Buch

A Course in Model Theory
1
Katrin Tent and Martin Ziegler
Mnster and Freiburg 2010
1
v0.8-53-g4c86431, Thu Nov 18 14:57:21 2010 +0100
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1 The Basics 6
1 Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3 Theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2 Elementary extensions and compactness 25
4 Elementary substructures . . . . . . . . . . . . . . . . . . . . . . 25
5 The Compactness Theorem . . . . . . . . . . . . . . . . . . . . . 28
6 The LwenheimSkolem Theorem . . . . . . . . . . . . . . . . . . 34
3 Quantier elimination 36
7 Preservation theorems . . . . . . . . . . . . . . . . . . . . . . . . 36
8 Quantier elimination . . . . . . . . . . . . . . . . . . . . . . . . 42
9 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4 Countable models 58
10 The Omitting Types Theorem . . . . . . . . . . . . . . . . . . . . 58
11 The space of types . . . . . . . . . . . . . . . . . . . . . . . . . . 60
12
0
categorical theories . . . . . . . . . . . . . . . . . . . . . . . . 64
13 The amalgamation method . . . . . . . . . . . . . . . . . . . . . 69
14 Prime models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
1
CONTENTS 2
5
1
categorical theories 76
15 Indiscernibles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
16 stable Theories . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
17 Prime extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
18 Lachlans theorem . . . . . . . . . . . . . . . . . . . . . . . . . . 88
19 Vaughtian Pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
20 Algebraic formulas . . . . . . . . . . . . . . . . . . . . . . . . . . 94
21 Strongly minimal sets . . . . . . . . . . . . . . . . . . . . . . . . 97
22 The Baldwin-Lachlan theorem . . . . . . . . . . . . . . . . . . . . 103
6 Morley Rank 104
23 Saturated models and the Monster . . . . . . . . . . . . . . . . . 104
24 Morley Rank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
25 Countable models of
1
-categorical theories . . . . . . . . . . . . 117
26 Computation of Morley Rank . . . . . . . . . . . . . . . . . . . . 121
7 Simple theories 125
27 Dividing and Forking . . . . . . . . . . . . . . . . . . . . . . . . . 125
28 Simplicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
29 The Independence Theorem . . . . . . . . . . . . . . . . . . . . . 136
30 Lascar strong types . . . . . . . . . . . . . . . . . . . . . . . . . . 142
31 Example: Pseudo-nite elds . . . . . . . . . . . . . . . . . . . . 145
8 Stable Theories 148
32 Heirs and Coheirs . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
33 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
34 Denable types . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
35 Elimination of imaginaries and T
eq
. . . . . . . . . . . . . . . . . 159
36 Properties of forking in stable theories . . . . . . . . . . . . . . . 166
37 SU-rank and the stability spectrum . . . . . . . . . . . . . . . . . 173
9 Prime extensions 178
CONTENTS 3
38 Indiscernibles in stable theories . . . . . . . . . . . . . . . . . . . 178
39 Totally transcendental theories . . . . . . . . . . . . . . . . . . . 181
40 Countable stable theories . . . . . . . . . . . . . . . . . . . . . . 185
10 The ne structure of
1
categorical theories 187
41 Internal types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
42 Analysable types . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
43 Locally modular strongly minimal sets . . . . . . . . . . . . . . . 195
44 Hrushovskis Examples . . . . . . . . . . . . . . . . . . . . . . . . 199
A Set Theory 208
1 Sets and classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
2 Cardinals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
3 Ordinals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
B Fields 216
4 Ordered elds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
5 Dierential Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
6 Separable and regular eld extensions . . . . . . . . . . . . . . . 224
7 Pseudo-nite elds and pronite groups . . . . . . . . . . . . . . 228
C Combinatorics 232
8 Pregeometries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
9 The ErdsMakkai Theorem . . . . . . . . . . . . . . . . . . . . . 238
10 The ErdsRado Theorem . . . . . . . . . . . . . . . . . . . . . . 239
D Solutions of Exercises 240
Bibliography 258
Index 262
CONTENTS 4
1 Introduction
This book aims to be an introduction to model theory which can be used with-
out any background in logic. We start from scratch, introducing rst-order
logic, structures, languages etc. but move on fairly quickly to the fundamental
results in model theory and stability theory. We also decided to cover simple
theories and Hrushovski constructions, which over the last decade developed
into an important subject. We try to give the necessary background in algebra,
combinatorics and set theory either in the course of the text or in the corre-
sponding section of the appendices. The exercises form an integral part of the
book. Some of them are used later on, others complement the text or present
aspects of the theory that we felt should not be completely omitted. For the
most important exercises (and the more dicult ones) we include (hints for)
solutions at the end of the book.
The book falls into four parts. The rst three chapters introduce the basics
as would be contained in a course giving a general introduction to model theory.
This rst part ends with Chapter 4 which introduces and explores the notion of
a type, the topology on the space of types and a way to make sure that a certain
type will not be realized in a model to be constructed. The chapter ends with
Frasss amalgamation method, a simple but powerful tool for constructing
models.
Chapter 5 is devoted to Morleys famous theorem that a theory with a unique
model in some uncountable cardinality has a unique model in every uncountable
cardinality. To prove this theorem, we describe the analysis of uncountably
categorical theories due to Baldwin and Lachlan in terms of strongly minimal
sets. These are in some sense the easiest examples of stable theories and serve
as an introduction to the topic. This chapter forms a unit with Chapter 6 in
which the Morley rank is studied in a bit more detail.
For the road to more general stable theories we decided to go via simplicity.
The notion of a simple theory was introduced by Shelah in [49]. Such theories
allow for a notion of independence which is presented in Chapter 7. Funda-
mental examples such as pseudo-nite elds make simple theories an important
generalisation of the stable ones. We specialize this notion of independence in
Chapter 8 to characterize forking in stable theories.
In Chapters 9 and 10 we go back to more classical topics of stability theory
such as existence and uniqueness of prime extensions and their analysis in the
uncountably categorical case due to Hrushovski. We end the exposition by
explaining a variant of Hrushovskis construction of a strongly minimal set.
Model theory does not exist independently of set theory or other areas of
mathematics. Many proofs require a knowledge of certain principles of innite
combinatorics which we were hesitant to assume as universally known. Similarly,
to study theories of elds we felt it necessary to explain a certain amount of
CONTENTS 5
algebra. In the three appendices we try to give the background on set theory
and algebra necessary to be able to follow the exposition in the text.
Other books, some emphasising particular aspects of the theory, some gen-
eral introductions that we recommend for further reading include the following
[39],[37], [35], [11], [19],[40], [41], [12], [53].
We would like to thank Manuel Bleichner, Juan-Diego Caycedo, Philipp
Doebler, Antongiulio Fornasiero, Nina Frohn, Zaniar Ghadernezhad, Guntram
Hainke, Immanuel Halupczok Franziska Jahnke, Alexander Kraut, Moritz Mller,
Alexandra Omar Aziz, Lars Scheele and Nina Schwarze for carefully reading ear-
lier versions of the manuscript. We also thank Andreas Baudisch for trying out
the book in a seminar and Bernhard Herwig, who translated early parts of the
lecture notes from which parts of this book evolved.
Chapter 1
The Basics
1 Structures
In this section we start at the very beginning, by introducing the prerequisites
for the objects of study. We deal with rst-order logic and its structures. To
this end we rst introduce the languages. These will be chosen in dierent ways
for the dierent mathematical structures that one wants to study.
Denition 1.1. A language L is a set of constants, function symbols and
relation symbols
1
.
Function symbols and relation symbols have an arity 1. One can think of
constants as 0ary function symbols
2
. This allows us to omit the constant
symbol case in many proofs.
The language per se has no inherent meaning. However, the choice of language
will reect the nature of the intended objects. Here are some standard examples:
L
= The empty language.

L
AG
= 0, +, The language of abelian groups.
L
R
= L
AG
1, The language of rings.
L
G
= e, ,
1
The language of groups.
L
O
= < The language of orders.
L
OR
= L
R
L
O
The language of ordered rings.
L
N
= 0, S, +, , < The language of the natural numbers.
L
sets
= The language of set theory.
The symbols are
1
we also use predicate for relation symbol.
2
By an unfortunate convention 0ary relation symbols are not considered.
6
CHAPTER 1. THE BASICS 7
constants: 0, 1, e
unary function symbols: ,
1
, S
binary function symbols: +, ,
binary relation symbols: <, .
The languages obtain their meaning only when interpreted in an appropriate
structure:
Denition 1.2. Let L be a language. An Lstructure is a pair A =
_
A, (Z
A
)
ZL
_
,
where
A is a nonempty set, the domain or universe of A,
Z
A
A if Z is a constant,
Z
A
: A
n
A if Z is an nary function symbol, and
Z
A
A
n
if Z is an nary relation symbol.
Z
A
is called the Ainterpretation of Z.
The requirement on A to be not empty is merely a (sometimes annoying)
convention. The cardinality of a structure is the cardinality of its universe. We
write [A[ or [A[ for the cardinality of A.
Denition 1.3. Let A and B be Lstructures. A map h: A B is called a
homomorphism if for all a
1
, . . . , a
n
A
h(c
A
) = c
B
h(f
A
(a
1
, . . . , a
n
)) = f
B
(h(a
1
), . . . , h(a
n
))
R
A
(a
1
, . . . , a
n
) R
B
(h(a
1
), . . . , h(a
n
))
for all constants c, n-ary function symbols f and relation symbols R from L.
We denote this by
h: A B.
If in addition h is injective and
R
A
(a
1
, . . . , a
n
) R
B
(h(a
1
), . . . , h(a
n
))
for all a
1
, . . . , a
n
A, then h is called an (isomorphic) embedding. An isomor-
phism is a surjective embedding. We denote isomorphisms by
h : A

B.
If there is an isomorphism between A and B, the two structures are called
isomorphic and we write
A
= B.
It is easy to see that being isomorphic is an equivalence relation between struc-
tures and that bijections can be used to transfer the structures between sets.
Denition 1.4. An automorphism of A is an isomorphism A

A. The set of
automorphisms Aut(A) forms a group under composition.
Denition 1.5. A is called substructure of B if A B and if the inclusion
map is an embedding from A to B. We denote this by
A B.
B is an extension of A if A is a substructure of B.
Remark 1.6. If B is an Lstructure and A a nonempty subset of B, then
A is the universe of a (uniquely determined) substructure A if and only if A
contains all c
B
and A is closed under all operations f
B
. In particular, if L does
not contain any constants or function symbols, then any nonempty subset of
an Lstructure is again an Lstructure. Also, if h: A B, then h(A) is the
universe of a substructure of B.
It is also clear that for any family A
i
of substructures of B, the intersection
of the A
i
is empty or a substructure of B. Therefore, if S is any nonempty
subset of B, then there exists a smallest substructure A = S
B
which contains
S. We call A the substructure generated by S. If S is nite, then A is said to
be nitely generated.
If L contains a constant, then the intersection of all substructures of B is
not empty as it contains the Binterpretation of this constant. Thus B has a
smallest substructure
B
. If L has no constants, we set
B
=
Lemma 1.7. If A is generated by S, then every homomorphism h: A B is
determined by its values on S.
Proof. If h
: A B is another homomorphism, then C = b [ h(b) = h
(b) is
empty or a substructure. If h and h
coincide on S, then S is a subset of C, and

therefore C = A.
Lemma 1.8. Let h : A

A
be an isomorphism and Ban extension of A. Then

there exists an extension B
of A
and an isomorphism g : B

B
extending
h.
Proof. First extend the bijection h: A A
to a bijection g : B B
and use
g to dene an Lstructure on B
.
Denition 1.9. Let (I, ) be a directed partial order. This means that for all
i, j I there exists a k I such that i k and j k. A family (A
i
)
iI
of
Lstructures is called directed if
i j A
i
A
j
.
If I is linearly ordered, we call (A
i
)
iI
a chain. If, for example, a structure
A
1
is isomorphic to a substructure A
0
of itself,
h
0
: A
0
A
1
,
then Lemma 1.8 gives an extension
h
1
: A
1
A
2
.
Continuing in this way, we obtain a chain A
0
A
1
A
2
. . . and an increasing
sequence h
i
: A
i
A
i+1
of isomorphisms.
Lemma 1.10. Let (A
i
)
iI
be a directed family of Lstructures. Then A =
iI
A
i
is the universe of a (uniquely determined) Lstructure
A =
_
iI
A
i
,
which is an extension of all A
i
.
Proof. Let R be an nary relation symbol and a
1
, . . . , a
n
A. As I is directed,
there exists k I such that all a
i
are in A
k
. We dene (and this is the only
possibility)
R
A
(a
1
, . . . , a
n
) R
A
k
(a
1
, . . . , a
n
).
Constants and function symbols are treated similarly.
A subset K of L is called a sublanguage. An Lstructure becomes a K
structure if we forget the interpretations of the symbols from L K:
A K =
_
A, (Z
A
)
ZK
_
We call this process restriction. Conversely we call A an expansion of A K.
Here are some examples:
Let A be an Lstructure.
a) Let R be an nary relation on A. We introduce a new relation symbol R
and we denote by
(A, R)
the expansion of A to an L Rstructure in which R is interpreted by R.
b) For given elements a
1
, . . . , a
n
we may introduce new constants a
1
, . . . , a
n
and
consider the L a
1
, . . . , a
n
structure
(A, a
1
, . . . , a
n
) .
c) Let B be a subset of A. By considering every element of B as a new constant,
we obtain the new language
L(B) = L B
and the L(B)structure
A
B
= (A, b)
bB
.
Note that Aut(A
B
) is the group of automorphisms of A xing B elementwise.
We denote this group by Aut(A/B).
Similarly, if C is a set of new constants, we write L(C) for the language
L C.
Manysorted structures
Without much eort, the concepts introduced here can be extended to many
sorted languages and structures, which we shall need to consider later on.
Let S be a set, which we call the set of sorts. An S-sorted language L is given
by a set of constants for each sort in S, and typed function and relation symbols
which carry the information about their arity and the sorts of their domain and
range. More precisely, for any tuple (s
1
, . . . s
n
) and (s
1
, . . . , s
n
, t) there is a set
of relation symbols and function symbols, respectively. An S-sorted structure
is a pair A =
_
A, (Z
A
)
ZL
_
, where
A is a family (A
s
)
sS
of nonempty sets.
Z
A
A
s
if Z is a constant of sort s S,
Z
A
: A
s1
. . . A
sn
A
t
if Z is a function symbol of type (s
1
, . . . , s
n
, t),
Z
A
A
s1
. . . A
sn
if Z is a relation symbol of type (s
1
, . . . , s
n
).
It should be clear how to dene homomorphisms between manysorted struc-
tures A and B: they are given by maps taking A
s
to B
s
for s S and behaving
as before with respect to constants, function and relation symbols.
Example:
Consider the two-sorted language L
PG
for permutation groups with a sort x for
the set and a sort g for the group. The constants and function symbols for L
PG
are those of L
G
restricted to the sort g and an additional function symbol
of type (x, g, x). Thus, an L
PG
-structure (X, G) is given by a set X and an
L
G
-structure G together with a function X G X.
Exercise 1.1 (Direct products).
Let A
1
, A
2
be Lstructures. Dene an Lstructure A
1
A
2
with universe A
1
A
2
such that the natural epimorphisms
i
: A
1
A
2
A
i
for i = 1, 2 satisfy the
following universal property: Given any Lstructure D and homomorphisms
i
: D A
i
, i = 1, 2 there is a unique homomorphism : D A
1
A
2
such
that
i
=
i
, i = 1, 2, i.e., this is the product in the category of Lstructures
with homomorphisms.
Exercise 1.2 (Afshordel).
Let f : A A be an embedding. Then there is an extension A B and an
extension of f to an automorphism g of B. We can nd B as the union of the
chain A g
1
(A) g
2
(A) . The pair (B, g) is uniquely determined by
that property.
2 Language
Starting from the inventory of the languages dened in Section 1 we now describe
the grammar which allows us to build well-formed terms and formulas which
will again be interpreted in the according structures.
Denition 2.1. Lterms are words (sequences of symbols) built from con-
stants, the function symbols of L and the variables v
0
, v
1
, . . . according to the
following rules:
1. Every variable v
i
and every constant c is an Lterm.
2. If f is an nary function symbol and t
1
, . . . , t
n
are Lterms, then also
ft
1
. . . t
n
is an Lterm.
The number of function symbols occurring in a term is called its complexity.
This will be used in induction arguments.
We often write f(t
1
, . . . , t
n
) instead of ft
1
. . . t
n
for better readability and
use the usual conventions for some particular function symbols. For example
(x +y) (z +w)
stands for
+xy +zw
and (x y)
1
for
1
xy.
Let A be an Lstructure and

b = (b
0
, b
1
, . . .) a sequence of elements which
we consider as assignments to the variables v
0
, v
1
, . . .. If we replace in t each
variable v
i
by a
i
, the term t determines an element t
A
[
b] of A in an obvious way:
Denition 2.2. For an Lterm t, an Lstructure A and an assignment

b we
dene the interpretation t
A
[
b] by
v
A
i
[
b] = b
i
c
A
[
b] = c
A
ft
1
. . . t
A
n
[
b] = f
A
(t
A
1
[
b], . . . , t
A
n
[
b])
This (recursive) denition is possible because every term has a unique decom-
position into its constituents: if ft
1
. . . t
n
= ft
1
. . . t
n
, then t
1
= t
1
, . . . , t
n
= t
n
.
This as well as the following lemma are easy to prove using induction on the
complexity of the terms involved.
Lemma 2.3. The interpretation t
A
[
b] depends on b
i
only if v
i
occurs in t.
If the variables x
1
, . . . , x
n
are pairwise distinct
3
and if no other variables
occur in t, we write
t = t(x
1
, . . . , x
n
).
According to the previous lemma, if

b is an assignment for the variables which
assigns a
i
to x
i
, we can dene
t
A
[a
1
, . . . , a
n
] = t
A
[
b].
If t
1
, . . . , t
n
are terms, we can substitute t
1
, . . . , t
n
for the variables x
1
, . . . , x
n
.
The resulting term is denoted by
t(t
1
, . . . , t
n
).
One easily proves:
Lemma 2.4 (Substitution Lemma).
t(t
1
, . . . , t
n
)
A
[
b] = t
A
_
t
A
1
[
b], . . . , t
A
n
[
b]
_
If we expand A to the L(A)structure A
A
, we get as a special case
t(a
1
, . . . , a
n
)
AA
= t
A
[a
1
, . . . , a
n
].
Lemma 2.5. Let h: A B be a homomorphism and t(x
1
, . . . , x
n
) a term.
Then we have for all a
1
, . . . , a
n
from A
t
B
[h(a
1
), . . . , h(a
n
)] = h
_
t
A
[a
1
, . . . , a
n
]
_
Proof. Induction on the complexity of t.
Lemma 2.6. Let S be a subset of the Lstructure A. Then
S
A
=
_
t
A
[s
1
, . . . , s
n
] [ t(x
1
, . . . , x
n
) Lterm, s
1
, . . . , s
n
S
_
.
Proof. We may assume that S is not empty or that L contains a constant since
otherwise both sides of the equation are empty. It follows from Lemma 2.5
that the universe of a substructure is closed under interpretations of terms
t
A
[, . . . , ]. Thus the right hand side is contained in S
A
. For the converse
we have to show that the right hand side is closed under the operations f
A
. The
assertion now follows using Remark 1.6.
3
Remember that x
i
{v
0
, v
1
, . . .}
A constant term is a term without variables. As a special case of Lemma 2.6
we thus have
A
=
_
t
A
[ t constant Lterm
_
.
The previous lemma implies:
Corollary 2.7.
[S
A
[ max[S[, [L[,
0
Proof. There are at most max[L[,

