partitions of the real line into Borel sets of bounded
rank, Annals of Mathematical Logic, 18 (1980), 2960. Cited on page:
129
[111] S.Todorcevic, Two examples of Borel partially ordered sets with the
countable chain condition, Proceedings of the American Mathematical
Society, 112(1991), 11251128. Cited on page: 124
[112] S.Williard, General Topology, AddisonWesley, 1970. Cited on page:
6
Index 152
(
α<ω
1
P
α
)∗
◦
Q 45
2
ω
1
A
α
53
A
<α
53
F/I 33
F
σ
5
G
δ
5
ILuzin set 35
L
∞
(P
α
: α < κ) 26
P(T, q) 130
Q
α
49
S(T, q) 130
T _
ˆ
T 92
T ≤
n
T
52
T ≺
ˆ
T 92
T
0
23
T
>0
23
T
α
49
WF 125
WF
<α
125
WO 113
[A]
I
33
[T] 22
[ω]
ω
2
[s] 1
∆
1
1
codes 109
∆
1
2
wellordering 82
∆
0
formulas 84
Π
0
α
(hyp) 130
Π
0
α
(semihyp) 137
Π
1
1
equivalence relations 113
Π
1
1
singleton 91
Π
1
1
Reduction 107
Π
1
1
Reﬂection 137
Π
0
1
78
Π
1
1
on Π
1
1
137
Π
β
sentence 26
Σ
0
α
(semihyp) 137
Σ
1
1
equivalence relations 125
Σ
1
1
78
Σ
1
1
(x) 78
Σ
1
2
82
Σ
1
2
= Σ
HC
1
84
Σ
1
formula 84
αcode 130
αforcing 23
αtopology 139
∆
0
α
¯
universal set 133
∆
0
α
¯
4
∆
0
α
¯
(F) 10
B
+
30
Π
0
α
¯
4
Π
0
α
¯
(F) 10
Π
1
1
¯
78
Π
1
2
¯
98
Σ
0
α
¯
4
Σ
0
α
¯
(F) 10
Σ
1
1
¯
78
Σ
1
2
¯
equivalence relation 129
ˇ x 16
cl
α
(A) 139
FIN(X, ω) 19
FIN(ℵ
ω
, 2) 75
FIN(c
+
, 2) 40
FIN(κ, 2) 49
ˆ
Q
α
60
κBorel 102
κSouslin 102
κSouslin 86
µ 65
◦
Un
16
ω
<ω
1
ω
ω
1
ω
CK
1
130
Index 153
ord(B) 29
P∗
◦
Q 40
P(T) 54
P
α
43
Q
α
60
σδlattice 6
σﬁeld 33
σideal 33
∼ B 4
α<ω
1
P
α
45
m
P
76
sˆn 1
s ⊂ t 92
x <
c
y 82
pp 17
Borel(F) 10
Borel(X) 4
Borel(X)/meager(X) 48
MA
κ
38
OR 22
cov(I) 73
cov(meager(2
ω
)) 73
meager(X) 49
ord(X) 8
rank(p) 43
stone(B) 33
[s[ 1
Aronszajn tree 52
Baire space 1
Borel metric space 119
BorelDilworth Theorem 123
Boundedness Theorem 126
Cantor space 1
Fusion 52
Gandy forcing 113
H
γ
39
HC 84
Kleene Separation Theorem 104
Louveau’s Theorem 130
MA
κ
(ctbl) 73
MansﬁeldSolovay Theorem 88
Martin’s Axiom 16
MartinSolovay Theorem 38
Mostowski’s Absoluteness 80
Normal form for Σ
1
1
80
Qsets 17
Sack’s real model 72
Section Problem 130
Shoenﬁeld Absoluteness 87
Sierpi´ nski set 65
Silver forcing 16
SouslinLuzin Separation 102
SpectorGandy Theorem 109
V=L 82
ZFC
∗
81
accumulation points 60
cBa 28
characteristic function 13
code for a hyperarithmetic set 104
collinear points 121
direct sum 45
ﬁeld of sets 10
hereditarily countable sets 84
hereditary order 59
hyperarithmetic subsets 104
nice αtree 22
perfect set forcing 72
perfect set 8
perfect tree 52
prewellorderings 91
prewellordering 107
rank function 22
scale property 90
second countable 1
separable 1
Index 154
separative 28
splitting node 52
super Luzin set 49
superILuzin 35
switcheroo 25
tree embedding 92
tree 22
two step iteration 39
uniformization property 90
universal for Σ
1
1
¯
sets 79
universal set 7
wellfounded 22
This page left blank.
Note to the readers Departing from the usual author’s statementI would like to say that I am not responsible for any of the mistakes in this document. Any mistakes here are the responsibility of the reader. If anybody wants to point out a mistake to me, I promise to respond by saying “but you know what I meant to say, don’t you?” These are lecture notes from a course I gave at the University of Wisconsin during the Spring semester of 1993. Some knowledge of forcing is assumed as well as a modicum of elementary Mathematical Logic, for example, the LowenheimSkolem Theorem. The students in my class had a one semester course, introduction to mathematical logic covering the completeness theorem and incompleteness theorem, a set theory course using Kunen [56], and a model theory course using Chang and Keisler [17]. Another good reference for set theory is Jech [44]. Oxtoby [90] is a good reference for the basic material concerning measure and category on the real line. Kuratowski [59] and Kuratowski and Mostowski [60] are excellent references for classical descriptive set theory. Moschovakis [89] and Kechris [54] are more modern treatments of descriptive set theory. The ﬁrst part is devoted to the general area of Borel hierarchies, a subject which has always interested me. The results in section 14 and 15 are new and answer questions from my thesis. I have also included (without permission) an unpublished result of Fremlin (Theorem 13.4). Part II is devoted to results concerning the low projective hierarchy. It ends with a theorem of Harrington from his thesis that is consistent to have Π1 sets of arbitrary size. 2 The general aim of part III and IV is to get to Louveau’s theorem. Along the way many of the classical theorems of descriptive set theory are presented “justintime” for when they are needed. This technology allows the reader to keep from overﬁlling his or her memory storage device. I think the proof given of Louveau’s Theorem 33.1 is also a little diﬀerent. 1 Questions like “Who proved what?” always interest me, so I have included my best guess here. Hopefully, I have managed to oﬀend a large number of
1 In a randomly inﬁnite Universe, any event occurring here and now with ﬁnite probability must be occurring simultaneously at an inﬁnite number of other sites in the Universe. It is hard to evaluate this idea any further, but one thing is certain: if it is true then it is certainly not original!– The Anthropic Cosmological Principle, by John Barrow and Frank Tipler.
AWM April 1995 Added April 2001: Several brave readers ignored my silly joke in the ﬁrst paragraph and sent me corrections and comments. . Since no kind act should go unpunished.mathematicians. let me say that any mistakes introduced into the text are their fault.
anyway? 1 I On the length of Borel hierarchies 4 4 10 13 16 18 22 28 33 35 38 43 48 52 65 73 2 Borel Hierarchy 3 Abstract Borel hierarchies 4 Characteristic function of a sequence 5 Martin’s Axiom 6 Generic Gδ 7 αforcing 8 Boolean algebras 9 Borel order of a ﬁeld of sets 10 CH and orders of separable metric spaces 11 MartinSolovay Theorem 12 Boolean algebra of order ω1 13 Luzin sets 14 Cohen real model 15 The random real model 16 Covering number of an ideal II Analytic sets 78 78 17 Analytic sets .Contents 1 What are the reals.
18 Constructible wellorderings 19 Hereditarily countable sets 20 Shoenﬁeld Absoluteness 21 MansﬁeldSolovay Theorem 22 Uniformity and Scales 23 Martin’s axiom and Constructibility 24 Σ1 wellorderings 2 25 Large Π1 sets 2 82 84 86 88 90 95 97 98 III Classical Separation Theorems 102 102 104 107 109 26 SouslinLuzin Separation Theorem 27 Kleene Separation Theorem 28 Π1 Reduction 1 29 ∆1 codes 1 IV Gandy Forcing 113 113 119 125 130 137 142 30 Π1 equivalence relations 1 31 Borel metric spaces and lines in the plane 32 Σ1 equivalence relations 1 33 Louveau’s Theorem 34 Proof of Louveau’s Theorem References .
Index 152 .
Let ω <ω be the set of all ﬁnite sequences of elements of ω. The basic open sets of ω ω are the sets of the form: [s] = {x ∈ ω ω : s ⊆ x} for s ∈ ω <ω . 4. . s is the length of s. A subset of ω ω is open iﬀ it is the union of basic open subsets. and 6. each Is is a nontrivial open interval in R with 2. and for s ∈ ω <ω and n ∈ ω let sˆn denote the sequence which starts out with s and has one more element n concatenated onto the end.1 WHAT ARE THE REALS. anyway? Let ω = {0. Theorem 1. 1}. y) = 0 1 n+1 if x = y if x n = y n and x(n) = y(n) Cantor space 2ω is the subspace of ω ω consisting of all functions from ω to 2 = {0.1 (Baire [4]) ω ω is homeomorphic to the irrationals P. {Isˆn : n ∈ Z} covers all of Is except for their endpoints. 1. for every s ∈ Z<ω and n ∈ Z Isˆn ⊆ Is . the right end point of Isˆn is the left end point of Isˆn+1 . . We will construct a mapping from Zω to P. . and for s = rational endpoints. 3. is the empty sequence. ANYWAY? 1 1 What are the reals. It is separable (has a countable dense subset) since it is second countable (has a countable basis). The following deﬁnes a complete metric on ω ω : d(x. I = R. the length of Is is less than 1 s for s = .} and let ω ω (Baire space) be the set of functions from ω to ω. Inductively construct a sequence of open intervals Is : s ∈ Z<ω satisfying the following: 1. 5. It is compact. . Proof: First replace ω by the integers Z. the nth rational qn is an endpoint of It for some t ≤ n + 1. Enumerate the rationals Q = {qn : n ∈ ω}.
Descriptive Set Theory and Forcing Deﬁne the function f : Zω → P as follows. It is nonempty because closure(Ix n+1 ) ⊆ Ix n . It is a singleton because their diameters shrink to zero. The function F is a onetoone onto map from ω ω to 2ω \ F . Given x ∈ Zω the set Ix n n∈ω 2 must consist of a singleton irrational.2 ω ω is homeomorphic to [ω]ω . Note that the map given is also an order isomorphism from Zω with the lexicographical order to P with it’s usual order. It is a homeomorphism because F ([s]) = [t] where t = 0s(0) ˆ1ˆ0s(1) ˆ1ˆ0s(2) ˆ1ˆ · · · ˆ0s(n) ˆ1 where s = n + 1. . Theorem 1. So we can deﬁne f by {f (x)} = n∈ω Ix n . The function f is onetoone because if s and t are incomparable then Is and It are disjoint. It is a homeomorphism because f ([s]) = Is ∩ P and the sets of the form Is ∩ P form a basis for P. Let F = {x ∈ 2ω : ∀∞ n x(n) = 0} (the quantiﬁer ∀∞ stands for “for all but ﬁnitely many n”). We can identify 2ω with P (ω). F corresponds to the ﬁnite sets and so 2ω \ F corresponds to the inﬁnite subsets of ω which we write as [ω]ω . Proof: Let f ∈ ω ω and deﬁne F (f ) ∈ 2ω to be the sequence of 0’s and 1’s determined by: F (f ) = 0f (0) ˆ1ˆ0f (1) ˆ1ˆ0f (2) ˆ1ˆ · · · where 0f (n) refers to a string of length f (n) of zeros. the set of all subsets of ω. It is onto since for every u ∈ P and n ∈ ω there is a unique s of length n with u ∈ Is . by identifying a subset with its characteristic function.
is clearly named so because it is homeomorphic to Cantor’s middle two thirds set.” 3 It is impossible for a man to learn what he thinks he already knows. I am indebted to John C. Brezinski [13] has more on the history of continued fractions. Just as coeﬃcients of linear equations evolved into matrices the sequences of natural numbers in continued fraction developments of irrational numbers were liberated by Baire’s mind to live in their own world.2 n The classical proof of Theorem 1. Sierpi´ski [99] gives very few references. Unfortunately. 2ω . “Baire introduced his space in Baire [3]. Sierpi´ski also says that Frechet was the ﬁrst to describe the metric n d given above. Euler [19] proved that every rational number gives rise to a ﬁnite continued fraction and every irrational number gives rise to an inﬁnite continued fraction.Descriptive Set Theory and Forcing 3 Note that sets of the form [t] where t is a ﬁnite sequence ending in a one form a basis for 2ω \ F . My proof of Theorem 1.1 is to use “continued fractions” to get the correspondence. Morgan II for supplying the following reference and comment. Cantor space.1 allows me to remain blissfully ignorant3 of even the elementary theory of continued fractions.Epictetus 2 . I wonder why ω ω is called Baire space? The earliest mention of this I have seen is in Sierpi´ski [99] where he refers to ω ω as the 0dimensional space of n Baire.
.1 Σ0 is closed under countable unions. i.4 Part I On the length of Borel hierarchies 2 Borel Hierarchy Deﬁnitions. α α+1 α+1 α Proof: That Σ0 is closed under countable unions is clear from its deﬁnition.e. Theorem 2. For α > 1 deﬁne A ∈ Σ0 iﬀ there exists a sequence Bn : n ∈ ω α with each Bn ∈ Σ0 n for some βn < α such that β A= n∈ω ∼ Bn where ∼ B is the complement of B in X. ∼ B = X \ B. Deﬁne Π0 = {∼ B : B ∈ Σ0 } α α and ∆0 = Σ0 ∩ Π0 . α α α The Borel subsets of X are deﬁned by Borel(X) = α<ω1 Σ0 (X). For X a topological space deﬁne Σ0 to be the open subsets of 1 X. For any α.2 If f : X → Y is continuous and A ∈ Σ0 (Y ). and ∆0 is closed under complements. Theorem 2. then f −1 (A) is α in Σ0 (X). α Π0 (X) ⊆ Σ0 (X) and Σ0 (X) ⊆ Π0 (X). α It follows from DeMorgan’s laws by taking complements that Π0 is closed α under countable intersections. Π0 is closed under α α countable intersections. α . It is clearly α the smallest family of sets containing the open subsets of X and closed under countable unions and complementation.
but 2 there is no ∆0 set Q in Y = [0. α α α For example. true for Π0 or ∆0 in place of Σ0 .1]. then 1. α 5. Σ0 is closed under ﬁnite intersections. 1] contains an interval. The class of sets Σ0 is also referred to as Fσ and the class Π0 as Gδ . and α 6. Π0 (X) ⊆ Π0 (X). α α α Theorem 2.3 is true for Π0 in place of Σ0 .2 is also.) Theorem 2. α α Proof: For Σ0 it follows from the deﬁnition of subspace. then Σ0 (X) = {A ∩ X : A ∈ Σ0 (Y )}. closed 2 1 sets are Gδ ). Π0 (X) ∪ Σ0 (X) ⊆ ∆0 (X) α+1 α α 4. α+1 α 2. If Z contained 2 2 in X is dense and codense then Z is ∆0 in X (every subset of X is). but a comeager Fσ in [0.. Theorem 2.2 BOREL HIERARCHY 5 This is an easy induction since it is true for open sets (Σ0 ) and f −1 passes 1 over complements and unions. and complements.3 Suppose X is a subspace of Y . 1] whose intersection with X is Z.4 For X a topological space and Π0 (X) ⊆ Π0 (X) (i. For α > 1 it is an easy 1 induction. ∆0 is closed under ﬁnite intersections. Then since X is countable every subset of X is Σ0 in X and hence ∆0 in X. α Proof: . of course. (If Q 2 is Gδ and Fσ and contains Z then its comeager. 1] and Y be [0. let X be the rationals in [0. Π0 is closed under ﬁnite unions. Σ0 (X) ⊆ Σ0 (X). and hence α α+1 3. 2 2 Theorem 2. but not in general for ∆0 . ﬁnite unions.e.
In metric spaces closed sets are Gδ . The assumption that closed sets are Gδ is necessary since if X = ω1 + 1 with the order topology. ∆0 is α α closed under ﬁnite intersections. The family of all sets of the form (B ∩ X) ∪ C where B. since C= n∈ω {x : ∃y ∈ C d(x. C are (ordinary) Borel subsets of 2ω is the σδlattice generated by the open subsets of 2ω . use that ( n∈ω Pn ) ∩ ( n∈ω Qn ) = n. and complements since it is the intersection of the two classes.Laczkovich has pointed out to me that the class Π0 (X) where for the 3 ordered space X = ω1 + 1 is not closed under ﬁnite unions: The elements of Π0 are of the form ∩∞ An where each An is either open 3 n=1 or Fσ .m∈ω (Pn ∩ Qm ) It follows by DeMorgan’s laws that Π0 is closed under ﬁnite unions. because: ∗ (Bn ∩ X) ∪ Cn = ( ( n n Bn ∪ Cn ) ∩ X) ∪ n Cn Cn . Williard [112] gives an example which is a second countable Hausdorﬀ space. then the closed set consisting of the singleton point {ω1 } is not Gδ . to see that Σ0 is closed under ﬁnite interα sections. ﬁnite unions. Let 2ω be the space 2ω with the ∗ smallest topology containing the usual topology and X as an open set. For example. y) < 1 } n+1 for C a closed set. in fact. it is not in the σδlattice generated by the open sets (the smallest family containing the open sets and closed under countable intersections and countable unions).2 BOREL HIERARCHY 6 Induction on α. (Bn ∩ X) ∪ Cn = (( n n Bn ) ∩ X) ∪ n Note that ∼ X is not in this σδlattice. This implies that Π0 contains the open sets and the closed sets. Let X ⊆ 2ω be any nonBorel set. 3 . M.
Now deﬁne U by: β U= n ∗ ∼ Un . then for every α with 1 ≤ α < ω1 there exists a universal Σ0 set U ⊆ 2ω × X. This map is continuous so if we deﬁne ∗ −1 ∗ Un = gn (Un ). We will need the following lemma.5 (Lebesgue [63]) For every α with 1 ≤ α < ω1 Σ0 (2ω ) = Π0 (2ω ). Let Un be a universal Σ0 n set. where / 3 n=1 each An is either open or Fσ .6 Suppose X is second countable (i. Let β n. Theorem 2. For α = 1 let U = {(x. For α > 1 let βn be a sequence which sups up to α if α a limit. and because the map x → xn is onto it is also β a universal Σ0 n set. For any n the map gn : 2ω × X → 2ω × X is deﬁned by (x.2 BOREL HIERARCHY 7 However.e.. Lemma 2. Thus A ∩ B contains a countable limit point. y) where xn (m) = x( n. then Un is Σ0 n . y) ∈ U }. m = 2n (2m + 1) − 1 be the usual pairing function which gives a recursive bijection between ω 2 and ω. y) → (xn . the union of an open set and a closed set is not necessarily in Π0 . If An is open then ω1 ∈ An implies that An contains an unbounded closed subset of ω1 . or equals α − 1 if α is a successor.e. a set U which is Σ0 (2ω × X) such that for every A ∈ Σ0 (X) there exists α α x ∈ 2ω such that A = Ux where Ux = {y ∈ X : (x. Let {Bn : n ∈ ω} be a countable base for X. Then A is open and B is closed. y) : ∃n (x(n) = 1 ∧ y ∈ Bn )} = n ({x : x(n) = 1} × Bn ). α i. Proof: The proof is by induction on α. Therefore ∩n An also contains an unbounded closed subset of ω1 . . Suppose A ∪ B = ∩∞ An . α α The proof of this is a diagonalization argument applied to a universal set. 3 Let A be the set of isolated points of ω1 + 1 and let B = {ω1 }. m ). But A ∪ B ∈ Π0 . which is impossible. If An is Fσ then A ⊆ An implies the same. has a countable base).
cl(Usˆ0 ) ∩ cl(Usˆ1 ) = ∅. and any uncountable complete separable metric space have ord = ω1 . 2. 3. Lebesgue’s α theorem says that ord(X) = ω1 . then there would be x ∈ 2ω with α ∼ D = Ux and then x ∈ D iﬀ x. x . then by Theorem 2.e. x ∈ U iﬀ x ∈ Ux iﬀ x ∈∼ D. Us is uncountable..3 it also collapses on all subspaces of X. diameter of Us less than 1/s . Construct a family of open sets Us : s ∈ 2<ω such that 1. D is the continuous preimage of U under the map x → x. To see this suppose X is an uncountable complete separable metric space.7 For any space X which contains a homeomorphic copy of 2ω (i. R. x ∈ U }.2 BOREL HIERARCHY 8 U is universal for Σ0 because given any sequence Bn ∈ Σ0 n for n ∈ ω there α β ∗ exists x ∈ 2ω such that for every n ∈ ω we have that Bn = (Un )x = (Un )xn (this is because the map x → xn : n < ω takes 2ω onto (2ω )ω . Deﬁne ord(X) to be the least α such that Borel(X) = Σ0 (X). a perfect set) we have that ord(X) = ω1 . so it is Σ0 . Let α D = {x : x. cl(Usˆi ) ⊆ Us for i=0. Note that ord(X) = 1 if X is a discrete space and that ord(Q) = 2.1. α but it cannot be Π0 because if it were. consequently ω ω . Proof: If the Borel hierarchy on X collapses.) But then Ux = ( n ∗ ∼ Un )x = n ∗ ∼ (Un )x = n ∼ (Bn ). Proof of Theorem 2. Corollary 2.5: Let U ⊆ 2ω × 2ω be a universal Σ0 set. Every uncountable complete separable metric space contains a perfect set (homeomorphic copy of 2ω ). and 4.
Π0 was ﬁrst popularized by Addison [1].2 BOREL HIERARCHY Then the map f : 2ω → X deﬁned so that {f (x)} = n∈ω 9 Ux n gives an embedding of 2ω into X. .5 of Louveau’s Theorem shows that there can be no Borel set which is universal for all ∆0 sets. I don’t know α β if the “bold face” and “light face” notation is such a good idea. but the proof is much the same. Consequently. Miller [82] contains examples α from model theory of Borel sets of arbitrary high rank. Corollary 33. Lebesgue [63] used universal functions instead of sets. some copy machines wipe it out. The notation Σ0 . I use Σ0 α which is blackboard boldface.
Bledsoe. Corollary 3. α Theorem 3. Proof: ˆ Suppose 2ω ⊆ X and let F = {A ∩ 2ω : A ∈ F }. Let Π0 (F ) = F and for every α 0 α > 0 deﬁne A ∈ Π0 (F ) iﬀ there exists Bn ∈ Π0 n for some βn < α such that α β A= n∈ω ∼ Bn . Most of the time we would like to think of F as a countable ﬁeld of sets (i.e.1 is a generalization of Lebesgue’s universal set argument. Then ord(F ) = ω1 .1 we know ord(F ) = ω1 . α .3 ABSTRACT BOREL HIERARCHIES 10 3 Abstract Borel hierarchies Suppose F ⊆ P (X) is a family of sets. The proof of Theorem 3. We deﬁne the classes Π0 (F ) analogously. Lemma 3. Deﬁne • Σ0 (F ) = {∼ B : B ∈ Π0 (F )}. α α • ∆0 (F ) = Π0 (F ) ∩ Σ0 (F ).2 Suppose X is any space containing a perfect set and F is a countable family of subsets of X with Borel(X) ⊆ Borel(F ).3 (Universal sets) Suppose H ⊆ P (X) is countable and deﬁne R = {A × B : A ⊆ 2ω is clopen and B ∈ H}. and α • let ord(F ) be the least α such that Borel(F ) = Σ0 (F ). We need to prove the following two lemmas. But this implies ord(F ) = ω1 . α α α • Borel(F ) = α<ω1 Σ0 (F ). Then for every α with 1 ≤ α < ω1 there exists U ⊆ 2ω × X with U ∈ Π0 (R) α such that for every A ∈ Π0 (H) there exists x ∈ 2ω with A = Ux . Mauldin [12]) Suppose F ⊆ P (2ω ) is a countable family such that Borel(2ω ) ⊆ Borel(F ).1 (Bing.3 we have ˆ ˆ that Borel(2ω ) ⊆ Borel(F ) and so by Theorem 3. closed under complements and ﬁnite intersections) and so analogous to the family of clopen subsets of some space. Then ord(F ) = ω1 . By Theorem 2.
1] .3. ∗ Note that when we replace Un by Un it is necessary to prove by induction on β that for every set A ∈ Π0 (R) and n ∈ ω that the set β A∗ = {(x. x) ∈ n An } = n {x : (x. x) ∈ An } and {x : (x. x) ∈ U } is in Borel(H) and hence its complement is in Borel(H) = Π0 (H). 1]. / Theorem 3. so the result follows by induction. Proof: Let f : X → [0. replacing the basis for X with H.6.4 the set D = {x : (x. Note that {x : (x. x) ∈ A}. By Lemma 3. Then for every set A ∈ Borel(R) the set D = {x : (x. Proof of Theorem 3. x) ∈ A} is in Borel(H). Let B be a countable base for X and let H = {f (B) : B ∈ B}. 1] be continuous and onto.3. Hence we get α the same old contradiction: if Ux =∼ D.4 Suppose H ⊆ P (2ω ). y) ∈ A} is also in Π0 (R). then x ∈ D iﬀ x ∈ D. then D = B ∩ C which is in Borel(H) by assumption. then ord(X) = ω1 . R is deﬁned as in Lemma 3.1: Suppose Borel(H) = Π0 (H) and let U ⊆ 2ω × 2ω be universal for Π0 (H) α α given by Lemma 3. [0. Proof: If A = B × C where B is clopen and C ∈ H. y) : (xn . Since the preimage of an open subset of [0.5 (Reclaw) If X is a second countable space and X can be mapped continuously onto the unit interval. x) ∈∼ A} =∼ {x : (x.Descriptive Set Theory and Forcing 11 Proof: This is proved exactly as Theorem 2. and Borel(2ω ) ⊆ Borel(H). β Lemma 3.
But f maps the Borel hierarchy of X directly over to the hierarchy generated by H.Descriptive Set Theory and Forcing 12 is open in X it is clear that Borel([0. 1].5 we may replace [0.2 it follows that ord(H) = ω1 . 1]. Note that if X is a discrete space of cardinality the continuum then there is a continuous map of X onto [0. but I did. 1] and ω ω can be mapped onto 2ω . . 1] by 2ω . The Cantor space 2ω can be mapped continuously onto [0. 2n+1 This map is even onetoone except at countably many points where it is twotoone. Myrna Dzamonja points out that any completely regular space Y which contains a perfect set can be mapped onto [0. Reclaw did not publish his result. so ord(X) = ω1 . But since Y is completely regular this map extends to Y . 1] via the map ∞ x→ n=0 x(n) . It is also easy to see that R can be mapped continuously onto [0. 1] but ord(X) = 1. see Miller [86] and [87]. then there is a continuous map f from P onto [0. It follows that in Theorem 3. So by Corollary 3. or R. This is true because if P ⊆ Y is perfect. ω ω . 1]) ⊆ Borel(H).
c maps X into 2F  . F is a family of sets which is closed under complements in X and ﬁnite intersections). In general.4 CHARACTERISTIC FUNCTION OF A SEQUENCE 13 4 Characteristic function of a sequence The idea of a characteristic function of a sequence of sets is due to Kuratowski and generalized the notion of a characteristic function of a set introduced by de le Vall´ePoussin.2 (Miller [75]) Suppose F ⊆ P (X) is a ﬁeld of sets and ord(F ) = λ where λ is a countable limit ordinal. α . The preimages of points of Y under c are exactly the equivalence classes of ≈. Deﬁne c : X → 2ω by c(x)(n) = 1 if x ∈ An 0 if x ∈ An / Let Y = c(X).e. Hence ord(F ) = ord(Y ). Then there exists B ∈ Borel(F ) which is not in Π0 (F ) for any α < λ. Suppose F ⊆ P (X) is a countable ﬁeld of sets (i. He also exploited it in Szpilrajn [107].1 (Szpilrajn [107]) If F ⊆ P (X) is a countable ﬁeld of sets. then we see that members of Borel(F ) respect ≈. Proof: If we deﬁne x ≈ y iﬀ ∀n (x ∈ An iﬀ y ∈ An ). The general notion was introduced by e Szpilrajn [106]. Theorem 4. The map c induces a bijection between X/ ≈ and Y which takes the family F exactly to a clopen basis for the topology on Y . then there exists a subspace Y ⊆ 2ω such that ord(F ) = ord(Y ). Szpilrajn changed his name to Marczewski soon after the outbreak of World War II most likely to hide from the Nazis. Theorem 4. Let F = {An : n ∈ ω}. He kept the name Marczewski for the rest of his life. The following theorem says that bounded Borel hierarchies must have a top. then there is a direct correspondence between F and {C ∩ Y : C ⊆ 2ω clopen }.
A set C ⊆ 2κ is clopen iﬀ it is a ﬁnite union of sets of the form [s] = {x ∈ 2κ : s ⊆ x} where s : D → 2 is a map with D ∈ [κ]<ω (i. α α Proof: This is proved by a LowenheimSkolem kind of an argument. s ⊆ t ∈ T implies s ∈ T . D is a ﬁnite subset of κ). Since ∈ T it must be that T has an inﬁnite branch.e.e. By permuting κ around we may assume without loss of generality that S = ω. Note that T is a tree.3 There exists a countable S ⊆ κ with the properties that α < λ and s : D → 2 with D ∈ [S]<ω if ord(Y ∩ [s]) > α then there exists A in Σ0 (F ) but not in ∆0 (F ) such that A ⊆ [s] and A is supported by S. membership in A is determined by restrictions to S..Descriptive Set Theory and Forcing 14 Proof: By the characteristic function of a sequence of sets argument we may assume without loss of generality that F = {C ∩ Y : C ⊆ 2κ clopen }. Deﬁne T = {s ∈ ω <ω : ord(Y ∩ [s]) = λ}. because [s] = [sˆ0] ∪ [sˆ1]. That is to say. i. Also for any s ∈ T either sˆ0 ∈ T or sˆ1 ∈ T . y ∈ 2κ if x S = y S then (x ∈ A iﬀ y ∈ A). Note that by induction for every A ∈ Borel(F ) there exists an S ∈ [κ]ω (called a support of A) with the property that for every x. Let x : ω → 2 be such that x n ∈ T for all n < ω. For each n deﬁne tn = (x n)ˆ(1 − x(n)) and note that 2κ = [x] ∪ n∈ω [tn ] . Lemma 4.
e. R0 = R and let Rn+1 be the family of countable unions (if n even) or family of countable intersections (if n odd) of sets from Rn . For n ∈ ω deﬁne Rn as follows. Question 4.e. Proof: Suppose not and let α and n witness this.. the smallest family containing R and closed under countable unions and countable intersections. α Then ∪n An is not at any level bounded below λ.Descriptive Set Theory and Forcing is a partition of 2κ into clopen sets and one closed set [x]. closed under ﬁnite unions and ﬁnite intersections.4 Suppose R ⊆ P (X) is a ring of sets. This proves the Claim. If R∞ = n<ω Rn . Since A is supported by ω either [x] ⊆ A or A is disjoint from [x]. Since ord([x n] ∩ Y ) = λ we know there exists A ∈ Σ0 (F ) \ ∆0 (F ) α+1 α+1 such that A ⊆ [x n] and A is supported by S = ω. The claim allows us to construct a set which is not at a level below λ as follows. Note that [x n] = [x] ∪ n≤m<ω 15 [tm ]. Let αn < λ be a sequence unbounded in λ and let kn be a distinct sequence with ord([tkn ] ∩ Y ) ≥ αn . Claim: For every α < λ and n ∈ ω there exists m > n with ord(Y ∩ [tm ]) > α. then must there be n < ω such that R∞ = Rn ? . But if ord([tm ] ∩ Y ) ≤ α for each m > n. then A0 = (A ∩ [tm ]) n≤m<ω is Σ0 (F ) and A = A0 or A = A0 ∪ [x] either of which is Σ0 (F ) (as long as α α α > 1).. i. Let An ⊆ [tkn ] be in Borel(F ) \ ∆0 n (F ). i. Let R∞ be the σring generated by R.
It follows that P is the countable union of ﬁlters and hence we cannot ﬁnd an uncountable set of pairwise incompatible conditions. The partial order P is deﬁned as follows. p is stronger ◦ than q. Proof: Note that since B is countable there are only countably many sentences ◦ of the type “B ⊆U n ”. p ∈ P iﬀ p is a ﬁnite consistent set of sentences of the form 1. However. however to make it more readable we often just write x. “x ∈U n ” where x ∈ X \ A or / 2. it is probably just another view of Solovay’s ‘almost disjoint sets forcing’. The forcing we use is due to Silver. Theorem 5. in fact. Consistent means that there is not a pair of sentences “x ∈U n ”. For standard references on forcing see Kunen [56] or Jech [44]. “B ⊆U n ” where B ∈ B and n ∈ ω.1 Assuming Martin’s Axiom if X is any second countable Hausdorﬀ space of cardinality less than the continuum. i. The circle in the notation U n ’s means that it is the name for the set Un which will be determined by the generic ﬁlter.e. For an element x of the ground model we should use x to denote the canonical name ˇ of x. Also if p and q have exactly the same sentences of this type then p ∪ q ∈ P and hence p and q are compatible.5 MARTIN’S AXIOM 16 5 Martin’s Axiom The following result is due to Rothberger [94] and Solovay [45][74]. Proof: Let A ⊆ X be arbitrary and let B be a countable base for the topology on X. then ord(X) ≤ 2 and. p ≤ q iﬀ p ⊇ q. We call this forcing Silver forcing. The ordering on P is reverse containment. every subset of X is Gδ . For x ∈ X \ A deﬁne Dx = {p ∈ P : ∃n “x ∈U n ” ∈ p}. Claim: P satisﬁes the ccc. / ◦ ◦ ◦ ◦ ◦ . “B ⊆U n ” / in p where x ∈ B.
Choose n large enough so ◦ that Un is not mentioned in p. Since the cardinality of X is less than the continuum we can ﬁnd a Pﬁlter n G with the property that G meets each Dx for x ∈ X \ A and each Ex for x ∈ A and n ∈ ω. see Fleissner [24] [25]. Then is p > ω1 . .. and Balogh [5]. For more information on Martin’s Axiom see Fremlin [27]. Fleissner and Miller [23] show it is consistent to have a Qset whose union with the rationals is not a Qset. Proof: To see that Dx is dense let p ∈ P be arbitrary. Judah and Shelah [46] [47]. The following question was raised during an email correspondence with Zhou. For more on Qsets. ◦ Un and so A is Gδ in X. Then q = (p ∪ {“B ⊆U n ”}) ∈ P n and q ∈ Ex . is it true that for every family F ⊆ [ω]ω of size ω1 with the ﬁnite intersection property there exists an X ∈ [ω]ω with X ⊆∗ Y for all Y ∈ F? It is a theorem of Bell [11] that p is the ﬁrst cardinal for which MA for σcentered forcing fails. Another result along this line due to Alan Taylor is that p is the cardinality of the smallest set of reals which is not a γset. Przymusinski [92]. i.e. / n To see that Ex is dense let p be arbitrary and let Y ⊆ X \ A be the set of elements of X \ A mentioned by p. Question 5. then (p ∪ {“x ∈U n ”}) ∈ P.Descriptive Set Theory and Forcing For x ∈ A and n ∈ ω deﬁne n Ex = {p ∈ P : ∃B ∈ B x ∈ B and “B ⊆U n ” ∈ p}. see Galvin and Miller [30]. Miller [83] [87]. Spaces X in which every subset is Gδ are called Qsets. Now deﬁne Un = Note that A = n∈ω ◦ ◦ {B : “B ⊆U n ” ∈ G}. Since x ∈ A and X is Hausdorﬀ there exists B ∈ B with B ∩ Y = ∅ and x ∈ B.2 Suppose every set of reals of cardinality ℵ1 is a Qset. ◦ 17 n Claim: Dx is dense for each x ∈ X \ A and Ex is dense for each x ∈ A and n ∈ ω.
For suppose p ∈ P and A ⊆ X is in V is such that ◦ ˇ p  U = A. / The ordering is reverse inclusion. Let X be a Hausdorﬀ space with a countable base B. p ∈ P iﬀ it is a ﬁnite consistent set of sentences of the form: 1. or / 3. “x ∈U n ” where x ∈ X and n ∈ ω. Since X is inﬁnite there exist x ∈ X which is not mentioned in p.) But p0  x ∈U and p1  x ∈U . Consider the following forcing notion. A P ﬁlter G determines a Gδ set U as follows: Let ◦ Un = {B ∈ B : “B ⊆U n ” ∈ G}.e. like most of the demonstrations of the higher mathematics. ‘Gentlemen. “x ∈ n<ω ◦ ◦ U n ” where x ∈ X. ◦ Note that U is not in V (as long as X is inﬁnite). certainly for Un not mentioned in / p. Let U = n Un . or 2. “B ⊆U n ” where B ∈ B and n ∈ ω. and since x is either in A or not in A we arrive at a contradiction.’ Baron Kelvin 4 ◦ ◦ . Note that ◦ ◦ p0 = p ∪ {“x ∈ n<ω U n ”} is consistent and also p1 = p ∪ {“x ∈U n ”} is / consistent for all suﬃciently large n (i.6 GENERIC Gδ 18 6 Generic Gδ “What are the possibly lengths of Borel hierarchies?” It is natural4 to ask In this section we present a way of forcing a generic Gδ . a density argument shows that for every x ∈ X we have that x ∈ U iﬀ “x ∈ n<ω U n ” ∈ G. the great thing about this. is that it can be of no earthly use to anybody. ◦ ◦ ◦ Consistency means that we cannot say that both “B ⊆U n ” and “x ∈U n ” if it / ◦ ◦ happens that x ∈ B and we cannot say both “x ∈U n ” and “x ∈ n<ω U n ”. If G is Pgeneric over V .
1 P has ccc.) If V [G] is a generic extension of a model V which contains a topological space X. For each p ∈ A let Dp be the ﬁnitely many elements of X mentioned by p and let sp : Dp → ω be deﬁned by sp (x) = 0 Un ” ∈ p ◦ n + 1 if “x ∈U n ” ∈ p / if “x ∈ n<ω ◦ But now {sp : p ∈ A} is an uncountable family of pairwise incompatible ﬁnite partial functions from X into ω which is impossible. ω) has the ccc. Consequently for each distinct pair p. Recall that there are three types of sentences: 1. then since there are only countably many sentences of type 1 above we may assume that all p ∈ A have the same set of type 1 sentences. Now if for contradiction A were an uncountable antichain. Theorem 6. . Then in V [G] the Gδ set U is not Fσ .Descriptive Set Theory and Forcing 19 In fact. (FIN(X. To see this we will ﬁrst need to prove that P has ccc. then we let X also refer to the space in V [G] whose topology is generated by the open subsets of X which are in V .2 (Miller [75]) Suppose X in V is an uncountable Hausdorﬀ space with countable base B and G is Pgeneric over V . Lemma 6. Proof: Note that p and q are compatible iﬀ (p ∪ q) ∈ P iﬀ (p ∪ q) is a consistent set of sentences. B ⊆U n 2. q ∈ A there must be an ◦ ◦ x ∈ X and n such that either “x ∈U n ” ∈ p and “x ∈ n<ω U n ” ∈ q / or viceversa. n ∈ ω. and x ∈ X. U is not Fσ in the extension (assuming X is uncountable). x ∈ ◦ n<ω ◦ ◦ Un where B ∈ B. see Kunen [56]. x ∈U n / 3.
Suppose for contradiction p n∈ω 20 U n= n∈ω ◦ C n where C n are closed in X . clearly q x∈ n∈ω ◦ Cn so there exists r ≤ q and n ∈ ω so that r  x ∈C n .Descriptive Set Theory and Forcing Proof: We call this argument the old switcheroo. / Consequently there exist s ∈ P(Y ) and B ∈ B such that ◦ . Let t = r0 ∪ {“x ∈U m ”} / where m is chosen suﬃciently large so that t is a consistent condition. Let Y ⊆ X be countable such that 1. Let q = p ∪ {“x ∈ n∈ω ◦ U n ”}. Now we do the switch. ◦ ◦ For Y ⊆ X let P(Y ) be the elements of P which only mention y ∈ Y in type 2 or 3 statements. p ∈ P(Y ) and 2. Since X is uncountable there exists x ∈ X \ Y . for every n and B ∈ B there exists a maximal antichain A ⊆ P(Y ) which decides the statement “B∩ C n = ∅”. Let r = r0 ∪ {“x ∈ n∈ω ◦ U n ”} ◦ where r0 does not mention x. ◦ Since q extends p. Since t x∈ / n∈ω ◦ ◦ Un we know that t  x ∈C n .
s and t are compatible. s  B∩ C n = ∅. 2. and 3. This is a contra◦ ◦ diction since s ∪ r  x ∈C n and s ∪ r  x ∈C n . x ∈ B. because s does not mention x.Descriptive Set Theory and Forcing 1. / . ◦ 21 But s and r are compatible.
3. Now deﬁne r(s) = α iﬀ r(s) ≥ α but not r(s) ≥ α + 1 and r(s) = ∞ iﬀ r(s) ≥ α for every ordinal α. T ⊆ ω <ω is a tree. if r( ) = α ∈ OR and x ∈ [T ].7 αFORCING 22 7 αforcing In this section we generalize the forcing which produced a generic Gδ to arbitrarily high levels of the Borel hierarchy. r(s) ≥ 0 iﬀ s ∈ T . Hence x ∈ [T ]. T is a nice αtree iﬀ 1. Deﬁne [T ] = {x ∈ Qω : ∀n x n ∈ T }. r : T → (α + 1) is its rank function (r( ) = α).1 T is wellfounded iﬀ r( ) ∈ OR. Deﬁne T ⊆ Q<ω to be a tree iﬀ s ⊆ t ∈ T implies s ∈ T . Let OR denote the class of all ordinals. then r(x (n + 1)) < r(x n) is a descending sequence of ordinals. On the other hand. Deﬁnition. We say that T is wellfounded iﬀ [T ] = ∅. Deﬁne the rank function r : T → OR ∪ {∞} of T as follows: 1. 2. So if r( ) = ∞ we can construct (using the axiom of choice) a sequence sn ∈ T with r(sn ) = ∞ and sn+1 = sn ˆx(n). r(s) ≥ λ (for λ a limit ordinal) iﬀ r(s) ≥ α for every α < λ. Hence. if r(s) = ∞ then for some q ∈ Q we must have r(sˆq) = ∞. Proof: It follows easily from the deﬁnition that if r(s) is an ordinal. . r(s) ≥ α + 1 iﬀ ∃q ∈ Q r(sˆq) ≥ α. Theorem 7. Before doing so we must prove some elementary facts about wellfounded trees. then r(s) = sup{r(sˆq) + 1 : q ∈ Q}. 2.
then x = y. if r(s) = λ is a limit ordinal. / The ordering on P is given by p ≤ q iﬀ tp ⊇ tq and Fp ⊇ Fq . But now for each p ∈ A let Hp : X → [T >0 ]<ω be the ﬁnite partial function deﬁned by Hp (x) = {s ∈ T >0 : (s. deﬁne the partial order P = P(X. (sˆn. sˆn ∈ T >0 such that (s. a countable set. if r(s) = β is a successor ordinal. then x ∈ B. Proof: Suppose A is uncountable antichain. and 4. F ⊆ T >0 × X is ﬁnite where T >0 = T \ T 0 = {s ∈ T : r(s) > 0}. x). F ) where 1. then r(sˆ0) ≥ 2 and r(sˆn) increases to λ as n → ∞. For X a Hausdorﬀ space with countable base. Then {Hp : p ∈ A} is an uncountable antichain in the order of ﬁnite partial functions from X to [T >0 ]<ω . if r(s) > 0. B. and 5.Descriptive Set Theory and Forcing 3. x) ∈ Fp } where domain Hp is {x : ∃s ∈ T >0 (s. x) ∈ F and t(sˆn) = B. 23 4. Consequently for p. if (s. t : D → B where D ⊆ T 0 = {s ∈ T : r(s) = 0} is ﬁnite. x). x) ∈ (Fp ∪ Fq ). if (s.2 P has ccc. Deﬁne for G a Pﬁlter the set Us ⊆ X for s ∈ T as follows: . y) ∈ F . 2. 3. x) ∈ Fp }. (sˆn. then for every n ∈ ω r(sˆn) = β − 1. B. T ) which we call αforcing as follows: p ∈ P iﬀ p = (t. q ∈ A the only thing that can keep p ∪ q from being a condition is that there must be an x ∈ X and an s. then for every n ∈ ω sˆn ∈ T . Since there are only countably many diﬀerent tp without loss we may assume that there exists t such that tp = t for all p ∈ A. Lemma 7. and T a nice αtree (α ≥ 2). It is easy to see that for every α < ω1 nice αtrees exist.
If (s. Choose B ∈ B with x ∈ B and Y disjoint from B. On the other hand if (sˆn. and let q = (tq . x) ∈ Fp . x) ∈ Fp then p ∈ D and we are already done. Proof: First suppose that r(s) = 1 and note that the following set is dense: D = {p ∈ P : (s. (Note r(sˆn) > 0. So q ≤ p and q ∈ D. Choose sˆn not in the domain of tp . x) ∈ Fp then we know that for every q ∈ G and for every n. x) ∈ Fp then let / Y = {y : (s. On the other / hand if tp (sˆn) = B where x ∈ B. Fp ∪ {(sˆn. x) ∈ Fq (since otherwise p and q would be incompatible). If (s. So let G be a generic ﬁlter and p ∈ G ∩ D. To see this let p ∈ P be arbitrary. B). Fp ) be deﬁned by tq = tp ∪ (sˆn. To see this let p ∈ P be arbitrary. if tq (sˆn) = B then x ∈ B. x)}) ∈ E. Hence D is dense. x ∈ Us . In this case note that the following set is dense: E = {p ∈ P : (s. for s ∈ T >0 let Us = n∈ω 24 ∼ U sˆn Note that Us is a Π0 (X)set where r(s) = β. then x ∈ Us and for every q ∈ G it must / be that (s. x) ∈ Fp }.3 If G is Pgeneric over V then in V [G] we have that for every x ∈ X and s ∈ T >0 x ∈ Us ⇐⇒ ∃p ∈ G (s. If (s. x) ∈ Fp or ∃n∃B ∈ B x ∈ B and tp (sˆn) = B}.) Now assume the result is true for all Usˆn . β Lemma 7. Let p ∈ G ∩ E. x) ∈ Fp . If (s. x) ∈ Fp and already p ∈ E or by choosing n large enough q = (tp . y) ∈ Fp }. then by / . x) ∈ Fp then for every q ∈ G and n we have (sˆn. x) ∈ Fq and so by induction / x ∈ Usˆn and so x ∈ Us . for s ∈ T 0 let Us = B iﬀ ∃p ∈ G such that tp (s) = B and 2. Consequently. / Now suppose r(s) > 1.Descriptive Set Theory and Forcing 1. x) ∈ Fp or ∃n (sˆn. Now by deﬁnition x ∈ Us iﬀ x ∈ n∈ω ∼ U sˆn . Then either (s.
Q) < β. if r(s) = λ > β is a limit ordinal and r(sˆn) < β + 1. x) ∈ Fp and (sˆn. Also r(sˆnˆm) > 0 so this is possible to do. x) ∈ Fp0 : x ∈ Q or r(s) < β + 1}. Lemma 7. Q) < β + 1 and p 2. rank(ˆ.Descriptive Set Theory and Forcing 25 induction x ∈ Usˆn and so x ∈ Us . then there exist m ∈ ω such that (sˆnˆm. x) ∈ Fq and tp (sˆn) = B with x ∈ B. x) ∈ Fq and (sˆn. Given any Q ⊂ X deﬁne the rank(p. x) ∈ Fp and tq (sˆn) = B with x ∈ B. x) ∈ Fp for some x ∈ X \ Q}. or 2.4 (Rank Lemma). ˆ Suppose for contradiction that there exists q such that rank(q. p ˆ and q compatible. or 4. for any q ∈ P if rank(q. then p and q compatible implies p and q compatible. x) ∈ Fp . Since p and q are incompatible either 1. x) ∈ Fp and n ∈ ω. there exists (s. Q) as follows: rank(p. . x) and sˆn to worry about. ˆ Proof: Let p0 ≤ p be any extension which satisﬁes: for any (s. there exists (s. x) ∈ Fq . x) ∈ Fq .2. or 3. there exists (s. there exists (s. Note that since r(sˆn) is increasing to λ there are only ﬁnitely many (s. Q) < β. but p and q incompatible. Q) = max{r(s) : (s. / The following lemma is the heart of the old switcheroo argument used in Theorem 6. For any β ≥ 1 and p ∈ P there exists p ˆ compatible with p such that 1. and so again for every q ∈ G we have / (s. Now let p be deﬁned as follows: ˆ tp = tp ˆ and Fp = {(s. x) ∈ Fp0 .
ˆ ˆ Intuitively. q incompatible. q would be incompatible. {Pα : α < κ}. ψ∈Γ Models for this propositional language can naturally be regarded as subsets Y of κ where we deﬁne 1. ˆ ◦ . Q) < β we know / that r(sˆn) < β. it should be that statements of small rank are forced by conditions of small rank. Finally in case (4) since x ∈ Q and so r(s) < β ˆ / we have that r(sˆn) < β and so (sˆn. ˆ ˆ If λ > β then by our construction of p0 there exist m with (sˆnˆm. Let L∞ (Pα : α < κ) be the inﬁnitary propositional logic with {Pα : α < κ} as the atomic sentences. ˆ ˆ ˆ So assume x ∈ Q. x) ∈ Fp and so again p and ˆ ˆ q are incompatible. x) ∈ Fp and so p. In case (3) by the deﬁnition of rank(q. Now since T is a nice tree we know that either r(s) ≤ β and so (s.5 (Rank and Forcing Lemma) Suppose rank : P → OR is any function on a poset P which satisﬁes the Rank Lemma 7. x) ∈ Fp ˆ and so p. x) ∈ Fp or r(s) = λ a limit ordinal. x) ∈ Fp or (in case 4) (sˆn. x) ∈ Fp . Y = Γ iﬀ Y = θ for every θ ∈ Γ. Suppose  and for every p ∈ P and α < κ if p  α ∈Y ◦ κ then there exist p compatible with p such that rank(ˆ) = 0 and ˆ p p  α ∈Y . If (3) or (4) happens for x ∈ Q then again (in case 3) (s. and 3. q are incompatible. Y = Pα iﬀ α ∈ Y . Now if λ ≤ β then (s.Descriptive Set Theory and Forcing 26 (1) cannot happen since tp = tp and so p. ◦ PY ⊂ Lemma 7. q are incompatible. (2) cannot ˆ ˆ happen since r(s) = 1 and β ≥ 1 means that (s.4. Y = ¬θ iﬀ not Y = θ. The next lemma will make this more precise. For any β > 0 let θ be a Πβ sentence iﬀ there exists Γ ⊆ δ<β Πδ sentences and θ= ¬ψ. 2. Let Π0 sentences be the atomic ones. x) ∈ Fp and so p.
◦ ◦ ◦ ◦ Some earlier uses of rank in forcing arguments occur in Steel’s forcing. . which is a contradiction because ˆ ˆ ˆ θ implies ¬ψ and so ◦ p  “ Y = ¬ψ”. see Steel [108]. Then there exists r ≤ p and ψ ∈ Γ such that ˆ r  “ Y = ψ”. q compatible then p. q compatible. Friedman [29].Descriptive Set Theory and Forcing 27 Then for every Πβ sentence θ (in the ground model) and every p ∈ P. By the rank lemma there exists p compatible with p such that rank(ˆ) ≤ β and for ˆ p every q ∈ P with rank(q) < β if p. So suppose β > 0 and θ = ψ∈Γ ¬ψ where Γ ⊆ δ<β Πδ sentences. The proof is induction on β and for β = 0 the conclusion is true by assumption. ˆ Proof: This is one of those lemmas whose statement is longer than its proof. ˆ But r. p compatible. and Harrington [36]. We ˆ claim that ◦ p  “ Y = θ”. see Silver [101]. In Miller [77] αforcing for all α is used to construct generic Souslin sets. if p  “ Y = θ” then there exists p compatible with p such that rank(ˆ) ≤ β and ˆ p p  “ Y = θ”. It also occurs in Silver’s analysis of the collapsing algebra. p compatible implies r. ˆ Suppose not. By inductive assumption there exists r compatible with r such that ˆ rank(ˆ) < β r such that r  “ Y = ψ”.
for every d ∈ B if c ≤ d for every c ∈ C. for any sentence θ in the forcing language [ θ ] =  {b ∈ B : b  θ} = {p ∈ P : p  θ} where P is any dense subset of B.e. A partial order P is separative iﬀ for any p. or 2. i.e. for every b ∈ B if b > 0 then there exists p ∈ P with p ≤ b.8 BOOLEAN ALGEBRAS 28 8 Boolean algebras In this section we consider the length of Borel hierarchies generated by a subset of a complete boolean algebra. i. This means that in addition to being a boolean algebra. inﬁnite sums and products. then b ≤ d. r compatible). We start by presenting some background information. complete boolean algebra. so there exists s such that tq (s) = B and either s not in the domain of tp or tp (s) = C where C = B and so in either case we can ﬁnd r ≤ p with r. x) ∈ (Fq \ Fp ) and we can either add (sˆn. Fp does not contain Fq . for any C ⊆ B there exists b (denoted C) such that 1. It is easy to check that the αforcing P is separative (as long as B is inﬁnite): If p ≤ q then either 1.. Solovay see [44]) A partial order P is separative iﬀ there exists a cBa B such that P ⊆ B is dense in B. q incompatible. so there exists (s. The elegant (but as far as I am concerned mysterious) approach to forcing using complete boolean algebras contains the following facts: 1.1 (Scott. c ≤ b for every c ∈ C and 2. i.e. Let B be a cBa. x) for suﬃciently large n or add tr (sˆn) = B for some suﬃciently large n and some B ∈ B with x ∈ B and get r ≤ p which is incompatible with q. . q ∈ P we have p ≤ q iﬀ ∀r ∈ P(r ≤ p implies q. Similarly we deﬁne C = − c∈C −c where −c denotes the complement of c in B. We ﬁnd that the generators of the complete boolean algebra associated with αforcing generate it in exactly α + 1 steps. tp does not extend tq . also exist. Theorem 8.
  4.2 (Miller [75]) For every α ≤ ω1 there exists a countably generated ccc cBa B with ord(B) = α. p  θ iﬀ p ≤ [ θ ]. α Theorem 8.Descriptive Set Theory and Forcing 2.    6. {(s.  .3 we know for generic ﬁlters G that for every x ∈ X and s ∈ T >0 x ∈ Us ⇐⇒ ∃p ∈ G (s. x)} since if they are not equal.  x Deﬁnitions. [ θ ∨ ψ ] = [ θ ] ∨[ ψ ]. {(s. x)} > 0.  3. α+1 First note that for any s ∈ T with r(s) = 0 and x ∈ X. By Lemma 7.    5. [ ¬θ ] = −[ θ ]. Let C = {p ∈ P : Fp = ∅}. Hence [ x ∈ Us ] = ∅. [ x ∈ Us ] =  {p ∈ C : ∃B ∈ B tp (s) = B and x ∈ B}. We will show that ord(B) = α + 1. C is countable and we claim that P ⊆ Π0 (C). Proof: Let P be αforcing and B be the cBa given by the ScottSolovay Theorem 8.1. α β ord(B) = min{α : ∃C ⊆ B countable with Π0 (C) = B}. For B a cBa and C ⊆ B deﬁne: Π0 (C) = C and 0 Π0 (C) = { Γ : Γ ⊆ {−c : c ∈ β<α Π0 (C)}} for α > 0. then  b = [ x ∈ Us ] ∆ ∅. [ θ ∧ ψ ] = [ θ ] ∧[ ψ ]. ˇ [ ∀x ∈ X θ(x) ] =  x∈X 29 [ θ(ˇ ]. for any set X in the ground model. x) ∈ Fp . Since B = Σ0 (P) this will 1 α imply that B = Σ0 (C) and so ord(B) ≤ α + 1.
we see by  1 >0 induction that for every s ∈ T if r(s) = β then ∅. then for any a ∈ Π0 (E) and a = 0 there γ exists b ≤ a with r(b) ≤ γ. {(s. x)} . Proof: Easy induction. We use the following Lemmas.  Remembering that for r(sˆn) = 0 we have [ x ∈ Usˆn ] ∈ Σ0 (C).Descriptive Set Theory and Forcing 30 but letting G be a generic ultraﬁlter with b in it would lead to a contradiction. We get that for r(s) > 0: ∅.x)∈Fp ∅. x)} ∈ Π0 (C).3 If r : P → OR is a rank function. So we have that p ∈ Π0 (C). C and ∀p ∈ C r(p) ≤ β} . Lemma 8. α Now we will see that ord(B) > α. β For any p ∈ P p = tp .4 and in addition p ≤ q implies r(p) ≤ r(q). it satisﬁes the Rank Lemma 7. Proof: ◦ To see this let E = {en : n ∈ ω} and let Y be a name for the set in the generic extension Y = {n ∈ ω : en ∈ G}. then if P is dense in the cBa B then r extends to r∗ on B+ : r∗ (b) = min{β ∈ OR : ∃C ⊆ P : b = and still satisﬁes the Rank Lemma. {(s. Lemma 8. {(s. ∅ ∧ (s. x)} = [ x ∈ Us ] = [ x ∈   n∈ω ∼ U sˆn ] = n∈ω −[ x ∈ Usˆn ]. B+ are the nonzero elements of B. i.4 If r : B+ → ord is a rank function and E ⊆ B is a countable collection of rank zero elements.e.
5 gives us.4 we have that there exists p ˆ with p ≤ bn ≤ b = [ x ∈ U ] and rank(ˆ. ˆ p Now we use the lemmas to see that ord(B) > α.4 is satisﬁed. Suppose for contradiction that [ x ∈ U ] = b ∈ Σ0 (E). ˆ  This takes care of all countable successor ordinals. 1 for the reader to contemplate. x)} . But by the deﬁnition of ˆ  p rank(ˆ. by translating p  Y = θb into p ≤ [ Y = θb ] = b) that:  For any γ ≥ 1 and p ≤ b ∈ Π0 (E) there exists a p compatible ˆ γ with p such that p ≤ b and r(ˆ) ≤ γ. Q) = 0. Q) = 0 for any p ∈ E so the hypothesis of Lemma 8. x) is not in Fp . Given any countable E ⊆ B. The  γ Rank and Forcing Lemma 7. {( . Let x ∈ X \ Q be arbitrary. For elements b of B in the complete subalgebra  generated by E let us associate sentences θb of the inﬁnitary propositional logic L∞ (Pn : n ∈ ω) as follows: θen = Pn θ−b = ¬θb θ R = r∈R θr Note that [ Y = θb ] = b and if b ∈ Π0 (E) then θb is a Πγ sentence.  / α We have chosen Q so that r(p) = rank(p. Q) the pair ( .) For λ a limit ordinal take αn . By Lemma 8.Descriptive Set Theory and Forcing ◦ 31 Note that en = [ n ∈Y ].  α Let b = n∈ω bn where each bn is Π0n (C) for some γn < α. (We leave the case of α = 0. For some n and γ p ∈ P we would have p ≤ bn . let Q ⊆ X be countable so that for any e ∈ E there exists H ⊆ P countable so that e = H and for every p ∈ H we have rank(p. Q) ≤ γn . then we claim: [ x ∈ U ] ∈ Σ0 (E). but this contradicts p ˆ p ≤ bn ≤ b = [ x ∈ U ] = ∅.
P = {p ∈ Pλ : ¬∃x ∈ X ( .. Similarly to the 1 β λ other argument we see that for any countable E we can choose a countable Q ⊆ X such for any s ∈ T with 2 ≤ r(s) = β < λ (so s = ) we have that [ x ∈ Us ] is not Σ0 (E). Then take P to be the subposet of Pλ such that doesn’t occur. β P ⊆ β<λ Π0 (C) and so since B = Σ0 (P) we get B = Σ0 (C). i. Now if P is dense in the cBa B.  β For ord(B) = ω1 we postpone until section 12. This is easy to see. x) ∈ Fp }. where Pαn is αn forcing. Another way to describe essentially the same thing is as follows: Let Pλ be λforcing.Descriptive Set Theory and Forcing 32 increasing to λ and let P = n<ω Pαn be the direct sum. because for each p ∈ P there exists β < λ with p ∈ Π0 (C). then ord(B) = λ. . Consequently.e. Hence ord(B) = λ.
B is countably complete means that for any sequence {bn : n < ω} in B there exists b ∈ B such that b = n∈ω bn . stone(B). so it is nowhere dense. A ⊆ B ∈ I and A ∈ F implies A ∈ I. i. / F/I is the countably complete boolean algebra formed by taking F and modding out by I. Let I be the σideal generated by the closed nowhere dense subsets of F (i.9 BOREL ORDER OF A FIELD OF SETS 33 9 Borel order of a ﬁeld of sets In this section we use the SikorskiLoomis representation theorem to transfer the abstract Borel hierarchy on a complete boolean algebra into a ﬁeld of sets. the ideal . Proof: Recall that the Stone space of B. 3. I contains the empty set. A ≈ B iﬀ A∆B ∈ I. This space is a compact Hausdorﬀ space in which the ﬁeld of clopen sets exactly corresponds to B.1 (Sikorski. I is closed under countable unions.Loomis. hence it has no interior. is the space of ultraﬁlters u on B with the topology generated by the clopen sets of the form: [b] = {u ∈ stone(B) : b ∈ u}. X ∈ I. A family F ⊆ P (X) is a σﬁeld iﬀ it contains the empty set and is closed under countable unions and complements in X.e. Let F be the σﬁeld generated by the clopen subsets of stone(B). For A ∈ F we use [A] or [A]I to denote the equivalence class of A modulo I. Theorem 9. and 4.e. This translates to the fact that given any countable family of clopen sets {Cn : n ∈ ω} in stone(B) there exists a clopen set C such that n∈ω Cn ⊆ C and the closed set C \ n∈ω Cn cannot contain a clopen set. I ⊆ F is a σideal in F iﬀ 1. 2. see [100] section 29) For any countably complete boolean algebra B there exists a σﬁeld F and a σideal I such that B is isomorphic to F/I.
it is consistent (Miller [75]) that either ord(H) ≤ 2 or ord(H) = ω1 for every countable H. Kunen pointed out the following result. Theorem 9. By induction on β it is easy to prove that for any Q ∈ F : Q ∈ Σ0 (H) iﬀ [Q]I ∈ Σ0 (C).Descriptive Set Theory and Forcing 34 of meager sets). so st(B) ∈ I and the same / holds for any nonempty clopen subset of stone(B). Since the countable union of clopen sets is equivalent to a clopen set modulo I it follows that the map C → [C] is an isomorphism taking the clopen algebra of stone(B) onto F/I.2 (Kunen see [75]) For every α ≤ ω1 there exists a ﬁeld of sets H such that ord(H) = α. Proof: Clearly we only have to worry about α with 2 < α < ω1 . Let C ⊆ F/I be a countable set of generators. β β From which it follows that ord(H) = α. Let B be the complete boolean algebra given by Theorem 8. Deﬁne H = {A ∈ F : [A]I ∈ C}.2. The Baire category theorem implies that no nonempty open subset of a compact Hausdorﬀ space is meager. Shortly after I gave a talk about my boolean algebra result (Theorem 8. . In fact. Let B F/I where F is a σﬁeld of sets and I a σideal.2). Note that there is no claim that the family H is countable.
1 (Sikorski. see [100] section 31) B is a countably generated ccc cBa iﬀ there exists a ccc σideal I in Borel(2ω ) such that B Borel(2ω )/I. Since B is countably generated there exists Cn ∈ F for n ∈ ω such that {[Cn ] : n ∈ ω} generates F/J where [C] denotes the equivalence class of C modulo J. We say that X is superILuzin iﬀ X is ILuzin and for every B ∈ Borel(2ω ) \ I we have that B ∩ X = ∅. Mahlo’s paper was overlooked and hence these kinds of sets have always been referred to as Luzin sets. A general theorem of Tarski is that any κcomplete κcc boolean algebra is complete. For I a σideal in Borel(2ω ) we say that X ⊆ 2ω is an ILuzin set iﬀ for every A ∈ I we have that X ∩ A is countable. which we know is equivalent to existence of separable metric spaces of all possible orders. Deﬁne I = {A ∈ Borel(2ω ) : φ(A) ∈ J}. Proof: To see that Borel(2ω )/I is countably generated is trivial since the clopen sets modulo I generate it. Apparently. we may assume by using the SikorskiLoomis Theorem. Now let h : X → 2ω be deﬁned by h(x)(n) = and deﬁne φ : Borel(2ω ) → F by φ(A) = h−1 (A). The following Theorem was ﬁrst proved by Mahlo [70] and later by Luzin [69] for the ideal of meager subsets of the real line. we claim that ˆ ˆ φ : Borel(2ω )/I → F/I deﬁned by φ([A]I ) = [φ(A)]J is an isomorphism of the two boolean algebras. then this isomorphism can be taken so as to map the clopen sets mod I onto C. For the other direction. Finally. Furthermore if B is generated by the countable set C ⊆ B. that B is F/J where F is a σﬁeld and J a σideal in F . We will need a sharper form of the representation theorem.10 CH AND ORDERS OF SEPARABLE METRIC SPACES 35 10 CH and orders of separable metric spaces In this section we prove that assuming CH that there exists countable ﬁeld of sets of all possible Borel orders. 1 if x ∈ Cn 0 if x ∈ Cn / . Theorem 10.