0
many Lterms and for every term t at
most max[S[,
0
many assignments of elements of S to the variables of t.
We still need to dene Lformulas . These are sequences of symbols which
are built from the symbols of L, the parentheses ( and ) as auxiliary symbols
and the following logical symbols:
variables v
0
, v
1
, . . .
equality symbol
.
=
negation symbol
conjunction symbol
existential quantier
Denition 2.8. Lformulas are
1. t
1
.
= t
2
where t
1
, t
2
are Lterms,
2. Rt
1
. . . , t
n
where R is an nary relation symbol from L and
t
1
, . . . , t
n
are Lterms,
3. where is an Lformula,
4. (
1

2
) where
1
and
2
are Lformulas,
5. x where is an Lformula and x a variable.
Formulas of the form t
1
.
= t
2
or Rt
1
. . . , t
n
are called atomic.
As with terms, we dene the complexity of a formula as the number of
occurrences of , and . This allows us to do induction on (the complexity
of) formulas.
We use the following abbreviations:
(
1

2
) = (
1

2
)
(
1

2
) = (
1

2
)
(
1

2
) = ((
1

2
) (
2

1
))
x = x
for disjunction, implication, equivalence and universal quantier.
Sometimes we write t
1
Rt
2
for Rt
1
t
2
, x
1
. . . x
n
for x
1
. . . x
n
and x
1
. . . x
n
for x
1
. . . x
n
. To increase readability we might use superuous parentheses.
On the other hand we might omit parentheses with an implicit understanding
of the binding strength of logical symbols: , , and bind more strongly
than which in turn binds more strongly than . Finally and have the
least binding strength. For example
1

2

3
is understood to mean
((
1

2
)
3
).
Given an Lstructure A and an Lformula (x
1
, . . . x
n
) it should now be
clear what it means for to hold for
b. Here is the formal denition:

Denition 2.9. Let A be an Lstructure. For Lformulas and all assignments
b we dene the relation

A [= [
b]
recursively over :
A [= t
1
.
= t
2
[
b] t
A
1
[
b] = t
A
2
[
b]
A [= Rt
1
. . . t
n
[
b] R
A
_
t
A
1
[
b], . . . , t
A
n
[
b]
_
A [= [
b] A ,[= [
b]
A [= (
1

2
) [
b] A [=
1
[
b] and A [=
2
[
b]
A [= x [
b] there exists a A such that A [=

_
b
a
x
_
Here we use the notation
b
a
x
= (b
0
, . . . , b
i1
, a, b
i+1,...
) if x = v
i
.
If A [= [
b] holds we say holds in A for

b or
b satises (in A).

For this denition to work one has to check that every formula has a unique
decomposition
4
into subformulas: if Rt
1
, . . . , t
n
= Rt
1
, . . . , t
n
, then t
1
= t
1
, . . . , t
n
=
t
n
; and (
1

2
) = (
2
) implies
1
=
1
and
2
=
2
.
It should be clear that our abbreviations have the intended meaning, e.g.
A [= (
1

2
)[
b] if and only if (A [=
1
[
b] implies A [=
2
[
b]).
Whether holds in A for
b depends only on the free variables of :

Denition 2.10. The variable x occurs freely in the formula if it occurs at a
place, which is not within the scope of a quantier x. Otherwise its occurrence
4
It is precisely for this uniqueness that brackets were introduced when dening formulas.
is called bound. Here is the formal denition (recursive in ):
x free in t
1
.
= t
2
x occurs in t
1
or in t
2
.
x free in Rt
1
. . . t
n
x occurs in one of the t
i
.
x free in x free in
x free in (
1

2
) x free in
1
or x free in
2
x free in y x ,= y and x free in
For example the variable v
0
does not occur freely in v
0
(v
1
R(v
0
, v
1
)
P(v
1
)); v
1
occurs both freely and bound
5
.
Lemma 2.11. Suppose
b and c agree on all variables which are free in . Then

A [= [
b] A [= [c].
Proof. By induction on the complexity of .
If we write a formula in the form (x
1
, . . . , x
n
), we mean:
the x
i
are pairwise distinct,
all free variables in are among x
1
, . . . , x
n
.
If furthermore a
1
, . . . , a
n
are elements of the structure A, we dene
A [= [a
1
, . . . , a
n
]
by A [= [
b], where

b is an assignment satisfying

b(x
i
) = a
i
. Because of
Lemma 2.11 this is well dened.
Thus (x
1
, . . . , x
n
) denes an nary relation
(A) = a [ A [= [a]
on A, the realisation set of . Such realisation sets are called 0denable
subsets of A
n
, or 0denable relations.
Let B be a subset of A. A Bdenable subset of A is a set of the form (A)
for an L(B)formula (x). We also say that (and (A)) are dened over B and
that the set (A) is dened by . Often we dont explicitly specify a parameter
set B and just talk about denable subsets. 0denable sets are denable over
the empty set. We call two formulas equivalent if in every structure they dene
the same set.
5
However, we usually make sure that no variable occurs both freely and bound. This can
be done by renaming the free occurrence by an unused variable.
Denition 2.12. A formula without free variables is called a sentence. We
write A [= if A [= [
b] for some (all)
b.
In that case A is called a model of . We also say holds in A. If is a set
of sentences, then A is a model of if all sentences of hold in A. We denote
this by
A [= .
Let = (x
1
, . . . , x
n
) and let t
1
, . . . , t
n
be terms. The formula
(t
1
, . . . , t
n
)
is the formula we obtain by rst renaming all bound variables by variables which
do not occur in the t
i
and then replacing every free occurrence of x
i
by t
i
.
Lemma 2.13 (Substitution lemma).
A [= (t
1
, . . . , t
n
)[
b] A [=
_
t
A
1
[
b], . . . , t
A
n
[
b]
Proof. The proof is an easy induction on .

Also note this (trivial) special case:
A
A
[= (a
1
, . . . , a
n
) A [= [a
1
, . . . , a
n
].
Frequently we suppress the subscript and simply write
A [= (a
1
, . . . , a
n
).
Atomic formulas and their negation are called basic. Formulas without quan-
tiers (or: quantier free formulas) are Boolean combinations of basic formulas,
i.e. they can be built from basic formulas by successively applying and . The
conjunction of formulas
i
is denoted by
_
i<m
i
and
_
i<m
i
denotes their dis-
junction. By convention
_
i<1
i
=
_
i<1
i
=
0
. It is convenient to allow the
empty conjunction and the empty disjunction. For that we introduce two new
formulas: the formula , which is always true, and the formula , which is
always false. We dene
i<0
i
=
i<0
i
=
A formula is in negation normal form if it is built from basic formulas using
, , , .
Lemma 2.14. Every formula can be transformed into an equivalent formula
which is in negation normal form.
Proof. Let denote equivalence of formulas. We consider formulas which are
built using , , , and and move the negation symbols in front of atomic
formulas using
( ) ( )
( ) ( )
x x
x x

Denition 2.15. A formula in negation normal form which does not contain
any existential quantier is called universal. Formulas in negation normal form
without universal quantiers are called existential.
Clearly an isomorphism h: A B preserves the validity of every formula:
A [= [a
1
, . . . , a
n
] B[= [h(a
1
), . . . , h(a
n
)].
Embeddings preserve the validity of existential formulas:
Lemma 2.16. Let h: A Bbe an embedding. Then for all existential formulas
(x
1
, . . . , x
n
) and all a
1
, . . . , a
n
A we have
A [= [a
1
, . . . , a
n
] =B[= [h(a
1
), . . . , h(a
n
)].
For universal , the dual holds:
B[= [h(a
1
), . . . , h(a
n
)] = A [= [a
1
, . . . , a
n
].
Proof. By an easy induction on : for basic formulas the assertion follows from
the denition of an embedding and Lemma 2.5. The inductive step is trivial for
the cases and . Let nally (x) be y (x, y). If A [= [a], there exists an
a A such that A [= [a, a]. By induction we have B [= [h(a), h(a)]. Thus
B[= [h(a)].
Let A be an Lstructure. The atomic diagram of A is
Diag(A) =
_
basic L(A)sentence
A
A
[=
_
,
the set of all basic sentences with parameters from A which hold in A.
Lemma 2.17. The models of Diag(A) are precisely the structures (B, h(a))
aA
for embeddings h: A B.
Proof. The structures (B, h(a))
aA
are models of the atomic diagram by Lemma 2.16.
For the converse note that a map h is an embedding if and only if it preserves
the validity of all formulas of the form
() x
1
.
= x
2
c
.
= x
1
f(x
1
, . . . , x
n
)
.
= x
0
() R(x
1
, . . . , x
n
).
Manysorted languages
In a manysorted language with sorts in S, terms and formulas are built
with respect to the sorts. For each sort s S we have variables v
s
0
, v
s
1
, . . . from
which we build the following terms of sort s.
Every variable v
s
i
is an Lterm of sort s.
Every constant c of sort s is an Lterm of sort s.
If f is a function symbol of type (s
1
, . . . , s
n
, s) and t
i
is an Lterm of sort
s
i
for i = 1, . . . n, then ft
1
. . . t
n
is an Lterm of sort s.
L-formulas are constructed as before with the following adjustments:
t
1
.
= t
2
where t
1
, t
2
are Lterms of the same sort,
Rt
1
. . . , t
n
where R is a relation symbol from L of type (s
1
, . . . , s
n
)
and t
i
is an Lterm of sort s
i
,
x where is an Lformula and x a variable (of some sort
s).
It should be clear how to extend the denitions of this section to the many
sorted situation and that the results presented here continue to hold without
change. In the sequel we will not deal separately with manysorted languages
until we meet them again in Section 35.
Exercise 2.1.
Let L be a language and P be a new nary relation symbol. Let = (P)
be an L(P) = L Psentence and (x
1
, . . . , x
n
) an Lformula. Now replace
every occurrence of P in by . More precisely, every subformula of the form
Pt
1
. . . t
n
is replaced by (t
1
. . . t
n
). We denote the resulting Lformula by ().
Show
A [= () if and only if(A, (A)) [= (P).
Exercise 2.2.
Every quantier free formula is equivalent to a formula of the form
i<m
j<mi
ij
and to a formula of the form

i<m
j<mi
ij
where the
ij
are basic formulas. The rst form is called conjunctive normal
form, the second disjunctive normal form.
Exercise 2.3.
Every formula is equivalent to a formula in prenex normal form:
Q
1
x
1
. . . Q
n
x
n
The Q
i
are quantiers ( or ) and is quantier free.
Exercise 2.4 (Ultraproducts and oss Theorem).
A lter on a set I is a nonempty set T P(1) which does not contain the
empty set and is closed under intersections and supersets, i.e. for A, B T, we
have A B T and if A T and A C I we have C T. A lter T is
called ultralter if for every A P we have A T or I A T. (By Zorns
Lemma, any lter can be extended to an ultralter.)
For a family (A
i
: i I) of Lstructures and T an ultralter on I we dene
the ultraproduct
iI
A
i
/T as follows. On the cartesian product
iI
A
i
, the
ultralter T denes an equivalence relation
F
by
(a
i
)
iI

F
(b
i
)
iI
i I : a
i
= b
i
T
On the set of equivalence classes (a
i
)
F
we dene an Lstructure
iI
A
i
/T.
For constants c L, put c
FAi
= (c
Ai
)
F
.
For n-ary function symbols f L put
f
F
Ai
((a
1
i
)
F
, . . . , (a
n
i
)
F
)) = (f
Ai
(a
1
i
, . . . , a
n
i
))
F
.
For n-ary relation symbols R L put
R
FAi
((a
1
i
)
F
, . . . , (a
n
i
)
F
)) i I : R
Ai
(a
1
i
, . . . , a
n
i
) T.
1. Show that the ultraproduct
iI
A
i
/T is well-dened.
2. Prove oss Theorem: For any Lformula we have
iI
A
i
/T [= ((a
1
i
)
F
, . . . , (a
n
i
)
F
) i I : A
i
[= (a
1
i
, . . . , a
n
i
) T.
3 Theories
Having dened a language, we can now take a closer look at which formulas
hold in a given structure. Conversely, we can start with a set of sentences and
consider those structures in which they hold. In this way, these sentences serve
as a set of axioms for a theory.
Denition 3.1. An Ltheory T is a set of Lsentences.
A theory which has a model is a consistent theory. More generally, we call
a set of L-formulas consistent if there is an Lstructure A and an assignment
b such that A [= (
b) for all . is consistent with T if T is consistent.

Lemma 3.2. Let T be an Ltheory and L
be an extension of L. Then T is
consistent as an Ltheory if and only if T is consistent as a L
theory.
Proof. This follows from the (trivial) fact, that every Lstructure is expandable
to an L
structure.
Examples:
To keep algebraic expressions readable we will write 0 and 1 for the symbols 0
and 1 in the following examples. We will omit the point for the multiplication
and brackets if they are implied by the order of operations rule.
T
AG
, the theory of abelian groups, has the axioms
x, y, z (x +y) +z
.
= x + (y +z)
x 0 +x
.
= x
x (x) +x
.
= 0
x, y x +y
.
= y +x
T
R
, the theory of commutative rings:
T
AG
x, y, z (xy)z
.
= x(yz)
x 1x
.
= x
x, y xy
.
= yx
x, y, z x(y +z)
.
= xy +xz
T
F
, the theory of elds:
T
R
0
.
= 1
x (x
.
= 0 y xy
.
= 1)
Denition 3.3. If a sentence holds in all models of T, we say that follows
from T (or that T proves ) and write
6
T .
By Lemma 3.2 this relation is independent of the language. Sentences which
follow from the empty theory are called valid.
The most important properties of are:
Lemma 3.4.
1. If T and T ( ), then T .
2. If T (c
1
, . . . , c
n
) and the constants c
1
, . . . , c
n
occur neither in T nor in
(x
1
, . . . , x
n
), then T x
1
. . . x
n
(x
1
, . . . , x
n
).
Proof. We prove 2: Let L
= L c
1
, . . . , c
n
. If the L
structure A is a model
of T and a
1
, . . . , a
n
are arbitrary elements, then (A, a
1
, . . . , a
n
) [= (c
1
, . . . , c
n
).
That means A [= x
1
. . . x
n
(x
1
, . . . , x
n
). Thus T x
1
. . . x
n
(x
1
, . . . , x
n
).
We generalise this relation to theories S: we write T S if all models of T
are models of S. S and T are called equivalent, S T, if S and T have the
same models.
Denition 3.5. A consistent Ltheory T is called complete if for all Lsentences
T or T .
This notion clearly depends on L. If T is complete and L
is an extension of
L, T will in general not be complete as a L
theory.
Denition 3.6. For a complete theory T we dene
[T[ = max([L[,
0
).
The typical (and, as we will see below, only) example of a complete theory
is the theory of a structure A
Th(A) = [ A [= .
Lemma 3.7. A consistent theory T is complete if and only if Ded(T) = [
T is maximal consistent, or equivalently, if T S for every consistent
extension S of T.
6
Note that sometimes this relation is denoted T |= to distinguish this notion from the
more syntactic notion of logical inference, see [50, section 2.6]
Proof. We call independent from T if neither nor follows from T. So
is independent from T exactly if T is a proper (in the sense of ) consistent
extension of T. From this the lemma follows directly.
Denition 3.8. Two Lstructures A and B are called elementarily equivalent
,
A B,
if they have the same theory, that is, if for all Lsentences
A [= B[= .
Isomorphic structures are always elementarily equivalent. The converse
holds only for nite structures, see Exercise 3.3 and Theorem 6.1.
Lemma 3.9. Let T be a consistent theory. Then the following are equivalent
a) T is complete.
b) All models of T are elementarily equivalent.
c) There exists a structure A with T Th(A).
Proof. a c: Let A be a model of T. If holds in A, then T , and thus
T . So Ded(T) = Th(A) holds.
c b: If B [= T, B [= Th(A) and therefore B A. Note that is an
equivalence relation.
b a: Let A be a model of T. If holds in A, then holds in all models of
T, i.e. T . Otherwise, holds in A and we have T .
For an L-theory T, we let Mod(T) denote the class of all models of T. A class
of Lstructures forms an elementary class if it is the class of models of some
L-theory T. By oss Theorem (Exercise 2.4), elementary classes are closed
under ultraproducts (see Exercise 4.2).
By the previous examples, the class of all abelian groups (commutative rings,
elds, respectively) is an elementary class as is the subclass of elementary abelian
p-groups for some prime p. However, the class of all nite abelian p-groups does
not form an elementary class (see Exercise 3.4).
Exercise 3.1.
Write down the axioms for the theory T
DLO
of dense linear orders without
endpoint in the language L
O
of orders and the axioms for the theory T
ACF
of
algebraically closed elds in L
R
. Is T
ACF
complete?
Exercise 3.2.
1. For a prime number p, let Z
p
denote the p-Prfer group, i.e. the group
of all p
k
-th roots of unity for all k N. Show that the groups Z
m
p
and
Z
n
p
are not elementarily equivalent for m ,= n.
2. Show that Z
n
, Z
m
in the language of groups if n ,= m.
Exercise 3.3.
Show that if A is a nite Lstructure and B is elementarily equivalent to A,
then they are isomorphic. (Show this rst for nite L.)
Exercise 3.4.
Show that the class of all nite groups (all torsion groups, all nilpotent groups,
respectively) does not form an elementary class.
Chapter 2
Elementary extensions and
compactness
4 Elementary substructures
As in other elds of mathematics, we need to compare structures and consider
maps from one structure to another. For this to make sense we consider a xed
language L. Maps and extensions are then required to respect this language:
Let A and B be two Lstructures. A map h: A B is called elementary
if it preserves the validity of arbitrary formulas (x
1
, . . . , x
n
)
1
. More precisely,
for all a
1
, . . . , a
n
A we have:
A [= [a
1
, . . . , a
n
] B[= [h(a
1
), . . . , h(a
n
)].
In particular, h preserves quantier free formulas and is therefore an embedding.
Hence h is also called elementary embedding. We write
h : A

B
The following lemma is clear.
Lemma 4.1. The models of Th(A
A
) are exactly the structures of the form
(B, h(a))
aA
for elementary embeddings h : A