3 (Kunen see [75]) Suppose B = Borel(2ω )/I is a cBa. Proof: Note that the ord(X) is the minimum α such that for every B ∈ Borel(2ω ) there exists A ∈ Π0 (2ω ) with A ∩ X = B ∩ X. Corollary 10. ord(C) = α > 2. F1 such that X ∩ B = X ∩ ((A \ F0 ) ∪ F1 ). But since α > 2 we have that ((A\F0 )∪F1 ) is also Π0 and hence ord(X) ≤ α. .4 (CH) For every α ≤ ω1 there exists a separable metric space X such that ord(X) = α. and X is superILuzin. But since X / β is superILuzin we have that for every Π0 set A that X ∩ (B∆A) = ∅ and β hence X ∩ B = X ∩ A. Inductively choose xα ∈ 2ω so that xα ∈ Bα \ ({xβ : β < α} ∪ β<α Aα ).Descriptive Set Theory and Forcing 36 Theorem 10. α On the other hand for any β < α we know there exists a Borel set B such that for every Π0 set A we have B∆A ∈ I (since ord(C) > β). C ⊆ B are the clopen mod I sets. Hence there exist countable sets F0 . Then ord(X) = α. Theorem 10. Since X is Luzin we know that X ∩ (A∆B) is countable. Suppose I is a σideal in Borel(2ω ) containing all the singletons. Proof: Let I = {Aα : α < ω1 } and let Borel(2ω ) \ I = {Bα : α < ω1 }. ord(X) > β. Consequently.2 (Mahlo [70]) CH. Then X = {xα : α < ω1 } is a superILuzin set. α Since ord(C) = α we know that given any Borel set B there exists a Π0 α set A such that A∆B ∈ I. Then there exists a superILuzin set.
I found out in the real world (outside of Berkeley) that they are considered very diﬀerently. In Miller [75] it is shown that for every α < ω1 it is consistent there exists a separable metric space of order β iﬀ α < β ≤ ω1 . Question 10. But the general question is open. my way of thinking at the time was that proving something from CH is practically the same as just showing it is consistent. In fact.Descriptive Set Theory and Forcing 37 While a graduate student at Berkeley I had obtained the result that it was consistent with any cardinal arithmetic to assume that for every α ≤ ω1 there exists a separable metric space X such that ord(X) = α.5 For what C ⊆ ω1 is it consistent that C = {ord(X) : X separable metric }? . It never occurred to me at the time to ask what CH implied.
for every bn : n ∈ ω ∈ F bn ∈ U iﬀ ∃n bn ∈ U n∈ω 3. For any ccc σideal I in Borel(2ω ) and I ⊂ I with I < κ we have that 2ω \ I = ∅.11 MARTINSOLOVAY THEOREM 38 11 MartinSolovay Theorem In this section we the theorem below. The technique of proof will be used in the next section to produce a boolean algebra of order ω1 .. The following are equivalent for an inﬁnite cardinal κ: 1. for any I ⊂ I with I < κ 2ω \ I=∅ Proof: To see that (1) implies (2) note that for any bn : n ∈ ω ∈ Bω the set D = {p ∈ P : p ≤ − n bn or ∃n p ≤ bn } is dense. for any poset P which is ccc and family D of dense subsets of P with D < κ there exists a Pﬁlter G with G∩D = ∅ for all D ∈ D 2. i. for any family F ⊆ Bω with F < κ there exists an ultraﬁlter U on B which is Fcomplete.2 Let B = Borel(2ω )/I for some ccc σideal I and let P = B \ {0}. for any family D of dense subsets of P with D < κ there exists a Pﬁlter G with G ∩ D = ∅ for all D ∈ D 2. Lemma 11.e. Note also that any ﬁlter extends to an ultraﬁlter.e. i.. Theorem 11.1 (MartinSolovay [74]) The following are equivalent for an inﬁnite cardinal κ: 1. . MAκ .
I ⊆ M . I ∈ M . D ⊆ M . Let M be an elementary substructure of Hγ for suﬃciently large γ with M  < κ. The set Hγ is a natural model of all the axioms of set theory except possibly the power set axiom.Descriptive Set Theory and Forcing 39 To see that (2) implies (3) do as follows. they are hereditarily of cardinality less than γ.. Let F be all the ωsequences of Borel sets which are in M .2. This proves Lemma 11. Proof: This is true for subbasic clopen sets by deﬁnition. Let I = M ∩ I. To see that (3) implies (1). Since F < κ we know there exists U an Fcomplete ultraﬁlter on B.e. Deﬁne x ∈ 2ω by the rule: x(n) = i iﬀ [{y ∈ 2ω : y(n) = i}] ∈ U.  P Q is a poset. To prove the theorem it necessary to do a two step iteration. Then deﬁne G = {[B] ∈ BM : x ∈ B}.e. let M be an elementary substructure of Hγ for suﬃciently large γ with M  < κ. I ∈ M . Claim: For every Borel set B ∈ M : x ∈ B iﬀ [B] ∈ U. Inductive steps just use that U is an Mcomplete ultraﬁlter. Let Hγ stand for the family of sets whose transitive closure has cardinality less than the regular cardinal γ. . By (3) there exists x ∈ 2ω \ Let BM = B ∩ M . a poset and Q∈ V P be the Pname of a poset. Check G is a P ﬁlter which meets every D ∈ D. see Kunen [56]. i. i. Let P be ◦ I.
Finally we prove Theorem 11. and for every p.1. Theorem 11. . then G = {p : ∃q (p. P∗ Q is a class. First. q G ∈ H}. (1) implies (2) follows immediately from Lemma 11. Second. then P∗ Q is ccc.3 If G is Pgeneric over V and H is QG generic over V [G]. see Kunen [56] or Jech [44]. To see (2) implies (1) proceed as follows. We may also assume that the ccc poset P has cardinality less than κ. D ∩ Q is dense in Q for every D ∈ D. 2). q) ∈ K} is QG generic over V [G]. q ∈ Q if p and q are compatible (in P) then there exists r ∈ Q with r ≤ p and r ≤ q. Now replace P by Q. ◦ is a P∗ Q ﬁlter generic over V .5 (SolovayTennenbaum [104]) If P is ccc and  ◦ ◦ P “Q is ccc”. q) ∈ K} is Pgeneric over V and H = {q G : ∃p (p. it does not satisfy antisymmetry: x ≤ y and y ≤ x implies x = y. then ◦ G ∗ H = {(p. Use a LowenheimSkolem argument to obtain a set Q ⊆ P with the properties that Q < κ. The last condition on Q guarantees that Q has the ccc. q ) ≤ (p. q) iﬀ p ≤ p and p  p ˆ ˆ ˆ ◦ ◦ ◦ ◦ ◦ 40 q ≤ q. q) ∈ P∗ Q: p ∈ G. In general there are ˆ two problems with this. Note that κ ≤ c. q ) : p  q ∈Q} ordered by (ˆ. since (1) fails for FIN(c+ . ◦ Theorem 11.Descriptive Set Theory and Forcing Then we form the poset P∗ Q= {(p. For proofs of these results. Three of the main theorems are: Theorem 11.2. These can be solved by cutting down to a suﬃciently large set of nice names and modding out by the appropriate equivalence relation.4 If K is a P∗ Qﬁlter generic over V .
e. note that for each p ∈ P [ xp ∈ ∩n Un ] =  n∈ω [ xp ∈ Un ]  [ B ⊆ Un ]  [ xp ∈ Un ] =  B∈B. i. n∈ω Un . By Lemma 11. / Note that if D ⊆ P is dense in P..xp ∈B furthermore (p. To see this. q) ∈ P∗ Q iﬀ p ∈ P and q ∈ V is a ﬁnite set of consistent sentences of the form: 1. A simple description of P∗ Q can be given by: (p. then p and r are incompatible (so p  r ∈ G). “B ⊆U n ” where B ∈ B and n ∈ ω.1 it is enough to see that if P∗ Q⊆ B is dense in the ccc cBa algebra B. Consequently.Descriptive Set Theory and Forcing 41 Choose X = {xp : p ∈ P} ⊆ 2ω distinct elements of 2ω . ∅) = [ xp ∈ ∩n Un ]  . n ∈ ω}. every r ∈ P∗ Q is compatible with an element of D.2 and Sikorski’s ◦ Theorem 10. then B is countably generated. “x ∈U n ” where x ∈ X or / 2.  We claim that C generates B. Let C = {[ B ⊆ Un ] : B ∈ B. If G is Pﬁlter generic over V let Q be Silver’s forcing for forcing a Gδ set. it is enough ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ to ﬁnd suﬃciently generic ﬁlters for P∗ Q. with the additional requirement that whenever the sentence “x ∈U n ” is in / q and x = xr . n∈ω Let B ∈ V be a countable base for X. then D is predense in P∗ Q. in X such that G = {p ∈ P : xp ∈ Un }.
Descriptive Set Theory and Forcing ◦ 42 and so it follows that every element of P∗ Q is in the boolean algebra gen◦ erated by C and so since P∗ Q is dense in B it follows that C generates B. It follows from the MartinSolovay Theorem 11. Miller and Prikry [84] show that it is necessary to assume the continuum is regular in the above observation. Deﬁne X ⊆ 2ω to be a generalized ILuzin set for an ideal I in the Borel sets iﬀ X = c and X ∩ A < c for every A ∈ I. .1 that (assuming that the continuum is regular) MA is equivalent to for every ccc ideal I in the Borel subsets of 2ω there exists a generalized ILuzin set.
p is compatible with p. Let T be a nice tree of rank α (2 ≤ α < ω1 ). Note that rT (sˆn) < β < λ = rT (s) can happen only when λ is a limit ordinal and for any such s there can be at most ﬁnitely many n (because T is a nice tree). Now let E = {s ∈ domain(p0 ) : rT (s) ≤ β} and deﬁne p = p0 E.4. and p 3. So let q ∈ Pα have rank(q) < β and be incompatible with p. Before doing so we introduce a countable version of αforcing which will be useful for other results also. Deﬁne Pα = {p : D → ω : D ∈ [ω]<ω . then p and ˆ q are compatible.e. ∀s. for every p ∈ Pα and β ≥ 1 there exists p ∈ Pα such that ˆ 1. sˆn ∈ D p(s) = p(sˆn)}. Proof: First let p0 ≤ p be such that for every s ∈ domain(p) and n ∈ ω if rT (sˆn) < β < λ = rT (s) then there exists m ∈ ω with p0 (sˆnˆm) = p(s). for any q ∈ Pα if rank(q) < β and p and q are compatible. We need to show it is incompatible with p.1 rank : Pα → α+1 satisﬁes the Rank Lemma 7. rank(ˆ) ≤ β. ˆ From its deﬁnition it has rank ≤ β. Lemma 12. It is similar to one used in Miller [76] to give a simple proof about generating sets in the category algebra. For p ∈ Pα deﬁne rank(p) = max{rT (s) : s ∈ domain(p)} where rT is the rank function on T . i.12 BOOLEAN ALGEBRA OF ORDER ω1 43 12 Boolean algebra of order ω1 Now we use the MartinSolovay technique to produce a countably generated ccc cBa with order ω1 . This is ordered by p ≤ q iﬀ p ⊇ q. It is compatible with p since p0 is stronger than both. ˆ 2. There are ˆ only three ways for q and p to be incompatible: .
or 3. ∃s ∈ domain(q) ∃sˆn ∈ domain(p) q(s) = p(sˆ n). This map has the property that for every s ∈ T >0 the value of x(s) is the unique element of ω not in {x(sˆn) : n ∈ ω}. ∃s ∈ domain(p) ∩ domain(q) p(s) = q(s). In this case we have arranged p so that there exists m with ˆ p(s) = p(sˆnˆm) and so again q and p are incompatible. If rT (s) = β. ∃s ∈ domain(p) ∃sˆn ∈ domain(q) p(s) = q(sˆ n). 2.Descriptive Set Theory and Forcing 1. since if p ≤ q then either p and q are incompatible or there exists s ∈ domain(q) \ domain(p) in which case we can ﬁnd p ≤ p with p(s) = q(s). For (3) since sˆn ∈ domain(q) we know rT (sˆn) < β. 44 For (1) since rank(q) < β we know rT (s) < β and hence by construction s is in the domain of p and so q and p are incompatible. The poset Pα is separative.2 There exists a countable family D of dense subsets of Pα such that for every G a Pα ﬁlter which meets each dense set in D the ﬁlter G determines a map x : T → ω by p ∈ G iﬀ p ⊆ x. Otherwise p ˆ since T is a nice tree. ˆ ˆ . For (2) since ˆ ˆ rT (sˆn) < rT (s) < β we get the same conclusion. Also for each s ∈ T >0 and k ∈ ω the set k Es = {p : p(s) = k or ∃n p(sˆn) = k} is dense. ˆ ˆ Lemma 12. rT (sˆn) < β < rT (s) = λ a limit ordinal. Proof: For each s ∈ T the set Ds = {p : s ∈ domain(p)} is dense. then s ∈ domain(ˆ) and so q and p are incompatible.
with the additional proviso that whenever “xα. deﬁne ( iﬀ p∈ α<ω1 Pα )∗ Q by (p. Let 0 X = {xα. For G = Gα : α < ω1 which is α<ω1 Pα generic over V . it follows that for each p ∈ Pα p = [ p ⊆ x ]  and for any s ∈ T >0 and k [ x(s) = k ] =  m∈ω 45 [ x(sˆ m) = k ]. i.e.n ∈U n ” ∈ q then s is in the / domain of pα and pα (s) = n. This is a little stronger than saying p  q ∈ Q.. i. so the usual deltalemma argument shows that it is ccc.Descriptive Set Theory and Forcing Now if Pα ⊆ B is dense in the cBa B. Now let α<ω1 Pα be the direct sum. use X and Silver forcing to code the rank zero parts of each Gα . q) = max{rank(pα ) : α < ω1 } which means we ignore q entirely. p = pα : α < ω1 with pα ∈ Pα and pα = 1α = ∅ for all but ﬁnitely many α. s ∈ Tα ..s. n ∈ ω} be distinct elements of 2ω .  Consequently if C = {p ∈ Pα : domain(p) ⊆ T 0 } then C ⊆ B has the property that ord(C) = α + 1. “B ⊆U n ” where B is clopen and n ∈ ω. q) ∈ ( ◦ ◦ α<ω1 Pα )∗ Q α<ω1 ◦ Pα and q is a ﬁnite set of consistent sentences of the form: 1.e. . “x ∈U n ” where x ∈ X or / 2. The rank function ◦ ◦ ◦ rank : ( α<ω1 Pα )∗ Q:→ ω1 is deﬁned by rank( pα : α < ω1 .n ∈ 2ω : α < ω1 .s. This forcing is equivalent to adding ω1 Cohen reals. ˇ but would be true for a dense set of conditions.
and pα (s) = n. s ∈ Tα and n ∈ ω if p = ( pα : α < ω1 . p is compatible with p. then ◦ p = [ xα. ◦ ◦ ◦ Let ( α<ω1 Pα )∗ Q⊆ B be a dense subset of the ccc cBa B.1 to each pα to obtain pα and then let ˆ p = ( pα : α < ω1 . ˆ 2. q).Descriptive Set Theory and Forcing Lemma 12. . then all countable sets of generators have order ω1 . 0 We know that for each α < ω1 . rank(ˆ) ≤ β. ˆ then p and q are compatible. This is because any countable set will be generated by a countable stage. Proof: Apply Lemma 12. A strange thing about ω1 is that if one countable set of generators has order ω1 . n ∈ ω}. q) is the condition for which pα is the function with domain {s}. and p 3.  ˇ Note that [ x ∈ ∩n∈ω Un ] =  n∈ω ◦ ◦ [ x ∈ Un ] =  n∈ω {[ B ⊆Un ] : x ∈ B}  ˇ and also each Pα is generated by 0 {p ∈ Pα : domain(p) ⊆ Tα }.s. and the rest of p is the trivial condition. for any q ∈ ( α<ω1 Pα )∗ Q if rank(q) < β and p and q are compatible. We show that B is countably generated and ord(B) = ω1 . One set of generators for B is C = {[ B ⊆Un ] : B clopen .n ∈ ˇ n∈ω Un ]. ˆ ˆ This is still a condition because pα retains all the rank zero part of pα which ˆ is needed to force q ∈Q.3 For every p ∈ ( ◦ ◦ 46 ˆ α<ω1 Pα )∗ Q and β ≥ 1 there exists p in the poset ( α<ω1 Pα )∗ Q such that 1.
Descriptive Set Theory and Forcing 47 From these facts it follows that C generates B. / β Otherwise by Lemma 8.2. . Then b ∈ Σ0 (C). there would be some a ≤ b with rank(a. β This proves the ω1 case of Theorem 8. C) < β.4 that the order of C is ω1 . It follows from Lemma 8.4. q) be the condition all of whose components are trivial except for pβ . but then pa would not have in its domain. and pβ any the function with domain . For any β < ω1 let b = ( pα : α < ω1 .
13 LUZIN SETS
48
13
Luzin sets
In this section we use Luzin sets and generalized ILuzin sets to construct separable metric spaces of various Borel orders. Before doing so we review some standard material on the property of Baire. Given a topological space X the σideal of meager sets is deﬁned as follows. Y ⊆ X is nowhere dense iﬀ the interior of the closure of Y is empty, i.e, int(cl(Y )) = ∅. A subset of X is meager iﬀ it is the countable union of nowhere dense sets. The Baire category Theorem is the statement that nonempty open subsets of compact Hausdorﬀ spaces or completely metrizable spaces are not meager. A subset B of X has the Baire property iﬀ there exists U open such that B∆U is meager. Theorem 13.1 (Baire) The family of sets with the Baire property forms a σﬁeld. Proof: If B∆U is meager where U is open, then B∆cl(U ) = (B \ cl(U )) ∪ (cl(U ) \ B) and (B \ cl(U )) ⊆ B \ U is meager and (cl(U ) \ B) ⊆ (U \ B) ∪ (cl(U ) \ U ) is meager because cl(U ) \ U is nowhere dense. Therefore, ∼ B ∆ ∼ cl(U ) = B∆cl(U ) is meager. If Bn ∆Un is meager for each n, then (
n∈ω
Bn )∆(
n∈ω
Un ) ⊆
n∈ω
Bn ∆Un
is meager. Hence every Borel set has the property of Baire. Theorem 13.2 Suppose that every nonempty open subset of X is nonmeager, then B = Borel(X)/meager(X) is a cBa.
Descriptive Set Theory and Forcing
49
Proof: It is enough to show that it is complete. Suppose Γ ⊆ B is arbitrary. Let U be a family of open sets such that Γ = {[U ]meager(X) : U ∈ U}. Let V = U and we claim that [V ] is the minimal upper bound of Γ in B. Clearly it is an upper bound. Suppose [W ] is any upper bound for Γ with W open. So U \ W is meager for every U ∈ U. We need to show that V \ W is meager (so [V ] ≤ [W ]). V \ W ⊆ cl(V ) \ W and if the latter is not nowhere dense, then there exists P a nonempty open set with P ⊆ cl(V ) \ W . Since V = U we may assume that there exists U ∈ U with P ⊆ U . But P is a nonempty open set and [P ] ≤ [W ] so it is impossible for P to be disjoint from W . We say that X ⊆ 2ω is a super Luzin set iﬀ for every Borel set B the set X ∩ B is countable iﬀ B is meager. (This is equivalent to superILuzin where I is the ideal of meager sets.) It is easy to see that if X is an ordinary Luzin set, then in some basic clopen set C it is a super Luzin set relative to C. Also since ω ω can be obtained by deleting countably many points from 2ω it is clear that having a Luzin set for one is equivalent to having it for the other. With a little more work it can be seen that it is equivalent to having one for any completely metrizable separable metric space without isolated points. The generic set of Cohen reals in the Cohen real model is a Luzin set. Let FIN(κ, 2) be the partial order of ﬁnite partial functions from κ into 2. If G is FIN(κ, 2)generic over V and for each α < κ we deﬁne xα by xα (n) = G(ω ∗ α + n), then X = {xα : α < κ} is a Luzin set in V [G]. Theorem 13.3 (Miller [75]) If there exists a Luzin set in ω ω , then for every α with 3 ≤ α < ω1 there exists Y ⊆ ω ω with ord(Y ) = α. Proof: Let Tα be the nice αtree used in the deﬁnition of Pα = {p : p : D → ω, D ∈ [Tα ]<ω , ∀s, sˆn ∈ D p(s) = p(sˆn)}. Let Qα be the closed subspace of ω Tα Qα = {x ∈ ω Tα : ∀s, sˆn ∈ Tα x(s) = x(sˆn).
Descriptive Set Theory and Forcing For each p ∈ Pα we get a basic clopen set [p] = {x ∈ Qα : p ⊆ x}.
50
It easy to check that Qα is homeomorphic to ω ω . Hence there exists a super 0 Luzin set X ⊆ Qα . Consider the map r : Qα → ω Tα deﬁned by restriction, 0 i.e., r(x) = x Tα . Note that by Lemma 12.2 there exists a countable sequence of dense open subsets of Qα , Dn : n ∈ ω , such that r is onetoone on n∈ω Dn . Since n∈ω Dn is a comeager set in Qα and X is Luzin we may assume that X⊆ Dn .
n∈ω
Y is just the image of X under r. (So, in fact, Y is the onetoone continuous image of a Luzin set.) An equivalent way to view Y is just to imagine X with the topology given by
0 B = {[p] : p ∈ Pα , domain(p) ⊆ Tα }.
We know by Lemma 8.4 that ord{[B] : B ∈ B} = α + 1 as a subset of Borel(Qα )/meager(Qα ) which means that: α+1 is minimal such that for every B ∈ Borel(Qα ) there exists a Σ0 (B) α+1 set A such that such that B∆A is meager in Qα . This translates (since X is superLuzin) to: α+1 is minimal such that for every B ∈ Borel(Qα ) there exists a Σ0 (B) α+1 set A such that such that (B∆A) ∩ X is countable. Which means for Y that: α + 1 is minimal such that for every B ∈ Borel(Y ) there exists a Σ0 (Y ) α+1 set A such that such that B∆A is countable. But since countable subsets of Y are Σ0 and α > 2, this means ord(Y ) = 2 α + 1. To get Y of order λ for a limit λ < ω1 just take a clopen separated union of sets whose order increases to λ. Now we clean up a loose end from Miller [75]. In that paper we had shown that assuming MA for every α < ω1 there exists a separable metric space X with α ≤ ord(X) ≤ α + 2 or something silly like that. Shortly afterwards, Fremlin supplied the missing arguments to show the following.
5 (MA) For any Z ⊆ Qα of cardinality less than c. pn : n < N ) : F ∈ [Z]<ω . Dn is predense in Pα .2 gives us a set X ⊆ Qα of cardinality c such that for every Borel set B ∈ Borel(Qα ) we have that B is meager iﬀ B ∩ X has cardinality less than c. pn : n < N ) ≤ (H. the Mahlo construction 10.4. Proof: Force with the following poset 0 P = {(F. This proves Theorem 2 13. What we need to see to complete the proof is that: for every Z ⊆ X of cardinality less than c there exists a Σ0 (B) set F such 2 that F ∩ X = Z. pn = qn for n < M .3 we see that: α+1 is minimal such that for every B ∈ Borel(Qα ) there exists a Σ0 (B) α+1 set A such that such that (B∆A) ∩ X has cardinality less than c. By applying the Lemma we get that for every Z ⊆ X of cardinality less than c there exists a Σ0 (B) set F which is meager in Qα and such that 2 Z ⊆ F ∩ X. and for each x ∈ H and M ≤ n < N we have x ∈ [pn ]. so there 2 exists an F0 a Σ0 (B) set such that Z = (F ∩ X) ∩ F0 . qn : n < M ) iﬀ F ⊇ H.4 (Fremlin [26]) MA implies that for every α with 2 ≤ α ≤ ω1 there exists a second countable Hausdorﬀ space X with ord(X) = α. Since / this forcing is ccc we can apply MA with the appropriate choice of family of dense sets to get Dn = {pm : m > n} to do the job. By Theorem 5. Lemma 13. p ∈ Dn implies domain(p) ⊆ Tα .Descriptive Set Theory and Forcing 51 Theorem 13. and 3. N ≥ M . domain(p) ∈ [Tα ]<ω } where (F. Z ∩ n∈ω s∈Dn [s] = ∅. .1 every subset of r(F ∩ X) is a relative Σ0 in Y . 0 2. Proof: Since the union of less than continuum many meager sets is meager. N < ω. Letting B be deﬁned as in the proof of Theorem 13. there exists Dn : n ∈ ω such that: 1. But since F is meager we know F ∩ X has cardinality less than c.
In order to insure the construction can keep going at limit ordinals we will need to use a fusion argument.2 (Fusion) Suppose (Tn : n ∈ ω) is a sequence of perfect sets such that Tn+1 ≤n Tn for every n ∈ ω. Given two perfect trees T and T and n ∈ ω deﬁne T ≤n T iﬀ T ⊆ T and the ﬁrst 2<n splitting nodes of T remain in T . a tree with the property that for every s ∈ T there exist a t ∈ T such that both tˆ0 ∈ T and tˆ1 ∈ T . G) is FIN(κ. then the order of X is 2. 2) is isomorphic to FIN(κ. The main idea of the proof is to construct an Aronszajn tree of perfect sets.1 Suppose G is FIN(κ. 2) × FIN(ω1 . Recall that a perfect set corresponds to the inﬁnite branches [T ] of a perfect tree T ⊆ 2<ω . ¢) and a family of perfect sets ([Ts ] : s ∈ A) such that ⊇ is the same order as ¢. 2)generic over V . This is because FIN(κ. 2) × FIN(ω1 . Then T = n∈ω Tn is a perfect tree and T ≤n Tn for every n ∈ ω. I thought there weren’t any. We are going to use the fact that forcing with FIN(ω1 . Such a T is called a splitting node of T . We will then show that if X = {xs : s ∈ A} is such that xs ∈ [Ts ].3 that since there is an uncountable Luzin set in Cohen real model that for every α with 3 ≤ α ≤ ω1 there is an uncountable separable metric space X with ord(X) = α. 2) and so by the product lemma we may replace V by V [H] where (H. Proof: We may assume that κ = ω1 . We already know from Theorem 13. Lemma 14. We construct an Aronszajn tree (A. 2)generic over V where κ ≥ ω1 .14 COHEN REAL MODEL 52 14 Cohen real model I have long wondered if there exists an uncountable separable metric space of order 2 in the Cohen real model. Proof: . i. a technique ﬁrst used by Todorcevic (see Galvin and Miller [30]).. Then in V [G] there is a separable metric space X of cardinality ω1 with ord(X) = 2. There is a natural correspondence of the splitting nodes of a perfect tree T and 2<ω . Theorem 14. 2) is equivalent to any ﬁnite support ω1 iteration of countable posets.e.
Given an Aronszajn tree A we let Aα be the nodes of A at level α. then [Ts ] and [Tt ] are disjoint. 2) we may assume that G = G α : α < ω1 where Gβ is FIN(ω. Item (5) is what we do a successor levels and guarantees the set we are building has order 2. By identifying FIN(ω1 . {Ts : s ∈ A<α } ∈ V [Gβ : β < α]. 2. 3. 2)generic over V [Gα : α < β] for each β < ω1 . Here are the details of our construction. The ﬁrst three items simply say that {[Ts ] : s ∈ A} and its ordering by ⊆ determines (A. for each s ∈ Aα and t ∈ Aα+1 with s ¡ t.Descriptive Set Theory and Forcing 53 If T = n<ω Tn . 2) with α<ω1 FIN(ω. Item (4) is there in order to allow the construction to proceed at limits levels. then T is a perfect tree because the ﬁrst 2<n splitting nodes of Tn are in Tm for every m > n and thus in T . i. ¢) and a family of perfect sets ([Ts ] : s ∈ A) such that 1. if s and t are distinct elements of Aα .e. . We use Gα : α < ω1 to construct an Aronszajn tree (A. 4. s ¢ t implies Ts ⊇ Tt . Item (6) is a consequence of the construction and would be true for a closed unbounded set of ordinals no matter what we did anyway. every s ∈ Aα has inﬁnitely many distinct extensions in Aα+1 . we have that [Tt ] is a generic perfect subset of [Ts ] obtained by using Gα (explained below in Case 2). and 6. ). so what we really have here is an Aronszajn tree of perfect sets. Aα = {s ∈ A : {t ∈ A : t ¡ s} has order type α} and A<α = β<α Aβ . for each s ∈ A<α and n < ω there exists t ∈ Aα such that Tt ≤n Ts . 5.
i. we can obtain q ≥ p such that for every terminal node s of q either [s] ∩ B is . Now let {Tt : t ∈ Aα } be a countable collection of perfect trees so that for every n and s ∈ Aαn there exists t ∈ Aα with Tt ≤n Ts .Descriptive Set Theory and Forcing 54 Case 1. [TG ] ⊆ D. i. for any D ⊆ [T ] dense open in [T ] and coded in M . p ∈ P(T ) iﬀ p is a ﬁnite subtree of T and p ≤ q iﬀ p ⊇ q and p is an end extension of q. if necessary. then TG = {p : p ∈ G} is a perfect subtree of T . Working in V [Gβ : β < α] choose a sequence αn for n ∈ ω which strictly increases to α. To see why this is true let p ∈ P(T ) and B Borel. The construction is done uniformly enough so that we already have that {Ts : s ∈ A<α } ∈ V [Gβ : β < α]. This means that for any Borel set B ⊆ [T ] coded in M . If T = m>n Tsm . there exists an clopen set C ∈ M such that C ∩ [TG ] = B ∩ [TG ]. This implies item (4) because for any s ∈ A<α and n < ω there exists some m ≥ n with s ∈ A<αm hence by inductive hypothesis there exists s ∈ Aαm with Ts ≤n Ts and by construction ˆ ˆ there exists t ∈ Aα with Tt ≤m Ts and so Tt ≤n Ts as desired. every new node of p extends a terminal node of q. Given a perfect tree T ⊆ 2<ω deﬁne the countable partial order P(T ) as follows. Furthermore. It is easy to see that if G is P(T )generic over a model M . then by Lemma 14.e. α a limit ordinal. Since B has the Baire property relative to [T ] by extending each terminal node of p.e. Given any sn ∈ Aαn we can choose by inductive hypothesis a sequence sm ∈ Aαm for m ≥ n such that Tsm+1 ≤m Tsm ..2 we have that T ≤n Tsn . the branches of TG are Cohen reals (relative to T ) over M .. ˆ Case 2. Successor stages. Suppose we already have constructed {Ts : s ∈ A<α+1 } ∈ V [Gβ : β < α + 1].
Descriptive Set Theory and Forcing 55 meager in [T ] or [s] ∩ B is comeager in [T ] ∩ [s]. Note that genericity also guarantees that corresponding perfect sets will be disjoint. n) = {p ∈ P(T ) : p end extends the ﬁrst 2<n splitting nodes of T }. To verify this. m ∈ ω of generic perfect trees such that Ts. This poset is countable and hence Gα+1 determines a sequence Ts. m) : s ∈ Aα . B∩ t∈Aα+1 [Tt ] = t∈Aα+1 (Ct ∩ [Tt ]). Finally to determine Aα+1 consider {P(Ts . To get TG ≤n T we could instead force with P(T. If we let C be union of all [s] for s a terminal node of q such that [s] ∩ B is comeager in [T ] ∩ [s]. Then X has order 2. then q  B ∩ TG = C ∩ TG . This ends the construction. By ccc there exists a countable α such that B is coded in V [Gβ : β < α + 1]. Hence B ∩ X is equal to a Σ0 set intersected X: 2 X∩ t∈Aα+1 (Ct ∩ [Tt ]) . Hence. For any Borel set B coded in V [Gβ : β < α+1] and Tt for t ∈ Aα+1 there exists a clopen set Ct such that Ct ∩ [Tt ] = B ∩ [Tt ].m : s ∈ Aα . m ∈ ω}.m ≤m Ts . Suppose X = {xs : s ∈ A} is such that xs ∈ [Ts ] for every s ∈ A. By taking generic perfect sets at successor steps we have guaranteed the following. We deﬁne Aα+1 to be this set of generic trees. let B ⊆ 2ω be any Borel set.
Consequently. in what I think of as “the Cohen real model”. To prove it we will need the following lemma. If the ground model satisﬁes CH. then X has order ω1 in V [G].3 Suppose G is FIN(κ. then there exists a Sierpi´ski set. i. Such a set n has order 2 (see Theorem 15. For α = 1 let B ∈ V be a countable base for the closed subsets of X. 2)generic over V and V = ord(X) = α Then V [G] = ord(X) = α By the usual ccc arguments it is clearly enough to prove the Theorem for FIN(ω. 2). there are separable metric spaces of cardinality ω1 and order α for every α with 2 ≤ α ≤ ω1 .e. Suppose p  “B is {x ∈ X : p  x ∈B } ˇ ◦ ◦ ◦ .Descriptive Set Theory and Forcing union a countable set: (B ∩ X) \ t∈Aα+1 56 [Tt ] and therefore B ∩ X is Σ0 in X. Theorem 14. ◦ i. α Proof: This is proved by induction on α. X is a second countable Hausdorﬀ space in V . see [57]) Suppose p ∈ FIN(ω.4 (Kunen. Lemma 14.1) in V and therefore by the next theorem it has order 2 in V [G]. α Then the set is a Π0 set in X. and B is a name such that ˇ p  B ⊆ X is a Π0 set. 2 Another way to get a space of order 2 is to use the following argument. the model obtained by adding ω2 Cohen reals to a model of CH.. 2).e. every closed set is the intersection of elements of B. It also follows from the next theorem that if X = 2ω ∩ V .
Let βn be a sequence which α is either constantly α − 1 if α is a successor or which is unbounded in α if α is a limit. Then in V [G] for any Borel set B ∈ Borel(X) B= {x ∈ X : p  x ∈B }. By the usual forcing facts there exists a sequence of names Bn : n ∈ ω such that for each n. it is a Σα set. each of the sets {x ∈ X : p  x ∈B } is a Borel set in V . it follows that B is a Σ0 set. ˇ {x ∈ X : ∀q ≤ p ∀B ∈ B (q  “ B ⊆ C” → x ∈ C)} is closed in X.Descriptive Set Theory and Forcing ◦ 57 ˇ a closed set in X”. ˇ p∈G ◦ ◦ By the lemma. and ˇ 0 since ord(X) = α. Hence. ˇ Now let us prove the Theorem. So. p  Bn ∈ Π0 n . But for every C ∈ B if q  “B ⊆ C”. Then for every x ∈ X p x∈B ˇ iﬀ ∀n ∈ ω p  x ∈∼ Bn iﬀ ∀n ∈ ω ∀q ≤ p q  x ∈ Bn . Then for every x ∈ X p  “ˇ ∈B ” iﬀ for every q ≤ p and x ◦ ˇ then x ∈ C. Consequently. β and p B= n<ω ◦ ∼ Bn . ◦ Now suppose α > 1 and p  B ∈ Π0 (X). Suppose V =“ord(X) = α”. α . {x ∈ X : p  x ∈B } = n∈ω q≤p ◦ ◦ ∼ {x : q  x ∈Bm }.
Lemma 14. 2)generic over V where V is a model of CH and κ ≥ ω2 . The reason is we can just replace Aα by n ˆ Aα = {p (∼ R) : p ∈ Aα and p R ∈ G} n n . For α each α and n choose a maximal antichain Aα ∪ Bn such that n p  τα (n) = 0 for each p ∈ Aα and n α p  τα (n) = 1 for each p ∈ Bn . β Part of this argument is similar to one used by Judah and Shelah [46] who showed that it is consistent to have a Qset which does not have strong measure zero. there exists R such that Xα ∩ Xβ = R for every α = β.Descriptive Set Theory and Forcing 58 V [G] = ord(X) ≤ α. α Let Xα ⊆ κ be union of domains of elements from n∈ω Aα ∪ Bn . i. Proof: This will follow easily from the next lemma. Then V [G] models that for every X ⊆ 2ω with X = ω2 there exists a Luzin set Y ⊆ 2ω and a onetoone continuous function f : Y → X. Theorem 14.6 (Miller [81]) Suppose G is FIN(κ. It is natural to ask if there are spaces of order 2 of higher cardinality.5 Suppose G is FIN(κ. we have ord(X) ≥ 3. 2)generic over V where V is a model of CH and κ ≥ ω2 . Proof: Let τα : α < ω2 be a sequence of names for distinct elements of X. To see that ord(X) ≥ α let β < α and suppose in V the set A ⊆ X is Σ0 but not Π0 . We can assume that R is the empty set. This must remain true in V [G] otherwise β β there exists a p ∈ G such that ˇ p  “A is Π0 ” β but by the lemma {x ∈ X : p  x ∈ A} = A ˇ ˇ is Π0 which is a contradiction. Since each n Xα is countable we may as well assume that the Xα ’s form a ∆system. Then in V [G] for every separable metric space X with X > ω1 .e.
Let Xn ⊆ Pn be a homeomorphic copy of X for each n < ω.Descriptive Set Theory and Forcing α and similarly for Bn . But note that f (D) is a countable set which cannot be Gδ in X. Bn ) : n ∈ ω . Let jα : Xα → ω : α < ω2 59 be a sequence of bijections in the ground model. It is natural to ask about the cardinalities of sets of various orders in this model. . n −1 The Luzin set is Y = {yα : α < ω2 } where yα (n) = G(jα (n)). (An . The continuous function. we may as well assume that there exists a single sequence. For example. We begin with a stronger version of Theorem 13. Bn : n ∈ ω to (An . Then D is not Gδ in Y . With this cheat in mind let us deﬁne a stronger notion of order. Then let V [G R] be the new ground model. Take a clopen separated union of a large Luzin set (which has order 3) and a set of size ω1 with order β. But this is also open to a simple trick of combining a small set of order β with a large set of small order. f . 2). This is because any Gδ set containing D is comeager and therefore must meet Y in an uncountable set. let X ⊆ 2ω be a clopen separated union of a Luzin set of cardinality κ and set of cardinality ω1 of order β ≥ 3. By CH. If f : Y → X is onetoone continuous function from a Luzin set Y . A separable metric space X has hereditary order β iﬀ every uncountable Y ⊆ X has order β. Bn ) : n ∈ ω . because f −1 (f (D)) would be Gδ in Y and since f is onetoone we have D = f −1 (f (D)). then ord(X) ≥ 3. 2) to FIN(ω. One possible way to strengthen the notion of order is to say that a space X of cardinality κ has essential order β iﬀ every nonempty open subset of X has order β and cardinality κ. Then n∈ω Xn is a set of cardinality κ which has essential order β. Let Pn : n ∈ ω be a sequence of disjoint nowhere dense perfect subsets of 2ω with the property that for every s ∈ 2<ω there exists n with Pn ⊆ [s]. But note that there is a trivial way to get a large set of order β. This proves the Theorem. Bn ) : n ∈ ω : f (x)(n) = 0 iﬀ ∃m x m ∈ An .3. such that α every jα maps Aα . To see this assume that Y is dense and let D ⊆ Y be a countable dense subset of Y . This proves the Lemma. Each jα extends to an order preserving map from FIN(Xα . is the map determined by (An .
Let Qα be the following partial order. is closed and uncountable. Let G be FIN(κ. then since {[B ∩ C] : B ∈ B} is α. Let ˆ Qα = Qαn be the closed subspace of ω Tαn n∈ω and let B be the collection of clopen subsets of Qα which are given by rank zero conditions of Q(α). then for every α with 2 < α < ω1 there exists a separable metric space Y of cardinality κ which is hereditarily of order α. Proof: This is a slight modiﬁcation of the proof of Theorem 13.e.8 Suppose that in V there is a separable metric space. Then in V [G] the space X has hereditary order β. we have by the proof of Theorem 13. . The rest of the proof is same except we use Qα+1 instead of Pα for successors and Qα for limit α instead of taking a clopen separated union. i. therefore must have nonempty interior. the set of all points all of whose neighborhoods contain uncountably many points of Y . As in the proof of Theorem 13. all rectangles of the form n∈ω [pn ] such that 0 pn ∈ Qαn with domain(p) ⊆ Tα and pn the trivial condition for all but ﬁnitely many n. (The accumulation points of Y .Descriptive Set Theory and Forcing 60 Theorem 14. Theorem 14. that the order of Y is α. Because we took the direct sum ˆ we get the stronger property that for any nonempty clopen set C in Qα the order of {[B ∩ C] : B ∈ B} is α.) If C is a nonempty clopen set in the interior of the accumulation points of Y .7 If there exists a Luzin set X of cardinality κ. with hereditary order β for some β ≤ ω1 . By using the direct sum (even in the successor case) we get a stronger property for the order. ˆ But know given X a Luzin set in Qα we know that for any uncountable ˆ Y ⊆ X there is a nonempty clopen set C ⊆ Qα such that Y ∩ C is a superLuzin set relative to C. then αn is a coﬁnal increasing sequence in α and if α = β + 1 then αn = β for every n. X.3.3 we get that the order of {[B] : B ∈ B} ˆ ˆ as a subset of Borel(Qα )/meager(Qα ) is α. Let αn : n ∈ ω be a sequence such that if α is a limit ordinal. 2)generic over V for any κ ≥ ω.3..
if x ∈ T \ R. ˇ According to Lemma 14. (assuming α α ≥ 3) there are uncountably many x ∈ A with x ∈ R∆T . then there exists r ≤ r in FIN(Γ. If not. For contradiction. Consequently.4 the set T is Π0 (A). Working in V by the usual ∆system argument we can get q ≤ p and px : x ∈ A for some A ∈ [X]ω1 such that and px ≤ q and px  x ∈Y ˇ for each x ∈ A and dom(px ) ∩ dom(py ) = dom(q) for distinct x and y in A. We claim that in V [G] the set R ∩ Y is not Π0 (Y ). then ˇ r ∪ px  “ˇ ∈Y ∩ S and x ∈Y ∩R”. Since A is uncountable we know that in V the order of A is ω1 . 2) such ˆ that ◦ r  x ∈S ˆ ˇ / ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ . But now. x / On the other hand. there α α exists r ≤ q and S such that r  “ Y ∩R =Y ∩ S and S ∈ Π0 (A)”.Descriptive Set Theory and Forcing Proof: In V [G] let Y ⊆ X be uncountable. if x ∈ R \ T . suppose that p  ord(Y ) ≤ α and  Y  = ω1 ◦ ◦ 61 for some p ∈ FIN(κ. Deﬁne ◦ T = {x ∈ A : q  x ∈S }. Consequently. there exists R ⊆ A which is Σ0 (A) but not α Π0 (A). α Since Borel sets are coded by reals there exists Γ ∈ [κ]ω ∩ V such that for any x ∈ A the statement “ˇ ∈S ” is decided by conditions in FIN(Γ. This can be done since the px form a ∆ system. Choose such an x which also has the property that dom(px ) \ dom(q) is disjoint from Γ. 2) and α < β. 2) and x also let Γ be large enough to contain the domain of r.
Of course X is really just the same as X but retopologized using B as a family of basic open sets. Note that in fact B has the property that for every b ∈ B we have ord(b) = ω1 . Since Q has ccc. it easy to show that any ccc cBa of order ω1 would have to contain an element b such that every c ≤ b has order ω1 . Hence by Mahlo’s construction (Theorem 10.9 (CH) There exists X ⊆ 2ω such that X has hereditary order ω1 . By the proof of the SikorskiLoomis Theorem 9. Alternatively. ˜ If X ⊆ 2ω is the image of X under the characteristic function of the sequence ˜ ˜ B (see Theorem 4.Descriptive Set Theory and Forcing and consequently. Also note that for any nonmeager Borel set B in Q there exists a basic open set C and F ∈ F with C \F ⊆ B. there exists a family F of meager sets with F = ω1 such that every meager set is a subset of one in F.2 there exists a countably generated ccc cBa B which has order ω1 .2) there exists a set X ⊆ Q with the property that for any Borel subset B of Q B ∩ X ≤ ω iﬀ B meager.1). ◦ ◦ ◦ 62 Theorem 14. then X has hereditary order ω1 . ˇ r ∪ px  “ˇ ∈Y ∩ S and x ∈Y ∩R”. Consequently.1 we know that B is isomorphic to Borel(Q)/meager(Q) where Q is a ccc compact Hausdorﬀ space with a basis of cardinality continuum. For any b ∈ B with b = 0 let ord(b) be the order of the boolean algebra you get by looking only at {c ∈ B : c ≤ b}. by using CH. Proof: By Theorem 8. Let B be a countable ﬁeld of clopen subsets of Q such that {[B]meager(Q) : B ∈ B} generates B. . Let R = {X ∩ B : B ∈ B}. every open dense set contains an open dense set which is a countable union of basic open sets. ˆ x/ Either way we get a contradiction and the result is proved.
To prove the other direction. Finally. Proof: Immediate from Theorem 14. 2)generic over V where V is a model of CH and κ ≥ ω2 . Theorem 14. B ∩ Y  ≤ ω iﬀ B meager. we show that there are no large spaces of hereditary order ω1 in the Cohen real model.8 and 14. Theorem 14. Proof: Let p and q stand for nonempty basic clopen sets. Obviously if B is meager then B ∩ Y is countable. since B ∩ X is countable. Then in V [G] there exists a separable metric space X with X = ω1 and hereditarily of order ω1 .9. Then in V [G] for every separable metric space X with X = ω2 there exists Y ∈ [X]ω2 with ord(Y ) < ω1 . But B= {Bq : q ∈ Σ} is a comeager Borel set which meets Y in a countable set.6 we can ﬁnd Gα : α < ω2 ∈ V [G] .10 Suppose G is FIN(κ.Descriptive Set Theory and Forcing 63 Let Y ∈ [X]ω1 .11 Suppose G is FIN(κ. Claim: There exists a basic clopen p in Q such that for every Borel B ⊆ p. Since ord(p) = ω1 for any basic clopen set the following claim shows that the order of Y (or rather the image of Y under the characteristic function of the sequence B) is ω1 . Proof: By the argument used in the proof of Lemma 14. By using ccc there exists a countable dense family Σ and Bq for q ∈ Σ with Bq ⊆ q Borel and comeager in q such that Bq ∩ Y is countable. This implies that Y is countable since X is contained in B except for countable many points. suppose for contradiction that for every p there exists q ≤ p and Borel Bq ⊆ q such that Bq is comeager in q and Bq ∩ Y is countable. 2)generic over V where V is a model of CH and κ ≥ ω.
Consequently. I claim that the order of Y is essentially less than or equal to the order of F in F.Descriptive Set Theory and Forcing 64 which is α<ω2 FIN(ω. 2)generic over V and a FIN(ω.12 Let B be a cBa.  ˇ  Then for each B ⊆ 2ω a Π0 set coded in V the boolean value [ τ ∈ B ] is α ω 0 0 Πα (F) and conversely. Let F be the complete subalgebra of B which is generated by F. Lemma 14. for every c ∈ Πα (F) there exists a B ⊆ 2 a Π0 set α ˇ coded in V such that c = [ τ ∈ B ]. Now suppose the order of F in F is α. 2) is dense. This is because F contains a countable dense set: D={ {c ∈ F : p ≤ c} : p ∈ FIN(ω. Let F = {[ τ ∈ C ] : C ⊆ 2ω clopen }  where boolean values are in the unique complete boolean algebra B in which FIN(ω. since we could replace V with V [H] and delete countably many elements of Y we may as well assume that B is coded in the ground model. 2)name τ for an element of 2ω such that Y = {τ Gα : α < ω2 } is subset of X. Since the order of F is α we have by the lemma that there exists a Π0 set A such that α ˇ ˇ [ τ ∈ A ] = [ τ ∈ B ]. We claim that ord(Y ) < ω1 . it follows that the order of F is 1 countable. .  Proof: Both directions of the lemma are simple inductions.   It follows that Y ∩A=Y ∩B and hence order of Y is less than or equal to α (or three since we neglected countably many elements of Y ). Since D is countable and Σ0 (D) = F. 2)}. First note that the order of F in F is less than ω1 . τ a Bname for an element of 2ω . By ccc there exists H = G Σ where Σ ⊆ κ is countable set in V such that B is coded in V [H]. and F = {[ τ ∈ C ] : C ⊆ 2ω clopen }. Let B ⊆ 2ω be any Borel set coded in V [G].
So B ∩ X = (F ∪ F0 ) ∩ X.John Owen (15601622) 5 . then ord(X) = 2. Also it is easy to see that n in the random real model. Presumably. hence n Fσ . the set of reals given by the generic ﬁlter is a Sierpi´ski set. We conclude with a few remarks about perfect set forcing. The result will easily follow from the next two lemmas. n Theorem 15. Then V [G] = α ≤ ord(X) ≤ α + 1. In the following lemmas boolean values [ θ ] will  ω be computed in the measure algebra B on 2 . leaving this gap here restores a certain cosmic balance of ignorance. but now confess.2 we know that under CH Sierpi´ski sets exists. i.15 THE RANDOM REAL MODEL 65 15 The random real model In this section we consider the question of Borel orders in the random real model. Fremlin’s proof (Theorem 13. n Proof: For any Borel set B ⊆ 2ω there exists an Fσ set with F ⊆ B and B \ F measure zero. but I haven’t been able to prove this.2 Suppose X ∈ V is a subspace of 2ω of order α and G is measure algebra 2κ generic over V . Since X is Sierpi´ski (B \ F ) ∩ X = F0 is countable.e. A set X ⊆ 2ω is a Sierpi´ ski set iﬀ X is uncountable and for every n measure zero set Z we have X ∩ Z countable. ord(X) = α in V [G]. Theorem 15. the more I know I know.1 (Poprougenko [91]) If X is Sierpi´ski. by the usual ccc arguments.. I know the less. adjoin κ many random reals. Let µ be the usual product measure on 2ω . All things I thought I knew. I had been rather hoping that every uncountable separable metric space in the random real model has order either 2 or ω1 . The following result shows that this is deﬁnitely not the case. Note that by Mahlo’s Theorem 10.4) having ﬁlled up one such missing gap. we may assume that κ = ω and G is just a random real.5 Clearly.
◦ ◦ ◦ Suppose U = n∈ω ∼ Un where each Un is a name for a Π0 n set for some α αn < α. ˇ ◦ ∀n ∈ ω not ∃m ∈ ω µ(b ∧[ x ∈ Un ]) ≥ µ(b) − + 1/m ˇ . Consequently. µ(b ∧[ x ∈U ]) ≥ ˇ ∀n ∈ ω µ(b ∧[ [x n]∩ U = ∅ ]) ≥  ◦ ◦ Case α > 1. α Proof: The proof is by induction on α.  ◦ Consequently. then [ x ∈U ] = ˇ n∈ω ◦ [ [x n]∩ U = ∅ ]. iﬀ and the set is closed. and U the name of a Π0 subset of 2ω α {x ∈ 2ω : µ(b ∧[ x ∈U ]) ≥ } ˇ ◦ is Π0 in V . ◦ If U is a name for a closed set. Case α = 1. b ∈ B.Descriptive Set Theory and Forcing Lemma 15.3 Suppose in V [G]. We can assume that the sequence ∼ Un is descending. Then the set ◦ 66 a real. µ(b ∧[ x ∈ U ]) ≥ ˇ ˇ iﬀ µ(b ∧[ x ∈ ˇ n∈ω ◦ ∼ Un ]) ≥ ◦ iﬀ iﬀ iﬀ ∀n ∈ ω µ(b ∧[ x ∈∼ Un ]) ≥ ˇ ◦ ∀n ∈ ω not µ(b ∧[ x ∈ Un ]) > µ(b) − .
which is a contradiction. then V [G] =“ord(X) > α”. Lemma 15. Then for any b ∈ B 3 b ∈ G iﬀ ∀∞ n µ([r n] ∧ b) ≥ µ([r n]). Suppose for contradiction that α ˇ ˇ ˇ b  “ U ∩X = F ∩ X and U is Π0 ”. α ◦ 67 It follows from this lemma that if X ⊆ 2ω and V =“ord(X) > α”. For suppose F ⊆ 2ω is Σ0 such that for every H ⊆ 2ω α which is Π0 we have F ∩ X = H ∩ X. To prove this it suﬃces to show: Claim: For any b ∈ B+ and for every d ≤ b in B+ there exists a tree T ⊆ 2<ω with [T ] of positive measure. 4 Let B+ be the nonzero elements of B.4 Let G be Bgeneric (where B is the measure algebra on 2ω ) and r ∈ 2ω is the associated random real. α To prove the other direction of the inequality we will use the following lemma. α But then {x ∈ 2ω : µ(b ∧[ x ∈ U ]) = µ(b)} ˇ ˇ is a Π0 set which must be equal to F on X. and 3 µ([s] ∩ b) ≥ µ([s]) 4 for all but ﬁnitely many s ∈ T .Descriptive Set Theory and Forcing By induction. [T ] ≤ d. 4 Proof: Since G is an ultraﬁlter it is enough to show that b ∈ G implies 3 ∀∞ n µ([r n] ∧ b) ≥ µ([r n]). each of the sets {x ∈ 2ω : µ(b ∧[ x ∈ Un ]) ≥ µ(b) − + 1/m} ˇ is Π0 n and so the result is proved. Proof: ◦ ◦ .
what we have done in Lemma 15. So suppose for contradiction that µ([T ]) = 0. 4 So we only need to see that [T ] has positive measure.4 is reprove the Lebesgue density theorem. I have been teaching a lot of Math 99 “College Fractions”. Then for some suﬃciently large N µ( s∈T ∩2N [s]) ≤ 1 µ([t0 ]). Let Σ be a maximal antichain of t like this. 10 4 20 10 This proves the claim and the lemma. see Oxtoby [90]. Let t0 ∈ Td be such that µ([t0 ] ∩ [Td ]) ≥ 9 µ([t0 ]). s∈2N ∩T t∈Σ By choice of Σ µ( and by choice of N µ( s∈2N ∩T 3 [s] ∩ b) ≤ µ([t0 ]) 4 s∈Σ [s]) ≤ 1 µ([t0 ]) 10 which contradicts the choice of t0 : µ([t0 ] ∩ [Td ]) ≤ ( 1 3 17 9 + )µ([t0 ]) = 6 µ([t0 ]) < µ([t0 ]). 6 Trust me on this. But note that [t0 ] ∩ [Td ] ⊆ [s] ∪ ([t] ∩ b). In eﬀect. if s ∈ T then there exists t with t0 ⊆ t ⊆ s / 3 and µ([t] ∩ b) < 4 µ([t]).Descriptive Set Theory and Forcing 68 Without loss we may assume that d is a closed set and let Td be a tree such that d = [Td ]. . 10 For every s ∈ Td ∩ 2N with t0 ⊆ s. 10 Deﬁne a subtree T ⊆ Td by r ∈ T iﬀ r ⊆ t0 or t0 ⊆ r and 3 ∀t (t0 ⊆ t ⊆ r implies µ([t] ∩ b) ≥ µ([t]) ).
Also a Luzin set X from the ground model has order 3 in the random real extension. Then in V [G] for every X ⊆ 2ω of cardinality ω2 there exists Y ∈ [X]ω2 with ord(Y ) = 2. Then we know that x ∈ U G iﬀ [ x ∈U ] ∈ G. α+1 This concludes the proof of Theorem 15.2.4 we have that for any x ∈ X x ∈ U iﬀ ∀∞ n x ∈ Br and so U is Σ0 (X) in V[G]. Let U be any name for a Borel subset of X in the extension. Since subtracting and adding a countable set from a Σ0 (X) is still Σ0 (X) we have that B is Σ0 (X) and so 3 3 3 the order of X is ≤ 3 in V [G].3 we know that for ˇ <ω any s ∈ 2 the set ◦ 3 Bs = {x ∈ X : µ([s] ∩ [ x ∈U ]) ≥ µ([s])} ˇ 4 ◦ is a Borel subset of X in the ground model and hence is Π0 (X).Descriptive Set Theory and Forcing 69 So now suppose that the order of X in V is ≤ α. Since (ord(X) = 3)V we know that (3 ≤ ord(X) ≤ 4)V [G] . We show that it is ≤ α+1 ◦ in V [G]. If we let C= Um n∈ω m>n n then C is and B∆C is countable. By Lemma α 15. Theorem 15. Σ0 (X) 3 . The above proof shows that there exists Borel sets Bn each coded in V (but the sequence may not be in V ) such that x ∈ B iﬀ ∀∞ n x ∈ Bn .5 Suppose V models CH and G is measure algebra on 2κ generic over V for some κ ≥ ω2 . To see that (ord(X) ≤ 3)V [G] suppose that B ⊆ X is Borel in V [G]. By Lemma 15. Note that this result does allow us to get sets of order λ for any countable limit ordinal λ by taking a clopen separated union of a sequence of sets whose order goes up λ. For each Bn there exists an open set Un ⊆ X such that Bn ∆Un is countable.
by Lemma 14. Let a = > 0 there exists a ∈ F with b ≤ a and n<N an for some suﬃciently large N . then b ≤ a and µ(a − b) = 0 and so a = b. Let F = {[ τ ∈ C ] : C ⊆ 2ω clopen }  where boolean values are in the measure algebra B on 2ω .11 we can get a Sierpi´ski set S ⊆ 2ω of cardinality ω2 and a term τ for any element of 2ω n such that Y = {τ r : r ∈ S} is a set of distinct elements of X.12 we have that for any Borel set B ⊆ Y there exists a Σ0 (Y ) set F such that y ∈ B iﬀ y ∈ F for all but 2 7 Pronounced ‘puﬀ’. n We will show that the order of Y is 2. an . Applying (1) we get an ∈ F with bn ≤ an and µ(an − bn ) < Then let a = n∈ω 2n .6 Let F ⊆ B be any subset of a measure algebra B closed under ﬁnite conjunctions. Then by applying (2) there exists an ∈ Σ0 (F) 2 1 with b ≤ an and µ(an − b) < 1/n. Then Π0 (F) = Σ0 (F).e. it is a super Sierpi´ski set. This Sierpi´ski n set has two additional properties: every element of it is random over the ground model and it meets every set of positive measure. Now suppose b ∈ Σ0 (F). i. Consequently. i. (1) For any b ∈ Π0 (F) and real 1 µ(a − b) < .Descriptive Set Theory and Forcing 70 Proof: Using the same argument as in the proof of Theorem 14. if a = n∈ω an . F has order ≤ 2. (2) For any b ∈ Σ0 (F) and real > 0 there exists a ∈ Σ0 (F) with b ≤ a and 1 2 µ(a − b) < . pf: b = n<ω bn . 2 2 Proof: Let µ be the measure on B. Since the order of F is 2. pf:7 b = n∈ω an . Lemma 15.e. . Let F be the complete subalgebra of F which is generated by F.
For each n choose m so that if sn = zn m. 2 For any s ∈ 2<ω deﬁne bs = [ s ⊆ τ ].Descriptive Set Theory and Forcing 71 countably many y ∈ Y . This proves the Lemma and Theorem 15. 2 Proof: It suﬃces to show that every countable subset of Y can be covered by a countable Π0 (Y ) since one can always subtract a countable set from a Π0 (Y ) 2 2 and remain Π0 (Y ). Now let Z = {zn : n < ω} ⊆ Y be arbitrary but listed with inﬁnitely many repetitions.5. To get it down to 2 we use the following lemma. Since the Sierpi´ski set consists only of reals random over the ground model we n know that for every r ∈ S r ∈ Bs iﬀ s ⊆ τ r . Also since the Sierpi´ski set meets every Borel set of positive measure we n know that for any z ∈ Y the set n<ω Bz n has measure zero.7 Every countable subset of Y is Π0 (Y ). Thus we see that the order of Y is ≤ 3. then µ(Bsn ) < 1/2n .  Working in the ground model let Bs be a Borel set with [Bs ]B = bs . It has countable intersec2 tion with Y because the set n<ω m>n Bsm has measure zero. Perfect Set Forcing . then by the lemma F0 would be Π0 and thus 2 2 B would be Σ0 . 2 Lemma 15. If B = (F \ F0 ) ∪ F1 where F0 and F1 are countable and F is Σ0 . Now for every r ∈ S we have that r∈ n<ω m>n Bsm iﬀ τ r ∈ [sm ]. n<ω m>n The set H = n<ω m>n [sm ] covers Z and is Π0 .
Consequently.5 that in this model every separable metric space of cardinality ω2 has order ω1 . Therefore in the iterated Sacks real model there are separable metric spaces of cardinality ω1 of every order α with 2 ≤ α < ω1 . we get a model. To see the ﬁrst fact note that ω1 length ﬁnite support iteration always adds a Luzin set. On the other hand if X has cardinality ω2 in V [Gα : α < ω1 ]. V [Gα : α < ω1 ]. Hence Y will have order ω1 in V [Gα : α < β + 1] and by examining the proof it is easily seen that it remains of order ω1 in V [Gα : α < ω1 ]. It follows from Reclaw’s Theorem 3.Descriptive Set Theory and Forcing 72 In the iterated Sack’s real model the continuum is ω2 and every set X ⊆ 2ω of cardinality ω2 can be mapped continuously onto 2ω (Miller [81]).7. I do not know if there is an uncountable separable metric space which is hereditarily of order ω1 in this model. Another way to obtain the same orders is to use the construction of Theorem 22 of Miller [75]. then for some β < ω1 there exists and uncountable Y ∈ V [Gα : α < β] with Y ⊆ X. but every separable metric space of cardinality ω2 has order ω1 . in this model there are Sierpi´ski sets n and Luzin sets of cardinality ω1 . What was done there implies the following: For any model V there exists a ccc extension V [G] in which every uncountable separable metric space has order ω1 . Consequently. On the other hand this forcing (and any other with the Sack’s property) has the property that every meager set in the extension is covered by a meager set in the ground model and every measure set in the extension is covered by a measure zero set in the ground model (see Miller [78]). . If we apply this result ω1 times with a ﬁnite support extension.1. where there are separable metric spaces of all orders of cardinality ω1 . for each α with 2 < α < ω1 there exists a separable metric space of cardinality ω1 which is hereditarily of order α. by Theorem 14. Also there is such an X of order 2 by the argument used in Theorem 14.
Pn+1 ⊆ Pn . P.1 of Martin and Solovay. . cov(meager(2ω )) ≥ κ.1 For any cardinal κ the following are equivalent: 1. The following theorem is wellknown. We will construct a sequence Pn ⊆ 2ω of perfect sets with the properties that 1. Let Cα : α < κ be a family of closed nowhere dense sets which cover 2ω . Theorem 16. P. MAκ (ctbl).16 COVERING NUMBER OF AN IDEAL 73 16 Covering number of an ideal This section is a small diversion..2 (Miller [79]) cof(cov(meager(2ω ))) > ω. All men’s gains are the fruit of venturing. then D = {s ∈ 2<ω : [s] ⊆ U } is dense in 2<ω . i. for any countable poset. Proof: MAκ (ctbl) implies cov(meager(2ω )) ≥ κ.e.g. and 2. Proof: Suppose for contradiction that κ = cov(meager(2ω )) has countable coﬁnality and let κn for n ∈ ω be a coﬁnal sequence in κ. and family D of dense subsets of P with D < κ there exists a Pﬁlter G with G ∩ D = ∅ for every D ∈ D.8 It is motivated by Theorem 11. cov(meager(2ω )) ≥ κ implies MAκ (ctbl) follows from the fact that any countable poset. either contains a dense copy of 2<ω or contains a p such that every two extensions of p are compatible. e. Herodotus BC 484425. Deﬁne for any ideal I in Borel(2ω ) cov(I) = min{I : I ⊆ I. 8 I = 2ω }. it is impossible to have cov(meager(2ω )) = ℵω . is easy because if U ⊆ 2ω is a dense open set. Theorem 16.
This easily gives a contradiction. For G a Pﬁlter. n ≥ m. contradicting the fact that the Cα ’s cover 2ω .Descriptive Set Theory and Forcing 2. since the argument easily relativizes to show how to obtain Pn+1 given Pn . m > n and 2. H ⊇ K. for every x ∈ H there exists y ∈ K with x m = y Note that P is countable. n) : n ∈ ω and H ∈ [D]<ω } This is ordered by (H. since n<ω Pn is nonempty and disjoint from all Cα . P = {(H. n) ≤ (K. ∀x ∈ H∃y ∈ H x n = y and for each α < κ0 let Fα = {(H. Pn ∩ {Cα : α < κn } = ∅. 2. m) ∈ P : ∀x ∈ H [x m] ∩ Cα = ∅}. m) iﬀ 1. We show how to obtain P0 . For each n ∈ ω deﬁne En ⊆ P by (H. . and 74 3. ∀α < κ Cα ∩ Pn is nowhere dense in Pn . Since cov(meager(2ω )) > κn there exists a countable sequence D = {xn : n ∈ ω} ⊆ 2ω such that D is dense and for every n xn ∈ / α<κn Cα . m. Consider the following forcing notion P. n) ∈ G} n but x m = y m. m) ∈ En iﬀ 1. deﬁne X ⊆ D by X= {H : ∃n (H. and 3.
Descriptive Set Theory and Forcing
75
and let P = cl(X). It easy to check that the En ’s are dense and if G meets each one of them, then P is perfect (i.e. has no isolated points). The Fα for α < κ0 are dense in P. This is because D ∩ Cα = ∅ so given (H, n) ∈ P there exists m ≥ n such that for every x ∈ H we have [x m] ∩ Cα = ∅ and thus (H, m) ∈ Fα . Note that if G ∩ Fα = ∅, then P ∩ Cα = ∅. Consequently, by Theorem 16.1, there exists a Pﬁlter G such that G meets each En and all Fα for α < κ0 . Hence P = cl(X) is a perfect set which is disjoint from each Cα for α < κ0 . Note also that for every α < κ we have that Cα ∩ D is ﬁnite and hence Cα ∩ X is ﬁnite and therefore Cα ∩ P is nowhere dense in P . This ends the construction of P = P0 and since the Pn can be obtained with a similar argument, this proves the Theorem.
Question 16.3 (Fremlin) Is the same true for the measure zero ideal in place of the ideal of meager sets? Some partial results are known (see Bartoszynski, Judah, Shelah [7][8][9]). Theorem 16.4 (Miller [79]) It is consistent that cov(meager(2ω1 )) = ℵω . Proof: In fact, this holds in the model obtained by forcing with FIN(ℵω , 2) over a model of GCH. cov(meager(2ω1 )) ≥ ℵω : Suppose for contradiction that {Cα : α < ωn } ∈ V [G] is a family of closed nowhere dense sets covering 2ω1 . Deﬁne Eα = {s ∈ FIN(ω1 , 2) : [s] ∩ Cα = ∅}. Using ccc, there exists Σ ∈ [ℵω ]ωn in V with {Eα : α < ωn } ∈ V [G Σ]. Let X ⊆ ℵω be a set in V of cardinality ω1 which is disjoint from Σ. By the product lemma G X is FIN(X, 2)generic over V [G Σ]. Consequently, if H : ω1 → 2 corresponds to G via an isomorphism of X and ω1 , then H ∈ Cα / for every α < ωn .