B.
Th(A
A
) is called the elementary diagram of A.
A substructure A of Bis called elementary if the inclusion map is elementary,
i.e. if
A [= [a
1
, . . . , a
n
] B[= [a
1
, . . . , a
n
]
1
This only means that formulas which hold in A also hold in B. But taking negations, the
converse follows.
25
CHAPTER 2. ELEMENTARY EXTENSIONS AND COMPACTNESS 26
for all a
1
, . . . , a
n
A. In this case we write
A B
and B is called an elementary extension of A.
Theorem 4.2 (Tarskis Test). Let B be an Lstructure and A a subset of B. A
is the universe of an elementary substructure if and only if every L(A)formula
(x) which is satisable in B can be satised by an element of A.
Proof. If A B and B [= x(x), we also have A [= x(x) and there exists
a A such that A [= (a). Thus B[= (a).
Conversely, suppose that the condition of Tarskis test is satised. First we
show that A is the universe of a substructure A. The L(A)formula x
.
= x is
satisable in B, so A is not empty. If f L is an nary function symbol (n 0)
and a
1
, . . . , a
n
from A, we consider the formula
(x) = f(a
1
, . . . , a
n
)
.
= x.
Since (x) is always satised by an element of A, A is closed under f
B
.
Now we show, by induction on , that
A [= B[=
for all L(A)sentences . This is clear for atomic sentences. The induction
steps for = and = (
1

2
) are trivial.
It remains to consider the case = x(x). If holds in A, there exists a A
such that A [= (a). The inductive hypothesis yields B [= (a), thus B [= .
For the converse suppose holds in B. Then (x) is satisable in B and by
Tarskis test we nd a A such that B [= (a). By induction A [= (a) and
A [= holds.
We use Tarskis Test to construct small elementary substructures:
Corollary 4.3. Suppose S is a subset of the Lstructure B. Then B has an
elementary substructure A containing S and of cardinality at most
max[S[, [L[,
0
.
Proof. We construct A as the union of an ascending sequence S
0
S
1
. . .
of subsets of B. We start with S
0
= S. If S
i
is already dened, we choose
an element a
B for every L(S

i
)formula (x) which is satisable in B and
dene S
i+1
to be S
i
together with these a
. It is clear that A is the universe of

an elementary substructure. It remains to prove the bound on the cardinality
of A.
An Lformula is a nite sequence of symbols from L, auxiliary symbols and
logical symbols. These are [L[ +
0
= max([L[,
0
) many symbols and therefore
there are exactly max([L[,
0
) many Lformulas (see Appendix A.2.6).
Let = max([S[, [L[,
0
). There are many L(S)formulas, thus [S
1
[ .
Inductively it follows for every i that [S
i
[ . Finally we have [A[
0
=
.
A directed family (A
i
)
iI
of structures is elementary if A
i
A
j
for all
i j. The following lemma is called Chain Lemma as it is mainly applied to
elementary chains.
Theorem 4.4 (Tarskis Chain Lemma). The union of an elementary directed
family is an elementary extension of all its members.
Proof. Let A =
iI
(A
i
)
iI
. We prove by induction over (x) that for all i
and a A
i
A
i
[= (a) A [= (a).
If is atomic, nothing is to be proved. If is a negation or a conjunction, the
claim follows directly from the inductive hypothesis.
If (x) = y(x, y), (a) holds in A exactly if there exists b A with A [=
(a, b). As the family is directed, there always exists a j i with b A
j
. By
the inductive hypothesis we have A [= (a, b) A
j
[= (a, b). Thus (a)
holds in A exactly if it holds in an A
j
(j i). Now the claim follows from
A
i
A
j
.
Exercise 4.1.
Let A be an Lstructure and ES(A) = B: B _ A. Show that if (A
i
)
iI
is a
chain of elements in ES(A) and B their union, then B ES(A).
Exercise 4.2.
Show that a class c of Lstructure is an elementary class if and only if c is
closed under ultraproducts and elementary equivalence.
5 The Compactness Theorem
In this section we prove the Compactness Theorem, one of the fundamental
results in rst-order logic. It states that a rst-order theory has a model if
every nite part of it does. Its name is motivated by the results in Section 11
which associate to each theory a certain compact topological space.
We call a theory T nitely satisable if every nite subset of T is consistent.
Theorem 5.1 (Compactness Theorem). Finitely satisable theories are consis-
tent.
Let L be a language and C a set of new constants. An L(C)theory T
is
called a Henkintheory if for every L(C)formula (x) there is a constant c C
such that
x(x) (c) T
.
The elements of C are called Henkinconstants of T
.
Let us call an Ltheory T nitely complete if it is nitely satisable and if
every Lsentence satises T or T. This terminology is only prelim-
inary (and not standard): by the compactness theorem a theory is equivalent
to a nitely complete one if and only if it is complete.
The compactness theorem follows from the following two lemmas.
Lemma 5.2. Every nitely satisable Ltheory T can be extended to a nitely
complete Henkintheory T
.
Note that conversely the lemma follows directly from the compactness theo-
rem: choose a model A of T. Then Th(A
A
) is a nitely complete Henkintheory
with A as a set of Henkinconstants.
Proof. We dene an increasing sequence = C
0
C
1
. . . of new constants by
assigning to every L(C
i
)formula (x) a constant c
(x)
and
C
i+1
=
_
c
(x)
(x) L(C
i
)formula
_
.
Let C be the union of the C
i
and T
H
the set of all Henkinaxioms
x(x) (c
(x)
)
for L(C)formulas (x). It is easy to see that one can expand every Lstructure
to a model of T
H
. Hence T T
H
is a nitely satisable Henkintheory. Using
that the union of a chain of nitely satisable theories is again nitely satisable,
we can apply Zorns Lemma and get a maximal nitely satisable L(C)theory
T
which contains TT
H
. As in Lemma 3.7 we show that T
is nitely complete:
if neither nor belongs to T
, neither T
nor T
would be
nitely satisable. Hence there would be a nite subset of T
which would
be consistent neither with nor with . Then itself would be inconsistent
and T
would not be nitely satisable. This proves the lemma.

Lemma 5.3. Every nitely complete Henkintheory T
has a model A (unique

up to isomorphism) consisting of constants; i.e.,
(A, a
c
)
cC
[= T
with A = a
c
[ c C.
Proof. Let us rst note that since T
is nitely complete, every sentence which

follows from a nite subset of T
belongs to T
. Otherwise the negation of that

sentence would belong to T
, but would also be inconsistent together with a

nite part of T
.
We dene for c, d C
c d c
.
= d T
As c
.
= c is valid, and d
.
= c follows from c
.
= d, and c
.
= e follows from c
.
= d
and d
.
= e, is an equivalence relation. We denote the equivalence class of c by
a
c
and set
A = a
c
[ c C.
We expand A to an Lstructure A by dening
R
A
(a
c1
, . . . , a
cn
) R(c
1
, . . . , c
n
) T
(1)
f
A
(a
c1
, . . . , a
cn
) = a
c0
f(c
1
, . . . , c
n
)
.
= c
0
T
(2)
for relation symbols R and function symbols f (n 0ary) from L.
We have to show that this is well-dened: For (1) we have to show that
a
c1
= a
d1
, . . . , a
cn
= a
dn
, R(c
1
, . . . , c
n
) T
implies R(d
1
, . . . , d
n
) T
. But clearly R(d

1
, . . . , d
n
) holds in any structure
satisfying
c
1
.
= d
1
, . . . , c
n
.
= d
n
, R(c
1
, . . . , c
n
).
Similarly for (2) we rst notice that a
c0
= a
d0
follows from
a
c1
= a
d1
, . . . , a
cn
= a
dn
, f(c
1
, . . . , c
n
)
.
= c
0
T
, f(d
1
, . . . , d
n
)
.
= d
0
T
.
For (2) we also have to show that for all c
1
, . . . , c
n
there exists c
0
with f(c
1
, . . . , c
n
)
.
=
c
0
T
. As T
is a Henkintheory, there exists c

0
with
xf(c
1
, . . . , c
n
)
.
= x f(c
1
, . . . , c
n
)
.
= c
0
T
.
Now the valid sentence xf(c
1
, . . . , c
n
)
.
= x belongs to T
, so f(c
1
, . . . , c
n
)
.
= c
0
belongs to T
. This shows that everything is well dened.

Let A
be the L(C)structure (A, a

c
)
cC
. We show by induction on the
complexity of that for every L(C)-sentence
A
[= T
.
There are four cases:
a) is atomic.
If has the form c
.
= d or R(c
1
, . . . , c
n
), the statement follows from the
construction of A
. Other atomic sentences contain function symbols f or

constants (which we consider in this proof as 0ary function symbols) from
L. We inductively reduce the number of such occurrences and apply the
previous case. Suppose contains a function symbol from L. Then can
be written as
= (f(c
1
, . . . , c
n
))
for a function symbol f L and an L(C)formula (x). Choose c
0
satisfying
f(c
1
, . . . , c
n
)
.
= c
0
T
. By construction f(c
1
, . . . , c
n
)
.
= c
0
holds in A
.
Thus A
[= A
[= (c
0
) and T
(c
0
) T
. By induction
hypothesis on the number of occurrences we have A
[= (c
0
) (c
0
)
T
. This suces.
b) = .
As T
is nitely complete, T
, T
holds and by the induction

hypothesis we have
A
[= A
,[= , T
.
c) = (
1

2
).
As T
contains all sentences which follow from a nite subset of T
, belongs
to T
if and only if
1
and
2
belong to T
. Thus
A
[= A
[=
i
(i = 1, 2)
i
T
(i = 1, 2) T
.
d) = x(x)
We have
A
[= A
[= (c) for some c C (c) T
for some c C T
.
The second equivalence is the induction hypothesis and for the third we argue
as follows: if T
, we choose c satisfying (c) T
. As T
we
also have (c) T
.
Corollary 5.4. T if and only if for a nite subset of T.
Proof. The formula follows from T if and only if T is inconsistent.
Corollary 5.5. A set of formulas (x
1
, . . . , x
n
) is consistent with T if and only
if every nite subset of is consistent with T.
Proof. Introduce new constants c
1
, . . . , c
n
. Then is consistent with T if and
only if T (c
1
, . . . , c
n
) is consistent. Now apply Theorem 5.1.
Denition 5.6. Let A be an Lstructure and B A. Then a A realises a
set of L(B)formulas (x) (containing at most the free variable x), if a satises
all formulas from (x). We write
A [= (a).
(x) is called nitely satisable in A if every nite subset of is realised in A.
Lemma 5.7. (x) is nitely satisable in A if and only if there is an elementary
extension of A in which (x) is realised.
Proof. By Lemma 4.1, is realised in an elementary extension of A if and only
if is consistent with Th(A
A
). So the lemma follows from the easy observation
that a nite set of L(A)formulas is consistent with Th(A
A
) if and only if it is
realised in A.
Denition 5.8. Let A be an Lstructure and B a subset of A. A set p(x) of
L(B)formulas is a type over B if p(x) is maximal nitely satisable in A. We
call B the domain of p. Let
S(B) = S
A
(B)
denote the set of types over B.
Every element a of A determines a type
tp(a/B) = tp
A
(a/B) = (x) [ A [= (a), an L(B)formula.
So an element a realises the type p S(B) exactly if p = tp(a/B). Note that if
A
is an elementary extension of A, then

S
A
(B) = S
A
(B) and tp
A
(a/B) = tp
A
(a/B).
We will use the notation tp(a) for tp(a/).
Similarly, maximal nitely satisable sets of formulas in x
1
, . . . , x
n
are called
ntypes and
S
n
(B) = S
A
n
(B)
denotes the set of ntypes over B. For an ntuple a from A, there is an obvious
denition of tp
A
(a/B) S
A
n
(B). Very much in the same way, we can dene the
type tp(C/B) of an arbitrary set C over B. This will be convenient in later
chapters. In order to do this properly we allow free variables x
c
indexed by
c C and dene
tp(C/B) = (x
c1
, . . . , x
cn
) [ A [= (c
1
, . . . , c
n
), an L(B)formula.
Corollary 5.9. Every structure A has an elementary extension B in which all
types over A are realised.
Proof. We choose for every p S(A) a new constant c
p
. We have to nd a
model of
Th(A
A
)
_
pS(A)
p(c
p
).
It is easy to see that this theory is nitely satisable using that every p is nitely
satisable in A. The compactness theorem now shows that the model exists.
We give a second proof which only uses Lemma 5.7: Let (p
)
<
be an
enumeration of S(A), where is an ordinal number (see Appendix A.3). Using
Theorem A.3.1, we construct an elementary chain
A = A
0
A
1
. . . A
. . . ( )
such that each p
is realised in A
+1
.
Let us suppose that the elementary chain (A
<
is already constructed. If
is a limit ordinal, we let A
<
A
.
2
The longer chain (A
is
elementary because of Lemma 4.4. If = + 1 we rst note that p
is also
nitely satisable in A
. Therefore we can realise p
in a suitable elementary
extension A
~ A
. Then B= A
is the model we were looking for.

Exercise 5.1.
Prove the Compactness Theorem using ultraproducts (see Exercise 2.4).
Exercise 5.2.
A class c of Lstructure is nitely axiomatisable if it is the class of models of a
nite theory. Show that c is nitely axiomatisable if and only if both c and its
complement form an elementary class.
Exercise 5.3.
Show that the class of connected graphs is not an elementary class. A graph
(V, R) is a set V with a symmetric, irreexive binary relation. It is connected
if for any x, y V there is a sequence of elements x
0
= x, . . . x
k
= y such that
(x
i1
, x
i
) R for i = 1, . . . n.
2
We call a chain (A) indexed by ordinal numbers continuous if A
<
A for all
limit ordinals .
Exercise 5.4.
Let A = (R, 0, <, f
A
) where f is a unary function symbol. Call an element
x A
~ A innitesimal if r < x < r for all r A

>0
. Show: If f
A
(0) = 0,
then f
A
is continuous in 0 if and only if for any elementary extension A
of A
the map f
A
takes innitesimal elements to innitesimal elements.

Exercise 5.5.
Let T be an L
R
-theory containing the theory of elds T
F
. Show:
1. If T has models of arbitrary large characteristic, then it has a model of
characteristic 0.
2. The theory of elds of characteristic 0 is not nitely axiomatisable.
6 The LwenheimSkolem Theorem
One of the consequences of the Compactness Theorem is the fact that a rst-
order theory cannot pin down the size of an innite structure. This is the
content of the following theorem:
Theorem 6.1 (LwenheimSkolem). Let B be an Lstructure, S a subset of
B and an innite cardinal.
1. If
max[S[, [L[ [B[,
then B has an elementary substructure of cardinality containing S.
2. If B is innite and
max[B[, [L[ ,
then B has an elementary extension of cardinality .
Proof. 1: Choose a set S S
B of cardinality and apply Corollary 4.3.

2: We rst construct an elementary extension B
of cardinality at least .
Choose a set C of new constants of cardinality . As B is innite, the theory
Th(B
B
) c
.
= d [ c, d C, c ,= d
is nitely satisable (even in B: just interpret the nitely many new constants
in a nite subset by elements of B). By Lemma 4.1, any model (B
B
, b
c
)
cC
is
an elementary extension of B with many dierent elements (b
c
).
Finally we apply the rst part of the theorem to B
and S = B.
Corollary 6.2. A theory which has an innite model has a model in every
cardinality max([L[,
0
).
Thus, no theory with an innite model can describe this model up to isomor-
phism. So the best we can hope for is a unique model for a given cardinality:
Denition 6.3 (preliminary, cf. 6.5). Let be an innite cardinal. A theory
T is called categorical if all models of T of cardinality are isomorphic.
Theorem 6.4 (Vaughts Test). A categorical theory T is complete if in ad-
dition the following conditions are satised:
a) T consistent,
b) T has no nite model,
c) [L[ .
Proof. We have to show that all models A and B of T are elementarily equiv-
alent. As A and B are innite, Th(A) and Th(B) have models A
and B
of
cardinality . By assumption A
and B
are isomorphic, and follows that

A A
B.
Examples:
1. (Theorem of Cantor, see Exercise 3.1) The theory T
DLO
of dense linear
orders without endpoints is
0
categorical and by Vaughts test complete.
To see this let A and B be countable dense linear orders without end-
points, and let A = a
i
: i , B = b
i
: i . We inductively
dene sequences (c
i
)
i<
, (d
i
)
i<
exhausting A and B, respectively, and
such that the assignment c
i
d
i
is the required isomorphism. Assume
that (c
i
)
i<m
, (d
i
)
i<m
have been dened so that c
i
d
i
, i < m is an order
isomorphism. If m = 2k, let c
m
= a
j
where a
j
is the element with minimal
index in a
i
: i not occurring in (c
i
)
i<m
. Since B is a dense linear
order without endpoints, there is some element d
m
b
i
: i such that
(c
i
)
im
and (d
i
)
im
are order isomorphic. If m = 2k + 1 we interchange
the roles of A and B.
2. For any prime p or p = 0, the theory T
ACFp
of algebraically closed elds
of characteristic p is -categorical for any >
0
. Any two algebraically
closed eld of the same characteristic and of cardinality >
0
have
transcendence bases (over the algebraic closure of the prime eld) of car-
dinality , see Appendix 2.6 and Appendix 8. Any bijection between these
transcendence bases induces an isomorphism of the elds. It follows that
T
ACFp
is complete.
Considering Theorem 6.4 we strengthen our denition:
Denition 6.5. Let be an innite cardinal. A theory T is called categorical
if it is complete, [T[ and, up to isomorphy, has exactly one model of
cardinality .
Exercise 6.1.
1. The theory of Kvector spaces T
M(K)
(see p. 49) is -categorical for
> [K[.
2. Is T
ACFp

0
categorical?
Exercise 6.2.
Show that an sentence, which holds in all nite elds, is true in all alge-
braically closed elds.
Chapter 3
Quantier elimination
7 Preservation theorems
In general, it can be dicult to tell which sentences belong to a given theory or
which extensions are consistent. Therefore it is helpful to know that in certain
theories one can restrict attention to sentences of a certain class, e.g., quantier-
free or, say, existential formulas. We consider a xed language L and rst prove
some separation results:
Lemma 7.1 (Separation Lemma). Let T
1
and T
2
be two theories. Assume 1
is a set of sentences which is closed under and and contains and (true
and false). Then the following are equivalent:
a) There is a sentence 1 which separates T
1
from T
2
. This means
T
1
and T
2
.
b) All models A
1
of T
1
can be separated from all models A
2
of T
2
by a
sentence 1. This means
A
1
[= and A
2
[= .
Proof. a) b):
If separates T
1
from T
2
, it separates all models of T
1
from all models of T
2
.
b) a):
For any model A
1
of T
1
let 1
A1
be the set of all sentences from 1 which are
true in A
1
. b) implies that 1
A1
and T
2
cannot have a common model. By the
Compactness Theorem there is a nite conjunction
A1
of sentences from 1
A1
inconsistent with T
2
. Clearly,
T
1

A1
[ A
1
[= T
1
36
CHAPTER 3. QUANTIFIER ELIMINATION 37
is inconsistent since any model A
1
of T
1
satises
A1
. Again by compactness T
1
implies a disjunction of nitely many of the
A1
. This is in 1 and separates
T
1
from T
2
.
For structures A, B and a map f : A B preserving all formulas from a set
of formulas , we use the notation
f : A
B.
We also write
A
B
to express that all sentences from true in A are also true in B.
Lemma 7.2. Let T be a theory, A a structure and a set of formulas, closed
under existential quantication, conjunction and substitution of variables. Then
the following are equivalent:
a) All sentences which are true in A are consistent with T.
b) There is a model B[= T and a map f : A
B.
Proof. b) a):
Assume f : A
B[= T. If is true in A it is also true in Band therefore

is consistent with T.
a) b):
Consider Th
(A
A
), the set set of all sentences (a), ((x) ), which are true in
A
A
. The models (B, f(a)
aA
) of this theory correspond to maps f : A
B.
This means that we have to nd a model of T Th
(A
A
). To show nite
satisability it is enough to show that T D is consistent for every nite subset
D of Th
(A
A
). Let (a) be the conjunction of the elements of D. Then A is a
model of = x (x), so by assumption T has a model B which is also a model
of . This means that there is a tuple b such that (B, b) [= (a).
Note that Lemma 7.2 applied to T = Th(B) shows that A
B if and only if
there exist a map f and a structure B
B such that f : A
.
Theorem 7.3. Let T
1
and T
2
be two theories. Then the following are equivalent:
a) There is a universal sentence which separates T
1
from T
2
.
b) No model of T
2
is a substructure of a model of T
1
.
Proof. a) b):
Let be a universal sentence which separates T
1
from T
2
. Let A
1
be a model
of T
1
and A
2
a substructure of A
1
. Since A
1
is a model of , A
2
is also model
of by Lemma 2.16. Therefore A
2
cannot be a model of T
2
.
b) a):
We use the Separation Lemma. Let A
1
and A
2
be models of T
1
and T
2
, which
cannot be separated by a universal sentence. This can be denoted by
A
2

A
1
.
Now Lemma 7.2 implies that A
2
has an extension A
1
A
1
. Then A
1
is again a
model of T
1
contradicting b).
Denition 7.4. For any L-theory T, the formulas ( x), ( x) are said to be
equivalent modulo T (or relative to T) if T x(( x) ( x)).
Corollary 7.5. Let T be a theory.
1. Consider a formula (x
1
, . . . , x
n
). The following are equivalent:
a) (x
1
, . . . , x
n
) is, modulo T, equivalent to a universal formula.
b) If A B are models of T and a
1
, . . . , a
n
A, then B[= (a
1
, . . . , a
n
)
implies A [= (a
1
, . . . , a
n
).
2. We call a theory which consists of universal sentences universal. T is
equivalent to a universal theory if and only if all substructures of models
of T are again models of T.
Proof. 1): Assume b). We extend L by an ntuple c of new constants c
1
, . . . , c
n
and consider the theory
T
1
= T (c) and T
2
= T (c).
b) says that substructures of models of T
1
cannot be models of T
2
. By Theo-
rem 7.3 T
1
and T
2
can be separated by a universal L(c)sentence (c). T
1
(c)
implies by Lemma 3.4.2
T x ((x) (x))
and from T
2
(c) we see
T x ((x) (x)).
2): It is clear that substructures of models of a universal theory are models
again. Now suppose that a theory T has this property. Let be an axiom of
T. If A is a substructure of B, it is not possible that B is a model of T and A
is a model of at the same time. By 7.3 there is a universal sentence with
T and . Hence all axioms of T follow from
T
= [ T , universal.
An formula is of the form
x
1
. . . x
n
where is existential (see p. 18). The following is clear:

Lemma 7.6. Suppose is an sentence, (A
i
)
iI
is a directed family of models
of and B the union of the A
i
. Then B is also a model of .
Proof. Write
= x (x),
where is existential. For any a B there is an A
i
containing a. Since A
i
[= ,
clearly (a) holds in A
i
. As (a) is existential it must also hold in B.
Denition 7.7. We call a theory T inductive if the union of any directed family
of models of T is again a model.
Theorem 7.8. Let T
1
and T
2
be two theories. Then the following are equivalent:
a) There is an sentence which separates T
1
from T
2
.
b) No model of T
2
is the union of a chain (or of a directed family) of models
of T
1
.
Proof. a) b):
Assume is a sentence which separates T
1
from T
2
, (A
i
)
iI
is a directed
family of models of T
1
and B the union of the A
i
. Since the A
i
are models of
, B is also a model of by Lemma 7.6. Since B [= , B cannot be a model
of T
2
.
a) b):
If a) is not true, the Separation Lemma gives us models A and B
0
of T
1
and T
2
which cannot be separated by an sentence. Since formulas are equivalent
to negated formulas, we have
B
0
A.
By Lemma 7.2 there is a map
f : B
0
A
0
with A
0
A. We can assume that B
0
A
0
and f is the inclusion map. Then
A
0
B

B
0
B
.
Applying Lemma 7.2 again, we obtain an extension B
1
B
of A
0
B
with B
1
B
B
0
B
,
i.e. B
0
B
1
.
B
0
A
0
B
1
The same procedure applied to A and B

1
gives us two extensions A
1
B
2
with
A
1
A and B
1
B
2
. This results in an innite chain
B
0
A
0
B
1
A
1
B
2
. . .
with A
i
A and B
i
B
i+1
. Let B be the union of the A
i
. Since B is also the
union of the elementary chain of the B
i
, B is an elementary extension of B
0
and hence a model of T
2
. But the A
i
are models of T
1
, so b) does not hold.
Corollary 7.9. Let T be a theory.
1. For each sentence the following are equivalent:
a) is, modulo T, equivalent to an sentence.
b) If
A
0
A
1
. . .
and their union B are models of T, then holds in B if it is true in
all the A
i
.
2. T is inductive if and only if T can be axiomatised by sentences.
Proof. 1): Theorem 7.8 shows that formulas are preserved by unions of
chains. Hence a) b). For the converse consider the theories
T
1
= T and T
2
= T .
b) says that the union of a chain of models of T
1
cannot be a model of T
2
. By
Theorem 7.8 we can separate T
1
and T
2
by an sentence . T
1
implies
T and T
2
implies T .
2): Clearly axiomatised theories are inductive. For the converse assume that
T is inductive and an axiom of T. If B is a union of models of T, it cannot
be a model of . By Theorem 7.8 there is an sentence with T and
. Hence all axioms of T follow from
T
= [ T , formula.
Exercise 7.1.
Let X be a topological space, Y
1
and Y
2
quasi-compact
1
subsets, and 1 a set
of clopen subsets. Then the following are equivalent:
a) There is a positive Boolean combination B of elements from 1 such that
Y
1
B and Y
2
B = .
b) For all y
1
Y
1
and y
2
Y
2
there is an H 1 such that y
1
H and y
2
, H.
(This, in fact, is a generalisation of the Separation Lemma 7.1.)
1
i.e. compact but not necessarily Hausdor
8 Quantier elimination
Having quantier elimination in a reasonable language is a property which makes
a theory tame. In this section we will collect some criteria and extensions of this
concept. They will be applied in Section 9 to show that a number of interesting
theories do have quantier elimination in the appropriate language.
Denition 8.1. A theory T has quantier elimination if every Lformula
(x
1
, . . . , x
n
) is equivalent modulo T to some quantier free formula
(x
1
, . . . , x
n
).
For n = 0 this means that modulo T every sentence is equivalent to a
quantier free sentence. If L has no constants, and are the only quantier
free sentences. T is then either inconsistent or complete.
Note that it is easy to transform any theory T into a theory with quantier
elimination if one is willing to expand the language: just enlarge L by adding an
nplace relation symbol R
for every Lformula (x

1
, . . . , x
n
) and T by adding
all axioms
x
1
, . . . , x
n
(R
(x
1
, . . . , x
n
) (x
1
, . . . , x
n
)).
The resulting theory, the Morleyisation T
m
of T, has quantier elimination.
Many other properties of a theory are not aected by Morleyisation. So T is
complete if and only if T
m
is, similarly for categoricity and other properties
we will dene in later chapters.
A prime structure of T is a structure which embeds into all models of T.
The following is clear:
Lemma 8.2. A consistent theory T with quantier elimination which possesses
a prime structure is complete.
Denition 8.3. A simple existential formula has the form
= y
for a quantier free formula . If is a conjunction of basic formulas, is called
primitive existential.
Lemma 8.4. T has quantier elimination if and only if every primitive exis-
tential formula is, modulo T, equivalent to a quantier free formula.
Proof. We can write every simple existential formula in the form y
_
i<n
i
for
i
which are conjunctions of basic formulas. This shows that every simple
existential formula is equivalent to a disjunction of primitive existential formulas,
namely to
_
i<n
(y
i
). We can therefore assume that every simple existential
formula is, modulo T, equivalent to a quantier free formula.
We are now able to eliminate the quantiers in arbitrary formulas in prenex
normal form (see Exercise 2.3)
Q
1
x
1
. . . Q
n
x
n
.
If Q
n
= , we choose a quantier free formula
0
which, modulo T, is equivalent
to x
n
. Then we proceed with the formula Q
1
x
1
. . . Q
n1
x
n1
0
. If Q
n
= ,
we nd a quantier free
1
which is, modulo T, equivalent to x
n
and proceed
with Q
1
x
1
. . . Q
n1
x
n1
1
.
The following theorem gives useful criteria for quantier elimination.
Theorem 8.5. For a theory T the following are equivalent:
a) T has quantier elimination.
b) For all models M
1
and M
2
of T with a common substructure A we have
M
1
A
M
2
A
.
c) For all models M
1
and M
2
of T with a common substructure A and for
all primitive existential formulas (x
1
, . . . , x
n
) and parameters a
1
, . . . , a
n
from A
M
1
[= (a
1
, . . . , a
n
) M
2
[= (a
1
, . . . , a
n
).
If L has no constants, A is allowed to be the empty structure.
Proof. a) b):
Let (a) be an L(A)sentence which holds in M
1
. Choose a quantier free (x)
which is, modulo T, equivalent to (x). Then
M
1
[= (a) M
1
[= (a)
A [= (a)
M
2
[= (a) M
2
[= (a).
b) c):
Clear.
c) a):
Let (x) be a primitive existential formula. In order to show that (x) is
equivalent, modulo T, to a quantier free formula (x) we extend L by a ntuple
c of new constants c
1
, . . . , c
n
. We have to show that we can separate T (c)
and T (c) by a quantier free sentence (c). We apply the Separation
Lemma. Let M
1
and M
2
be two models of T with two distinguished ntuples
a
1
and a
2
. Suppose that (M
1
, a
1
) and (M
2
, a
2
) satisfy the same quantier free
L(c)sentences. We have to show that
(1) M
1
[= (a
1
) M
2
[= (a
2
).
Consider the substructures A
i
= a
i
M
i
, generated by a
i
. If we can show that
there is an isomorphism
f : A
1
A
2
taking a
1
to a
2
, we may assume that A
1
= A
2
= A and a
1
= a
2
= a. Then (1)
follows directly from c).
Every element of A
1
has the form t
M
1
[a
1
] for an Lterm t(x), (2.6). The iso-
morphism f to be constructed must satisfy:
f
_
t
M
1
[a
1
]
_
= t
M
2
[a
2
].
We now dene f by this equation, and we have to check that f is welldened
and injective. Assume
s
M
1
[a
1
] = t
M
1
[a
1
].
Then s(c)
.
= t(c) holds in (M
1
, a
1
), and by our assumption also in (M
2
, a
2
),
which means
s
M
2
[a
2
] = t
M
2
[a
2
].
This shows that f is welldened. Swapping the two sides yields injectivity.
That f is surjective is clear. It remains to show that f commutes with the
interpretation of the relation symbols.
M
1
[= R
_
t
M
1
1
[a
1
], . . . , t
M
1
m
[a
1
]
,
is equivalent to (M
1
, a
1
) [= R(t
1
(c), . . . , t
m
(c)), which is equivalent to (M
2
, a
2
) [=
R(t
1
(c), . . . , t
m
(c)), which in turn is equivalent to
M
2
[= R
_
t
M
2
1
[a
2
], . . . , t
M
2
m
[a
2
]
.
Note that part b) of Theorem 8.5 is saying that T is substructure complete ;
i.e., for any model M[= T and substructure A M the theory T Diag(A) is
complete.
Denition 8.6. T is model complete if for all models M
1
and M
2
of T
M
1
M
2
M
1
M
2
.
Note that T is model complete if and only if for any M [= T the theory
T Diag(M) is complete.
Clearly, by 8.5.b) applied to A = M
1
all theories with quantier elimination
are model complete.
Lemma 8.7 (Robinsons Test). Let T be a theory. Then the following are
equivalent:
a) T is model complete.
b) For all models M
1
M
2
of T and all existential sentences from L(M
1
)
M
2
[= M
1
[= .
c) Each formula is, modulo T, equivalent to a universal formula.
Proof. a)b) is trivial.
a)c) follows from 7.5.1.
b) implies that every existential formula is, modulo T, equivalent to a universal
formula. As in the proof of 8.4 this implies c).
If M
1
M
2
satises b), we call M
1
existentially closed in M
2
. We denote
this by
M
1
1
M
2
.
Denition 8.8. Let T be a theory. A theory T
is a model companion of T if
the following three conditions are satised.
a) Each model of T can be extended to a model of T
.
b) Each model of T
can be extended to a model of T.

c) T
is model complete.
Theorem 8.9. A theory T has, up to equivalence, at most one model companion
T
.
Proof. If T
+
is another model companion of T, every model of T
+
is contained
in a model of T
and conversely. Let A

0
be a model of T
+
. A
0
can be embedded
in a model B
0
of T
. B
0
again is contained in a model A
1
of T
+
. In this way
we nd two elementary chains, (A
i
) and (B
i
), which have a common union C.
A
0
C and B
0
C implies A
0
B
0
. Thus A
0
is a model of T
. Interchanging
T
and T
+
yields that every model of T
is a model of T
+
.
Digression: Existentially closed structures and the Kaiser hull
Let T be an Ltheory. It follows from 7.2 that the models of T
are the substruc-

tures of models of T. The conditions a) and b) in the denition of model companion
can therefore be expressed as
T
= T
.
Hence the model companion of a theory T depends only on T
.
Denition 8.10. An Lstructure A is called Texistentially closed (or T-e.c), if
a) A can be embedded in a model of T.
b) A is existentially closed in every extension which is a model of T.
A structure A is Te.c exactly if it is T
e.c. This is clear for condition a) since every

model B of T
can be embedded in a model M of T. For b) this follows from the fact

that A B M and A 1 M implies A 1 B.
Lemma 8.11. Every model of a theory T can be embedded in a Te.c structure.
Proof. Let A be a model of T
. We choose an enumeration ()< of all existential

L(A)sentences and construct an ascending chain (A)
of models of T
. We begin
with A0 = A. Let A be constructed. If holds in an extension of A which is a
model of T, we let A+1 be such a model. Otherwise we set A+1 = A. For limit
ordinals we dene A
to be the union of all A, ( < ). Note that A
is again a
model of T
.
The structure A
1
= A has the following property: Every existential L(A)sentence
which holds in an extension of A
1
which is a model of T holds in A
1
. Now, in the same
manner, we construct A
2
from A
1
, etc. The union M of the chain A
0
A
1
A
2
. . . is
the desired Te.c structure.
The structure M constructed in the proof can be very big. On the other hand, it
is easily seen that every elementary substructure N of a Te.c structure M is again
Te.c: let N A be a model of T. Since MN
AN, there is an embedding of M in

an elementary extension B of A which is the identity on N. Since M is existentially
closed in B, N is existentially closed in B and therefore also in A.
N
A M
B
`
`
` `
`
`
` `
1
1
Lemma 8.12 ([28]). Let T be a theory. Then there is a biggest inductive theory T
KH
with T
= T
KH
. T
KH
is called the Kaiser hull of T.
Proof. Let T
1
and T
2
be two inductive theories with T
1
= T
2
= T
. We have to show
that (T
1
T
2
)
= T
. Let M be a model of T. As in the proof of 8.9 we extend M

by a chain A0 B0 A1 B1 . . . of models of T
1
and T
2
. The union of this chain
is a model of T
1
T
2
.
Lemma 8.13. T
KH
is the part of the theory of all T-e.c structures.
Proof. Let T
be the part of the theory of all T-e.c structures. Since Te.c struc-
tures are models of T
, we have T
. It follows from 8.11 that T
. Hence
T
is contained in the Kaiser hull.

It remains to show that every Te.c structure Mis a model of the Kaiser hull. Choose
a model N of T
KH
which contains M. Then M 1 N. This implies N
M and
therefore M|= T
KH
.
The previous lemma implies immediately that Te.c structures are models of T
.
Theorem 8.14. For any theory T the following are equivalent:
a) T has a model companion T
b) All models of T
KH
are Te.c.
c) The Te.c structures form an elementary class.
If T
exists, we have
T
= T
KH
= theory of all Te.c structures.
Proof. a) b):
Let T
be the model companion of T. As a model complete theory, T
is inductive.
So T
is contained in the Kaiser hull and it suces to show that every model Mof T
is Te.c. Let A be a model of T which extends M. A can be embedded in a model N

of T
. Now M N implies M1 A.
b) c):
By the last lemma all Te.c structures are models of T
KH
. Thus b) implies that Te.c
structures are exactly the models of T
KH
.
c) a):
Assume that the T-e.c structures are exactly the models of the theory T
+
. By 8.11 we
have T
= T
+
. Criterion 8.7 implies, that T

+
is model complete. So T
+
is the model
companion of T.
The last assertion of the theorem follows easily from the proof.
Exercise 8.1.
A theory T with quantier elimination is axiomatisable by sentences of the form
x1 . . . xn
where is primitive existential formula.
Exercise 8.2.
Let L be the language containing an unary function f and a binary relation symbol R
and consider the Ltheory T = {xy(R(x, y) (R(x, f(y))}. Show the following
1. For any Te.c structure M and a, b M with b / {a, f
M
(a), (f
M
)
2
(a), . . .} we
have M|= z(R(z, a) R(z, b)).
2. Let Mbe a model of T and a an element of M such that {a, f
M
(a), (f
M
)
2
(a), . . .}
is innite. Then in an elementary extension M
there is an element b with

M
|= z(R(z, a) R(z, b)).