Descriptive Set Theory and Forcing
76
cov(meager(2ω1 )) ≤ ℵω : Note that for every uncountable X ⊆ ω1 such that X ∈ V [G] there exists n ∈ ω and Z ∈ [ω1 ]ω1 ∩ V [G ωn ] with Z ⊆ X. To see this, note that for every α ∈ X there exists p ∈ G such that p  α ∈ X and p ∈ FIN(ωn , 2) for some n ∈ ω. Consequently, by ccc, some n works for uncountably many α. Consider the family of all closed nowhere dense sets C ⊆ 2ω1 which are coded in some V [G ωn ] for some n. We claim that these cover 2ω1 . This follows from above, because for any Z ⊆ ω1 which is inﬁnite the set C = {x ∈ 2ω1 : ∀α ∈ Z x(α) = 1} is nowhere dense. Theorem 16.5 (Miller [79]) It is consistent that there exists a ccc σideal I in Borel(2ω ) such that cov(I) = ℵω . Proof: ◦ ◦ Let P = FIN(ω1 , 2)∗ Q where Q is a name for the Silver forcing which codes up generic ﬁlter for FIN(ω1 , 2) just like in the proof of Theorem 11.1. Let α<ℵω P be the direct sum (i.e. ﬁnite support product) of ℵω copies of P. Forcing with the direct sum adds a ﬁlter G = Gα : α < ℵω where each Gα is Pgeneric. In general, a direct sum is ccc iﬀ every ﬁnite subproduct is ccc. This follows by a deltasystem argument. Every ﬁnite product of P has ccc, because P is σcentered, i.e., it is the countable union of centered sets. Let V be a model of GCH and G = Gα : α < ℵω be α<ℵω P generic over V . We claim that in V [G] if I is the σideal given by Sikorski’s Theorem 9.1 such that α<ℵω P is densely embedded into Borel(2ω )/I then cov(I) = ℵω . First deﬁne, mP , to be the cardinality of the minimal failure of MA for P, i.e., the least κ such that there exists a family D = κ of dense subsets of P such that there is no Pﬁlter meeting all the D ∈ D. Lemma 16.6 In V [ Gα : α < ℵω ] we have that mP = ℵω . Proof: Note that for any set D ⊂ P there exists a set Σ ∈ [ℵω ]ω1 in V with D ∈ V [ Gα : α ∈ Σ ]. So if D = ωn then there exists Σ ∈ [ℵω ]ωn in V with
Descriptive Set Theory and Forcing
77
D ∈ V [ Gα : α ∈ Σ ]. Letting α ∈ ℵω \ Σ we get Gα a Pﬁlter meeting every D ∈ D. Hence mP ≥ ℵω . On the other hand: Claim: For every X ∈ [ω1 ]ω1 ∩ V [ Gα : α < ℵω ] there exists n ∈ ω and Y ∈ [ω1 ]ω1 ∩ V [ Gα : α < ℵn ] with Y ⊆ X. Proof: ◦ For every α ∈ X there exist p ∈ G and n < ω such that p  α ∈X ˇ and domain(p) ⊆ ℵn . Since X is uncountable there is one n which works for uncountably many α ∈ X. It follows from the Claim that there is no H which is FIN(ω1 , 2) generic over all the models V [ Gα : α < ℵn ], but forcing with P would add such an H and so mP ≤ ℵω and the Lemma is proved. Lemma 16.7 If P is ccc and dense in the cBa Borel(2ω )/I, then mP = cov(I). Proof: This is the same as Lemma 11.2 equivalence of (1) and (3), except you have to check that m is the same for both P and Borel(2ω )/I. Kunen [58] showed that least cardinal for which MA fails can be a singular cardinal of coﬁnality ω1 , although it is impossible for it to have coﬁnality ω (see Fremlin [27]). It is still open whether it can be a singular cardinal of coﬁnality greater than ω1 (see Landver [61]). Landver [62] generalizes Theorem 16.2 to the space 2κ with basic clopen sets of the form [s] for s ∈ 2<κ . He uses a generalization of a characterization of cov(meager(2ω )) due to Bartoszynski [6] and Miller [80].
Σ1 (x) are obtained by 1 allowing R to be recursive in x.78 Part II Analytic sets 17 Analytic sets Analytic sets were discovered by Souslin when he encountered a mistake of Lebesgue. y) : ∀n R(x n. y n)}. For example. For the whole area of descriptive set theory and its history. A good reference is the volume devoted to analytic sets edited by Rogers [93]. e. The relativized classes. The boldface classes.e.. I think the mistake he made was to think that the countable intersection commuted with projection. are obtained by taking arbitrary R’s. y n). 1 Π1 . see Moschovakis [89]. The Π classes are the complements of the Σ’s and ∆ is the class of sets which are both Π and Σ.g. A set A ⊆ ω ω is Σ1 iﬀ there exists a recursive 1 R⊆ n∈ω (ω n × ω n ) such that for all x ∈ ω ω x ∈ A iﬀ ∃y ∈ ω ω ∀n ∈ ω R(x n. That means basically that C is a recursive closed set. see Kuratowski [59]. A ⊆ ω is Σ1 iﬀ there exists a recursive R ⊆ ω × ω <ω such that 1 for all m ∈ ω m ∈ A iﬀ ∃y ∈ ω ω ∀n ∈ ω R(m.. y n). A similar deﬁnition applies for A ⊆ ω and also for A ⊆ ω × ω ω and so forth. A set C ⊆ ω ω × ω ω is Π0 iﬀ there exists a recursive predicate 1 R⊆ n∈ω (ω n × ω n ) such that C = {(x. i. For the more classical viewpoint of operationA. Σ1 . e.g. R ≤T x. 1 . Lebesgue had erroneously proved that the Borel sets were closed under projection. Deﬁnition.
3. y) ∈ U }.2 The following are all true: 1..1 A ⊆ ω ω is Σ1 iﬀ there exists set C ⊆ ω ω × ω ω which is Π0 1 1 and A = {x ∈ ω ω : ∃y ∈ ω ω (x. then f −1 (A) is 1 1 Σ1 . if An : n ∈ ω is sequence of Σ1 sets given by the recursive predicates 1 Rn and Rn : n ∈ ω is (uniformly) recursive. 1 6.e. if A ⊆ ω ω × ω ω is Σ1 . if A ⊆ ω × ω ω is Σ1 . y) ∈ C}. 1 2. the above lemma is true with ω or ω × ω ω . . 1 1 1 Theorem 17. etc. if the graph of f : ω ω → ω ω is Σ1 and A ⊆ ω ω is Σ1 . It follows from the Lemma that every 1 Borel subset of ω ω is Σ1 and that the continuous preimage of Σ1 set is Σ1 . 5.. x) ∈ A}.17 ANALYTIC SETS 79 Lemma 17. for every Σ1 set A ⊆ ω ω there exists x ∈ ω ω with 1 1 A = {y : (x. then so is 1 B = {x ∈ ω ω : ∃y ∈ ω (x.3 There exists a Σ1 set U ⊆ ω ω × ω ω which is universal for 1 all Σ1 sets. then both An and n∈ω n∈ω An are Σ1 . then so is 1 B = {x ∈ ω ω : ∀n ∈ ω (n. then so is 1 B = {x ∈ ω ω : ∃n ∈ ω (n. x) ∈ A}. in place of ω ω . It also relativizes to any class Σ1 (x). 1 Of course. 4. y) ∈ A}. Lemma 17. i. if A ⊆ ω × ω ω is Σ1 . For every s ∈ ω <ω the basic clopen set [s] = {x ∈ ω ω : s ⊆ x} is Σ1 .
17 ANALYTIC SETS 80 Proof: There exists C ⊆ ω ω × ω ω × ω ω a Π0 set which is universal for Π0 subsets 1 1 of ω ω × ω ω . s i)}. The connection between Σ1 and wellfounded trees. {e}x computes the characteristic function of Tx . Let U be the projection of C on its second coordinate. Proof: Suppose x ∈ A iﬀ ∃y ∈ ω ω ∀n ∈ ω R(x n.4 (Normal form) A set A ⊆ ω ω is Σ1 iﬀ there exists a recursive 1 map <ω ω ω → 2ω x → Tx such that Tx ⊆ ω <ω is a tree for every x ∈ ω ω . gives us the following: 1 Theorem 17.5 (Mostowski’s Absoluteness) Suppose M ⊆ N are two transitive models of ZFC∗ and θ is Σ1 sentence with parameters in M . 1 The usual diagonal argument shows that there are Σ1 subsets of ω ω which 1 are not Π1 and Σ1 subsets of ω which are not Π1 . A is Σ1 iﬀ there is a uniformly 1 recursive list of recursive trees Tn : n < ω such that n ∈ A iﬀ Tn is illfounded. A similar thing is true for A ⊆ ω.e.. 1 1 1 Theorem 17. Then 1 M = θ iﬀ N = θ. Proof: . i. y Deﬁne Tx = {s ∈ ω <ω : ∀i ≤ s R(x i. n). and x ∈ A iﬀ Tx is illfounded. Similarly we can get Σ1 sets contained in ω × ω (or ω × ω ω ) which are 1 universal for Σ1 subsets of ω (or ω ω ). By recursive map we mean that there is a Turing machine {e} such that for x ∈ ω ω the machine e computing with an oracle for x.
This means that for some tree T ⊆ ω <ω in M θ is equivalent to “T has an inﬁnite branch”.17 ANALYTIC SETS 81 ZFC∗ is a nice enough ﬁnite fragment of ZFC to know that trees are wellfounded iﬀ they have rank functions (Theorem 7. On the other hand if M = ¬θ. So if M = θ then N = θ since a branch T exists in M . . then M = ∃r : T → OR a rank function” and then for this same r ∈ M N = r : T → OR is a rank function” and so N = ¬θ. θ is Σ1 sentence with 1 parameters in M means there exists R in M such that θ = ∃x ∈ ω ω ∀n R(x n).1).
y ∈ ω ω we have that x <c y iﬀ there exists E ⊆ ω ×ω ◦ ◦ and x. M =x<c y . 2. A set is Σ1 iﬀ it is the projection of a Π1 set.18 CONSTRUCTIBLE WELLORDERINGS 82 18 Constructible wellorderings G¨del proved the axiom of choice relatively consistent with ZF by producing a o deﬁnable wellorder of the constructible universe. Here “least” basically boils down to lexicographical order. m ∈ ω (y(n) = m iﬀ M =y (n) =m). L = α∈OR Lα . Deα<λ ﬁnable means with parameters from Lα . (Actually. for all n. y ∈ ω such that letting M = (ω. m ∈ ω (x(n) = m iﬀ M =x (n) =m). it is probably enough for α to be a limit ordinal. for x. then there exists an uncountable Π1 set without perfect subsets. The ﬁrst clause guarantees (by the Mostowski collapsing lemma) that M is isomorphic to a transitive set. He announced in G¨del [32] o 1 that if V=L. The second. The set x is constructed before y. for all n. 2 1 Theorem 18. or α = β and the “least” deﬁning formula for x is less than the one for y. (x <c y) iﬀ the least α such that x ∈ Lα is less than the least β such that y ∈ Lβ . E is extensional and wellfounded. L0 = ∅. Lλ = o Lα for λ a limit ordinal. Kuratowski wrote down a proof of the theorem below but the manuscript was lost during World War II (see Addison [2]).) Assuming V = L. Whatever the exact formulation of x <c y is it satisﬁes: x <c y iﬀ Lα = x <c y where x. E) then 1. and 5. and Lα+1 is the deﬁnable subsets of Lα .1 [V=L] There exists a ∆1 wellordering of ω ω . 4. 2 Proof: Recall the deﬁnition of G¨del’s Constructible sets L. M =ZFC∗ + V=L 3. y ∈ Lα and Lα =ZFC∗ where ZFC∗ is a suﬃciently large ﬁnite fragment of ZFC. that this transitive set will be ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ .
we have given a Σ1 deﬁnition of <c . It follows that <c is also n n Π1 and hence ∆1 .Descriptive Set Theory and Forcing 83 of the form Lα . But a total ordering < 2 which is Σ1 is ∆1 . The last two clauses guarantee that the image under the ◦ ◦ collapse of x is x and y is y. Wellfoundedness of E is Π1 . Hence. The remaining clauses are all Π0 for some 1 n n ∈ ω. 2 2 . since x < y iﬀ y = x or y < x.
for all n. Similarly. Hence. . θ ∧ φ. M =ZFC∗ . E is extensional and wellfounded. HC The theorem says that Σ1 = ΣHC . {x ∈ ω ω : HC = θ(x)} is a Σ1 set. ◦ ◦ ◦ ◦ . and bounded quantiﬁcation. ◦ HC = θ(x) iﬀ there exists E ⊆ ω ×ω and x∈ ω such that letting M = (ω. A formula of set theory is ∆0 iﬀ it is in the smallest family of formulas containing the atomic formulas of the form “x ∈ y” or “x = y”. y) such that A = {x : ∃yθ(x. 2. y)}.1 A set A ⊆ ω ω is Σ1 iﬀ there exists a Σ1 formula θ(·) of set 2 theory such that A = {x ∈ ω ω : HC = θ(x)}. un ψ where ψ is ∆0 . m ∈ ω x(n) = m iﬀ M =x (n) =m. . negation. A formula θ of set theory is Σ1 iﬀ it of the form ∃u1 . Then for every x ∈ ω ω we have that HC = θ(x) iﬀ there exists a countable transitive set M ∈ HC with x ∈ M such that M = θ(x). On the other hand given a Σ1 2 2 set A there exists a Π1 formula θ(x. ¬θ. then M = ψ(y) iﬀ V = ψ(y). E) then 1. But this is a Σ1 formula for HC. and closed under conjunction. if ψ(· · ·) is ∆0 formula. i. 4. Σ1 2 1 n+1 = Σn . There is a close connection between the projective hierarchy above level 2 and a natural hierarchy on the subsets of HC.. But 1 then by Mostowski absoluteness (Theorem 17. . Theorem 19. . Let us illustrate this with an example construction.e. ∀x ∈ y or ∃x ∈ y. M = θ(x).19 HEREDITARILY COUNTABLE SETS 84 19 Hereditarily countable sets HC is the set consisting of all hereditarily countable sets. Suppose that θ(·) is a Σ1 formula of set theory. y). Therefore. M a transitive set and y a ﬁnite sequence of elements of M . Proof: We note that ∆0 formulas are absolute between transitive sets.5) we have that x ∈ A iﬀ there exists a countable transitive set M with x ∈ M and there exists y ∈ M such that M =ZFC∗ and M = θ(x. (or just that ω exists) 3.
. M models the ﬁrst paragraph of this proof). then M =“x ∈ X”. then x will never get put into X after this. So X is a Luzin set. 2 85 Proof: Let {Tα : α < ω1 } (ordered by <c ) be all subtrees T of ω <ω whose branches [T ] are a closed nowhere dense subset of ω ω . Deﬁne xα to be the least constructed (<c ) element of ω ω which is not in [Tβ ] ∪ {xβ : β < α}. To see that X is ΠHC note that x ∈ X iﬀ for all M if M is a transitive 1 countable model of ZFC∗ +V=L with x ∈ M and M =“∃y ∈ X x <c y”.Descriptive Set Theory and Forcing Theorem 19.e. β<α Deﬁne X = {xα : α < ω1 }.2 If V=L. This is true because the nature of the construction is such that if you put a real into X which is constructed after x. So x will be in X iﬀ it is already in X. then there exists a ∆1 Luzin set X ⊆ ω ω . To see that X is ΣHC note that x ∈ X iﬀ there exists a transitive count1 able M which models ZFC∗ +V=L such that M =“x ∈ X”(i.
and N ˆ (r. (t. Let T ∗ be the tree deﬁned by T∗ = n∈ω {(s. So deﬁning Tx = {t : (x t. To see this. m < N [(t. s1 (i) >. sm ) ∈ T and sn ⊂ sm ] implies r(n) > r(m). t) ∈ T iﬀ ∀n. To see this let <. .4). A = p[T ] where T ∈ L. r) ∈ [T ] and ˆ ˆ so x ∈ p[T ]. Proof: From the construction of T ∗ it is clear that is enough to see this for A which is Π1 . t) ∈ κn × ω n : (s0 . sn ). 1 We know that a countable tree is wellfounded iﬀ there exists a rank function r : T → ω1 . i. y) ∈ A} is κSouslin.1 (Shoenﬁeld [98]) If A is a Σ1 set. then Tx is wellfounded and ˆ so it has a rank function and therefore there exists r with (x. then r determines a rank function on Tx and so Tx is wellfounded and hence x ∈ A. For s ∈ κn let s0 ∈ κn and s1 ∈ ω n be deﬁned by s(i) =< s0 (i). y n) ∈ T / where T is a recursive tree. Let {sn : n ∈ ω} be a recursive listing of ω <ω with sn  ≤ n. Thus Σ1 sets are precisely the 1 ωSouslin sets.e. Theorem 20. On the other hand if (x. Then p[T ∗ ] = {y : ∃x ∈ ω ω (x. y) ∈ A}. Then for every N < ω. Note that if A ⊆ ω ω × ω ω and A = p[T ] then the projection of A. note that if x ∈ A. s) ∈ ω1 × ω N we have (r. t) ∈ T } we have that x ∈ A iﬀ Tx is wellfounded (Theorem 17. Suppose x ∈ A iﬀ ∀y∃n (x n. ˆ The ω1 tree T is just the tree of partial rank functions. t) ∈ T }. p[T ] = {y ∈ ω ω : ∃x ∈ κω ∀n(x n. then A is ω1 Souslin set 2 coded in L. s1 . r) ∈ [T ]. ˆ Then A = p[T ]. y A set deﬁned this way is called κSouslin.20 SHOENFIELD ABSOLUTENESS 86 20 Shoenﬁeld Absoluteness n<ω For a tree T ⊆ κn × ω n deﬁne n) ∈ T }. {y : ∃x ∈ ω ω (x. >: κ × ω → κ be a pairing function.
i.Descriptive Set Theory and Forcing 87 Theorem 20.. Note that r codes a witness to a Π1 (x) predicate and a rank function 1 showing the tree corresponding to this predicate is wellfounded. because Π1 sentences are absolute. But Tα ∈ M (since by assumption (ω1 )N ⊆ M ) and so by the absoluteness of wellfounded trees. and consequently. then N = θ.1 we get <ω a tree T ⊆ ω1 with T ∈ L[x] such that T is illfounded. r ∈ αω we see that Tα = T ∩ α<ω is illfounded. M thinks that Tα is illfounded. Working in N using the proof of Theorem 20.2 (Shoenﬁeld Absoluteness [98]) If M ⊆ N are transitive models of ZFC∗ and ω1 N ⊆ M . Since for some α < ω1 .e. On the other 1 hand suppose N = θ. But a branch thru [T ] gives a witness and a rank function showing that θ is true. there exists r ∈ [T ]. M = θ. then for any Σ1 (x) sentence θ with parameter 2 x∈M M = θ iﬀ N = θ. Proof: If M = θ. .
sσ ) : σ ∈ 2<ω (r. This allows us to build by induction (working in L): (rσ . Since A = p[T ] and A is not a subset of L. s0 ). Since Tα0 +1 = Tα0 . y0 ) ∈ L. s1 ) extend (r.s) . (r1 . it must be that there exists (u. s1 ) ∈ Tα0 such that (r0 . s) ⊆ (r. Prove by induction on α that (x.1 (Mansﬁeld [72]. y) ∈ [T ] be any pair with y not constructible. and for α < β we have Tβ ⊆ Tα . y) ∈ [Tα+1 ].s) Tα = {(ˆ. s)} r ˆ r ˆ r ˆ which has the property that p[Tα ] = {y}. s). such a pair must exist. s) or (r. Solovay [103]) If A ⊆ ω ω is a Σ1 set with 2 constructible parameter which contains a nonconstructible element of ω ω . By the deﬁnition it must be that there exists n < ω such / that (x n. This is easy for α a limit ordinal. Proof: Let (x. (r1 . s) ∈ Tα : (ˆ. y) ∈ [Tα ].1. This proves the claim. y0 ) ∈ [Tα ] with (u.s) founded trees. s) ∈ Tα : ∃(r0 .21 MANSFIELDSOLOVAY THEOREM 88 21 MansﬁeldSolovay Theorem Theorem 21. But in L we can deﬁne the tree: / (r. (r1 . Proof: By Shoenﬁeld’s Theorem 20. Working in L deﬁne the following decreasing sequence of subtrees as follows. and s0 and s1 are incompatible}. we may assume A = p[T ] where T ∈ L and n T ⊆ n<ω ω1 × ω n . s) ∈ Tα0 there exist (r0 . Claim: [Tα0 ] is nonempty. But then y0 = y ∈ L which is a contradiction. s1 ) ∈ Tα such that (r0 . Tλ = β<λ Tβ . s0 ). s) ∈ Tα+1 . s) ⊆ (ˆ. Each Tα is tree. But by absoluteness of well(r. y) ∈ [Tα ] but (x. then A contains a perfect set which is coded in L. (r1 . if λ a limit ordinal. So suppose (x. s) and s0 and s1 are incompatible. and Tα+1 = {(r. it follows that for every (r. Thus there exists some α0 such that Tα0 +1 = Tα0 . s0 ). s0 ). y n) = (r. T0 = T . s1 ) extend (r.
9 . For any q ∈ 2ω deﬁne xq = n<ω rq n and yq = n<ω sq n . “Consistency is the hobgoblin of little minds. Thus we have departed9 from our theme of giving forcing proofs. With consistency a great soul has simply nothing to do. sσ ) and sσ0 and sσ1 are incompatible. Solovay’s proof used forcing. sσ ) ∈ Tα0 and for each σ ∈ 2<ω (rσ0 . Then we have that (xq .Descriptive Set Theory and Forcing 89 with (rσ . This proof is due to Mansﬁeld. (rσ1 . sσ0 ). sσ1 ) extend (rσ . yq ) ∈ [Tα0 ] and therefore P = {yq : q ∈ 2ω } is a perfect set such that P ⊆ p[Tα0 ] ⊆ p[T ] = A and P is coded in L.” Ralph Waldo Emerson.
for all x ∈ X and y. this amounts to saying for any nonempty Π1 set A ⊆ ω ω 1 there exists a Π1 set B ⊆ A such that B is a singleton.22 UNIFORMITY AND SCALES 90 22 Uniformity and Scales Given R ⊆ X × Y we say that S ⊆ X × Y uniformizes R iﬀ 1. then y = z. i. Lemma 22.e. S ⊆ R. 1 Theorem 22. 2. y). In this last case. The Π1 sets have the uniformization property. and / . y) iﬀ φi (x) ≤ φi (y) and for all x ∈ A.1 (Kondo [49]) Every Π1 set R can be uniformized by a Π1 1 1 set S. X and Y can be taken to be either ω or ω ω or even a singleton {0}.. for every x. x. y ∈ A if φi+1 (x) ≤ φi+1 (y). and 3. then φi (x) ≤ φi (y). y ∈ A and i 1 P (i. z). x. then there exists y ∈ Y such that S(x. Another way to say the same thing is that S is a subset of R which is the graph of a function whose domain is the same as R’s. y). for all i and x. y). i ∈ ω P (i. for all xn : n < ω ∈ Aω and αi : i < ω ∈ ORω if for every i and for all but ﬁnitely many n φi (xn ) = αi . y ∈ A if ∀i φi (x) = φi (y). 4. 2. for all x ∈ X if there exists y ∈ Y such that R(x. there exists P a Π1 set such that for all x. The 1 proof of this Theorem is to use a property which has become known as the scale property. 3. then there exists x ∈ A such that limn→∞ xn = x and for each i φi (x) ≤ αi . 5. z ∈ Y if S(x. each φi : A → OR. y ∈ A. B = 1.2 (scale property) For any Π1 set A there exists φi : i < ω 1 such that 1. then x = y. y) and S(x. Here.
there exists S a Σ1 set such that for all x. The third item says that in the “limit” we get a linear order. for every y ∈ A we have that φi (xn ) ≤ φi (y) and hence φi (xn ) = αi . And the last two items are the deﬁnability properties of the scale. then x ∈ A. let us deduce uniformity from it. Now to prove a more general case of uniformity suppose that R ⊆ ω ω ×ω ω is Π1 . y) iﬀ φi (x) ≤ φi (y) and for all y ∈ A. Before proving the lemma.Descriptive Set Theory and Forcing 6. Deﬁne αi = min{φi (x) : x ∈ A}. x. y). So it remains to show that B is nonempty. Let φi : R → OR be scale given by the lemma and 1 P ⊆ ω × (ω ω × ω ω ) × (ω ω × ω ω ) . y ∈ A and i 1 S(i. x. First let us show that for any nonempty Π1 1 set A ⊆ ω ω there exists a Π1 singleton B ⊆ A. Since P is Π1 the set B is Π1 . Deﬁne 1 x ∈ B iﬀ x ∈ A and ∀n∀y P (n. The second item says that these relations get ﬁner and ﬁner as i increases. The fourth item is some sort of continuity condition. Proof: By choice of xn for every y ∈ A we have φn (xn ) ≤ φn (y). x. y). These are called prewellorderings. We do not use the last item in the lemma. i ∈ ω if S(i. Claim: If n > i then φi (xn ) = αi . and also by item (2) of the 1 1 lemma. 91 Another way to view a scale is from the point of view of the relations on A deﬁned by x ≤i y iﬀ φi (x) ≤ φi (y). Clearly B ⊆ A. By the minimality of αi it must be that φi (x) = αi . For each i choose xi ∈ A such that φi (xi ) = αi . By item (4) there exists x ∈ A such that limn→∞ xn = x and φi (x) ≤ αi all i. x. By item (2) in the lemma. So x ∈ B and we are done. B can have at most one element. They are well orderings if we mod out by x ≡i y which is deﬁned by x ≡i y iﬀ x ≤i y and y ≤i x.
T subtrees of Q<ω we say that σ : T → T is a tree embedding iﬀ for all s.4). x(1). We write T T iﬀ there exists a tree ˆ.Descriptive Set Theory and Forcing 92 be the Π1 predicate given by item (5). Given a wellfounded tree T ⊆ ω <ω with rank function rT extend rT to all of ω <ω by deﬁning rT (s) = −1 if s ∈ T .. Recall that r : T → OR is a rank function iﬀ for all s. Then r ◦ σ is a rank function on T . Also note that tree ˆ embeddings need not be onetoone. y). Also the rank of T is the minimal ordinal α such that there exists a rank function r : T → α + 1.e. [T ] = ∅. y) ∈ S iﬀ (x. z)). (x.4 Suppose T and T are well founded trees and rank of T is less ˆ ˆ than or equal rank of T . i. ˆ ˆ ˆ Lemma 22. then T ˆ is wellfounded and rank of T is less than or equal to rank of T . Note that s ⊂ t means that s is a proper initial segment of t. then T T . rTx (sn ). t ∈ T if s ⊂ t then σ(s) ⊂ σ(t). For T. For each − − n < ω deﬁne ψn : A → ω1 × ω × · · · ω1 × ω by ψn (x) = rTx (s0 ). t ∈ T if s ⊂ t then r(s) > r(t). (x. Let − ω1 = {−1} ∪ ω1 be wellordered in the obvious way. The same proof shows that S uniformizes R. Let {sn : n < ω} be a recursive listing of ω <ω with s0 = . x(0). Proof: ˆ ˆ Let σ : T → T be a tree embedding and r : T → OR a rank function. We write T T iﬀ there is a tree embedding which ˆ embedding from T into T ˆ takes the root node of T to a nonroot node of T . y) ∈ R and ∀z∀n P (n.2).3 Suppose T T and T is wellfounded. so as to satisfy rT (s) ≤ rT (σ(s)). The proof of the Lemma will need the following two elementary facts ˆ ˆ about wellfounded trees. x(n) . ˆ Let rT and rT be the canonical rank functions on T and T (see Theorem 7.1). ˆ ˆ Inductively deﬁne σ : T ∩ Qn → T ∩ Qn . . . Then deﬁne the Π1 set S ⊆ ω ω × ω ω 1 1 by (x. . . ˆ Lemma 22. / ω 1 Now suppose A ⊆ ω is Π1 and x ∈ A iﬀ Tx is wellfounded (see Theorem 17. ˆ Now we a ready to prove the existence of scales (Lemma 22. rTx (s1 ). .
because ψi (x) = ψi (y) implies x i = y i. to say that rT (s) ≤ rT (s) ˆ is equivalent to saying there exists a tree embedding which takes s to s. This is true.. For item (3): if ∀i ψi (x) = ψi (y). The scale φi is just obtained by mapping the range of ψi order isomorphically to the ordinals. Consequently x ∈ A. Note that if (n. Note that this is Σ1 . T are wellfounded. x(n) is lexicographically less than or equal to rTy (s0 ). rTy (sn ). then x = y. it follows that ψi (x) ≤lex ti . T ) : T. x(1). y) ∈ S iﬀ rTx (s0 ). y(0). ˆ Consequently. For item (4): Suppose xn : n < ω ∈ Aω and for every i and for all but ﬁnitely many n ψi (xn ) = ti .) Now we verify the properties.i. since Ty is a wellfounded tree and S implies Tx Ty . y ∈ A we have (n. . rTx (sn ). assuming that T. .e.(6): The following set is Σ1 : 1 ˆ ˆ {(T. rTy (s1 ). . This shows that it is possible to deﬁne a Σ1 set S ⊆ ω×ω ω ×ω ω 1 1 such that for every x. y(1). For item (5). T T or .Descriptive Set Theory and Forcing 93 − − The set ω1 × ω × · · · ω1 × ω is wellordered by the lexicographical order. This is true because we are just taking the lexicographical order of a longer sequence. For item (2): if ψi+1 (x) ≤lex ψi+1 (y). . then ψi (x) ≤lex ψi (y). x. and so the range of ψi is less than or equal to ω1 . . y) ∈ S and y ∈ A then x ∈ A. . rTx (s1 ). T ˆ T }. Note that since {sn : n ∈ ω} lists every element of ω <ω . Then since ψi (xn ) contains xn i there must be x ∈ ω ω such that limn→∞ xn = x. (Remark: by choosing s0 = . T are subtrees of ω <ω . Now since rTx (s) ≤ r(s). we guarantee that the ﬁrst coordinate is always the largest coordinate. . instead of saying T can be embedded 1 ˆ ˆ ˆ into T we say that T cannot be embedded properly into T . Using this and n→∞ lim Txn = Tx it follows that r is a rank function on Tx . so Tx is wellfounded and so x ∈ A. . x(0). To get the Π1 relation P (item (5)). y(n) . we have that for every s ∈ ω <ω there exists r(s) ∈ OR such that rTxn (s) = r(s) for all but ﬁnitely many n. x.
4 this is equivalent to the nonexistence of such an embedding. there does not exists a tree embedding σ : T → T such 1 ˆ that σ( ) = . which. Note also that if x ∈ A and y ∈ A.2. The scale property was invented by Moschovakis [88] to show how determinacy could be used to get uniformity properties10 in the higher projective classes. y) for every n. This ﬁnishes the proof of the Scale Lemma 22. This is a Π1 statement. and Martin. Poul Anderson 10 . The 500 page book by Kuratowski and Mostowski [60] ends with a proof of the uniformization theorem. did not become still more complicated. For T and T wellfounded trees ˆ is equivalent to saying rank saying that rank of T is less than or equal to T ˆ of T is not strictly smaller than the rank of T . He was building on earlier ideas of Blackwell. however complicated. This is because Ty is not / wellfounded and so cannot be embedded into the wellfounded tree Tx . x. I have yet to see any problem. But by Lemma 22. Addison. then we will have P (n.Descriptive Set Theory and Forcing 94 ˆ in other words. when you looked at it in the right way.