3. The class of Te.c structures is not elementary, so T does not have a model
companion.
9 Examples
In this section we present a number of theories with quantier elimination,
or at least elimination down to some well-understood formulas, among them
the theories of vector spaces and of algebraically, dierentially, and real closed
elds. Since such theories are comparatively easy to understand, they form a
core inventory of the working model theorist.
9.1 Innite sets
The models of the theory T
of innite sets are all innite sets without addi-

tional structure. The language L
is empty, the axioms are (for n = 1, 2, . . .)

x
0
. . . x
n1
_
i<j<n
x
i
.
= x
j
Theorem 9.1. The theory T
of innite sets has quantier elimination and

is complete.
Proof. Clear.
9.2 Dense linear orderings
Theorem 9.2. T
DLO
has quantier elimination.
Proof. Let A be a nite common substructure of the two models O
1
and O
2
.
We choose an ascending enumeration A = a
1
, . . . , a
n
. Let y (y) be a simple
existential L(A)sentence, which is true in O
1
and assume O
1
[= (b
1
). We
want to extend the order preserving map a
i
a
i
to an order preserving map
Ab
1
O
2
. For this we have to nd an image b
2
of b
1
. There are four cases
i) b
1
A. We set b
2
= b
1
.
ii) b
1
lies between a
i
and a
i+1
. We choose b
2
in O
2
with the same property.
iii) b
1
is smaller than all elements of A. We choose a b
2
O
2
of the same kind.
iv) b
1
is bigger that all a
i
. Choose b
2
in the same manner.
This denes an isomorphism A b
1
A b
2
, which shows that O
2
[=
(b
2
).
Since L
O
has no constants, we have another proof that T
DLO
is complete.
9.3 Modules
Let R be a (possibly noncommutative) ring with 1. An Rmodule
M= (M, 0, +, , r)
rR
is an abelian group (M, 0, +, ) together with operations r : M M for every
ring element r R which satises certain axioms. We formulate the axioms
in the language L
M(R)
= L
AG
r [ r R. The theory T
M(R)
of Rmodules
consists of
T
AG
x, y r(x +y)
.
= rx +ry
x (r +s)x
.
= rx +sx
x (rs)x
.
= r(sx)
x 1x
.
= x
for all r, s R. Then T
T
M(R)
is the theory of all innite Rmodules.
We start with the case where the ring is a eld K. Of course, a Kmodule
is just a vector space over K.
Theorem 9.3. Let K be a eld. Then the theory of all innite Kvector spaces
has quantier elimination and is complete.
Proof. Let A be a common nitely generated substructure (i.e. a subspace) of
the two innite Kvector spaces V
1
and V
2
. Let y (y) be a simple existential
L(A)sentence which holds in V
1
. Choose a b
1
from V
1
which satises (y). If
b
1
belongs to A, we are nished since then V
2
[= (b
1
). If not, we choose a
b
2
V
2
A. Possibly we have to replace V
2
by an elementary extension. The
vector spaces A + Kb
1
and A + Kb
2
are isomorphic by an isomorphism which
maps b
1
to b
2
and xes A elementwise. Hence V
2
[= (b
2
).
The theory is complete since a quantier free sentence is true in a vector space
if and only if it is true in the zerovector space.
For arbitrary rings R, we only get a relative elimination result down to
positive primitive formulas:
Denition 9.4. An equation is a L
M(R)
formula ( x) of the form
r
1
x
1
+r
2
x
2
+... +r
m
x
m
= 0.
A positive primitive formula (ppformula) is of the form
y(
1
. . .
n
)
where the
i
( x y) are equations.
Theorem 9.5. For every ring R and any R-module M, every L
M(R)
-formula
is equivalent (modulo the theory of M) to a Boolean combination of positive
primitive formulas.
Remark 9.6. 1. We assume the class of positive primitive formulas to be
closed under .
2. A ppformula (x
1
, . . . , x
n
) denes a subgroup (M
n
) of M
n
:
M [= (0) and M [= (x) (y) (x y).
Lemma 9.7. Let (x, y) be a ppformula and a M. Then (M, a) is empty
or a coset of (M, 0).
Proof. M [= (x, a) ((y, 0) (x +y, a)).
Corollary 9.8. Let a, b M, (x, y) a ppformula. Then (in M) (x, a) and
(x, b) are equivalent or contradictory.
For the proof of Theorem 9.5 we need two further lemmas:
Lemma 9.9 (B.H. Neumann). Let H
i
denote subgroups of some abelian group.
If H
0
+a
0

n
i=1
H
i
+a
i
and H
0
/(H
0
H
i
) is innite for i > k, then H
0
+a
0

k
i=1
H
i
+a
i
.
For a proof see Exercise 23.15. The following is an easy calculation:
Lemma 9.10. Let A
i
, i k, be any sets. If A
0
is nite, then A
0

k
i=1
A
i
if
and only if
{1,...k}
(1)
||
[A
0

i
A
i
[ = 0.
Proof of 9.5. Fix M. It is enough to show that if (x, y) is in M equivalent
to a Boolean combination of ppformulas, then so is x. Since ppformulas
are closed under conjunction, is M-equivalent to a conjunction of formulas
0
(x, y)
1
(x, y) . . .
n
(x, y) where the
i
(x, y) are ppformulas.
We may assume that itself is of this form. Let H
i
=
i
(M, 0), so the
i
(M, y) are empty or cosets of H
i
. (Think of y as being xed in M.) Let
H
0
/(H
0
H
i
) be nite for i = 1, . . . , k and innite for i = k + 1, . . . , n, k 0.
By Neumanns Lemma we have
M [= x x(
0
(x, y)
1
(x, y) . . .
k
(x, y)) .
We apply Lemma 9.10 to the sets A
i
=
i
(M, y)/(H
0
. . . H
k
): So
(M, y)
i
(M, y) is empty or consists of N
cosets of H
0
. . . H
k
where
N
H
0

i
H
i
/(H
0
. . . H
k
)
.
Whence
M [= x
N
(1)
||
N
= 0.
where
A =
_
1, . . . k: x
_
0
(x, y)
i
(x, y)
__
.
9.4 Algebraically closed elds
As the next group of examples we consider elds:
Theorem 9.11 (Tarski). The theory T
ACF
of algebraically closed elds has
quantier elimination.
Proof. Let K
1
and K
2
be two algebraically closed elds and R a common sub-
ring. Let y (y) be a simple existential sentence with parameters in R which
holds in K
1
. We have to show that y (y) is also true in K
2
.
Let F
1
and F
2
be the quotient elds of R in K
1
and K
2
, respectively, and
let f : F
1
F
2
be an isomorphism which is the identity on R (see e.g. [31],
Ch. II.4). Then f extends to an isomorphism g : G
1
G
2
between the relative
algebraic closures G
i
of F
i
in K
i
, (i = 1, 2) (see e.g. [31], Ch. V.2) .
R R
id
F
1
F
2
f
G
1
G
2
g
G
1
(b
1
)
K
1
K
2
Choose an element b
1
of K
1
which satises (y). There are two cases:
Case 1: b
1
G
1
Then b
2
= g(b
1
) satises the formula (y) in K
2
.
Case 2: b
1
, G
1
Then b
1
is transcendental over G
1
and the eld extension G
1
(b
1
) is isomorphic
to the rational function eld G
1
(X). If K
2
is a proper extension of G
2
, we
choose any element from K
2
G
2
for b
2
. Then g extends to an isomorphism
between G
1
(b
1
) and G
2
(b
2
) which maps b
1
to b
2
. Hence b
2
satises (y) in K
2
.
In case that K
2
= G
2
we take a proper elementary extension K
2
of K
2
. (K
2
exists by 6.1.2 since K
2
is innite.) Then (for the same reason) y (y) holds
in K
2
and therefore in K
2
.
Corollary 9.12. T
ACF
is model complete.
Obviously, T
ACF
is not complete:
For prime numbers p let
T
ACFp
= T
ACF
p 1
.
= 1 +. . . + 1
. .
p times
.
= 0
be the theory of algebraically closed elds of characteristic p and
T
ACF0
= T
ACF
n 1 = 1 +. . . + 1
. .
n times
.
= 0 [ n = 1, 2, . . .
the theory of algebraically closed elds of characteristic 0.
Corollary 9.13. The theories T
ACFp
and T
ACF0
are complete.
Proof. This follows from 8.2 since the prime elds are prime structures for these
theories.
Corollary 9.14 (Hilberts Nullstellensatz). Let K be a eld. Then any proper
ideal I in K[X
1
, . . . , X
n
] has a zero in the algebraic closure acl(K).
Proof. As a proper ideal, I is contained in a maximal ideal P. Then L =
K[X
1
, . . . , X
n
]/P is an extension eld of K in which the cosets of the X
i
are a
zero of I. If I is generated by f
0
, . . . , f
k1
and
= x
1
, . . . x
n
i<k
f
i
(x
1
, . . . , x
n
)
.
= 0,
then holds in L and therefore in acl(L). We can assume that acl(K) lies in
acl(L). Since acl(K) acl(L), holds in acl(K).
9.5 Real closed elds
The theory of real closed elds, T
RCF
, will be discussed in section B.4. It is
axiomatised in the language L
OR
of ordered rings.
Theorem 9.15 (Tarski-Seidenberg). T
RCF
has quantier elimination and is
complete.
Proof. Let (K
1
, <) and (K
2
, <) be two real closed eld with a common subring
R. Consider an L
OR
(R)sentence y (y) (for a quantier free ) which holds
in (K
1
, <). We have to show y (y) also holds in (K
2
, <).
We build rst the quotient elds F
1
and F
2
of R in K
1
and K
2
. By B.4.1 there
is an isomorphism f : (F
1
, <) (F
2
, <) which xes R. The relative algebraic
closure G
i
of F
i
in K
i
is a real closure of (F
i
, <), (i = 1, 2). By B.4.5 f extends
to an isomorphism g : (G
1
, <) (G
2
, <).
Let b
1
be an element of K
1
which satises (y). There are two cases:
Case 1: b
1
G
1
Then b
2
= g(b
1
) satises (y) in K
2
.
Case 2: b
1
, G
1
:
Then b
1
is transcendental over G
1
and the eld extension G
1
(b
1
) is isomorphic to
the rational function eld G
1
(X). Let G
1
be the set of all elements of G
1
which
are smaller than b
1
, G
r
1
the set of all elements of G
1
which are larger than b
1
.
Then all elements of G
2
= g(G
1
) are smaller than all elements of G
r
2
= g(G
r
1
).
Since elds are densely ordered, we nd in an elementary extension (K
2
, <) of
(K
2
, <) an element b
2
which lies between the elements of G
2
and the elements
of G
r
2
. Since b
2
is not in G
2
, it is transcendental over G
2
. Hence g extends to
an isomorphism h: G
1
(b
1
) G
2
(b
2
) which maps b
1
to b
2
.
In order to show that h is order preserving it suces to show that h is order
preserving on G
1
[b
1
] (B.4.1). Let p(b
1
) be an element of G
1
[b
1
]. Corollary 4.8
gives us a decomposition
p(X) =
i<m
(X a
i
)
j<n
_
(X c
j
)
2
+d
j
_
with positive d
j
. The sign of p(b
1
) depends only on the signs of the factors ,
b
1
a
0
, . . . , b
1
a
m1
. The sign of h(p(b
1
)) depends in the same way on the
signs of g(), b
2
g(a
0
), . . . , b
2
g(a
m1
). But b
2
was chosen in such a way
that
b
1
< a
i
b
2
< g(a
i
).
Hence p(b
1
) is positive if and only if h(p(b
1
)) is positive.
Finally we have
(K
1
, <) [= (b
1
) (G
1
(b
1
), <) [= (b
1
) (G
2
(b
2
), <) [= (b
2
)
(K
2
, <) [= y (y) (K
2
, <) [= y (y),
which proves quantier elimination.
T
RCF
is complete since the ordered eld of the rationals is a prime structure.
Corollary 9.16 (17. Hilberts Problem). Let (K, <) be a real closed eld. A
polynomial f K[X
1
, . . . , X
n
] is a sum of squares
f = g
2
1
+. . . +g
2
k
of rational functions g
i
K(X
1
, . . . , X
n
) if and only if
f(a
1
, . . . , a
n
) 0
for all a
1
, . . . , a
n
K.
Proof. Clearly a sum of squares cannot have negative values. For the converse
assume that f is not a sum of squares. Then, by Corollary 4.3, K(X
1
, . . . , X
n
)
has an ordering in which f is negative. Since in K the positive elements are
squares, this ordering, which we denote also by <, extends the ordering of K.
Let (L, <) be the real closure of
_
K(X
1
, . . . , X
n
), <
_
. In (L, <) the sentence
x
1
, . . . , x
n
f(x
1
, . . . , x
n
) < 0
is true. Hence it is also true in (K, <).
9.6 Separably closed elds
A eld is separably closed if every nonconstant separable polynomial has a zero.
Clearly, separably closed elds which are also perfect are algebraically closed.
For any eld K of characteristic p > 0, K
p
= a
p
[ a K is a subeld of
K. If the degree [K : K
p
] is nite, it has the form p
e
and e is called the degree
of imperfection of K. If the degree [K : K
p
] is innite, then we say that K has
innite degree of imperfection. See also page 234.
For any natural number e we denote by SCF
p,e
the theory of separably closed
elds with degree of imperfection e. SCF
p,
is the theory of separably closed
closed eld of characteristic p with innite degree of imperfection. We will prove
below that SCF
p,e
is complete. A proof of the completeness of SCF
p,
can be
found in [15].
To study SCF
p,e
we consider an expansion of it: SCF
p
(c
1
, , c
e
), the theory
of separably closed elds of characteristic p in the language L(c
1
, . . . , c
e
) of rings
with constants c
1
, . . . , c
e
for a distinguished nite pbasis. We show
Proposition 9.17. SCF
p
(c
1
, , c
e
) is model complete.
For the proof we need the following lemma:
Lemma 9.18. Let K and L be extensions of F. Assume that K/F is sepa-
rable and that L is separably closed. Then K embeds over F in an elementary
extension of L.
Proof. Since L is innite, it has arbitrarily large elementary extensions. So
we may assume that tr. deg(L/F) is innite. Let K
be a nitely generated
subeld of K over F. By compactness it suces to show that all such K
/F
can be embedded into L/F. By LemmaB 6.12 K
/F has a transcendence basis

x
1
, . . . , x
n
so that K
/F(x
1
, . . . , x
n
) is separably algebraic. F(x
1
, . . . , x
n
)/F
can be embedded into L/F. Since L is separably closed this embedding extend
to K
.
Proof of Proposition 9.17. Let (F, b
1
, . . . , b
e
) (K, b
1
, . . . , b
e
) be an extension
of models of SCF
p
(c
1
, , c
e
). Since F and K have the same pbasis, K is
separable over F by Remark B6.9. By Lemma 9.18, K embeds over F into an
elementary extension of F, showing that F is existentially closed in K. Now
the claim follows by Robinsons Test (Lemma 8.7).
Corollary 9.19 (Ershov). SCF
p,e
is complete.
Proof. Consider the polynomial ring R = F
p
[x
1
, . . . , x
e
]. It is easy to see that
x
1
, . . . , x
e
is a pbasis of R in the sense of Lemma B 6.11. The same lemma
implies that x
1
, . . . , x
e
is a pbasis of F = F
p
(x
1
, . . . , x
e
)
sep
(where L
sep
denotes
the separabel algebraic closure of the eld L). So (F, x
1
, . . . , x
e
) is a model of
SCF
p
(c
1
, , c
e
). If (L, b
1
, . . . , b
e
) is another model of SCF
p
(c
1
, , c
e
), then
by Lemma B 6.10 the b
i
are algebraically independent over F
p
, so we can embed
(F, x
1
, . . . , x
e
) into (L, b
1
, . . . , b
e
). This embedding is elementary by Proposi-
tion 9.17, so L is elementarily equivalent to F.
Let K be a eld with pbasis b
1
, . . . , b
e
. The -functions
: K K are
dened by
x =
(x)
p
b
,
where the are multi-indices (
1
, . . . ,
e
) with 0
i
< p and b
denotes
b
1
1
b
e
e
.
Theorem 9.20 (Delon). SCF
p
(c
1
, , c
e
) has quantier elimination in the lan-
guage L(c
1
, . . . , c
e
,
)
p
e.
It can be shown that SCF
p
(c
1
, , c
e
) has quantier elimination already in the
language L(
)
p
e without naming a pbasis.
Proof. Let / = (K, b
1
, . . . , b
e
) and L = (L, b
1
, . . . , b
e
) be models of
SCF
p
(c
1
, , c
e
) and let R be a common subring which contains the b
i
and
is closed under the -functions of / and L.
2
Since R is closed under the -
functions, the b
i
form a pbasis of R in the sense of Lemma B 6.11. Let F
be the separable closure of the quotient eld of R. By Lemma B 6.11 the b
i
also form a pbasis of F. So F is an elementary subeld of K and of L by
Proposition 9.17, and hence K
R
and L
R
are elementarily equivalent. Now the
claim follows from Theorem 8.5.
2
Note that the -functions of K and L agree automatically on R.
9.7 Dierentially closed elds
We next consider elds with a derivation in the language of elds expanded
by a function symbol d. Dierential elds are introduced and discussed in
Section B.5.
Denition 9.21. The theory of dierentially closed elds, T
DCF0
, is the theory
of dierential elds (K, d) in characteristic 0 satisfying the following property:
For f K[x
0
, . . . , x
n
] K[x
0
, . . . , x
n1
] and g K[x
0
, . . . , x
n1
], g ,= 0, there
is some a K such that f(a, da, . . . , d
n
a) = 0 and g(a, da, . . . , d
n1
a) ,= 0.
Clearly, models of T
DCF0
are algebraically closed.
Theorem 9.22.
1. Any dierential eld can be extended to a model of T
DCF0
.
2. T
DCF0
is complete and has quantier elimination.
Proof. Let (K, d) be a dierential eld and f and g as in the denition. We
may assume that f is irreducible. Then f determines a eld extension F =
K(t
0
, . . . , t
n1
, b) where the t
i
are algebraically independent over K and
f(t
0
, . . . , t
n1
, b) = 0
(see Remark B 6.7). By Lemma B.5.2 and B.5.3 there is an extension of
the derivation to F with dt
i
= t
i+1
and dt
n1
= b. For a = t
0
we have
f(a, da, . . . , d
n
a) = 0 and g(a, da, . . . , d
n1
a) ,= 0. Let K
0
denote the dieren-
tial eld that we obtain from K by doing this for all pairs f, g as above with
coecients from K. Inductively, we dene dierential elds K
i+1
satisfying the
required condition for polynomials with coecients in K
i
. Their union
i<
K
i
is a model of T
DCF0
.
Since the rational numbers with trivial derivation are a prime structure for
T
DCF0
, by Lemma 8.2 it suces for part 2. to prove quantier elimination.
For this, let K be a dierential eld with two extensions F
1
and F
2
which are
models of T
DCF0
. Let a be an element of F
1
and let Ka = K(a, da, d
2
a, . . . )
be the dierential eld generated by K and a. We have to show that Ka can
be embedded over K into an elementary extension of F
2
. We distinguish two
cases.
1. The derivatives a, da, d
2
a, . . . are algebraically independent over K: Since
F
2
is a model of T
DCF0
, there is an element b in some elementary extension
such that g(a, da, . . . , d
n1
a) ,= 0 for all n and all g K[x
0
, . . . , x
n1
] 0.
The isomorphism from Ka to Kb dened by d
i
a d
i
b is the required
embedding.
2. Let d
n
a be algebraic over K(a, da, . . . , d
n1
a) and n minimal: Choose an
irreducible f K[x
0
, . . . , x
n
] such that f(a, da, . . . , d
n
a) = 0 (cf. Remark B 6.7).
We may nd some b with f(b, db, . . . , d
n
b) = 0 and g(b, ba, . . . , d
n1
b) ,= 0 for all
g K[x
0
, . . . , x
n1
] 0 in an elementary extension of F
2
. The eld isomorphism
from K
1
= K(a, . . . , d
n
a) to K
2
= K(b, . . . , d
n
b) xing K and taking d
i
a to d
i
b
takes the derivation of F
1
restricted to K(a, . . . , d
n1
a) to the derivation of F
2
restricted to K(b, . . . , d
n1
b). The uniqueness part of Lemma B.5.3 implies that
K
1
and K
2
are closed under the respective derivations, and that K
1
and K
2
are
isomorphic over K as dierential elds.
Exercise 9.1.
The theory T
RG
of the random graph is axiomatised by the following axiom
scheme:
x
0
, . . . x
m1
, y
1
, . . . y
n1
_
i=j
x
i
.
= y
j

z(
i<m
zRx
i
z
.
= x
i
) (
j<n
zRy
j
z
.
= y
j
)
_
Show that this theory has quantier elimination.
Exercise 9.2.
In models of T
ACF
, T
RCF
and T
DCF0
, the model theoretic algebraic closure of
a set A coincides with the algebraic closure of the (dierential) eld generated
by A.
Exercise 9.3.
Show that the following is true in any algebraically closed eld K: every injective
polynomial map of a denable subset of K
n
in itself is surjective.
In fact, more is true: use Exercise 23.13 to show that the previous statement
holds for every injective denable map.
Chapter 4
Countable models
10 The Omitting Types Theorem
As we have seen in Corollary 5.9, it is not hard to realise a given type or in fact
any number of them. But as Sacks [44] pointed out, it needs a model theorist
to avoid realizing a given type.
Denition 10.1. Let T be an Ltheory and (x) a set of Lformulas. A model
A of T not realizing (x) is said to omit (x). A formula (x) isolates (x) if
a) (x) is consistent with T.
b) T x ((x) (x)) for all (x) in (x).
A set of formulas is often called a partial type. This explains the name of
the following theorem.
Theorem 10.2 (Omitting Types). If T is countable
1
and consistent and if
(x) is not isolated in T, then T has a model which omits (x).
If (x) is isolated by (x) and A is a model of T, then (x) is realised in
A by all realisations of (x). Therefore the converse of the theorem is true for
complete theories T: if (x) is isolated in T, then it is realised in every model
of T.
Proof. We choose a countable set C of new constants and extend T to a theory
T
with the following properties:

1
An Ltheory is countable if L is at most countable.
58
CHAPTER 4. COUNTABLE MODELS 59
a) T
is a Henkintheory: for all L(C)formulas (x) there exists a constant

c C with x(x) (c) T
.
b) For all c C there is a (x) (x) with (c) T
.
We construct T
inductively as the union of an ascending chain

T = T
0
T
1
. . .
of consistent extensions of T by nitely many axioms from L(C), in each step
making an instance of a) or b) true.
Enumerate C = c
i
: i < and let
i
(x): i < be an enumeration of
the L(C)-formulas.
Assume that T
2i
is already constructed. Choose some c C which does not
occur in T
2i
and set T
2i+1
= T
2i
x
i
(x)
i
(c). Clearly T
2i+1
is consistent
Up to equivalence T
2i+1
has the form T (c
i
, c) for an Lformula (x, y)
and a tuple c C which does not contain c
i
. Since y (x, y) does not isolate
(x), for some the formula y (x, y) (x) is consistent with T. Thus,
T
2i+2
= T
2i+1
(c
i
) consistent.
Take a model (A
, a
c
)
cC
of T
. Since T
is a Henkin theory, Tarskis Test 4.2

shows that A = a
c
[ c C is the universe of an elementary substructure A
(cf. Lemma 5.3). By property b), (x) is omitted in A.
Corollary 10.3. Let T be countable and consistent and let
0
(x
1
, . . . , x
n0
),
1
(x
1
, . . . , x
n1
), . . .
be a sequence of partial types. If all
i
are not isolated, then T has a model
which omits all
i
.
Proof. Generalise the proof of the Omitting Types Theorem.
Exercise 10.1.
Prove Corollary 10.3.
11 The space of types
We now endow the set of types of a given theory with a topology. The Com-
pactness Theorem 5.1 then translates into the statement that this topology is
compact, whence its name.
Fix a theory T. An ntype is a maximal set of formulas p (x
1
, . . . , x
n
) con-
sistent with T. We denote by S
n
(T) the set of all ntypes of T. We also write
S(T) for S
1
(T).
2
If B is a subset of an Lstructure A, we recover S
A
n
(B) (see p. 31) as
S
n
(Th(A
B
)). In particular, if T is complete and A is any model of T, we
have S
A
() = S(T).
For any Lformula (x
1
, . . . , x
n
), let [] denote the set of all types contain-
ing .
Lemma 11.1.
1. [] = [] if and only if and are equivalent modulo T.
2. The sets [] are closed under Boolean operations. In fact [][] = [],
[] [] = [ ], S
n
(T) [] = [], S
n
(T) = [] and = [].
Proof. For the rst part just notice that if and are not equivalent modulo
T, then or is consistent with T and hence [] ,= []. The rest is
clear.
It follows that the collection of sets of the form [] is closed under nite
intersections and includes S
n
(T). So these sets form a basis of a topology on
S
n
(T).
Lemma 11.2. S
n
(T) is a 0dimensional compact space.
Proof. Being 0dimensional means having a basis of clopen sets. Our basic open
sets are clopen since their complements are also basic open.
If p and q are two dierent types, there is a formula contained in p but
not in q. It follows that [] and [] are open sets which separate p and q. This
shows that S
n
(T) is Hausdor.
To show compactness consider a family [
i
], (i I), with the nite intersec-
tion property. This means that all
i1

i
k
are consistent with T. So, by
Corollary 5.5,
i
[ i I is consistent with T and can be extended to a type
p, which then belongs to all [
i
].
Lemma 11.3. All clopen subsets of S
n
(T) have the form [].
2
S
0
(T) can be considered as the set of all complete extensions of T, up to equivalence.
Proof. It follows from Exercise 7.1 that we can separate any two disjoint closed
subsets of S
n
(T) by a basic open set.
Remark. The Stone duality theorem asserts that the map
X C [ C clopen subset of X
yields an equivalence between the category of 0dimensional compact spaces
and the category of Boolean algebras. The inverse map assigns to every Boolean
algebra B its Stone space S(B), the set of all ultralters (see Exercise 2.4) of B.
For more on Boolean algebras see [17].
Denition 11.4. A map f from a subset of a structure A to a structure B is
elementary if it preserves the truth of formulas; i.e., f : A
0
B is elementary
if for every formula (x
1
, . . . , x
n
) and a A
0
we have
A [= ( a) B[= (f( a)).
Note that the empty map is elementary if and only if A and B are elementarily
equivalent. An elementary embedding of A is an elementary map which is
dened on all of A.
Lemma 11.5. Let A and B be Lstructures, A
0
and B
0
subsets of A and B,
respectively. Any elementary map A
0
B
0
induces a continuous surjective
map S
n
(B
0
) S
n
(A
0
).
Proof. If q(x) S
n
(B
0
), we dene
S(f)(q) = (x
1
, . . . , x
n
, a) [ a A
0
, (x
1
, . . . , x
n
, fa) q.
It is easy to see that S(f) denes a map from S
n
(B
0
) to S
n
(A
0
). S(f) is surjective
since (x
1
, . . . , x
n
,

fa) [ (x
1
, . . . , x
n
, a) p is nitely satisable for every
p S
n
(A
0
). And S(f) is continuous since [(x
1
, . . . , x
n
, fa)] is the preimage of
[(x
1
, . . . , x
n
, a)] under S(f).
There are two main cases:
An elementary bijection f : A
0
B
0
denes a homeomorphism S
n
(A
0
)
S
n
(B
0
). We write f(p) for the image of p.
If A = Band A
0
B
0
, the inclusion map induces the restriction S
n
(B
0
)
S
n
(A
0
). We write q A
0
for the restriction of q to A
0
. We call q an ex-
tension of q A
0
.
We leave the following lemma as an exercise (see Exercise 11.1).
Lemma 11.6. A type p is isolated in T if and only if p is an isolated point in
S
n
(T). If fact, isolates p if and only if [] = p. I.e. [] is an atom in the
Boolean algebra of clopen subsets of S
n
(T).
We call a formula (x) complete if
(x) [ T x ((x) (x))
is a type. We have shown:
Corollary 11.7. A formula isolates a type if and only if it is complete.
Exercise 11.1.
Show that a type p is isolated if and only if it is isolated as an element in the
Stone space.
Exercise 11.2.
a) Closed subsets of S
n
(T) have the form p S
n
(T) [ p, where is any
set of formulas.
b) Let T be countable and consistent. Then any meager
3
subset X of S
n
(T)
can be omitted, i.e. there is model which omits all p X.
Exercise 11.3.
Consider the space S
(T) of all complete types in variables v

0
, v
1
, . . .. Note that
S
(T) is again a compact space and therefore not meager by Baires theorem.
1. Show that tp(a
0
, a
2
, . . .) [ the a
i
enumerate a model of T is comeager
in S
(T).
2. Use this to give a purely topological proof the Omitting Types Theorem
(10.3).
Exercise 11.4.
Let L L
, T an Ltheory, T
an L
theory and T T
. Show that there is a

natural continuous map S
n
(T
) S
n
(T). This map is surjective if and only if
T
/T is a conservative extension, i.e. if T
and T prove the same Lsentences.

Exercise 11.5.
Let B be a subset of A. Show that the restriction map S
m+n
(B) S
n
(B) is
open, continuous and surjective. Let a be an ntuple in A. Show that the bre
over tp(a/B) is canonically homeomorphic to S
m
(aB).
Exercise 11.6.
A theory T has quantier elimination if and only if every type is implied by its
quantier free part.
3
A subset of a topological space is nowhere dense if its closure has no interior. A countable
union of nowhere dense sets is meager.
Exercise 11.7.
Consider the structure M = (Q, <). Determine all types in S
1
(Q). Which of
these types are realised in R? Which extensions does a type over Q have to a
type over R?
12
0
categorical theories
In this section, we consider theories with a unique countable model (up to
isomorphism, of course). These theories can be characterised by the fact that
they have only nite many n-types for each n, see Exercise 12.3. We show the
following equivalent statement:
Theorem 12.1 (RyllNardzewski). Let T be a countable complete theory with
innite models. Then T is
0
categorical if and only if for every n there are
only nitely many formulas (x
1
, . . . , x
n
) up to equivalence relative to T.
The proof will make use of the following notion:
Denition 12.2. An Lstructure A is saturated if all types over nite subsets
of A are realised in A.
The types in the denition are meant to be 1types. On the other hand, it
is not hard to see that an -saturated structure realises all n-types over nite
sets (cp. Exercise 12.9), for all n 1. The following lemma is a generalisation
of the
0
categoricity of T
DLO
. The proof is essentially the same, see p. 35
Lemma 12.3. Two elementarily equivalent countable and saturated struc-
tures are isomorphic.
Proof. Suppose A and B are as in the lemma. We choose enumerations A =
a
0
, a
1
, . . . and B = b
0
, b
1
, . . .. Then we construct an ascending sequence
f
0
f
1
. . . of nite elementary maps
f
i
: A
i
B
i
between nite subset of A and B. We will choose the f
i
in such a way that A
is the union of the A
i
and B the union of the B
i
. The union of the f
i
is then
the desired isomorphism between A and B.
The empty map f
0
= is elementary since A and B are elementarily equiv-
alent. Assume that f
i
is already constructed. There are two cases:
i = 2n:
We will extend f
i
to A
i+1
= A
i
a
n
.Consider the type
p(x) = tp(a
n
/A
i
).
Since f
i
is elementary, f
i
(p)(x) is in B a type over B
i
. Since B is saturated,
there is a realisation b
of this type. So for a A

i
A [= (a
n
, a) B[= (b
, f
i
(a)).
This shows that f
i+1
(a
n
) = b
denes an elementary extension of f

i
.
i = 2n + 1:
We exchange A and B: since A is saturated, we nd an elementary map f
i+1
with image B
i+1
= B
i
b
n
.
Proof of Theorem 12.1. Assume that there are only nitely many (x
1
, . . . , x
n
)
relative to T for every n. By Lemma 12.3 it suces to show that all models of
T are saturated. Let Mbe a model of T and A an nelement subset. If there
are only N many formulas, up to equivalence, in the variables x
1
, . . . , x
n+1
,
there are, up to equivalence in M, at most N many L(A)formulas (x). Thus,
each type p(x) S(A) is isolated (with respect to Th(M
A
)) by a smallest
formula
p
(x). Each element of M which realises
p
(x) also realises p(x), so M
is saturated.
Conversely, if there are innitely many (x
1
, . . . , x
n
) modulo T for some n,
then as the type space S
n
(T) is compact there must be some non-isolated
type p. By the Omitting Types Theorem (10.2) there is a countable model of
T in which this type is not realised. On the other hand, there also exists a
countable model of T realizing this type. So T is not
0
categorical.
Remark 12.4. The proof shows that a countable complete theory with innite
models is
0
categorical if and only if all countable models are saturated.
In Theorem 16.10 this characterisation will be extended to theories categor-
ical in uncountable cardinalities.
Remark 12.5. The proof of Lemma 12.3 also shows that -saturated models
are homogeneous in the following sense:
Denition 12.6. An Lstructure Mis homogeneous if for every elementary
map f
0
dened on a nite subset A of M and for any a M there is some
b M such that
f = f
0
a, b
is elementary.
Note that f = f
0
a, b is elementary if and only if b realises f
0
(tp(a/A)).
Corollary 12.7. Let A be a structure and a
1
, . . . , a
n
elements of A. Then
Th(A) is
0
categorical if and only if Th(A, a
1
, . . . , a
n
) is
0
categorical.
Examples:
The following theories are
0
categorical:
T
, the theory of innite sets.

For every nite eld F
q
, the theory of innite F
q
vector spaces. (Indeed, this
theory is categorical in all innite cardinals. This follows directly from the
fact that vector spaces over the same eld and of the same dimension are
isomorphic.)
The theory T
RG
of the random graph (see Exercise 9.1): This follows from
Theorem 12.1 since T
RG
has quantier elimination and for any n there are
only nitely many graphs on n elements.
The theory T
DLO
of dense linear orders without endpoints. This follows from
Theorem 12.1 since T
DLO
has quantier elimination: For every n there are only
nitely many (say N
n
) ways to order n (not necessarily distinct) elements.
For n = 2 for example there are the three possibilities a
1
< a
2
, a
1
= a
2
and a
2
< a
1
. Each of these possibilities corresponds to a complete formula
(x
1
, . . . , x
n
). Hence there are, up to equivalence, exactly 2
Nn
many formulas
(x
1
, . . . , x
n
).
Next we study the existence of countable saturated structures:
Denition 12.8. A theory T is small if S
n
(T) are at most countable for all
n < .
A countable theory with at most countably many nonisomorphic at most
countable models is always small. The converse is not true.
Lemma 12.9. A countable
4
complete theory is small if and only if it has a
countable saturated model.
Proof. If T has a nite model A, T is small and A is saturated. So we may
assume that T has innite models.
If all types can be realised in a single countable model, there can be at most
countably many types.
If conversely all S
n+1
(T) are at most countable, then over any nelement
subset of a model of T there are at most countably many types. We construct
an elementary chain
A
0
A
1
. . .
of models of T. For A
0
we take any countable model. If A
i
is already con-
structed, we use, Corollary 5.9 and Theorem 6.1.1 to construct a countable
model A
i+1
in such a way that all types over nite subsets of A
i
are realised in
A
i+1
. This can be done since there are only countable many such types. The
union A =
i
A
i
is countable and saturated since every type over a nite
subset B of A is realised in A
i+1
if B A
i
.
Theorem 12.10 (Vaught). A countable complete theory cannot have exactly
two countable models.
Proof. We can assume that T is small and not
0
categorical. We will show that
T has at least three nonisomorphic countable models. T has an saturated
4
The statement is true even for uncountable L.
countable model A and there is a nonisolated type p(x), which can be omitted
in a countable model B. Let p(x) be realised in A by a. Since Th(A, a) is not
0
categorical, Th(A, a) has a countable model (C, c) which is not saturated.
C is not saturated and therefore not isomorphic to A. But C realises p(x) and
is therefore not isomorphic to B.
Exercise 12.5 shows, that for any n ,= 2, n there is a countable com-
plete theory with exactly n countable models. Vaughts Conjecture states
that if a complete countable theory has fewer than continuum many countable
non-isomorphic models, the number of countable models is at most countable
(see [43] for a survey on what is currently known).
Exercise 12.1. 1. If T is
0
categorical, then in any model M the algebraic
closure of a nite set is nite (see Denition p. 94). In particular, M is
locally nite, i.e., any substructure generated by a nite subset is nite.
(In manysorted structures we mean that in each sort the trace of the
algebraic closure is nite.)
2. There is no
0
categorical theory of elds, i.e., if T is a complete L
R
theory containing T
F
, then T is not
0
categorical.
Exercise 12.2.
A theory T is small exactly if T has at most countably many completions, each
of which is small.
Exercise 12.3.
Show that T is
0
categorical if and only if S
n
(T) is nite for all n.
Exercise 12.4.
Write down a theory with exactly two countable models.
Exercise 12.5.
Show that for every n > 2 there is a countable complete theory with exactly n
countable models. (Consider (Q, <, P
0
, . . . , P
n2
, c
0
, c
1
, . . .), where the P
i
form
a partition of Q into dense subsets and the c
i
are an increasing sequence of
elements of P
0
.)
Exercise 12.6.
Give an example of an uncountable complete theory with exactly one countable
model which does not satisfy the conclusion of Theorem 12.1.
Exercise 12.7.
Suppose M is countable and
0
categorical. Show that if X M
n
is invariant
under all automorphisms of M, then X is denable.
Exercise 12.8.
Let M be a structure and assume that for some n only nitely many n-types
are realised in M. Then any structure elementarily equivalent to M satises
exactly the same n-types.
Exercise 12.9.
If A is saturated, all ntypes over nite sets are realised. More generally
prove the following: If A is saturated i.e. if all 1-types over sets of cardinality
less than are realised in A, then the same is true for all n-types. See also
Exercise 23.5.
Exercise 12.10.
Show:
1. The theory of (R, 0, +) has exactly two 1-types but
0
many 2-types.
2. The theory of (R, 0, +, <) has exactly three 1-types but 2
0
many 2-types.
13 The amalgamation method
In this section we will present one of the main methods to construct new and
interesting examples of rst order structures. It goes back to Frass, but has
more recently been modied by Hrushovski [24]. We here focus mainly on the
0
categorical examples and return to the fancier version in Section 44.
Denition 13.1. For any language L, the skeleton /
5
of an Lstructure M
is the class of all nitely generated Lstructures which are isomorphic to a
substructure of M. We say that an Lstructure Mis /saturated if its skeleton
is / and if for all A, B in / and all embeddings f
0
: A M and f
1
: A B
there is an embedding g
1
: BM with f
0
= g
1
f
1
.
Theorem 13.2. Let L be a countable language. Any two countable /-saturated
structures are isomorphic.
Proof. Let M and N be countable Lstructures with the same skeleton /, and
assume that M and N are /saturated. As in the proof of Lemma 12.3 we
construct an isomorphism between M and N as the union of an ascending se-
quence of isomorphisms between nitely generated substructures of M and N.
This can be done because if f
1
: A N is an embedding of a nitely generated
substructure A of M into N, and a is an element of M, then by /saturation
f
1
can be extended to an embedding g
1
: A
N where A
= Aa
M
. Now
interchange the roles of M and N.
Remark 13.3. The proof shows that any countable /-saturated structure M
is ultrahomogeneous i.e. any isomorphism between nitely generated substruc-
tures extends to an automorphism of M.
Theorem 13.4. Let L be a countable language and / a countable class of
nitely generated Lstructures. There is a countable /-saturated Lstructure
M if and only if
a) (Heredity) If A
0
belongs to /, then all elements of the skeleton of A
0
also
belong to /.
b) (Joint Embedding) For B
0
, B
1
/ there are some D / and embeddings
g
i
: B
i
D.
c) (Amalgamation) If A, B
0
, B
1
/ and f
i
: A B
i
, (i = 0, 1) are embed-
dings, there is some D / and two embeddings g
i
: B
i
D such that
g
0
f
0
= g
1
f
1
.
5
This is also called the age of M.
A
B
0
B
1
D
`
`
` `
`
`
` `
f
0
f
1
g
0
g
1
In this case, M is unique up to isomorphism and is called the Frass limit
of /.
Proof. Let / be the skeleton of a countable /-saturated structure M. Clearly,
/ has the Hereditary Property. To see that / has the Amalgamation Property
let A, B
0
, B
1
, f
0
and f
1
be as in c). We may assume that B
0
M and
f
0
is the inclusion map. Furthermore we can assume A B
1
and that f
1
is
the inclusion map. Now the embedding g
1
: B
1
M is the extension of the
isomorphism f
0
: A f
0
(A) to B
1
and satises f
0
= g
1
f
1
. For D we choose
a nitely generated substructure of M which contains B
0
and the image of g
1
.
For g
0
: B
0
D take the inclusion map. The Joint Embedding Property is
proved similarly.
For the converse assume that / has properties a), b), and c). Choose an
enumeration (B
i
)
i
of all isomorphism types in /. We construct M as the
union of an ascending chain
= M
0
M
1
. . . M
of elements of /. Suppose that M
i
is already constructed. If i = 2n is even, we
choose M
i+1
as the top of the diagram
M
i
B
n
M
i+1
`
`
` `
g
0
g
1
where we can assume that g
0
is the inclusion map. If i = 2n + 1 is odd, let A
and B from / and two embeddings f
0
: A M
i
and f
1
: A B be given. We
construct M
i+1
using the diagram
A
M
i B
M
i+1
`
`
` `
`
`
` `
f
0
f
1
g
0
g
1
To ensure that M is /-saturated we do the following. Assume that we
have A, B / and embeddings f
0
: A M and f
1
: A B. For large j the
image of f
0
will be contained in M
j
. During the construction of the the M
i
, in
order to guarantee the /-saturation of M, we have to ensure that eventually,
for some odd i j, the embeddings f
0
: A M
i
and f
1
: A B were used
in the construction of M
i+1
. This can be done since for each j there are up
to isomorphism at most countably many possibilities. Thus there exists an
embedding g
1
: BM
i+1
with f
0
= g
1
f
1
.
Clearly, / is the skeleton of M: the nitely generated substructures of M are
the substructures of the M
i
. Since the M
i
belong to /, their nitely generated
substructures also belong to /. On the other hand each B
n
is isomorphic to a
substructure of M
2n+1
.
Uniqueness follows from Theorem 13.2
For nite relational languages L, any non-empty nite subset is itself a (-
nitely generated) substructure. For such languages, the construction yields
0
categorical structures. We now take a closer look at