Arbitrarily wellorder the two sets. Then every A ⊆ 2ω of cardinality ω1 is Π1 . L Corollary 23. This is because if P ⊆ S is perfect. there exists uncountable o Π1 set A ⊆ ω ω which contains no perfect subsets. The set B cannot contain a perfect 1 set since the sentence: ∃T T is a perfect tree and ∀x (x ∈ [T ] implies θ(x)) is a Σ1 and false in L and so by Shoenﬁeld absoluteness (Theorem 20. Actually. 1 Proof: Let A ⊆ 2ω be a uncountable Π1 set of constructible reals and let B be an 1 arbitrary subset of 2ω of cardinality ω1 . Then S is an uncountable Π1 set which contains 1 no perfect subset. It follows then by the MansﬁeldSolovay Theorem 21. y) = x.2 If ω1 = ω1 . so one can avoid using the SolovayMansﬁeld Theorem. Note that it is suﬃcient to assume that ω1 = (ω1 )L .1) to 1 get S ⊆ R with the property that X is the onetoone image of S via the projection map π(x. For 2 example.1 that B cannot contain a nonconstructible real and so A = B. Theorem 23. Suppose A ∈ L is deﬁned by the Π1 formula θ. 1 Proof: Let X be any uncountable Σ1 set containing no perfect subsets. Use Π1 uniformization (Theorem 22.3 (MartinSolovay [74]) Suppose MA + ¬CH + ω1 = (ω1 )L .2) must 2 be false in V . . So by Π1 absoluteness A = B ∩ L. Then let B be the set which is deﬁned by θ in 1 V . by tracing thru the actual deﬁnition of X one can see that the elements of the uniformizing set S (which is what A is) consist of pairs (x. These pairs are reals which witness their own constructibility.1 (G¨del see Solovay [103]) If V=L. A = {aα : α < ω1 } and B = {bα : α < ω1 }. then there exists a Π1 set of constructible reals 1 which contains no perfect set. y). then π(P ) is a perfect subset of X.23 MARTIN’S AXIOM AND CONSTRUCTIBILITY 95 23 Martin’s axiom and Constructibility Theorem 23. Let R ⊂ ω ω × ω ω be Π1 2 1 such that x ∈ X iﬀ ∃y R(x. a ∆1 Luzin set would do (Theorem 18.1). y) where y is isomorphic to some Lα and x ∈ Lα .
is relatively Gδ by Theorem 5. To see this let Bα : α < ω1 be any sequence of Borel sets. although it is an arbitrary subset of A. Then y∈ α<ω1 Bα iﬀ ∃x x ∈ {xα : α < ω1 } ∧(x.Descriptive Set Theory and Forcing 96 By Theorem 5.1 there exists two sequences of Gδ sets Un : n < ω and Vn : n < ω such that for every α < ω1 for every n < ω aα (n) = 1 iﬀ bα ∈ Un and bα (n) = 1 iﬀ aα ∈ Vn . Since A is Π1 this deﬁnition of B has the form: 1 ∀a([Π0 ] implies [Π1 and Π0 ]) 3 1 3 So B is Π1 . y) ∈ U }. 2 1 . But {xα : α < ω1 } is Π1 and so the union is Σ1 .1. This is because the set {aα : bα (n) = 1}. But note that b ∈ B iﬀ ∀a ∈ 2ω [∀n (a(n) = 1 iﬀ b ∈ Un )] implies [a ∈ A and ∀n (b(n) = 1 iﬀ a ∈ Un )]. 1 Note that if every set of reals of size ω1 is Π1 then every ω1 union of Borel 1 sets is Σ1 . Let 2 U be a universal Π1 set and let xα : α < ω1 be a sequence such that 1 Bα = {y : (xα . y) ∈ U.
24 Σ1 WELLORDERINGS 2
97
24
Σ1 wellorderings 2
F ⊆ ω ω and ¡ ⊆ F 2
Theorem 24.1 (Mansﬁeld [71]) If (F, ¡) is a Σ1 wellordering, i.e., 2
are both Σ1 , then F is a subset of L. 2 Proof: We will use the following: Lemma 24.2 Assume there exists z ∈ 2ω such that z ∈ L. Suppose f : P → / F is a 11 continuous function from the perfect set P and both f and P are coded in L, then there exists Q ⊆ P perfect and g : Q → F 11 continuous so that both g and Q are coded in L and for every x ∈ Q we have g(x) ¡ f (x). Proof: (Kechris [52]) First note that there exists σ : P → P an autohomeomorphism coded in L such that for every x ∈ P we have σ(x) = x but σ 2 (x) = x. To get this let c : 2ω → 2ω be the complement function, i.e., c(x)(n) = 1−x(n) which just switches 0 and 1. Then c(x) = x but c2 (x) = x. Now if h : P → 2ω is a homeomorphism coded in L, then σ = h−1 ◦ c ◦ h works. Now let A = {x ∈ P : f (σ(x)) ¡ f (x)}. The set A is a Σ1 set with code 2 in L. Now since P is coded in L there must be a z ∈ P such that z ∈ L. / Note that σ(z) ∈ L also. But either / f (σ(z)) ¡ f (z) or f (z) = f (σ 2 (z)) ¡ f (σ(z)) and so either z ∈ A or σ(z) ∈ A. In either case A has a nonconstructible member and so by the MansﬁeldSolovay Theorem 21.1 the set A contains a perfect set Q coded in L. Let g = f ◦ σ. Assume there exists z ∈ F such that z ∈ L. By the MansﬁeldSolovay / Theorem there exists a perfect set P coded in L such that P ⊆ F . Let P0 = P and f0 be the identity function. Repeatedly apply the Lemma to obtain fn : Pn → F so that for every n and Pn+1 ⊆ Pn , for every x ∈ Pn+1 fn+1 (x) ¡ fn (x). But then if x ∈ n<ω the sequence fn (x) : n < ω is a descending ¡ sequence with contradicts the fact that ¡ is a wellordering. Friedman [28] proved the weaker result that if there is a Σ1 wellordering 2 of the real line, then ω ω ⊆ L[g] for some g ∈ ω ω .
25 LARGE Π1 SETS 2
98
25
Large Π1 sets 2
A set is Π1 iﬀ it is the complement of a Σ1 set. Unlike Σ1 sets which cannot 2 2 2 have size strictly in between ω1 and the continuum (Theorem 21.1), Π1 sets 2 can be practically anything.11 Theorem 25.1 (Harrington [35]) Suppose V is a model of set theory which L satisﬁes ω1 = ω1 and B is arbitrary subset of ω ω in V . Then there exists a ccc extension of V , V [G], in which B is a Π1 set. 2 Proof: Let PB be the following poset. p ∈ PB iﬀ p is a ﬁnite consistent set of sentences of the form: 1. “[s]∩ C n = ∅”, or 2. “x ∈C n , where x ∈ B. This partial order is isomorphic to Silver’s view of almost disjoint sets forcing (Theorem 5.1). So forcing with PB creates an Fσ set n∈ω Cn so that ∀x ∈ ω ω ∩ V (x ∈ B iﬀ x ∈
n<ω ◦ ◦
Cn ).
α<ω1
Forcing with the direct sum of ω1 copies of PB ,
PB , we have that
α ∪n<ω Cn ).
∀x ∈ ω ω ∩ V [ Gα : α < ω1 ](x ∈ B iﬀ x ∈
α<ω1
One way to see this is as follows. Note that in any case B⊆
α<ω1 α ∪n<ω Cn .
So it is the other implication which needs to be proved. By ccc, for any x ∈ V [ Gα : α < ω1 ] there exists β < ω1 with x ∈ V [ Gα : α < β ]. But considering V [ Gα : α < β ] as the new ground model, then Gβ would be PB β generic over V [ Gα : α < β ] and hence if x ∈ B we would have x ∈ ∪n<ω Cn . / / Another argument will be given in the proof of the next lemma.
11
It’s life Jim, but not as we know it. Spock of Vulcan
Descriptive Set Theory and Forcing
99
Lemma 25.2 Suppose cα : α < ω1 be a sequence in V of elements of ω ω and aα : α < ω1 is a sequence in V [ Gα : α < ω1 ] of elements of 2ω . Using Silver’s forcing add a sequence of Π0 sets Un : n < ω such that 2 ∀n ∈ ω∀α < ω1 (aα (n) = 1 iﬀ cα ∈ Un ). Then V [ Gα : α < ω1 ][ Un : n < ω ] = ∀x ∈ ω ω (x ∈ B iﬀ x ∈
α<ω1 α ∪n<ω Cn ).
Proof: The lemma is not completely trivial, since adding the Un : n < ω adds new elements of ω ω which may somehow sneak into the ω1 intersection. Working in V deﬁne p ∈ Q iﬀ p is a ﬁnite set of consistent sentences of the form: 1. “[s] ⊆ Un,m ” where s ∈ ω <ω , or 2. “cα ∈ Un,m ”. Here we intend that Un = ∩m∈ω Un,m . Since the c’s are in V it is clear that the partial order Q is too. Deﬁne P = {(p, q) ∈ (
α<ω1
PB ) × Q : if “cα ∈ Un,m ”∈ q, then p  aα (n) = 1}.
Note that P is a semilowerlattice, i.e., if (p0 , q0 ) and (p1 , q1 ) are compatible elements of P, then (p0 ∪ p1 , q0 ∪ q1 ) is their greatest lower bound. This is another way to view the iteration, i.e, P is dense in the usual iteration. Not every iteration has this property, one which Harrington calls “innocuous”. Now to prove the lemma, suppose for contradiction that (p, q)  x∈
α<ω1 ◦ α ∪n<ω Cn and x∈ B. / ◦
To simplify the notation, assume (p, q) = (∅, ∅). Since P has the ccc a sequence of Working in V let An : n ∈ ω be a sequence of maximal antichains ◦ of P which decide x, i.e. for (p, q) ∈ An there exists s ∈ ω n such that (p, q)  x n = s. ˇ
◦
q ) ∈ Am . q ) ∈ G ∩ Am . Lemma 25. Faα = n∈ω Let C = cα : α < ω1 be a Π1 1 L set in V . and p ˆ ˇ 3. Let (ˆ. the condition α (p ∪ p ∪ {[s] ∩ Cn = ∅}. so there must be an m. ˆ Since s was chosen so that xi ∈ [s] for every i < N .)) p ˆ Now consider (p ∪ p. α Let “xi ∈ Cn ” for i < N be all the sentences of this type which occur in p(α). ◦ ((To get (ˆ. (ˆ. q ∪ q ) ∈ P. q ) and (p. such that p ˆ ◦ 1. Since α was not in the support of p. q )  x m = s. q)  x∈ Cn . q) and n ∈ ω such that α (p. Since we are assuming x is being forced not in B it must be diﬀerent than all the xi . Such a set exists since ω1 = ω1 . hence an element of PB . q). ˆ ˆ ˆ (p ∪ p)(α) = p(α). q ) and s let G be a generic ﬁlter containing (p. (ˆ. Since x is α forced to be in ∪n<ω Cn there exists (p. q) are compatible.3 In V [ Gα : α < ω1 ][ Un : n < ω ] the set B is Π1 . / α p(α) ∪ {[s] ∩ Cn = ∅} is a consistent set of sentences. 2 .Descriptive Set Theory and Forcing 100 Since P has the ccc. This is a contradiction. / α Let F be a universal Σ0 set coded in V and let aα ∈ 2ω : α < ω1 be 2 such that α Cn . q ∪ q ) ˆ ˆ α forces x ∈ Cn and also x ∈ Cn . and s ∈ ω m . (ˆ. p ˆ 2. q) in n∈ω An . the An are countable and we can ﬁnd an α < ω which does not occur in the support of any p for any (p. then since p ˆ G x = xi for every i < N there must be m < ω and s ∈ ω m such that xG m = s and s = xi m for every i < N . xi m = s for every i < N .
e. He also shows how to take a further innocuous 3 extensions to make B a ∆1 set and to get a ∆1 wellordering. Note that • “c ∈ C” is Π1 . and • “(a. x) ∈ F ” is Borel. 2 Harrington [35] also shows how to choose B so that the generic extension has a ∆1 wellordering of ω ω . 3 3 . i. c if c ∈ C and ∀n (a(n) = 1 iﬀ c ∈ Un ). x) ∈ F . 1 • “∀n (a(n) = 1 iﬀ c ∈ Un )” is Borel.Descriptive Set Theory and Forcing 101 Proof: α x ∈ B iﬀ x ∈ α<ω1 n∈ω Cn iﬀ x ∈ α<ω1 Faα iﬀ ∀a. then (a. and so this ﬁnal deﬁnition for B has the form: ∀((Π1 ∧ Borel)) → Borel) 1 Therefore B is Π1 . (x ∈ Fa ).
t sˆα.t Cα is a κBorel set separating p[TA ] and p[TB ]. Proof: Let A = p[TA ] and B = p[TB ]. 1 Deﬁne the κBorel sets to be the smallest family of subsets of ω ω containing the usual Borel sets and closed under intersections or unions of size κ and complements.t r. then for every α we would have a κBorel set Cα with sˆα. deﬁne n n<ω (κ × ω n ). Given a tree T ⊆ s ∈ κ<ω .t p[TA ] cannot be separated from p[TB ] by a κBorel set. But then α<κ s.t = {(ˆ. If there were no such α. Then for some α < κ the set sˆα. × ωn) In this case we write A = p[T ].. and ˆ ˆ T s. A ⊆ C and C ∩ B = ∅.t Note that p[TA ] = α<κ p[TA ]. Proof: s.t p[TA ] ⊆ Cα and Cα ∩ p[TB ] = ∅.t Lemma 26. y n) ∈ T. Then there exists a κBorel set C which separates A and B. t ∈ ω <ω (possibly of diﬀerent lengths).t r.1 Suppose A and B are disjoint κSouslin subsets of ω ω .102 Part III Classical Separation Theorems 26 SouslinLuzin Separation Theorem n n<ω (κ Deﬁne A ⊆ ω ω to be κSouslin iﬀ there exists a tree T ⊆ such that y ∈ A iﬀ ∃x ∈ κω ∀n < ω (x n. s ˆ ˆ ˆ s. Note that ωSouslin is the same as Σ1 .e. Theorem 26.t r.2 Suppose p[TA ] cannot be separated from p[TB ] by a κBorel set. the projection of the inﬁnite branches of the tree T .t r. t) ∈ T : (s ⊆ s or s ⊆ s) and (t ⊆ t or t ⊆ t)}. i. .
t Cβ is a κBorel set separating p[TA ] and p[TB ].t n<ω Cn would separate p[TA ] from p[TB ]. Proof: r.t p[TA ] ⊆ Cβ and Cβ ∩ p[TB ] = ∅.x n ] = ∅ and they could be separated. Then for some n < ω r.4 Suppose p[TA ] cannot be separated from p[TB ] by a κBorel set. Proof: Note that s.t Lemma 26.t s.tˆn p[TA ] cannot be separated from p[TB ] by a κBorel set. v ∈ κω and x ∈ ω ω so that for every n.t rˆβ.t r. then s. s. x n) ∈ TB u v otherwise either p[TA n.x n separated from p[TB ].x n ] = ∅ or p[TB n.3 Suppose p[TA ] cannot be separated from p[TB ] by a κBorel set.26 SOUSLINLUZIN SEPARATION THEOREM 103 s. if there were no such β then for every β we would have κBorel set Cβ with rˆβ. u. for every n (u n.tˆn s. But necessarily.t p[TB ] = p[TB ] ∩ [tˆn].t p[TA ] = p[TA ] ∩ [tˆn] and r. r.t Since p[TB ] = β<κ p[TB ]. Then for some β < κ s.tˆn s.t s.x n ] cannot be v n. .tˆn r.t Lemma 26. p[TA n.t r. But this means that x ∈ p[TA ] = A and x ∈ p[TB ] = B contradicting the fact that they are disjoint.t rˆβ.t r.t p[TA ] cannot be separated from p[TB ] by a κBorel set.tˆn r. To prove the separation theorem apply the lemmas iteratively in rotation u to obtain.tˆn Thus if Cn ⊆ [tˆn] were to separate p[TA ] and p[TB ] for each n. x n) ∈ TA and (v n. But then β<κ s.
p. p : T >0 → 2 is recursive. q) we deﬁne Cs : s ∈ T as follows. p. p. Then there exists a hyperarithmetic set C which separates them. y n) ∈ TA } . the set C coded by (T. Proof: This amounts basically to a constructive proof of the classical Separation Theorem 26. then Cs = and • if s is a not a terminal node and p(s) = 1. A ⊆ C and C ∩ B = ∅. then Cs = q(s) • if s is a not a terminal node and p(s) = 0. • if s is a terminal node of T . q) where T is a recursive wellfounded subtree of ω <ω . p. A code for a hyperarithmetic set is a triple (T. and q : T 0 → B is a recursive map. then Cs = {Csˆn : sˆn ∈ T }. q). Let A = p[TA ] and B = p[TB ] where TA and TB are recursive subtrees of n n n∈ω (ω × ω ). i. where B is the set of basic clopen subsets of ω ω including the empty set. {Csˆn : sˆn ∈ T }. Theorem 27. q) is the set C .. A set C ⊆ ω ω is hyperarithmetic iﬀ it is coded by some recursive (T.27 KLEENE SEPARATION THEOREM 104 27 Kleene Separation Theorem We begin by deﬁning the hyperarithmetic subsets of ω ω .1. Here we are being a little more ﬂexible by allowing unions and intersections at various nodes.e. Finally. We continue with our view of Borel sets as wellfounded trees with little dohickey’s (basic clopen sets) attached to its terminal nodes. Given a code (T.1 (Kleene [55]) Suppose A and B are disjoint Σ1 subsets of 1 ω ω . and p[TA ] = {y : ∃x∀n (x n.
(u. t) ∈ T / either (u. .t) = ∅ or (u. then by induction on rank of u. v(0). v. . t) ∈ TB }. Then since (u. t(0). u(0). . v. u(n) Algebraic symbols are used when you do not know what you are talking about (Philippe Schnoebelen).t v. v.Descriptive Set Theory and Forcing and similarly for p[TB ]. Note that in either case / u.v.e.v. To get a hyperarithmetic code use the tree consisting of all subsequences of sequences of the form. Then both n<ω m<ω Cnm and m<ω n<ω Cnm separate n<ω An from m<ω Bm . Suppose (u. t n) ∈ T where u = v = t = n + 1.v. Lemma 27.v. t) ∈ T + as follows. Any inﬁnite branch thru T would give a point in the intersection of A and B which would contradict the fact that they are disjoint.t v.vˆm. i. v(n). Bm : m < ω Cnm : n. . Hence. t) ∈ TA and / (v.t) = k<ω m<ω n<ω C(uˆn. C = C( . t) in T that C(u. t) ∈ T + iﬀ (u n. Proof: Left to reader. t) ∈ TA and (v.t + (u.t C(u.t) ⊆ [t] separates p[TA ] from p[TB ]. v n.v. m < ω are such that for every n and m Cnm separates An from Bm . Let T + be the tree of all nodes which are either “in” or “just out” of T . t(n). t) ∈ T + is a terminal node of T + . Now we deﬁne the family of sets C(u.tˆk) (or any other permutation12 of and ).v. 105 Notice that T is recursive tree which is wellfounded. t) ∈ TA in which case we deﬁne C(u. ) separates A = p[TA ] from B = p[TB ].. v. v. It follows from the Lemma that if we let C(u.t) ⊆ [t] separates p[TA ] from p[TB ].2 Suppose An : n < ω . Now deﬁne the tree T = {(u. t) ∈ TB in which case we deﬁne C(u.t) = [t]. 12 . v. . t) : (u.t) : (u.
then in(s) = 1 iﬀ for all n with sˆ n ∈ T we have in(sˆ n) = 1. Then x ∈ C iﬀ there exists a function in : T → {0. 2. Corollary 27.4 A set is ∆1 iﬀ it is hyperarithmetic. The deﬁnition of hyperarithmetic code (T. then there exists a ∆1 set 1 1 which separates them. in( ) = 1. q) is changed only by letting q map into the ﬁnite subsets of ω. Consequently. then in(s) = 1 iﬀ there exists n with sˆ n ∈ T and in(sˆ n) = 1. then in(s) = 1 iﬀ x ∈ q(s). if s ∈ T and not terminal and p(s) = 1. See Barwise [10] for a model theoretic or admissible sets approach to the hyperarithmetic hierarchy. 106 The theorem also holds for A and B disjoint Σ1 subsets of ω. then C is ∆1 . 1 Proof: This is true whether C is a subset of ω ω or ω. q) be a hyperarithmetic code for C. and (4) where 4 in( ) = 0. see Hinman [40]. 4. 1 Corollary 27.Descriptive Set Theory and Forcing where (u. etc). t) ∈ T + . . v. For more on the eﬀective Borel hierarchy. We just do the case C ⊆ ω ω . Details are left to the reader. C is Σ1 . p. One way 1 to see this is to identify ω with the constant functions in ω ω . 1} such that 1. p. if s a terminal node of T . Let (T. 3. It is clear that in is just coding up whether or not x ∈ Cs for 1 s ∈ T . if s ∈ T and not terminal and p(s) = 0. (2). Note that (1) thru (4) are all ∆1 (being a terminal node in a recursive tree 1 is Π0 . and ﬁnally.3 If C is a hyperarithmetic set. Theorem 27.5 If A and B are disjoint Σ1 sets. (3). To see that ∼ C is Σ1 note that x ∈ C iﬀ / 1 1 there exists in : T → {0. 1} such that (1).
The sets can be either subsets of ω or of ω ω . letting A0 = {x ∈ X : (x. which is a weakening of the scale property used in the proof of Π1 uniformity. A0 ∩ B0 = ∅. 1 So. B ⊆ X are Π1 where X = ω or X = ω ω . Let 1 P = (A × {0}) ∪ (B × {1}). Π1 reduction is the property that every pair of Π1 sets can be reduced by 1 1 a pair of Π1 sets. suppose A and B are Π1 subsets of ω ω . a b x → Tx and y → Ty which are “recursive” and a x ∈ A iﬀ Tx is wellfounded and b y ∈ B iﬀ Ty is wellfounded. for example.1) 1 1 there exists Q ⊆ P which is Π1 and for every x ∈ X. Now deﬁne .B iﬀ 1. 1 1 Proof: Suppose A. 0) ∈ Q} and B0 = {x ∈ X : (x. Then we know there 1 are maps from ω ω to trees. if there exists i ∈ {0. 1 Theorem 28. and 3. A0 ⊆ A and B0 ⊆ B.1 Π1 uniformity implies Π1 reduction.28 Π1 REDUCTION 1 107 28 Π1Reduction 1 We say that A0 .B0 reduce A. i) ∈ Q. 1} such that (x. Hence. 1) ∈ Q} gives a pair of Π1 sets which reduce A and B. then there exists a unique i ∈ {0. 1} 1 such that (x. A0 ∪ B0 = A ∪ B. 1 There is also a proof of reduction using the prewellordering property. Then P is a Π1 subset of X × ω ω and so by Π1 uniformity (Theorem 22. i) ∈ P . 2.
for any two dis1 1 joint Σ1 sets A and B there exists a ∆1 set C which separates them. in which case x ∈ B0 / and x ∈ A0 . then ∼ A0 = B0 . x ∈ B0 iﬀ x ∈ B and not (Tx a Tx ). / Theorem 28. or Tx Tx . then Tx is wellfounded and Tx is / b a illfounded and so not (Tx Tx ) and a ∈ A0 . Proof: Note that ∼ A ∪ ∼ B = X. If we set C = B0 .e. / b a then x ∈ B0 .e. then 1 C = B0 =∼ A0 ⊆∼ A so C ⊆∼ A and therefore A ⊆ C. so they are both ∆1 . 1 1 A ⊆ C and C ∩ B = ∅.. x ∈ A0 iﬀ x ∈ A and not (Tx a 2. if x ∈ B and x ∈ A. . then both Tx and Tx are wellfounded and either a b b a Tx Tx .Descriptive Set Theory and Forcing b 1. and b Tx ). On the other hand C = B0 ⊆∼ B implies C ∩ B = ∅. If A0 and B0 are Π1 sets reducing ∼ A and 1 ∼ B..2 Π1 reduction implies Σ1 separation. that A0 and B0 are Π1 subsets of A 1 1 a b and B respectively. in which case x ∈ A0 and x ∈ B0 . Similarly. If x ∈ A and x ∈ B. i. If x ∈ A ∩ B. i. 108 Since and are both Σ1 it is clear.
for every u ∈ ω ω and A ∈ Π1 (u) with A ⊆ X there exists e ∈ ω such 1 that A = {x ∈ X : (e. Here is what we mean by ∆1 codes for subsets of X where X = ω or 1 X = ωω .[31] ) ∆1 codes exist. u) ∈ C such 1 that D = {x ∈ X : (e.. u. u.1 (SpectorGandy Theorem [105]. u. u) ∈ C and remember that it is a Π1 predicate.e. x) ∈ S} • for any u ∈ ω ω and ∆1 (u) set D ⊆ X there exists a (e. Then deﬁne (e. u) ∈ C {x ∈ X : (e. x) ∈ S}. x) ∈ U iﬀ {e}u is the characteristic function of a tree T ⊆ n<ω (ω n × ω n ) and Tx = {s : (s. Theorem 29. 1 Proof: Let U ⊆ ω × ω ω × X be a Π1 set which is universal for all Π1 (u) sets. x) ∈ P } = {x ∈ X : (e.29 ∆1 CODES 1 109 29 ∆1codes 1 Using Π1 reduction and universal sets it is possible to get codes for ∆1 subsets 1 1 of ω and ω ω . using either 1 1 S or P . u. x) ∈ P } = {x ∈ X : (e. Note that x ∈ D can be said in either a Σ1 (u) way or Π1 (u) way. x s) ∈ T } is wellfounded. to get such a U proceed as follows. From now on we will write “e is a ∆1 (u)code for a subset of X” 1 to mean (e. Let {e}u be the partial function you get by using the eth Turing machine with oracle u. x) ∈ U }. For example. There exists a Π1 sets C ⊆ ω × ω ω and P ⊆ ω × ω ω × X and a Σ1 set 1 1 S ⊆ ω × ω ω × X such that • for any (e. x) ∈ P } = {x ∈ X : (e. x) ∈ S}. u. 1 We also write “D is the ∆1 (u) set coded by e” if “e is a ∆1 (u)code for a 1 1 subset of X” and D = {x ∈ X : (e. 1 1 i. . u.
x) : (e0 . Now apply reduction to obtain P 0 ⊆ U 0 and P 1 ⊆ U 1 which are Π1 sets. 1 0 1 0 Note that the by the nature of taking cross sections. 1 . u. u. For this e it must be that Ue. ∀n if n is the ∆1 (u)set coded by e.u ). then n is in the ∆1 (u)set coded by e. So this e is a ∆1 (u) code which 1 codes the set D. ∀n if n ∈ x.u or x ∈ Pe. Furthermore if D ⊆ X is a 1 ∆1 (u) set.Descriptive Set Theory and Forcing 110 Now get a doubly universal pair.u 1 1 and Ue.u reduce Ue. The pair of sets U 0 and U 1 are Π1 and doubly universal. u). u. x) ∈ U 0 } and B = {x : (e.u) = Se.u = Pe. and 1 3.u .u and Pe. u. x) ∈ U }.u = Pe. i..e. and 1 • P = P 0 and S =∼ P 1 .u = D and Ue. Note that e is a ∆1 (u) code is a Π1 statement in (e. Corollary 29. Also if e is a ∆1 (u) 1 1 1 code.u =∼ D. u. u. e1 ) be the usual recursive unpairing function from ω to ω × ω and deﬁne U 0 = {(e. there exists e 1 0 0 1 0 such that Ue. x) : (e1 . u) ∈ P (ω) × ω ω : x ∈ ∆1 (u)} is Π1 . then n ∈ x. x) ∈ U } and U 1 = {(e. 1 1 Proof: x ∈ ∆1 (u) iﬀ ∃e ∈ ω such that 1 1.2 {(x. then since U 0 and U 1 were a doubly universal pair. Let e → (e0 . then P(e. Now we deﬁne 1 0 • “e is a ∆1 (u) code” iﬀ ∀x ∈ X(x ∈ Pe.u 1 and Ue.u since the P ’s reduce the U ’s. Pe. 1 2. e is a ∆1 (u) code. x) ∈ U 1 }.u and so its a ∆1 (u) set. for any u ∈ ω ω 1 and A and B which are Π1 (u) subsets of X there exists e ∈ ω such that 1 A = {x : (e.
While clause (3) is Π1 if we 1 1 use that (e. m) : y(n) = m} is ∆1 (u). z) = ∃x ∈ ∆1 (y) θ(x. . y.g. then 1 ψ(y. z) where occurrences of the “q ∈ D” in the formula θ have been replaced by either (e. Clause (2) is Π1 if we use that (e. y. ∀x if x is the set coded by (e. z). 1 . n) ∈ S is equivalent to “n is in the ∆1 (u)set coded by e”. y. and 1 2. 1 1 Since being the graph a function is a Π0 property it is easy to see how to 2 obtain ∆1 (u) codes for functions y ∈ ω ω . n) ∈ P . e. q) ∈ P or (e. y. then θ(x. if n ∈ x then (e. z). y.. q) ∈ S. y).Descriptive Set Theory and Forcing 111 Note that clause (1) is Π1 . whichever is necessary to makes θ come out Π1 . The method of this corollary also works for the quantiﬁer ∃D ⊆ ω ω such that D ∈ ∆(y) θ(D. y. u. u. if (e. z) 1 is a Π1 formula. A similar argument 1 1 1 works for x ∈ ω ω . 1 We say that y ∈ ω ω is ∆1 (u) iﬀ its graph {(n. 1 Corollary 29. for x ⊆ ω we just say: 1 1 ∀n ∈ ω 1. Both of these clauses are Σ1 since S is Σ1 and P is Π1 . z) is a Π1 formula. 1 1 But this is ∆1 provided that e is a ∆1 (y) code. This will be Π1 just in case the clause “x is the set coded by (e. y. 1 Proof: ψ(y. n) ∈ S and 2. then n ∈ x. It is equivalent to say ∃e ∈ ω such that e is a ∆1 (y) code for a subset of 1 ω ω and θ(.3 Suppose θ(x. y)” is Σ1 . y. y. . . n) ∈ P is 1 1 equivalent to “n is in the ∆1 (u)set coded by e”. e is a ∆1 (y) code. z) iﬀ ∃e ∈ ω such that 1.
it is shown (assuming V=L in all three cases) that 1. Proof: By relativizing the following argument to an arbitrary parameter we may assume that the graph of f and the set B are ∆1 .Descriptive Set Theory and Forcing 112 Corollary 29. It follows that y ∈ f (B) iﬀ ∃x R(x. and 1 3. y)} is a ∆1 (y) singleton (or empty). there exists an uncountable Π1 set which is concentrated on the ratio1 nals (Erdos. y) 1 and so f (B) is both Σ1 and Π1 . and f is onetoone on B. is Borel. Deﬁne 1 R = {(x. its unique element is ∆1 in 1 1 y.4 Suppose f : ω ω → ω ω is Borel. . Then for any y the set {x : R(x. there exists a Π1 Hamel basis (Miller [85]). y) : f (x) = y and x ∈ B}. and Mauldin [21]). Miller. 1 1 Many applications of the GandySpector Theorem exist. y) iﬀ ∃x ∈ ∆1 (y) R(x. and Steel 1 [18]). 2. Consequently. f (B). Then the image of B under f . B ⊆ ω ω is Borel. Kunen. For example. there exists a topologically rigid Π1 set (Van Engelen.
Before giving the proof consider the following example.1 (Silver [101]) Suppose (X. since the proof readily relativizes 1 1 to an arbitrary parameter. Consequently. This is a Π1 subset 1 of 2ω×ω . Clearly. 1 i. Silver’s theorem is the best possible. then there exists a ∈ ω ω such that G = {p ∈ P : a ∈ p} and {a} = G. we need to show a ∈ B. if we restrict to W O. y ∈ W O. see also Kechris and Martin [53]. ) is a Π1 equivalence relation (since wellorderings are isomorphic iﬀ neither is 1 isomorphic to an initial segment of the other). then (W O. Let B = p[T ]. Furthermore. We can assume that X is ∆1 and E is Π1 . The proof we are going to give is due to Harrington [33]. This is known as Gandy forcing. Deﬁne a ∈ ω ω by [a n] ∈ G for each n. . Mansﬁeld and Weitkamp [73] and Louveau [64]. ) is a Σ1 equivalence relation with exactly ω1 / 1 equivalence classes. E) is a Π1 equivalence relation. G ⊆ {a}.113 Part IV Gandy Forcing 30 Π1 equivalence relations 1 Theorem 30. Note that there are many trivial generic ﬁlters corresponding to Σ1 singletons. 1 Lemma 30. Let P be the partial order of nonempty Σ1 subsets of ω ω ordered by 1 inclusion.2 If G is Pgeneric over V . X is a Borel set and E ⊆ X 2 is a Π1 equivalence relation on X.e. Now suppose B ∈ G. Now deﬁne x y iﬀ there exists an isomorphism taking x to y or x. without loss. we may assume that X = ω ω since we just make the complement of X into one more equivalence class. A model theoretic proof is given in Harrington and Shelah [38]. Proof: For every n an easy density argument shows that there exists a unique s ∈ ω n such that [s] ∈ G where [s] = {x ∈ ω ω : s ⊆ x}. Note that (2ω×ω . Then 1 either E has countably many equivalence classes or there exists a perfect set of pairwise inequivalent elements. Let W O be the set of all characteristic functions of wellorderings of ω. Also.
a1 (where . Proof: This is by induction on n. Suppose p[T x n. then a0 and a1 are both Pgeneric over V . Lemma 30. By the Claim we have that (x.a n+1 ] and so by a density argument there exists m = x(n) such that p[T x nˆm. But this is impossible. Consequently. because a ∈ q ∩ p. This proves the Claim. is the standard pairing function).30 Π1 EQUIVALENCE RELATIONS 1 Claim: There exists x ∈ ω ω such that p[T x n. Now suppose that a ∈ p ∈ P and p ∈ G. 1 We say that a ∈ ω ω is Pgeneric over V iﬀ G = {p ∈ P : a ∈ p} is Pgeneric over V . . Then p[T x n.a n+1 ] ∈ G. But note that p[T x n. Let E = {p ∈ P : {x0 : x ∈ p} ∈ D}.a n+1 ] ∈ G since p[T x n. Proof: The proof is symmetric so we just do it for a0 . Note that we are not claiming that they are product generic only that each is separately generic. Suppose D ⊆ P is dense open.a n ] ∈ G. G = {a}.3 If a is Pgeneric over V and a = a0 .a n+1 ] = m∈ω 114 p[T x nˆm. Then since / {q ∈ P : q ≤ p or q ∩ p = ∅} is dense there must be q ∈ G with q ∩ p = ∅. a) ∈ [T ] (since elements of P are nonempty) and so a ∈ p[T ] = B. but q ∩ p = ∅ is a Π1 sentence and hence absolute.a n ] and both of these are in G.a n ] ∈ G for every n ∈ ω.a n+1 ] = [a n + 1] ∩ p[T x n.
∀x. y in the set coded by e we have xEy. 1 1 1 z ∈ B iﬀ ∃e ∈ ω such that 1. 115 Since q0 is a nonempty Σ1 set and D is dense. 1 B= {D ⊆ ω ω : D ∈ ∆1 and ∀x. Since all elements of B are 1 equivalent. Proof: Let A = {x ∈ ω ω : ∀y y ∈ B → xEy}. Now we come to the heart of Harrington’s proof. and 3. e is a ∆1 code for a subset of ω ω . Deﬁne q0 = {x0 : x ∈ q}. so are all elements of A and hence D is as required. consequently by the Separation Theorem 27.5 or 28. y ∈ D we have that xEy.2 1 there exists a ∆1 set D with B ⊆ D ⊆ A. . Lemma 30. Then r ∈ E and r ≤ q. z is in the set coded by e. y ∈ D xEy}. y ∈ B we have that 1 xEy.e. Then A is a Π1 set which 1 contains the Σ1 set B.30 Π1 EQUIVALENCE RELATIONS 1 To see that E is dense let q ∈ P be arbitrary. i. 1 Since E is Π1 we know that by using ∆1 codes that this union is Π1 .. Then there exists a ∆1 set D with B ⊆ D ⊆ ω ω and such that for 1 every x. Let r = {x ∈ q : x0 ∈ r0 }. a0 ∈ p0 = {x0 : x ∈ p} ∈ D. Since E is dense we have that there exists p ∈ E with a ∈ p and consequently. there exists r0 ≤ q0 with 1 r0 ∈ D. Let B be the union of all ∆1 subsets of ω ω which meet only one equivalence class of E. 1 2. i.e.4 Suppose B ⊆ ω ω is Π1 and for every x.
This contradicts the fact that E is an equivalence relation. But ai ∈ C and C  ai Ec means that a0 Ec. Note that Q is a nonempty Σ1 set. and a0 Ea1 . Let a ∈ Q be Pgeneric over V . To see this replace V by a countable transitive model M of ZFC∗ (a suﬃciently large fragment of ZFC) and use absoluteness. then a0 Ea1 . Proof: Suppose not. such that for every x = y we have that (Gx .30 Π1 EQUIVALENCE RELATIONS 1 116 Note that item (1) is Π1 and (2) and (3) are both ∆1 (see Theorem 29. 1 1 If B = ω ω .4). c1 ∈ C. . Gy ) is P × Pgeneric over M . Then A ◦ P cE a ˇ ◦ where a is a name for the generic real (Lemma 30. then since there are only countably many ∆1 sets. a1 ∈ C.1). Note also that we don’t need to assume that there are a which are Pgeneric over V . But we these are all disjoint from A. a1 Ec.5 Suppose c ∈ ω ω ∩ V . Then there exists {Gx : x ∈ 2ω }. This is because a1 is Pgeneric over V [a0 ] and so a0 can be regarded as an element of the ground model. Then by 1 Lemma 30. a1 ) is P × Pgeneric over V and a1 ∈ A.3 we have that both a0 and a1 are Pgeneric over V and a0 ∈ C. 1 We know that there must exists c0 . and c0 Ec1 }. c1 ∈ C with c0 Ec1 .2). Note that the lemma implies that if (a0 . Lemma 30. Otherwise there would exists a ∆1 superset of C which meets only one equivalence class 1 (Lemma 30. So assume A =∼ B is a nonempty Σ1 set and in this case we will prove that there is a 1 perfect set of Einequivalent reals.6 Suppose M is a countable transitive model of ZFC∗ and P is a partially ordered set in M . Note that “E is an equivalence relation” is a Π1 statement hence it is 1 absolute. and a0 Ea1 . and let C ⊆ A be a nonempty Σ1 set such that C  cEa. a “perfect” set of Pﬁlters. Let Q = {c : c0 ∈ C. Lemma 30. there would 1 only be countably many E equivalence classes and we are done.
Finally to prove Theorem 30. Let {Gx : x ∈ 2ω } be given by Lemma 30. By Lemma 30. y ∈ A or x=y. pt ) ∈ Dn . 1 .1 let M be a countable transitive set isomorphic to an elementary substructure of (Vκ . Construct ps : s ∈ 2<ω by induction on the length of s so that 1. if s = t = n + 1 and s and t are distinct.7 Every Σ1 set which contains a real which is not ∆1 contains 1 1 a perfect subset.8 Every uncountable Σ1 set contains a perfect subset. Any perfect set of Einequivalent elements can contain at most one element of ∼ A. Now deﬁne for any x ∈ 2ω Gx = {p ∈ P : ∃n px n ≤ p}.6 with A ∈ Gx for all x and let P = {ax : x ∈ 2ω } be the corresponding generic reals. hence Harrington’s set 1 1 1 B in the above proof must be nonempty. then (ps . s ⊆ t implies pt ≤ ps and 2. Then is E is a / 1 Π1 equivalence relation. ∈) for some suﬃciently large κ. This is because for any n ∈ ω there exists m < ω such that every ps with s ∈ 2m decides a n. Deﬁne xEy iﬀ x. Proof: Let A ⊆ ω ω be a Σ1 set. A ∆1 singleton is a ∆1 real. Note also that P is perfect because the map x → ax is continuous. Corollary 30.30 Π1 EQUIVALENCE RELATIONS 1 117 Proof: Let Dn for n < ω list all dense open subsets of P × P which are in M . Corollary 30.5 we know that for every x = y ∈ 2ω we have that ax Eay .
and use item 2. b) is a P × Pgeneric over V pair of reals a. Thus if (a. Indeed otherwise the nonempty set Q = {(a. b ∈ A then a Eb. 2. see also Kanovei [48]) that there is a shorter proof of If (a. Contradiction with item 5.30 Π1 EQUIVALENCE RELATIONS 1 118 Perhaps this is not such a farfetched way of proving this result. Therefore aEb for any pair of Pgeneric a ∈ X and b ∈ Y . .3) a pair (a. a ) is not generic over V .6. 6. Indeed take b ∈ Y which is Pgeneric over V [a. b) and (a . a ∈ X. b) ∈ Q such that a ∈ X and b ∈ Y are Pgeneric. 3. since one of the usual proofs looks like a combination of Lemma 30. b) ∈ X × Y : a Eb} produces (by Lemma 30.5 and absoluteness: 1. Finally aEb for all a ∈ X and b ∈ Y . It follows that aEa for any two Pgeneric a. 4.] Then both (a. Similarly bEb for any pair of Pgeneric b. which avoids Lemma 30. b) ∈ X × Y is P × P generic then aEb. [Or over V [a] ∪ V [a ] if you see diﬃculties with V [a.2 and 30. b ∈ Y . a ]. Y ⊆ A such that X ×Y  aEb. Then there exist conditions X. Assume not. 5. a ] when the pair (a.Kanovei has pointed out to me (email. b) are P × Pgeneric. V.
. y) ≤ } is Borel. . Lemma 31. δ) is separable (i. xn ). First let us begin by isolating a principal which we call overﬂow. .2 (Harrington [39]) If (B. x2 . . y) ∈ B 2 : δ(x. . . . For n = 2 deﬁne B = {x : ∀y(y ∈ A → θ(x. . xn ∈ D θ(x1 . Then C is a Π1 set which also contains A.1 (Overﬂow) Suppose θ(x1 . The 1 proof for n > 2 is similar. Hence by separation there exists a ∆1 1 1 set E with A ⊆ E ⊆ B.5. It is an easy consequence of the Separation Theorem. Then there exists a ∆1 set D ⊇ A such that 1 ∀x1 . Theorem 31. . and for every ∈ Q the set {(x. We say that (B. . y ∈ P . Proof: For n = 1 this is just the Separation Theorem 27. xn ∈ A θ(x1 . . .. . δ) is a Borel metric space. Now deﬁne C = {x : ∀y(y ∈ E → θ(x. By applying separation again we 1 get a ∆1 set F with A ⊆ F ⊆ C. y)}. xn ) is a Π1 formula and A 1 is a Σ1 set such that 1 ∀x1 . δ is a metric on B. then either (B. y))}. . . Letting D = E ∩ F does the job. .31 BOREL METRIC SPACES AND LINES IN THE PLANE 119 31 Borel metric spaces and lines in the plane We give two applications of Harrington’s technique of using Gandy forcing. Then B is Π1 set which contains A.e. . . xn ). . y) > for every distinct x. contains a countable dense set) or for some > 0 there exists a perfect set P ⊆ B such that δ(x. . . δ) is a Borel metric space iﬀ B is Borel.
1 Proof: This follows from Lemma 31. ◦ ◦ . a1 ) : a0 .Descriptive Set Theory and Forcing 120 Proof: By relativizing the proof to an arbitrary parameter we may assume that B and the sets {(x. y ∈ B is a Π1 formula. a1 ∈ P and δ(a0 . Lemma 31.3 For any ∈ Q+ if A ⊆ B is Σ1 and the diameter of A is less 1 than . since θ(x.4 For every c ∈ V ∩ B P  δ(a. Then there exists P ≤ P such that P  δ(a. Since P is disjoint from Q by Lemma 31. c) > /3 ˇ where  is Gandy forcing and a is a name for the generic real (see Lemma 30. y) < and x.2).3 we know that the diameter of P is ≥ . then there exists a ∆1 set D with diameter less than and A ⊆ D ⊆ B. On the 1 other hand suppose for some ∈ Q+ we have that P = B \ Q = ∅. Proof: Suppose not. (B. 1 Note that Q is a Π1 set. Let R = {(a0 . y) ≤ } are ∆1 . y) iﬀ δ(x. 1 Lemma 31. y) ∈ B 2 : δ(x. a1 ) > (2/3) }. c) ≤ /3. 1 For any ∈ Q+ look at Q = {D ∈ ∆1 : D ⊆ B and diam(D) < }. If for every ∈ Q+ Q = B. δ) is separable and we are done.1. then since there are 1 only countably many ∆1 sets.
a1 ) > (2/3) . . 1 It is clear that ∼ P is Π1 and therefore P is Σ1 . Then there 1 exists a ∆1 set D with B ⊆ D such that all points of D are collinear. if a is Pgeneric over V with a ∈ R. This contradicts the fact that δ must remain a metric by absoluteness. 1 Proof: This follows from Lemma 31. y. y ∈ P . But by absoluteness δ(a0 . By relativizing the proof we may as well assume that A is Σ1 . then a0 and a1 are each separately Pgeneric over V . So assume that Q = P ∩ A = ∅. Kunen. z) iﬀ x. 1 Lemma 31.2. y) > /3 for each distinct x. either A can be covered by countably many lines or there exists a perfect set P ⊆ A such that no three points of P are collinear.6 is now easy to get a perfect set P ⊆ B such that δ(x. q) be the unique line on which they lie.Descriptive Set Theory and Forcing 121 Then R is in P and by Lemma 30. But a0 ∈ R and a1 ∈ R means that δ(a0 .6 Suppose B is a Σ1 set lying on a line in the plane. let line(p. and z are collinear is Π1 (even Π0 ). This proves Theorem 31. 1 1 Deﬁne ∼P = {D ⊆ R2 : D ∈ ∆1 and all points of D are collinear}. c) ≤ /3. p and q.1 since θ(x. y.5 (van Engelen.3. c) ≤ /3 and δ(a1 . The proof in [20] is more elementary. Theorem 31. then A can 1 1 be covered by countably many lines. Jackson. and Kechris. For any two distinct points in the plane. If P ∩ A = ∅. Using this lemma and Lemma 30. Proof: This existence of this proof was pointed out to me by Dougherty. Miller [20]) For any Σ1 set A in the 1 plane.
Lemma 31.3 each of the ai is Pgeneric if a is. p ˇ Since R is disjoint from {D ⊆ R2 : D ∈ ∆1 and all points of D are collinear} 1 it follows from Lemma 31. such that for every x. a1 .1) a perfect set of triply generic points in the plane. ◦ ◦ . Gz ) is P × P × Pgeneric over M .7 and 31. z distinct. a1 . q) which makes them collinear. This proves Theorem 31. a2 are not collinear} where a = (a0 .8 it is easy to get (just as in the proof of Theorem 30. Using Lemma 31. But this is a contradiction since all ai ∈ line(p. q ). The following Lemma is an easy generalization of Lemma 30. we have that (Gx . Any Σ1 subset of Rn which cannot be covered by countably many lines 1 contains a perfect set all of whose points are collinear.5. Deﬁne the nonempty Σ1 set 1 S = {a : a0 . Obvious generalizations of Theorem 31.5 are: 1.7 For any two distinct points in the plane. q ∈ V Q  a∈ line(ˇ. a “perfect” set of Pﬁlters.6 so we leave the proof to the reader.8 Suppose M is a countable transitive model of ZFC∗ and P is a partially ordered set in M . Then S ∈ P and by the obvious generalization of Lemma 30. q ). a2 ∈ R and a0 . Then there exists {Gx : x ∈ 2ω }. hence no three of which are collinear. a2 ) via some standard tripling function. Gy .Descriptive Set Theory and Forcing 122 Lemma 31. with p. / p ˇ Proof: Suppose for contradiction that there exists R ≤ Q such that R  a∈ line(ˇ. p and q. y. a1 .6 that not all triples of points from R are collinear.
Higher dimensional version of the above involving spheres or other surfaces. Any Σ1 subset of R2 which cannot be covered by countably many 1 parabolas contains a perfect set which does not contain four points on the same parabola. linear orders. Any Σ1 subset of R2 which cannot be covered by countably many circles 1 contains a perfect set which does not contain four points on the same circle. Q. see Louveau [67]. y) iﬀ x. Louveau [68] is a survey paper on Borel equivalence relations. and partial orders. ≤) where P and ≤ are ∆1 . A very general statement of this type is due to Solecki [102]. 5. Given any Polish space X. then either P is the union of countably many chains or there exist a perfect set P of pairwise incomparable elements. 3. Then 1 θ(x. It is due to Harrington [39]. either A can be covered by countably many elements of Q or there exists a Gδ set B ⊆ A such that B cannot be covered by countably many elements of Q. Another result of this type is known as the BorelDilworth Theorem.9 Suppose A is a Σ1 chain in a ∆1 poset P.Descriptive Set Theory and Forcing 123 2. and analytic A ⊆ X. family of closed sets Q in X. Then there exists 1 1 a ∆1 superset D ⊇ A which is a chain.5 from this. One of the early Lemmas from [39] is the following: Lemma 31.1. Solecki deduces Theorem 31.Feng [22] has shown that given an open partition of the two element subsets of ω ω . 1 Proof: Suppose P = (P. y ∈ P and (x ≤ y or y ≤ x) is Π1 and so the result follows by Lemma 31. 1 For more on Borel linear orders. 4. that either ω ω is the union of countably many 0−homogenous . It says that if P is a Borel partially ordered set. For any n any Σ1 subset of R2 which cannot be covered by countably 1 many polynomials of degree < n contains a perfect set which does not contain n + 1 points on the same polynomial of degree < n.
Todorcevic [111] has given an example showing that this is false for Borel partitions (even replacing open by closed). .Descriptive Set Theory and Forcing 124 sets or there exists a perfect 1−homogeneous set.
32 Σ1 EQUIVALENCE RELATIONS 1 125 32 Σ1 equivalence relations 1 Theorem 32. Proof: We will need to prove the boundedness theorem for this result. Another way to prove this is take a tree T of rank α and note that ˆ ˆ W F<α = {T : T T } and this set is ∆1 and hence Borel by Theorem 26.2 For each α < ω1 the set W F<α is Borel. 1 . Lemma 32.4). Deﬁne W F = {T ⊆ ω <ω : T is a wellfounded tree}. On the other hand each of the W F<α are Borel. For λ a limit s s W F<λ = W F<α . s ∈ T. i. Consequently. Proof: Deﬁne for s ∈ ω <ω and α < ω1 s W F<α = {T ⊆ ω <ω : T is a tree. α<λ For a successor α + 1 s sˆn T ∈ W F<α+1 iﬀ s ∈ T and ∀n (sˆ n ∈ T → T ∈ W F<α ). The set of trees 0 is Π1 . W F is a complete Π1 set. rT (s) < α}.. s The fact that W F<α is Borel is proved by induction on α. Then 1 either E has ≤ ω1 equivalence classes or there exists a perfect set of pairwise Einequivalent reals.1.e. For α < ω1 deﬁne W F<α to the subset of W F of all wellfounded trees of rank < α. W F is not Borel.1 (Burgess [14]) Suppose E is a Σ1 equivalence relation. for every B ⊆ ω ω which is Π1 there 1 1 exists a continuous map f such that f −1 (W F ) = B (see Theorem 17.
While there is no reason to expect that any of the Eα are equivalence relations.e. E= α<ω1 ˆ T. then there exists 1 α < ω1 such that A ⊆ W Fα . Proof: Suppose no such α exists. / Deﬁne xEα y iﬀ Txy ∈ W F<α . Otherwise. if A is CK a Σ1 subset of W F . Lemma 32. i. then there exists a recursive ordinal α < ω1 such that 1 A ⊆ W F<α .3 Boundedness Theorem If A ⊆ W F is Σ1 . Note that Eα reﬁnes Eβ for α > β. By the Normal Form 1 Theorem 17.4 For unboundedly many α < ω1 the binary relation Eα is an equivalence relation.Descriptive Set Theory and Forcing 126 Lemma 32.4 we know there exists a continuous mapping (x..2 we know that the binary relation Eα is Borel. we use the boundedness theorem to show that many are. Then ˆ T ∈ W F iﬀ there exists T ∈ A such that T But this would give a Σ1 deﬁnition of W F . y) → Txy such that Txy is always a tree and xEy iﬀ Txy ∈ W F . 1 Now suppose that E is a Σ1 equivalence relation. Eα and for any limit ordinal λ Eλ = α<λ Eα . {e ∈ ω : e is the code of a recursive wellfounded tree } would be Σ1 . 1 There is also a lightface version of the boundedness theorem. . / By Lemma 32. contradiction. Clearly.
The following claim will allow us to handle symmetry. y. Claim: For every α < ω1 there exists β < ω1 such that for every x. Proof: Let A = {Txy : xEα y and y Eα x}.3 there exists β < ω1 such that A ⊆ W F<β .Descriptive Set Theory and Forcing 127 Proof: Note that every Eα must be reﬂexive. Proof: Let B = {Txy ⊕ Tyz : xEα y. Claim: For every α < ω1 there exists β < ω1 such that for every x. then x Eβ y. The operation ⊕ on a pair of trees T0 and T1 is deﬁned by T0 ⊕ T1 = {(s. by the Boundedness Theorem there is a β < ω1 such that B ⊆ W F<β and this proves the Claim. it follows that A ⊆ W F . (Deﬁne the rank function on T0 ⊕ T1 by taking the minimum of the rank functions on the two trees. then either x Ey or y Ez. then either x Eβ y or y Eβ z. and x Eα z}. yEα z. Note that the rank of T0 ⊕ T1 is the minimum of the rank of T0 and the rank of T1 . Since y Eα x implies y Ex and hence x Ey. Then A is a Borel set. and x Eα z. It follows that either Txy ∈ W F or Tyz ∈ W F and so in either case Txy ⊕ Tyz ∈ W F and so B ⊆ W F . Note also that since x Eα z implies x Ez and E is an equivalence relation. Using the usual LowenheimSkolem argument we can ﬁnd arbitrarily large countable ordinals λ such that .) The set B is Borel because the relation Eα is. since E is reﬂexive and E = α<ω1 Eα . t) : s ∈ T0 . t ∈ T1 . z if xEα y and yEα z. The next claim is to take care of transitivity. By the Boundedness Theorem 32. and s = t}. y if xEα y and y Eα x. Again. Now we use the Claims to prove the Lemma.
The following is a Σ1 2 statement: V [G] = ∃P perfect ∀x∀y (x. V by Silver’s Theorem 30.g it is possible to have c = ω23 and there exists a Π1 set B with 2 . then it still is one in V [G]. a contradiction. Since in V [G] the equivalence relation E reﬁnes Eω1 .1 it is consistent to have Π1 sets of arbitrary 2 cardinality.1 there exists a perfect set P of Eω1 inequivalent V reals. Then Q has cardinality ω1 in V [G] and V for every x = y ∈ Q there exists α < ω1 with x Eα y. Let M [G] be obtained as M in the above proof by Levy collapsing ω1 . it V follows that the Borel relation Eω1 is an equivalence relation. Hence it must be that V the elements of Q are in diﬀerent Eω1 equivalence classes. Consequently. By absoluteness. E is to ω1 (see Kunen [56] or Jech [44]). Then we can conclude as above that M thinks E has a perfect set of inequivalent elements. A similar argument using the second Claim works for transitivity. But by the Claim there exist β < λ such that x Eβ y and hence x Eλ y. By Harrington’s Theorem 25.Descriptive Set Theory and Forcing 128 for every α < λ there is a β < λ which satisﬁes both Claims for α. A way to avoid taking a generic extension of the universe is to suppose Burgess’s Theorem is false. Hence. For suppose xEλ y and y Eλ x. Then let M be the transitive collapse of an elementary substructure of some suﬃciently large Vκ (at least large enough to know about absoluteness and Silver’s Theorem). For example. Then ω1 still an equivalence relation and for any α if Eα was an equivalence relation in V . which contradicts the assumption that M thought Burgess’s Theorem was false. Then since Eλ = α<λ Eα there must be α < λ such that xEα y and y Eα x. Since V Eω1 = V α<ω1 Eα and the intersection of equivalence relations is an equivalence relation. Let G be any generic ﬁlter over V with the property that it collapses ω1 but not ω2 . Let Q be a set of size ω2 in V of pairwise Einequivalent reals. it must be that the elements of P are pairwise Einequivalent also. But this means that Eλ is an equivalence relation. So now suppose that E had more than ω2 equivalence classes in V . by Shoenﬁeld Absoluteness 20. V must think that there is a perfect set of Einequivalent reals.2. Levy forcing with ﬁnite partial functions from ω V [G] V = ω2 . y ∈ P and x = y) → x Ey. e.
. Marker. but 2 since the continuum is ω23 there is no perfect set of Einequivalent reals. y ∈ B / 129 then we get Σ1 equivalence relation with exactly ω17 equivalence classes. Sami [96]. Shelah [97]. Harrington and Shelah [38]. Hence. Kechris [51]. if we deﬁne xEy iﬀ x = y or x. and Harrington. and Shelah [39]. Stern [109] [110]. For further results concerning projective equivalence relations see Harrington and Sami [37].Descriptive Set Theory and Forcing B = ω17 . See Burgess [15] [16] and Hjorth [41] for more results on analytic equivalence relations.
q). say all sets of the form [s] for s ∈ ω <ω plus the empty set. ω1 is the ﬁrst nonrecursive ordinal. B ⊆ ω ω are Σ1 sets. q) the Π0 set and the Σ0 set coded by (T. then α B ∈ Σ0 ({D × C : D ∈ Borel(ω ω ) and Cis clopen}). α CK 5. q) = S (T. q) = 3. and 1 0 A and B can be separated by Πα set. q)sˆm : sˆm ∈ T } and S s (T.3 (Section Problem) If B ⊆ ω ω × ω ω is Borel and α < ω1 is such that Bx ∈ Σ0 for every x ∈ ω ω . Deﬁne (T. q) =∼ q(s). CK Theorem 33. q) for s ∈ T by induction on the rank of s as follows.2 ∆1 ∩ Π0 = Π0 (hyp) α 1 α Corollary 33. T 0 . q) and S(T. respectively. Deﬁne P (T. Deﬁne S s (T. Corollary 33. (S is short for α α Sigma and P is short for Pi. B. q) = P (T. q). q) and P s (T. rank zero nodes) to a nice base. α < ω1 . For s ∈ T >0 deﬁne P s (T. α . for the clopen sets of ω ω . Deﬁnitions 1. of T (i.e. then A and B can be separated by a Π0 (hyp)set. For s ∈ T 0 deﬁne P s (T. q) is an αcode iﬀ T ⊆ ω <ω is a tree of rank ≤ α and q : T 0 → B is a map from the terminal nodes.) 4.1 (Louveau [65]) If A. 2.33 LOUVEAU’S THEOREM 130 33 Louveau’s Theorem Let us deﬁne codes for Borel sets in our usual way of thinking of them as trees with basic clopen sets attached to the terminal nodes. q) = q(s) and S s (T. α {S(T. Deﬁne C ⊆ ω ω is Π0 (hyp) iﬀ it has an αcode which is hyperarithmetic. q) =∼ P s (T.
q) such that R(x. q) = Bx . In their paper [66] Louveau and SaintRaymond give a diﬀerent proof of it. By relativizing the proof CK to a parameter we may assume α < ω1 and that B is ∆1 . CK Lemma 33. y) : y is a βcode for some β < α and x ∈ S(T. q). z) iﬀ there exists (T. (T.3. R(x. q)).3 by induction on α. Let T be a recursive tree of rank α. q) ∈ ∆1 (x). q)}. q)). By taking 1 complements we may assume that the result holds for Π0 for all β < α.4 For α < ω1 the following sets are ∆1 : 1 {y : y is a βcode for some β < α}. q) ∈ ∆1 (x) such that 1 ˆ 1. 2. q) is the Π0 set coded by (T. By Π1 uniformization (Theorem 22. (T. β Deﬁne R(x. SaintRaymond proved it for α = 2 and Louveau and SaintRaymond independently proved it for α = 3 and then Louveau proved it in general. This result was proved by Dellecherie for α = 1 who conjectured it in general. rankT ( n ) = β and 3. q)). q).n (x. For the second set just use an argument similar to Theorem 27.1) there exist a Π1 set R ⊆ R 1 1 ˆ such that for every x there exists a unique (T. (T. Note that by the relativized α version of Louveau’s Theorem for every x there exists a (T. We will need the following lemma. and P (T. (T. and {(x.Descriptive Set Theory and Forcing 131 Note that the converse is trivial. The third set is just the complement of the second one. Proof: For the ﬁrst set it is enough to see that W F<α the set of trees of rank < α ˆ ˆ is ∆1 . Now we prove Corollary 33. Fix β < α and n < ω and deﬁne Bβ. (T. 1 where P (T. . q)} . q) such that ˆ R(x. But since T is wellfounded T T iﬀ ¬(T T ) and so 1 it is Π1 . Then T ∈ W F<α iﬀ T T shows 1 1 ˆ ˆ ˆ that W F<α is Σ1 . q) is an αcode. y) : y is a βcode for some β < α and x ∈ P (T. q)) iﬀ (T. z ∈ P n (T. {(x.
z ∈ [s]. q( n ) = s. R(x.n (x. 2. 1 . z ∈ S n (T. and 4. q)). rankT ( n ) = β. z) iﬀ there exists (T. α Now to do the case for α = 1. Note that every 1 1 cross section of Bβ. then deﬁne the 1 Borel set Cs.n = Cs. (T. But note that ¬Bβ. q).β<α Bβ. 3. α But then B= n<ω.n is Π1 .s∈ω <ω Bs. rankT ( n ) = 0.n is a Π0 set and so by induction (in case α > 1) β Bβ.n and so B ∈ Σ0 ({D × C : D ∈ Borel(ω ω ) and Cis clopen}).n ∈ Π0 ({D × C : D ∈ Borel(ω ω ) and Cis clopen}).n and so B ∈ Σ0 ({D × C : D ∈ Borel(ω ω ) and C clopen}). R is Π1 . Let z0 ∈ [s] be arbitrary. we see that Bβ. or 3. q) ∈ ∆1 (x) such that 1 ˆ 1.n = {x : (x.Descriptive Set Theory and Forcing 132 ˆ Since quantiﬁcation over ∆1 (x) preserves Π1 (Theorem 29.n }. q) ∈ ∆1 (x) such that 1 ˆ 1. As in the other case Bs.n is ∆1 . z) iﬀ there exists (T. (T. 2.n is Π1 and therefore Bβ.n × [s] where But now B= n<ω. Then Bs. R(x.4. q)).n is ∆1 . deﬁne for every n ∈ ω and s ∈ ω <ω Bs. z0 ) ∈ Bs.3). and consequently. and the 1 1 1 1 1 rest is ∆1 by Lemma 33.n (x. ∼ Bβ.
α In order to prove this corollary we will need the following lemmas. Now put (x. for each s ∈ T the clopen set Cs is the disjoint union of the clopen sets {Csˆn : sˆn ∈ T }. Proof: Build a tree Cs : s ∈ T of nonempty clopen sets indexed by a tree T ⊆ ω <ω such that 1. the diameter of Cs is less that 1/s for s = . 2. i. Lemma 33. let V be a Π0 set which is universal for Π0 α α sets. and 3. C = X. y) ∈ U iﬀ y ∈ Vx0 and ∀z(z ∈ Vx0 iﬀ z ∈ Vx1 ) / where x = (x0 . x1 ) is some standard pairing function.5 There can be no Borel set universal for all ∆0 sets. If Y = [T ] (the inﬁnite branch of T ). A space is Polish iﬀ it is a separable complete metric space. Louveau’s Theorem implies that there can be no α Borel set universal for all ∆0 sets. I think the argument for the next lemma comes from a theorem about Hausdorﬀ that lifts the diﬀerence hierarchy on the ∆0 sets to the ∆0 sets.. so there is a Σ1 which is α 1 universal for all ∆0 sets.Descriptive Set Theory and Forcing 133 Note that for every α < ω1 there exists a Π1 set U which is universal for 1 all ∆0 sets.13 13 Latin for plagiarized.6 If X is a 0dimensional Polish space. . then X and Y are homeomorphic. I am not sure who proved this ﬁrst. To see this. then there exists a closed set Y ⊆ ω ω such that X and Y are homeomorphic. Note also that the complement of U is also universal for all ∆0 sets. α Corollary 33. every cross section of U is ∆0 and every ∆0 set occurs as a α α α cross section of U . This presentation is taken from Kechris 2 α [54] mutatis mutandis.e.
Proof: This will follow easily from the next two claims. τn ).e. Since X0 is closed in X the complete metric on X is complete when restricted to X0 . σ) such that τn ⊆ σ for every n.Descriptive Set Theory and Forcing 134 Lemma 33.. Let (X. In this case the complete metric d on ∼ F would be deﬁned by ˆ d(x. (In fact n<ω τn is a subbase for σ. σ) = X0 ⊕ X1 be the discrete topological sum. y) + 1 1 − d(x. n∈ω . y) = d(x. Claim: If (X. τn ) for n ∈ ω are 0dimensional Polish topologies extending τ . τ ) is a 0dimensional Polish space and F ⊆ X is closed. τ ∪ {F } is a subbase for σ. F ) d(y. (In fact. which contains the standard topology and each Bn is a clopen set in τ . τ ) is a Hausdorﬀ space and (X. Claim: Suppose (X.) Proof: Consider the 0dimensional Polish space (X. i. τ . Since ∼ F is open there is another metric which is complete on X1 . then there exists a 0dimensional Polish topology σ ⊇ τ such that F is clopen in (X. F ) is the distance from x to the closed set F . U is open iﬀ U = U0 ∪U1 where U0 ⊆ X0 is open in X0 and U1 ⊆ X1 is open in X1 . then there exists a 0dimensional Polish topology (X. σ). F ) where d is a complete metric on X and d(x. This is a special case of Alexandroﬀ’s Theorem which says that a Gδ set in a completely metrizable space is completely ˆ metrizable in the subspace topology.7 For any sequence Bn : n ∈ ω of Borel subsets of ω ω there exists 0dimensional Polish topology.) Proof: Let X0 be F with the subspace topology given by τ and X1 be ∼ F with the subspace topology.
e. It follows that σ is a Polish topology on X. f (x) = xn : n ∈ ω where xn = x for every n). τn ) with U = f −1 (V ). τn ) be the range of f . Let B = n∈ω Bn where each Bn is α Π0 for some β < α. Apply the ﬁrst Claim to get a 0dimensional Polish topology in which B is clopen.7 there exists a 0dimensional Polish topology τ such that if X = (ω ω . Consequently. Let D ⊆ n∈ω (X. which is easily seen to be impossible by the α standard diagonal argument. τ ) .7 by induction on the rank of the Borel sets. The idea of this proof is to reduce it to the case of a 0 ∆α set universal for ∆0 . To see that τn ⊆ σ for every n let U ∈ τN and deﬁne V = n<N X ×U × n>N X.3 B ∈ ∆0 ({D × C : D ∈ Borel(ω ω ) and C is clopen}). α By Lemma 33. Let σ be the topology on X deﬁned by U ∈ σ iﬀ there exists V open in n∈ω (X. We prove Lemma 33. τn ).sets. Applying the second Claim gives us a 0dimensional topology σ extending τ in which each Bn is clopen and therefore B is open. Note that by the second Claim it is enough to prove it for one Borel set at a time.5). τ ) is a homeomorphism. Since each τn extends τ we get that D is a closed subset of n∈ω (X. τ ). So suppose B is a Σ0 subset of (X. D with the subspace topology inherited from n∈ω (X. Proof: (of Corollary 33. τ ) → (D. Then f −1 (V ) = U and so U ∈ σ. By induction on α there exists a 0dimensional Polish β topology τn extending τ in which each Bn is clopen.Descriptive Set Theory and Forcing 135 Let f : X → n∈ω (X. the set of constant sequences. τn ) is Polish. τn ) be the embedding which takes each x ∈ X to the constant sequence x (i. Then by the α Corollary 33. Note that f : (X.. Suppose B is a Borel set which is universal for all ∆0 sets.