0
categorical theories with
quantier elimination in a nite relational language.
Remark 13.5. A complete theory T in a nite relational language with quan-
tier elimination is
0
categorical. So all its models are homogeneous by
Remarks 12.4 and 12.5.
Proof. For every n, there is only a nite number of nonequivalent quantier
free formulas (x
1
, . . . , x
n
). If T has quantier elimination, this number is also
the number of all formulas (x
1
, . . . , x
n
) modulo T and so T is
0
categorical
by Theorem 12.1.
Clearly, if a theory has quantier elimination, any isomorphism between
substructures is elementary. For relational languages we can say more:
Lemma 13.6. Let T be a complete theory in a nite relational language and
M an innite model of T. The following are equivalent:
a) T has quantier elimination.
b) Any isomorphism between nite substructures is elementary.
c) The domain of any isomorphism between nite substructures can be extended
to any further element.
Proof. a) b) is clear.
b)a): If any isomorphism between nite substructures of Mis elementary,
all ntuples a which satisfy in M the same quantier free type
tp
qf
(a) = (x) [ M[= (a), (x) quantier free
satisfy the same simple existential formulas. We will show from this that every
simple existential formula (x
1
, . . . , x
n
) = y (x
1
, . . . , x
n
, y) is, modulo T,
equivalent to a quantier free formula. Let r
1
(x), . . . , r
k1
(x) be the quanti-
er free types of all ntuples in M which satisfy (x). Let
i
(x) be equivalent
to the conjunction of all formulas from r
i
(x). Then
T x
_
(x)
i<k
i
(x)
_
.
a) c): T is
0
categorical and hence all models are homogeneous. Since
any isomorphism between nite substructures is elementary by the equivalence
of a) and b) the claim follows.
c) b): If the domain of any nite isomorphism can be extended to any
further element, it is easy to see that every nite isomorphism is elementary.
We have thus established the following:
Theorem 13.7. Let L be a nite relational language and / a class of nite
Lstructures. If the Frass limit of / exists, its theory is
0
categorical and
Example:
The class of nite linear orders obviously has the Amalgamation Property. Their
Frass limit is the dense linear order without endpoints.
Exercise 13.1.
Prove Remark 13.3.
Exercise 13.2.
Let / be the class of nite graphs. Show that its Frass limit is the countable
random graph. This yields another proof that the theory of the random graph
14 Prime models
Some, but not all, theories have models which are smallest in the sense that they
elementarily embed into any other model of the theory. For countable complete
theories these are the models realizing only the necessary types. If they exist,
they are unique and -homogeneous.
In this section - unless explicitly stated otherwise - we let T be a
countable complete theory with innite models.
Denition 14.1. Let T be a countable theory with innite models, not neces-
sarily complete.
1. A
0
is a prime model of T if A
0
can be elementarily embedded into all
models of T.
2. A structure A is called atomic if all ntuples a of elements of A are atomic.
This means that the types tp(a) = tp(a/) are isolated in S
A
n
().
Prime models need not exist, see the example on p. 74. By Corollary 11.7,
a tuple a is atomic if and only if it satises a complete formula. For the termi-
nology cf. Lemma 11.6.
Since T has countable models, prime models must be countable and since
nonisolated types can be omitted in suitable models by Theorem 10.2, only
isolated types can be realised in prime models. Thus, one direction of the
following theorem is clear:
Theorem 14.2. A model of T is prime if and only if it is countable and atomic.
Proof. As just noted, a prime model has to be countable and atomic. For the
converse let M
0
be a countable and atomic model of T and M any model of T.
We construct an elementary embedding of M
0
to Mas a union of an ascending
sequence of elementary maps
f : A B
between nite subsets A of M
0
and B of M. We start with the empty map,
which is elementary since M
0
and M are elementarily equivalent.
It is enough to show that every f can be extended to any given Aa. Let p(x)
be the type of a over A and f(p) the image of p under f (cf. Lemma 11.5). We
will show that f(p) has a realisation b M. Then f a, b is an elementary
extension of f.
Let a be a tuple which enumerates the elements of A and (x, x) an L
formula which isolates the type of aa. Then p is isolated by (x, a): clearly
(x, a) tp(a/a) and if (x, a) tp(a/a), we have (x, y) tp(a, a). This
implies M
0
[= x, y ((x, y) (x, y)) and M[= x((x, a) (x, a)).
Thus f(p) is isolated by (x, f(a)) and, since (x, f(a)) can be realised in M,
so can be f(p).
Theorem 14.3. All prime models of T are isomorphic.
Proof. Let M
0
and M
0
be two prime models. Since prime models are atomic,
elementary maps between nite subsets of M
0
and M
0
can be extended to all
nite extensions. Since M
0
and M
0
are countable, it follows exactly as in the
proof of Lemma 12.3 that M
0
and M
0
are isomorphic.
The previous proof also shows the following:
Corollary 14.4. Prime models are -homogeneous.
Proof. Let M
0
be prime and a any tuple of elements from M
0
. By Theorem 14.2,
(M
0
, a) is a prime model of its theory. The claim follows now from Theorem 14.3.
Denition 14.5. The isolated types are dense in T if every consistent L
formula (x
1
, . . . , x
n
) belongs to an isolated type p(x
1
, . . . , x
n
) S
n
(T).
Remark 14.6. By Corollary 11.7 this denition is equivalent to asking that
every consistent Lformula (x
1
, . . . , x
n
) contains a complete formula (x
1
, . . . , x
n
):
T x ((x) (x))
Theorem 14.7. T has a prime model if and only if the isolated types are dense.
Proof. Suppose T has a prime model M(so Mis atomic by 14.2). Since consis-
tent formulas (x) are realised in all models of T, (x) is realised by an atomic
tuple a and (x) belongs to the isolated type tp( a).
For the other direction notice that a structure M
0
is atomic if and only if for
all n the set
n
(x
1
, . . . , x
n
) = (x
1
, . . . , x
n
) [ (x
1
, . . . , x
n
) complete
is not realised in M
0
. Hence, by Corollary 10.3, it is enough to show that the
n
are not isolated in T. This is the case if and only if for every consistent
(x
1
, . . . , x
n
) there is a complete formula (x
1
, . . . , x
n
) with T , x ((x)
(x)). Since (x) is complete, this is equivalent to T x ((x) (x)). We
conclude that
n
is not isolated if and only if the isolated ntypes are dense.
Notice that the last part shows in fact the equivalence directly. (Because if
n
is isolated for some n, then it is realised in every model and no atomic model
can exist.)
Example:
Let L be the language having a unary predicate P
s
for every nite 0 1
sequence s
<
2. The axioms of T
Tree
say that the P
s
, s
<
2, form a binary
decomposition of the universe:
xP
(x)
xP
s
(x)
x ((P
s0
(x) P
s1
(x)) P
s
(x))
x (P
s0
(x) P
s1
(x))
T
Tree
is complete and has quantier elimination. There are no complete formulas
and no prime model.
Denition 14.8. A family of formulas
s
(x), s
<
2, is a binary tree if for
all s
<
2 the following holds:
a) T x
_
(
s0
(x)
s1
(x))
s
(x)
_
b) T x
_
s0
(x)
s1
(x)
_
Theorem 14.9. Let T be a complete theory.
1. If T is small, it has no binary tree of consistent formulas. If T is count-
able, the converse holds as well.
2. If T has no binary tree of consistent formulas, the isolated types are dense.
Proof. 1. Let
_
s
(x
1
, . . . , x
n
)
_
be a binary tree of consistent formulas. Then,
for all

2, the set
_
s
(x)
s
_
is consistent and therefore is contained in some type p
(x) S
n
(T). The p
(x)
are all dierent showing that T is not small. We leave the converse as Exer-
cise 14.1.
2. If the isolated types are not dense, there is a consistent (x
1
, . . . , x
n
) which
does not contain a complete formula. Call such a formula perfect. Since perfect
formulas are not complete, they can be decomposed into disjoint
6
consistent
formulas, which again have to be perfect. This allows us to construct a binary
tree of perfect formulas.
Exercise 14.1.
Countable theories without a binary tree of consistent formulas are small.
Exercise 14.2.
Show that the isolated types being dense is equivalent to the isolated types
being (topologically) dense in the Stone space S
n
(T).
Exercise 14.3.
Let T be the theory of (R, <, Q) where Q is a predicate for the rational numbers.
Does T have a prime model?
6
We call two formulas disjoint if their conjunction is not consistent with T.
Chapter 5
1
We have already seen examples of
0
categorical theories (e.g. the theory of
dense linear orderings without endpoints) and of theories categorical in all in-
nite (e.g. the theory of innite dimensional vector spaces over nite elds)
and all uncountably innite (e.g. the theory of algebraically closed elds of
xed characteristic), respectively.
The aim of this chapter is to understand the structure of
1
-categorical the-
ories and to prove, in Corollary 22.2, Morleys theorem that a countable theory
categorical in some uncountable cardinality is categorical in all uncountable
cardinalities (but not necessarily countably categorical).
As in the case of
0
-categorical theories, we will see that the number of
complete types in an
1
-categorical theory is rather small (the theory is -
stable) albeit not always nite. We will dene a geometry associated to a
strongly minimal set whose dimension determines the isomorphism type of a
model of such a theory. This then implies Morleys theorem.
15 Indiscernibles
In this section we begin with a few facts about indiscernible elements. We will
see that structures generated by them realise only few types.
Denition 15.1. Let I be a linear order and A an Lstructure. A family (a
i
)
iI
of elements
1
of A is called a sequence of indiscernibles if for all Lformulas
(x
1
, . . . , x
n
) and all i
1
< . . . < i
n
and j
1
< . . . < j
n
from I
A [= (a
i1
, . . . , a
in
) (a
j1
, . . . , a
jn
).
1
or, more generally, of tuples of elements, all of the same length
76
CHAPTER 5.
1
CATEGORICAL THEORIES 77
If two of the a
i
are equal, all a
i
are the same. Therefore it is often assumed
that the a
i
are distinct.
Sometimes sequences of indiscernibles are also called order indiscernible to
distinguish them from totally indiscernible sequences in which the ordering of the
index set does not matter. However, in stable theories, these notions coincide.
So if nothing else is said, indiscernible elements will always be order indiscernible
in the sense just dened.
Denition 15.2. Let I be an innite linear order and 1 = (a
i
)
iI
a sequence
of k-tuples in M, A M. The Ehrenfeucht-Mostowski-type EM(1/A) of 1
over A is the set of L(A)formulas (x
1
, . . . , x
n
) with M [= (a
i1
, . . . , a
in
) for
all i
1
< . . . < i
n
I, n < .
Lemma 15.3 (The Standard Lemma). Let I and J be two innite linear orders
and 1 = (a
i
)
iI
a sequence of elements of a structure M. Then there is a struc-
ture N M with an indiscernible sequence (b
j
)
jJ
realizing the Ehrenfeucht-
Mostowski-type EM(1) of 1.
Corollary 15.4. Assume that T has an innite model. Then, for any linear
order I, T has a model with a sequence (a
i
)
iI
of distinct indiscernibles.
For the proof of the Standard Lemma we need Ramseys Theorem. Let [A]
n
denote the set of all nelement subsets of A.
Theorem 15.5 (Ramsey). Let A be innite and n . Partition the set of
n-element subsets [A]
n
into subsets C
1
, . . . C
k
. Then there is an innite subset
of A whose nelement subsets all belong to the same subset C
i
.
Proof. Thinking of the partition as a colouring on [A]
n
, we are looking for an
innite subset B of A such that [B]
n
is monochromatic. We prove the theorem
by induction on n. For n = 1, the statement is evident from the pigeonhole
principle. Assuming the theorem is true for n, we now prove it for n + 1.
Let a
0
A. Then any colouring of [A]
n+1
induces a colouring of the n-element
subsets of A
= Aa
0
: just colour x [A
]
n
by the colour of a
0
x [A]
n+1
.
By the induction hypothesis, there exists an innite monochromatic subset B
1
of
A
in the induced colouring. Thus, all the (n+1)-element subsets of A consisting

of a
0
and n elements of B
1
have the same colour. Now pick any a
1
B
1
. By the
same argument we obtain an innite subset B
2
of B
1
with the same properties.
Inductively, we thus construct an innite sequence A = B
0
B
1
B
2
. . .,
and elements a
i
B
i
B
i+1
such that the colour of each (n+1)-element subset
a
i(0)
, a
i(2)
, . . . , a
i(n)
with i(0) < i(2) < . . . < i(n) depends only on the value
of i(0). Again by the pigeonhole principle there are innitely many values of
i(0) for which this colour will be the same. These a
i(0)
s then yield the desired
monochromatic set.
CHAPTER 5.
1
Proof of 15.3. Choose a set C of new constants with an ordering isomorphic to
J. Consider the theories
T
= (c) [ (x) EM(1) (1)

T
= (c) (d) [ c, d C (2)

Here the (x) are Lformulas and c, d tuples in increasing order. We have to
show that T T
is consistent. It is enough to show that

T
C0,
= T (c) T
[ c C
0

_
(c) (
d)
(x) , c,

d C
0
_
is consistent for nite sets C
0
and . We can assume that the elements of
are formulas with free variables x
1
, . . . , x
n
and that all tuples c and

d have the
same length n.
For notational simplicity we assume that all a
i
are dierent. So we may
consider A = a
i
[ I as an ordered set. We dene an equivalence relation on
[A]
n
by
a b M[= (a) (b) for all (x
1
, . . . , x
n
)
where a, b are tuples in increasing order. Since this equivalence relation has at
most 2
||
many classes, by Ramseys Theorem there is an innite subset B of
A with all nelement subsets in the same equivalence class. We interpret the
constants c C
0
by elements b
c
in B ordered in the same way as the c. Then
(M, b
c
)
cC0
is a model of T
C0,
.
Lemma 15.6. Assume L is countable. If the Lstructure M is generated by
a wellordered sequence (a
i
) of indiscernibles, then M realises only countably
many types over every countable subset of M.
Proof. If A = a
i
[ i I, then every element b of M has the form b = t(a),
where t is an Lterm and a is a tuple from A.
Consider a countable subset S of M. Write
S = t
M
n
(a
n
) [ n .
Let A
0
= a
i
[ i I
0
be the (countable) set of elements of A which occur in
the a
n
. Then every type tp(b/S) is determined by tp(b/A
0
) since every L(S)
formula
_
x, t
M
n1
(a
n1
_
, . . .)
can be replaced by the L(A
0
)formula (x, t
n1
(a
n1
), . . .).
Now the type of b = t(a) over A
0
depends only on t(x) (countably many pos-
sibilities) and the type tp(a/A
0
). Write a = a
i
for a tuple

i from I. Since the
a
i
are indiscernible, the type depends only on the quantier free type tp
qf
(
i/I
0
)
in the structure (I, <). This type again depends on tp
qf
(
i) (nitely many pos-

sibilities) and on the types p(x) = tp
qf
(i/I
0
) of the elements i of

i. There are
three kinds of such types:
CHAPTER 5.
1
1. i is bigger than all elements of I
0
2. i is an element i
0
of I
0
3. For some i
0
I
0
, i is smaller than i
0
but bigger than all elements of
j I
0
[ j < i
0
.
There is only one type in the rst case, in the other cases the type is determined
by i
0
. This results in countably many possibilities for each component of

i.
Denition 15.7. Let L be a language. A Skolem theory T
Sk
(L) is a theory in
a bigger language L
Sk
with the following properties:
a) T
Sk
(L) has quantier elimination.
b) T
Sk
(L) is universal.
c) Every Lstructure can be expanded to a model of T
Sk
(L).
d) [L
Sk
[ max[L[,
0
Theorem 15.8. Every L has a Skolem theory.