Descriptive Set Theory and Forcing
136
then B is ∆0 (X × ω ω ). Now by Lemma 33.6 there exists a closed set Y ⊆ ω ω α and a homeomorphism h : X → Y . Consider C = {(x, y) ∈ X × X : (x, h(y)) ∈ B}. The set C is ∆0 in X × X because it is the continuous preimage of the set B α under the map (x, y) → (x, h(y)). The set C is also universal for ∆0 subsets α of X because the set Y is closed. To see this for α > 1 if H ∈ ∆0 (Y ), then α H ∈ ∆0 (ω ω ), consequently there exists x ∈ X with Bx = H. For α = 1 just α use that disjoint closed subsets of ω ω can be separated by clopen sets. Finally, the set C gives a contradiction by the usual diagonal argument: D = {(x, x) : x ∈ C} / would be ∆0 in X but cannot be a cross section of C. α Question 33.8 (Mauldin) Does there exists a Π1 set which is universal for 1 all Π1 sets which are not Borel?14 1 We could also ask for the complexity of a set which is universal for Σ0 \∆0 α α sets.
14
This was answered by Greg Hjorth [42], who showed it is independent.
34 PROOF OF LOUVEAU’S THEOREM
137
34
Proof of Louveau’s Theorem
Finally, we arrive at our last section. The following summarizes how I feel now. You are walking down the street minding your own business and someone stops you and asks directions. Where’s xxx hall? You don’t know and you say you don’t know. Then they point at the next street and say: Is that xxx street? Well by this time you feel kind of stupid so you say, yea yea that’s xxx street, even though you haven’t got the slightest idea whether it is or not. After all, who wants to admit they don’t know where they are going or where they are.
CK For α < ω1 deﬁne D ⊆ ω ω is Σ0 (semihyp) iﬀ there exists S a Π1 set α 1 of hyperarithmetic reals such that every element of S is a βcode for some β < α and D = {P (T, q) : (T, q) ∈ S}.
A set is Π0 (semihyp) iﬀ it is the complement of a Σ0 (semihyp) set. The α α Π0 (semihyp) sets are just the usual clopen basis ([s] for s ∈ ω <ω together 0 with the empty set) and Σ0 (semihyp) sets are their complements. 0 Lemma 34.1 Σ0 (semihyp) sets are Π1 and consequently Π0 (semihyp) sets α 1 α are Σ1 . 1 Proof: x ∈ {P (T, q) : (T, q) ∈ S} iﬀ there exists (T, q) ∈ ∆1 such that (T, q) ∈ 1 S and x ∈ P (T, q). Quantiﬁcation over ∆1 preserves Π1 ( see Corollary 29.3 1 1 ) and Lemma 33.4 implies that “x ∈ P (T, q)” is ∆1 . 1 We will need the following reﬂection principle in order to prove the Main Lemma 34.3. A predicate Φ ⊆ P (ω) is called Π1 on Π1 iﬀ for any Π1 set N ⊆ ω × ω 1 1 1 the set {e : Φ(Ne )} is Π1 (where Ne = {n : (e, n) ∈ N }). 1 Lemma 34.2 (Harrington [39] Kechris [50]) Π1 Reﬂection. Suppose Φ(X) 1 is Π1 on Π1 and Q is a Π1 set. 1 1 1 If Φ(Q), then there exists a ∆1 set D ⊆ Q such that Φ(D). 1
Descriptive Set Theory and Forcing
138
Proof: By the normal form theorem 17.4 there is a recursive mapping e → Te such that e ∈ Q iﬀ Te is wellfounded. Deﬁne for e ∈ ω
0 Ne = {ˆ : Te e ˆ 1 e Ne = {ˆ : ¬(Te
Te } Te )} ˆ
0 1 then N 0 is Σ1 and N 1 is Π1 . For e ∈ Q we have Ne = Ne = De ⊆ Q is ∆1 ; 1 1 1 1 and for e ∈ Q we have that Ne = Q. If we assume for contradiction that / 1 ¬Φ(Ne ) for all e ∈ Q, then 1 e ∈ Q iﬀ φ(Ne ). /
But this would mean that Q is ∆1 and this proves the Lemma. 1 Note that a Π1 predicate need not be Π1 on Π1 since the predicate 1 1 1 Φ(X) = “0 ∈ X” / is ∆0 but not Π1 on Π1 . Some examples of Π1 on Π1 predicates Φ(X) are 0 1 1 1 1 Φ(X) iﬀ ∀x ∈ X θ(x) / or Φ(X) iﬀ ∀x, y ∈ X θ(x, y) / where θ is a Π1 sentence. 1 Lemma 34.3 Suppose A is Σ1 and A ⊆ B ∈ Σ0 (semihyp), then there exists 1 α C ∈ Σ0 (hyp) with A ⊆ C ⊆ B. α Proof: Let B = {P (T, q) : (T, q) ∈ S} where S is a Π1 set of hyperarithmetic 1 ˆ < αcodes. Let S ⊆ ω be the Π1 set of ∆1 codes for elements of S, i.e. 1 1 ˆ e ∈ S iﬀ e is a ∆1 code for (Te , qe ) and (Te , qe ) ∈ S. 1 Now deﬁne the predicate Φ(X) for X ⊆ ω as follows: ˆ Φ(X) iﬀ X ⊆ S and A ⊆
e∈X
P (Te , qe ).
β But now deﬁne (T. β As usual. β By Lemma 34. The 1topology is just the standard topology on ω ω . q) is ∆1 and hence hyperarithmetic.2) there exists a ∆1 set D ⊆ S such that Φ(D).2. Since D is ∆1 it is easy to check that (T. Note that Q is a Π1 set and consequently. i. 1 1 Since Φ(D) holds it follows that C = S(T. Therefore by reﬂection 1 1 1 ˆ (Lemma 34. as desired. q) ∈ ∆1 . the partial order of all nonempty Σ1 subsets 1 ◦ of ω ω and let a be a name for the real obtained by forcing with P.4 If A is Σ1 . Let P be Gandy forcing. q) is a βcode for some β < α. α Note that it follows from the Lemmas that for A a Σ1 set. q) ∈ Q iﬀ (T.3 this same intersection can be written: clα (A) = {X ⊇ A : ∃β < α X ∈ Σ0 (hyp)}. so that by Lemma 30. 1 α Proof: Since the Σ0 (semihyp) sets for β < α form a basis for the αclosed sets. q) the Σ0 (hyp) set coded by (T. 1 and A ⊆ S(T. clα (A) denotes the closure of the set A in the αtopology. β clα (A) = {X ⊇ A : ∃β < α X ∈ Σ0 (semihyp)}. The αtopology has its basis certain special Σ1 sets so it is intermediate between the standard 1 topology and the Gandy topology corresponding to Gandy forcing. CK Deﬁne for α < ω1 the αtopology by taking for basic open sets the family {Π0 (semihyp) : β < α}. q). q) by T = {eˆs : e ∈ D and s ∈ Te } q(eˆs) = qe (s) for e ∈ D and s ∈ Te0 . (T. Deﬁne (T.Descriptive Set Theory and Forcing 139 ˆ The predicate Φ(X) is Π1 on Π1 and Φ(S).e. clα (A) is a Σ1 1 1 set which is a basic open set in the βtopology for any β > α. clα (A) is a 1 Π0 (semihyp) set. q) α ˆ has the property that A ⊆ C and since D ⊆ S it follows that C ⊆ B. Lemma 34. we have that p ∈ G iﬀ aG ∈ p.. for any G which is Pgeneric. . then clα (A) is Π0 (semihyp).
Proof: This is proved by induction on α. Suppose for contradiction that clα (p)  a∈ C ◦ Then for some n < ω and r ≤ clα (p) it must be that r  a∈ Cn .Descriptive Set Theory and Forcing CK Lemma 34. then clα (p)  a∈ C. But since clβ (r) is αopen this means that clβ (r) ∩ clα (p) = ∅ which contradicts the fact that r ≤ clα (p).1.5 For any α < ω1 . For α > 1 let C= ∼ Cn n<ω ◦ ◦ where each Cn is Π0 β for some β < α. else there exists s ∈ ω <ω with q = p ∩ [s] nonempty and [s] ∩ C = ∅. Since / cl(p)  a ∈ cl(p). and C ∈ Π0 (coded in V ) if α 140 p  a∈ C. If p  a∈ C. Then by induction β clβ (r)  a∈ Cn . ◦ ◦ .4 and hence a basic open β set in the αtopology. it follows that cl(p)  a ∈ C. For α = 1 recall that the αtopology is the standard topology and C ◦ is a standard closed set. Since A ⊆ C it α 1 follows that ◦ A  a∈ C. cl(p) ⊆ C. Note that since they force contradictory information ◦ ◦ (clβ (r)  a∈ C and p  a∈ C) it must be that clβ (r) ∩ p = ∅. Now we are ready to prove Louveau’s Theorem 33. p ∈ P. Suppose that Cn is Π0 for some β < α. (otherwise the / two conditions would be compatible in P). But clβ (r) is a Π0 (semihyp) set by Lemma 34. Suppose A and B are Σ1 sets and C is a Π0 set with A ⊆ C and C ∩ B = ∅. Hence p ⊆ C and since C is closed. But then ◦ q ≤ p and q  a∈ C. then it better be that p ⊆ C.
” “What do you mean?” says the man. but you can’t have an elephant sandwich. Its just that here it is 5pm already and I just don’t want to start another elephant.3 to its complement there exists a Π0 (hyp)set C with clα (A) ⊆ C and α C ∩ B = ∅. So clα (A) is a Πα (semihyp)set (Lemma 34.” Says the proprietor “Oh we have elephant. he walks in and says to the proprietor.4) which is disjoint from the set B and thus by applying Lemma 34. we have it”. For a generalization to higher levels of the projective hierarchy using determinacy. otherwise letting p = clα (A) ∩ B would be a condition of P such that p  a∈ C and p  a∈ B which would imply that B ∩ C = ∅ in the generic extension.” . you don’t have. The argument presented here is partially from Harrington [34]. Splashed all over are signs saying “Order any sandwich”.” The proprietor responds “Sorry.5 it follows that clα (A)  a∈ C.Descriptive Set Theory and Forcing By Lemma 34. but contains even more simpliﬁcation brought about by using forcing and absoluteness. Louveau’s Theorem is also proved in Sacks [95]. and “All kinds of sandwiches”. “All your signs say to order any sandwich. ◦ 141 Now it must be that clα (A) ∩ B = ∅. And here the ﬁrst thing I ask for. “I would like an elephant sandwich. But by absoluteness B and C must remain disjoint. ◦ ◦ Elephant Sandwiches A man walks by a restaurant. Mansﬁeld and Weitkamp [73] and Kanovei [48]. see Hjorth [43]. Intrigued. “Just ask us.
Fundamenta Mathematicae. R. H.Bartoszynski. 719726.Bartoszynski. Cited on page: 17 [6] T. Fundamenta Mathematicae. W. 97176. 1975. There is a Qset space in ZFC. 337357. Cited on page: 75 [8] T.Baire.Shelah.Balogh. Admissible sets and structures.D. Separation principles in the hierarchies of classical and eﬀective descriptive set theory. Cited on page: 75 [9] T. Cited on page: 75 [10] J.W. Cited on page: 3 [4] R. 46(1957). On the combinatorial principal P(c). S.REFERENCES 142 References [1] J. Some consequences of the axiom of constructibility.Judah.Judah. Paciﬁc J. Cited on page: 10 . Cited on page: 1 [5] Z. Sur la repr´sentation des fonctions discontinues (2me partie). SpringerVerlag. 131(1988).Addison. Proceedings of the American Mathematical Society. Sur la th´orie des ensembles.Barwise. 46(1959).Mauldin. Cited on page: 77 [7] T. 557561. On covering of real line by null sets.Bartoszynski. Cited on page: 9 [2] J.112. e Acta Mathematica. 113(1991). Journal of Symbolic Logic. 149157.Bartoszynski. Fundamenta Mathematicae.Addison. 946949. Cited on page: 106 [11] M. Journal of Symbolic Logic.Bledsoe. Math. 54(1989). Combinatorial aspects of measure and category. 129(1899). 127(1987). The coﬁnality of cardinal invariants related to measure and category. Comptes Rendus Academie Scie ence Paris. 51(1974. H.. Sets generated by rectangles. 225239. Cited on page: 17 [12] R. Cited on page: 82 [3] R.Bing.Bell. 32(1909). 2736. Fundamenta Mathematicae.H. 114(1981). 123135. 54(1989). On the coﬁnality of the smallest covering of the real line by meager sets.W. Paciﬁc Journal of Mathematics. 828832.W.Baire.
De fractionibus continuis dissertatio. 1991.Brezinski. 23(1980).van Engelen.Euler. A. Cited on page: 121 [21] P. Rigid Borel sets and better quasiorder theory. Cited on page: 17 . Lecture Notes in Mathematics.C. 323326. Cited on page: 3 [14] J.. Imp. K. St. Inﬁnitary languages and descriptive set theory. Budapest. Cited on page: 129 [16] J. Homogeneity for open partitions of pairs of reals.Mauldin. 199222. Cited on page: 3 e [20] F. Cited on page: 112 [22] Q.Burgess. Fundamenta Mathematicae. in Topology. 9(1737). Janos Bolyai.J. Cited on page: 112 [19] L. 69(1978).Miller. Contemporary Mathematics (American Mathematical Society). 353364.Chang. 1 Indiana University Mathematics Journal.Kunen. Two remarks about analytic sets. 78(1980). H. Current research on Q sets.Fleissner.van Engelen.Burgess. 659684. 65(1987). North Holland. Proceedings of the American Mathematical Society. P´tersbourg. Some additive properties of sets of real numbers. 339(1993). Fourth Colloq. Vol. Cited on page: 123 [23] W. Equivalences generated by families of Borel sets. Soc. A. 6872.Miller. On Qsets. A. 413431. K. Transactions of the American Mathematical Society.Burgess. Math.Erdos. Model Theory. PhD University of California. 1401(1989). 98137.Kunen. Comm.Keisler. Proceedings of the American Mathematical Society.Miller. 187199. Berkeley. 280284. SpringerVerlag. R. J. SpringerVerlag. Acad.REFERENCES 143 [13] C. Eﬀective enumeration of classes in a Σ1 equivalence relation.Feng.Steel. 1978). Sc. Cited on page: 129 [17] C. Cited on page: 0 [18] F. 1973. Cited on page: 125 [15] J. 113(1981). 1974. I (Proc. History of Continued Fractions and Pad´ Approxie mants. 28(1979).Fleissner. Cited on page: 17 [24] W. Colloq.
O. Cited on page: 82 [33] L. Topology and Its Applications.Friedman. Bulletin de L’Academie Polonaise des Science.Harrington.Harrington. Fundamenta Mathematicae.H. Squares of Q sets.Fleissner. The consistency of the axiom of choice and of the generalized o continuum hypothesis. Cited on page: 141 [35] L. Logic Colloquium 78. 8 (1960). On the Baire order problem. Cited on page: 129 . Louveau). 556557. Analytic determinacy and 0# . Cited on page: 27 [30] F. 22(1982). Cited on page: 27 [37] L. Journal of Symbolic Logic. Steel Forcing and Barwise compactness.Friedman. PCA wellorderings of the line. 39(1974). Long projective wellorderings. Cited on page: 17 77 [28] H.G¨del. 118(1983). 7980.Fremlin. 145155.Harrington. 17(1984). 685693.Gandy. A powerless proof of a Theorem of Silver. 1984. Proceedings of the National Academy of Science. 43(1978). ed Boﬀa et al. handwritten note undated.REFERENCES 144 [25] W.Harrington. 247263.Fremlin. 24(1938). 223231. Cambridge University Press. Consequences of Martin’s Axiom. projective and beyond. Cited on page: 98 101 [36] L. handwritten note dated 1176. Annals of Mathematical Logic. Cited on page: 97 [29] S.Galvin.82. Proof of Mostowski’s conjecture. Cited on page: 17 [26] D. Note of 29. Cited on page: 51 [27] D.H. Theorem (A. 3146. γsets and other singular sets of real numbers. 124. Cited on page: 17 52 [31] R.4. Cited on page: 109 [32] K.Miller. A. Annals of Mathematical Logic. Equivalence relations. R.Harrington. 12(1977). Cited on page: 113 [34] L.Sami. 571575. circa 1976. North Holland 1979. Journal of Symbolic Logic.
Bulletin of Symbolic Logic. 1978. Academic Press.Bar Hillel. 58(1993).G. 1978.Marker. Two applications of inner model theory to the study of Σ1 2 sets.Judah.Shelah.Hjorth. in Mathematical Logic and Foundations of Set Theory. Cited on page: 118 141 [49] M. ed by Y. 197230.Kondˇ. 111(1991). Cited on page: 141 [44] T. 293302.Shelah.Solovay.Shelah. Cited on page: 17 58 [47] H. 102 (1988). applications in descriptive set theory. Qsets do not necessarily have strong measure zero. Borel orderings. 147152. 11531164.REFERENCES 145 [38] L. Cited on page: 0 16 28 40 128 [45] R. 1982. in Logic Colloquium 80. Cited on page: 129 [42] G. Sur l’uniformisation des compl´mentaires analytiques et les o e ensembles projectifs de la seconde classe. 94107.Kanove˘ Topologies generated by eﬀectively Suslin sets and their ı. North Holland. Cited on page: 90 . Cited on page: 16 [46] H.Hjorth. Journal of Symbolic Logic. S.Judah. 38673873. Japan J.Harrington. Cited on page: 113 129 [39] L. SpringerVerlag. 51(1996). Thin equivalence relations and eﬀective decompositions.Jensen. Some applications of almost disjoint sets. 310(1988). Cited on page: 136 [43] G. Transactions of the American Mathematical Society. R. Qsets. Proceedings n of the American Mathematical Society. S. 84104. Cited on page: 17 [48] V. 1970. D.Harrington and S.Hinman.B.Jech. North Holland. Cited on page: 119 123 129 137 [40] P. Counting equivalence classes for coκSouslin equivalence relations. Set Theory. 124(1996). 2(1996). 15(1939). 821832. and rapid ﬁlters.Hjorth. 385417.M. Sierpi´ski sets. Universal coanalytic sets. Math. Proceedings of the American Mathematical Society. 681683. Russian Mathematical Surveys. S. Proceedings of the American Mathematical Society. Cited on page: 106 [41] G.Shelah. RecursionTheoretic Hierarchies.
Cited on page: 97 [53] A. 1966.S. Cited on page: 0 94 [61] A. 399409.A. Mathematical Proceedings of the Cambridge Philosophical Society. 1995. Inﬁnite games and eﬀective descriptive set theory.Mostowski.Kechris. Academic Press. Cited on page: 104 [56] K. New York. Studies Logic Foundations Math. Kunen. ed by C. North Holland.REFERENCES 146 [50] A. 156. in Generalized recursion theory.Kechris. 1978. North Holland.Kuratowski. Cited on page: 0 133 [55] S.Kuratowski. 1980. 43(1978). Journal of Symbolic Logic.A. Kechris.S. 193213. 1976. On transﬁnite sequences of projective sets with an application to Σ1 equivalence relations. Cited on page: 56 [58] K. vol 1. Cited on page: 0 78 [60] K. Borel and projective sets from the point of view of compact sets. Where MA ﬁrst fails. Classical descriptive set theory. 155160. 1980. Graduate Texts in Mathematics.Martin.. D. 14031416. 57(1992).Miller. North Holland 1978. Topology. 94(1983). Kechris. North Holland. Journal of Symbolic Logic. 63(1955). in Logic Colloquium ’77 ed by 2 Macintyre et al. Set theory. Spector second order classes and reﬂection. 429433. Academic Press. The perfect set theorem and deﬁnable wellorderings of the continuum. in Analytic sets.Kunen. Cited on page: 0 16 19 39 40 128 [57] K. A. Set Theory . Cited on page: 113 [54] A.Rogers. SpringerVerlag. A. II . Cited on page: 77 . Cited on page: 137 [51] A.Landver. Bulletin of the American Mathematical Society. 147183.Kleene. 53(1988). Cited on page: 129 [52] A.Kunen. 94(1978). Journal of Symbolic Logic.Kechris. Cited on page: 77 [59] K. with an introduction to descriptive set theory. 630634. Hierarchies of numbertheoretic predicates. Singular σdense trees.
2 Fundamenta Mathematicae. Cited on page: 97 [72] R. Transactions of the American Mathematical Society.Landver. Cited on page: 35 36 [71] R. Cited on page: 113 141 . 198283). Classifying Borel Structures. 86(1975). Cited on page: 7 e e 9 [64] A. Transactions of the Amer1 ican Mathematical Society. 139216.REFERENCES 147 [62] A. Sur un probl`me de M. Paris VII.Mansﬁeld. 431467. 158(1914). The nonexistence of Σ1 wellordering of the Baire space. Univ. Journal of Symbolic Logic. 103112. 1992. in Set Theory of the Continuum. 283315. Collection: General logic seminar. Baire.Weitkamp.Louveau. Borel classes and closed games: Wadgetype and Hurewicztype results. ed by Judah et al. Recursive Aspects of Descriptive Set Theory.Mansﬁeld.Louveau. Perfect subsets of deﬁnable sets of real numbers. 865874. Comptes Rendus Hebdomade daires Seances Academie Science Paris. Journal de e Math´matiques Pures et Appliqu´s. 304(1987). Uber Teilmengen des Kontinuums von dessen Machtigkeit. 279282. Paciﬁc Journal of Mathematics. Journal of Symbolic Logic. Sur les fonctions repr´sentables analytiquement. G. Cited on page: 123 [69] N. A separation theorem for Σ1 sets. Cited on page: 131 [67] A. Cited on page: 35 ¨ [70] P. 12581261. Cited on page: 123 [68] A. Two results on Borel orders. 113122.. 54(1989). Oxford University Press (1985). 363378. 10861107. Baire numbers.Louveau. 57(1992). uncountable Cohen sets and perfectset forcing. Sitzungsberichte der Sachsischen Akademie der Wissenschaften zu Leipsiz. J. Relations d’equivalence dans les espaces polonais. Louveau.Mansﬁeld. Math. 35(1970). Cited on page: 130 [66] A.Saint Raymond. Publ. 451457. Cited on page: 113 [65] A. (Paris. 1(1905).Luzin. SpringerVerlag. Cited on page: 77 [63] H. 260(1980).Mahlo. Louveau.Lebesgue. Mathematisch Naturwissenschaftliche Klasse 65. Cited on page: 88 [73] R. (1913).
171181.Miller. 41(1989). 27(1979). 1984. 48(1983). Paciﬁc Journal of Mathematics. Cited on page: 17 [84] A. Cited on page: 73 75 76 [80] A. On the length of Borel hierarchies.REFERENCES 148 [74] D. The Baire category theorem and cardinals of countable coﬁnality. Proceedings of the American Mathematical Society. Cited on page: 16 38 95 [75] A. Internal Cohen extensions. 24(1983).Miller.Miller.Miller.Miller. Generic Souslin sets. K. in Handbook of settheoretic topology. 2(1970). 575584. 47(1982). 271(1982). 16(1979). Annals of Pure and Applied Mathematical Logic.Miller. On generating the category algebra and the Baire order problem. Journal of Symbolic Logic. 266(1981). Prikry. R.Miller. 275288. The Borel classiﬁcation of the isomorphism class of a countable model.A.Miller. Annals of Mathematical Logic. Cited on page: 13 19 29 34 36 37 49 50 72 [76] A. Cited on page: 43 [77] A. North Holland. 347348. Transactions of the American Mathematical Society. Corrections and additions to some properties of measure and category. Cited on page: 9 [83] A. Journal of Symbolic Logic. 399407. Cited on page: 58 72 [82] A. Some properties of measure and category.Martin. Paciﬁc Journal of Mathematics. Transactions of the American Mathematical Society. Notre Dame Journal of Formal Logic. A characterization of the least cardinal for which the Baire category theorem fails. Special subsets of the real line.Miller. Annals of Mathematical Logic.Miller. 97(1981). 86(1982). 115(1984). 233267. 179203.Solovay. Bulletin de L’Academie Polonaise des Science. 751755. When the continuum has coﬁnality ω1 . Cited on page: 27 [78] A. Mapping a set of reals onto the reals.Miller. Cited on page: 77 [81] A.M. 143178. 201233. Inﬁnite combinatorics and deﬁnability. Cited on page: 72 [79] A. 2234. Cited on page: 42 [85] A. 498502. 93114. Cited on page: 112 .
5369. SpringerVerlag. 47(1984).Shoenﬁeld. 35(1948). Hebrew University. 1961. Jerusalem. Cited on page: 16 [95] G. North Holland 1980.Moschovakis. Mazurkiewicz.C. Cited on page: 78 [94] F. Israel Journal of Mathematics. Fundamenta Mathematicae. Cited on page: 86 87 . 626628. Annals of Pure and Applied Logic. Proceedings of the American Mathematical Society.Sacks. Cited on page: 17 [93] C.Sami. On some problems of Hausdorﬀ and of Sierpi´ski. Cited on page: 141 [96] R. 2946.Moschovakis.Miller. 12731283. Sur un probl/‘eme de M.Rogers et al. 1 Journal of Symbolic Logic. editors. 415432. 731736. The problem of predicativity. vol 6(1993). Magnes press. Cited on page: 12 17 [88] Y. 1980. Cited on page: 0 78 [90] J. Cited on page: 65 [92] T. Cited on page: 0 68 [91] G.Miller. in Set Theory of the Reals. The existence of Qsets is equivalent to the existence of strong Qsets. SpringerVerlag. 49(1984). Cited on page: 129 [97] S. Analytic Sets. 77(1971). Special sets of reals.Oxtoby. 139153. 132139. 1971. Higher Recursion Theory. Descriptive set theory. Projective subsets of separable metric spaces.Przymusinski. Bulletin of the American Mathematical Society. Cited on page: 129 [98] J. Shelah. 284286. Uniformization in a playful universe. On coκSouslin relations. 79(1980). American Math Society. Funn damenta Mathematicae. Cited on page: 12 [87] A. 50(1990). Essays on the foundations of mathematics.REFERENCES 149 [86] A. On Σ1 equivalence relations with Borel classes of bounded rank. Cited on page: 94 [89] Y.Rothberger.A. Israel Mathematical Conference Proceedings. 1990. Measure and category. Academic Press. 15(1930). ed Haim Judah.Poprougenko.
48(1960). The characteristic function of a sequence of sets and some of its applications. Iterated Cohen extensions and Souslin’s problem. Fundamenta Mathematicae. Counting the number of equivalence classes of Borel and coanalytic equivalence relations. Cited on page: 123 [103] R.Solecki. Boolean Algebras. SpringerVerlag New York. Fundamenta Mathematicae. 31(1938).Stern. Journal of Symbolic Logic. Biulof et al eds. Cited on page: 13 [107] E.Steel.M. Cited on page: 13 [108] J. in Foundations of 2 Mathematics. 128.Szpilrajn.R.Szpilrajn. S. Annals of Mathematical Logic. 10221031.M. Cited on page: 27 [109] J. 15(1978). 3rd edition. On the cardinality of Σ1 sets of reals.REFERENCES 150 [99] W. 26(1936).Sierpi´ski. Covering analytic sets by families of closed sets. Introduction to general topology. Sikorski. 1978. SpringerVerlag (1969). 207223.Silver. Perfect set theorems for analytic and coanalytic equivalence relations. Cited on page: 27 113 [102] S. North Holland. Cited on page: 3 [100] R. Cited on page: 109 [106] E. Cited on page: 129 .Solovay. 313320. Cited on page: 33 35 [101] J.Solovay. 94(1971). Covering analytic sets by families of closed sets. 201245. 302326. 5574. Forcing with tagged trees.Spector. in Logic Colloquium ’77 ed by Macintyre et al. Annals of Mathematics. University of n Toronto Press. 1969. 277284. Annals of Mathematical Logic. Hyperarithmetical quantiﬁers. Cited on page: 88 95 [104] R.Tennenbaum. Cited on page: 40 [105] C.. translated from original Polish version of 1928. Sur l’´quivence des suites d’ensembles et l’´quivence des e e fonctions. 18(1980). 1934. 59(1994). 5873. Fundamenta Mathematicae.
Stern. Cited on page: 124 [112] S. Eﬀective partitions of the real line into Borel sets of bounded rank. 112(1991). 2960. Two examples of Borel partially ordered sets with the countable chain condition. AddisonWesley. 1970. 11251128. Cited on page: 6 . Annals of Mathematical Logic. Proceedings of the American Mathematical Society.Todorcevic. Cited on page: 129 [111] S.REFERENCES 151 [110] J. General Topology. 18 (1980).Williard.
2) ˆ Qα κBorel κSouslin κSouslin µ ◦ ( α<ω1 Pα )∗ Q 2ω Aα A<α F/I Fσ Gδ ILuzin set L∞ (Pα : α < κ) P (T.Index ◦ 152 45 1 53 53 33 5 5 35 26 130 49 130 92 52 92 23 23 49 125 125 113 33 22 2 1 109 82 84 130 137 113 91 107 137 78 137 26 Σ0 (semihyp) α Σ1 equivalence relations 1 Σ1 1 Σ1 (x) 1 Σ1 2 Σ1 = ΣHC 2 1 Σ1 formula αcode αforcing αtopology ∆0 universal set α ∆0 α ∆0 (F ) α B+ Π0 α Π0 (F ) α Π1 1 Π1 2 Σ0 α Σ0 (F ) α Σ1 1 Σ1 equivalence relation 2 x ˇ clα (A) FIN(X. q) Qα S(T. 2) FIN(c+ . 2) FIN(κ. ω) FIN(ℵω . q) ˆ T T T ≤n T ˆ T T 0 T T >0 Tα WF W F<α WO [A]I [T ] [ω]ω [s] ∆1 codes 1 ∆1 wellordering 2 ∆0 formulas Π0 (hyp) α Π0 (semihyp) α Π1 equivalence relations 1 Π1 singleton 1 Π1 Reduction 1 Π1 Reﬂection 1 Π0 1 Π1 on Π1 1 1 Πβ sentence Un ω <ω ωω CK ω1 137 125 78 78 82 84 84 130 23 139 133 4 10 30 4 10 78 98 4 10 78 129 16 139 19 75 40 49 60 102 102 86 65 16 1 1 130 .
Index ord(B) ◦ 153 29 Kleene Separation Theorem 40 Louveau’s Theorem 54 MAκ (ctbl) 43 MansﬁeldSolovay Theorem 60 Martin’s Axiom 6 MartinSolovay Theorem 33 Mostowski’s Absoluteness 1 33 Normal form for Σ1 4 Qsets 45 Sack’s real model 76 Section Problem 1 Shoenﬁeld Absoluteness n 92 Sierpi´ski set 82 Silver forcing 17 SouslinLuzin Separation 10 SpectorGandy Theorem 4 V=L∗ 48 ZFC 38 accumulation points 22 cBa 73 characteristic function 73 code for a hyperarithmetic set 49 collinear points 8 direct sum 43 ﬁeld of sets 33 hereditarily countable sets 1 hereditary order 52 hyperarithmetic subsets 1 nice αtree 119 perfect set forcing 123 perfect set 126 perfect tree 1 prewellorderings 52 prewellordering 113 rank function 39 scale property 84 second countable separable 104 130 73 88 16 38 80 80 17 72 130 87 65 16 102 109 82 81 60 28 13 104 121 45 10 84 59 104 22 72 8 52 91 107 22 90 1 1 P∗ Q P(T ) Pα Qα σδlattice σﬁeld σideal ∼B α<ω1 Pα mP sˆn s⊂t x <c y pp Borel(F ) Borel(X) Borel(X)/meager(X) MAκ OR cov(I) cov(meager(2ω )) meager(X) ord(X) rank(p) stone(B) s Aronszajn tree Baire space Borel metric space BorelDilworth Theorem Boundedness Theorem Cantor space Fusion Gandy forcing Hγ HC .
Index separative splitting node super Luzin set superILuzin switcheroo tree embedding tree two step iteration uniformization property universal for Σ1 sets 1 universal set wellfounded 28 52 49 35 25 92 22 39 90 79 7 22 154 .
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.