Proof. We dene an ascending sequence of languages
L = L
0
L
1
L
2
. . . ,
by introducing for every quantier free L
i
formula (x
1
, . . . , x
n
, y) a new n
place Skolem function
2
f
and dening L
i+1
as the union of L
i
and the set of
these function symbols. L
Sk
is the union of all L
i
. We set
T
Sk
=
_
x
_
y (x, y) (x, f
(x))
_
(x, y) q.f. L
Sk
formula
_
.
Corollary 15.9. Let T be a countable theory with an innite model and let
be an innite cardinal. Then T has a model of cardinality which realises only
countably many types over every countable subset.
Proof. Consider the theory T
= T T
Sk
(L). T
is countable, has an innite

model and quantier elimination.
Claim: T
is equivalent to a universal theory.

Proof: Modulo T
Sk
(L) every axiom of T is equivalent to a quantier free
L
Sk
sentence
. Therefore T
is equivalent to the universal theory
[
T T
Sk
(L).
2
If n = 0, f is a constant.
CHAPTER 5.
1
Let I be a well-ordering of cardinality and N
a model of T
with indis-
cernibles (a
i
)
iI
. The claim implies that the substructure M
generated by the
a
i
is a model of T
. M
has cardinality . Since T
has quantier elimina-

tion, M
is an elementary substructure of N
and (a
i
) is indiscernible in M
.
By Lemma 15.6, there are only countably many types over every countable set
realised in M
. The same is then true for the reduct M= M
L.
Exercise 15.1.
A sequence of elements in (Q, <) is indiscernible if and only if it is either con-
stant, strictly increasing or strictly decreasing.
Exercise 15.2.
Prove Ramseys Theorem 15.5 by induction on n similarly to the proof of C 10.2
using a non-principal ultralter on A. (For ultralters see Exercise 2.4.)
CHAPTER 5.
1
16 stable Theories
In this section we x a complete theory T with innite models. In the previ-
ous section we saw that we may add indiscernible elements to a model without
changing the number of realised types. We will now use this to show that
1
-
categorical theories have a small number of types, i.e. they are -stable. Con-
versely, with few types it is easier to be saturated and since saturated structures
are unique we nd the connection to categorical theories.
Denition 16.1. Let be an innite cardinal. T is stable if in each model
of T, over every set of parameters of size at most , there are at most many
1-types, i.e.
[A[ [ S(A)[ .
Example 16.2 (Algebraically closed elds). The theories T
ACFp
for p a prime
or 0 are stable for all .
Note that by Theorem 16.5 below it would suce to prove that the theories
T
ACFp
are stable. The converse holds as well: any innite -stable eld is in
fact algebraically closed (see [34]).
Proof. Let K be a subeld of an algebraically closed eld. By quantier elimi-
nation the type of an element a over K is determined by the isomorphism type
of the extension K[a]/K. If a is transcendental over K, K[a] is isomorphic to
the polynomial ring K[X]. If a is algebraic with minimal polynomial f K[X],
K[a] is isomorphic to K[X]/(f). So there is one more 1type over K than there
are irreducible polynomials.
Exercise 16.5 shows that in stable theories there are only many ntypes
over a set of cardinality . This has a nice algebraic reason in algebraically closed
elds: the isomorphism type of K[a
1
, . . . , a
n
]/K is determined by the vanishing
ideal P of a
1
, . . . , a
n
(see Lemma B6.6). By Hilberts Basis Theorem, P is
nitely generated. So, if K has cardinality , the polynomial ring K[X
1
, . . . , X
n
]
has only many ideals.
Theorem 16.3. A countable theory T which is categorical in an uncountable
cardinal is stable
3
.
Proof. Let N be a model and A N countable with S(A) uncountable. Let
(b
i
)
iI
be a sequence of
1
many elements with pairwise distinct types over
A. (Note that we can assume that all types over A are realised in N.) We
choose rst an elementary substructure M
0
of cardinality
1
which contains A
and all b
i
. Then we choose an elementary extension M of M
0
of cardinality .
M is a model of cardinality which realises uncountably many types over the
3
stable and
0
stable are synonymous.
CHAPTER 5.
1
countable set A. By Corollary 15.9, T has another model in which this is not
the case. So T cannot be categorical.
Denition 16.4. T is totally transcendental if T has no model Mwith a binary
tree of consistent L(M)formulas.
Theorem 16.5.
1. stable theories are totally transcendental.
2. Totally transcendental theories are stable for all [T[.
It follows that a countable theory T is stable if and only if it is totally
transcendental.
Proof. 1. Let M be a model with a binary tree of consistent L(M)formulas.
The set A of parameters which occur in the trees formulas is countable but
S(A) has cardinality 2
0
. (cf. Theorem 14.9).
2. Assume that there are more than many types over some set A of cardinal-
ity . Let us call an L(A)formula big if it belongs to more than many types
over A and thin otherwise.
By assumption x
.
= x is big. If we can show that each big formula decomposes
into two big formulas, we can construct a binary tree of big formulas, which
nishes the proof.
So assume that is big. Since each thin formula belongs to at most many
types and since there are at most many formulas, there are at most many
types which contain thin formulas. Therefore belongs to two distinct types
p and q which contain only big formulas. If we separate p and q by p and
q, we decompose into the big formulas and .
The following denition generalises the notion of saturation.
Denition 16.6. Let be an innite cardinal. An Lstructure A is saturated
if in A all types over sets of cardinality less than are realised. An innite
structure A is saturated if it is [A[saturated.
Even though we ask for saturation only that 1-types are realised, as in the
-saturated case this easily implies that all n-types are realised as well (cp.
Exercise 12.9).
Lemma 12.3 generalises to sets
Lemma 16.7. Elementarily equivalent saturated structures of the same cardi-
nality are isomorphic.
CHAPTER 5.
1
Proof. Let A and B be elementarily equivalent saturated structures both of
cardinality . We choose enumerations (a
)
<
and (b
)
<
of A and B and
construct an increasing sequence of elementary maps f
: A
. Assume
that the f
are constructed for all < . The union of the f
is an elementary
map f
: A
. The construction will imply that A
and B
have at most
cardinality [[, which is smaller than .
We write = +n (as in p. 214) and distinguish two cases:
n = 2i:
In this case we consider p(x) = tp(a
+i
/A
). Realise f
(p) by b B and dene

f
= f
a
+i
, b.
n = 2i + 1:
Similarly. We nd an extension
f
= f
a, b
+i
.
Then

<
f
is the desired isomorphism between A and B.

Lemma 16.8. If T is stable, then for all regular there is a model of
cardinality which is saturated.
Proof. We construct a continuous elementary chain
M
0
M
1
. . . M
. . . ( < ),
of models of T with cardinality such that all p S(M
) are realised in M
+1
.
This is possible since by stability S(M
) has cardinality . Finally let M be

the union of this chain. Then Mis saturated. In fact, if [A[ < and if a A
is contained in M
(a)
then =
aA
(a) is an initial segment of of smaller
cardinality than . So has an upper bound < . It follows that A M
and all types over A are realised in M

+1
.
Remark 16.9. If T is -stable for a regular cardinal , the previous lemma
yields a saturated model of cardinality . Harnik [18] showed that this holds in
fact for arbitrary . See also Corollary 23.3 for more general constructions.
Theorem 16.10. A countable theory T is categorical if and only if all models
of cardinality are saturated.
Proof. If all models of cardinality are saturated, it follows from Lemma 16.7
that T is categorical.
Assume, for the converse, that T is categorical. For =
0
the theorem
follows from (the proof of) Theorem 12.1. So we may assume that is uncount-
able. Then T is totally transcendental by Theorems 16.3 and 16.5 and therefore
stable by Theorem 16.5.
CHAPTER 5.
1
By Lemma 16.8, all models of T of cardinality are
+
saturated for all < .
i.e. saturated.
Exercise 16.1.
Use Exercise 33.8 to show that a theory with an innite denable linear ordering
(like T
DLO
and T
RCF
) cannot be -stable for any .
Exercise 16.2.
Show that the theory of an equivalence relation with two innite classes has
quantier elimination and is stable. Is it
1
categorical?
Exercise 16.3.
If A is -saturated, then all denable subsets are either nite or have cardinality
at least .
Exercise 16.4.
An Ltheory T is totally transcendental if and only if T L
0
is -stable for all
at most countable L
0
L.
Exercise 16.5.
If T is stable, then in each model of T, over every set of parameters with at
most many elements, there are at most many ntypes.
CHAPTER 5.
1
17 Prime extensions
As with prime models, prime extensions are the smallest ones in the sense of
elementary embeddings. We will see here (and in Sections 39 and 40) that
prime extensions, if they exist, share a number of important properties with
prime models.
Denition 17.1. Let M be a model of T and A M.
1. M is a prime extension of A (or: prime over A) if every elementary map
A N extends to an elementary map MN.
A
M N
`
id
A
2. B M is constructible over A if B has an enumeration
B = b
[ < ,
where each b
is atomic over A B
, with B
= b
[ < .
So M is a prime extension of A if and only if M
A
is a prime model of
Th(M
A
).
Notice the following:
Lemma 17.2. If a model M is constructible over A, then M is prime over A.
Proof. Let (m
)
<
an enumeration of M, such that each m
is atomic over
A M
. Let f : A N be an elementary map. We dene inductively an

increasing sequence of elementary maps f
: A M
+1
N with f
0
= f.
Assume that f
is dened for all < . The union of these f
is an elementary
map f
: AM
N. Since p(x) = tp(a
/AM
) is isolated, f
(p) S(f
(A
M
)) is also isolated and has a realisation b in N. We set f
= f
, b.
Finally, the union of all f
( < ) is an elementary embedding MN.

We will see below that in totally transcendental theories prime extensions
are atomic.
Theorem 17.3. If T is totally transcendental, every subset of a model of T
has a constructible prime extension.
We will see in Section 40 that in totally transcendental theories prime extensions
are unique up to isomorphism (cf. Theorem 14.3).
For the proof we need the following lemma which generalises Theorem 14.7:
CHAPTER 5.
1
Lemma 17.4. If T is totally transcendental, the isolated types are dense over
every subset of any model.
Proof. Consider a subset A of a model M. Then Th(M
A
) has no binary tree of
consistent formulas (in one variable). By Theorem 14.9, the isolated 1types in
Th(M
A
) are dense.
We can now prove Theorem 17.3:
Proof. By Lemma 17.2 it suces to construct an elementary substructure M
0

Mwhich contains A and is contructible over A. An application of Zorns Lemma
gives us a maximal construction (a
)
<
, which cannot be prolonged by an
element a
M A
. Clearly A is contained in A
. We show that A
is the
universe of an elementary substructure M
0
using Tarskis Test. So assume that
(x) is an L(A
)formula and M [= x(x). Since isolated types over A
are
dense by Lemma 17.4, there is an isolated p(x) S(A
) containing (x). Let b

be a realisation of p(x) in M. We can prolong our construction by a
= b, thus
b A
by maximality and (x) is realised in A
.
To prove that in totally transcendental theories prime extensions are atomic,
we need the following:
Lemma 17.5. Let a and b be two nite tuples of elements of a structure M.
Then tp(ab) is atomic if and only if tp(a/b) and tp(b) are atomic.
Proof. First assume that (x, y) isolates tp(a, b). As in the proof of Theo-
rem 14.2, (x, b) isolates tp(a/b) and we claim that x(x, y) isolates p(y) =
tp(b): we have x(x, y) p(y) and if (y) p(y), then M[= x, y ((x, y)
(y)). This means that M[= y (x(x, y) (y)).
Now, conversely, assume that (x, b) isolates tp(a/b) and that (y) isolates
p(y) = tp(b). Then (x, y) (y) isolates tp(a, b). For, clearly, we have (x, y)
(y) tp(a, b). If, on the other hand, (x, y) tp(a, b), then (x, b) belongs to
tp(a/b) and
M[= x((x, b) (x, b)).
Hence
x((x, y) (x, y)) p(y)
and it follows that
M[= y ((y) x((x, y) (x, y))).
Thus M[= x, y ((x, y) (y) (x, y)).
Corollary 17.6. Constructible extensions are atomic.
CHAPTER 5.
1
Proof. Let M
0
be a constructible extension of A and let a be a tuple from
M
0
. We have to show that a is atomic over A. We can clearly assume that
the elements of a are pairwise distinct and do not belong to A. We can also
permute the elements of a so that
a = a
b
for some tuple b A
. Let (x, c) be an L(A
)formula which is complete over

A
and satised by a
. Then a
is also atomic over Abc. Using induction,

we know that bc is atomic over A. By Lemma 17.5 applied to (M
0
)
A
, a
bc is
atomic over A, which implies that a = a
b is atomic over A.
Corollary 17.7. If T is totally transcendental, prime extensions are atomic.
Proof. Let Mbe a model of T and A M. Since A has at least one constructible
extension M
0
and since all prime extensions of A are contained in M
0
4
, all prime
extensions are atomic.
A structure M is called a minimal extension of the subset A if M has no
proper elementary substructure which contains A.
Lemma 17.8. Let M be a model of T and A M. If A has a prime ex-
tension and a minimal extension, they are isomorphic over A, i.e., there is an
isomorphism xing A elementwise.
Proof. A prime extension embeds elementarily in the minimal extension. This
embedding must be surjective by minimality.
Exercise 17.1.
For every (countable) theory T the following are equivalent (cf. Theorem 14.7):
a) Every parameter set has a prime extension. (We say: T has prime exten-
sions.)
b) Over every parameter set A the isolated types are dense in the type space
S
1
(A).
c) Over every countable parameter set A the isolated types are dense in S
1
(A).
Exercise 17.2.
Lemma 17.5 follows from Exercise 11.5 and the following observation: Let
: X Y be a continuous open map between topological spaces. Then a
point x X is isolated if and only if (x) is isolated in Y and x is isolated in
1
((x)).
4
More precisely, they are isomorphic over A to elementary substructures of M
0
.
CHAPTER 5.
1
18 Lachlans theorem
Using the fact (established in Section 16) that uncountably categorical theories
are totally transcendental, we will prove the downward direction of Morleys
theorem. We use Lachlans result that in totally transcendental theories models
have arbitrary large elementary extensions realizing few types:
Theorem 18.1 (Lachlan). ([1] Lemma 10) Let T be totally transcendental and
M an uncountable model of T. Then M has arbitrarily large elementary ex-
tensions which omit every countable set of L(M)formulas which is omitted in
M.
Proof. For the proof, we call an L(M)formula (x) large if its realisation set
(M) is uncountable. Since there is no innite binary tree of large formulas,
there exists a minimal large formula
0
(x). This means that for every L(M)
formula (x) either
0
(x) (x) or
0
(x) (x) is at most countable. Now
it is easy to see that
p(x) = (x) [
0
(x) (x) large
is a type in S(M).
Clearly p(x) contains no formula of the form x
.
= a for a M, so p(x) is
not realised in M. On the other hand, every countable subset (x) of p(x) is
realised in M: since
0
(M) (M) is at most countable for every (x) (x),
the elements of
0
(M) which do not belong to the union of these sets realise
(x).
Let a be a realisation of p(x) in a (proper) elementary extension N. By
Theorem 17.3 we can assume that N is atomic over M a.
Fix b N. We have to show that every countable subset (y) of tp(b/M)
is realised in M.
Let (x, y) be an L(M)formula such that (a, y) isolates q(y) = tp(b/M
a). If b realises an L(M)formula (y), we have N [= y ((a, y) (y)).
Hence the formula
(x) = y ((x, y) (y))

belongs to p(x). Note that y (x, y) belongs also to p(x).
Choose an element a
M which satises
(x) [ y (x, y)
and choose b
M with M[= (a
, b
). Since
(a
) is true in M, (b
) is true
in M. So b
realises (y).
We have shown that Mhas a proper elementary extension which realises no
new countable set of L(M)formulas. By iteration we obtain arbitrarily long
chains of elementary extensions with the same property.
CHAPTER 5.
1
The corollary is the downwards part of Morleys Theorem, p. 76:
Corollary 18.2. A countable theory which is categorical for some uncount-
able , is
1
categorical.
Proof. Let T be categorical and assume that T is not
1
categorical. Then T
has a model Mof cardinality
1
which is not saturated. So there is a type p over
a countable subset of M which is not realised in M. By Theorems 16.3 and 16.5
T is totally transcendental. Theorem 18.1 gives an elementary extension N of
Mof cardinality which omits all countable sets of formulas which are omitted
in M. Thus also p is omitted. Since N is not saturated, T is not categorical,
a contradiction.
Exercise 18.1 (Afshordel).
Prove in a similar way: If a countable theory T is categorical for some un-
countable , it is categorical for every uncountable .
CHAPTER 5.
1
19 Vaughtian Pairs
A crucial fact about uncountably categorical theories is the absence of denable
sets whose size is independent of the size of the model in which they live (cap-
tured in the notion of a Vaughtian pair). In fact, in an uncountably categorical
theory each model is prime over any innite denable subset. This will allow us
in Section 21 to attach a dimension to the models of an uncountably categori-
cal theory. In this section let T be a countable complete theory with
innite models.
Denition 19.1. T has a Vaughtian pair if there are two models M N and
an L(M)formula (x) such that
a) M,= N,
b) (M) is innite,
c) (M) = (N)
If (x) does not contain parameters, we say that T has a Vaughtian pair for
(x).
Remark. Notice that T does not have a Vaughtian pair if and only if every
model M is a minimal extension of (M) A for any formula (x) with para-
meters in A M which denes an innite set in M.
Let N be a model of T where (N) is innite but has smaller cardinality
than N. The LwenheimSkolem Theorem yields an elementary substructure M
of N which contains (N) and has the same cardinality as (N). Then M N
is a Vaughtian pair for (x). The next theorem shows that a converse of this
observation is also true.
Theorem 19.2 (Vaughts Two-cardinal Theorem). If T has a Vaughtian pair,
it has a model M of cardinality
1
with (M) countable for some formula
(x) L(

M).
For the proof of Theorem 19.2 we need the following:
Lemma 19.3. Let T be complete, countable, and with innite models.
1. Every countable model of T has a countable homogeneous elementary
extension.
2. The union of an elementary chain of homogeneous models is homo-
geneous.
3. Two homogeneous countable models of T realizing the same ntypes for
all n < are isomorphic.
CHAPTER 5.
1
Proof. 1. Let M
0
be a countable model of T. We realise the countably many
types
f(tp(a/A)) [ a, A M
0
, A nite, f : A M
0
elementary
in a countable elementary extension M
1
. By iterating this process we obtain an
elementary chain
M
0
M
1
. . . ,
whose union is homogeneous.
2. Clear.
3. Suppose A and B are homogeneous, countable and realise the same n-
types. We show that we can extend any nite elementary map f : a
1
, . . . a
i

b
1
, . . . b
i
; a
j
b
j
to any a AA
i
. Realise the type tp(a
1
, . . . , a
i
, a) by some
tuple b
= b
1
, . . . , b
i+1
in B. Using the -homogeneity of B we may extend the
nite partial isomorphism g = (b
j
, b
j
): 1 j i by (b
i+1
, b) for some b B.
Then f
i+1
= f
i
(a, b) is the required extension. Reversing the roles of B and
A we construct the desired isomorphism.
Proof. (of Theorem 19.2) Suppose that the Vaughtian pair is witnessed (in
certain models) by some formula (x). For simplicity we assume that (x) does
not contain parameters (see Exercise 19.4). Let P be a new unary predicate. It
is easy to nd an L(P)theory T
VP
whose models (N, M) consist of a model N
of T and a subset M dened by the new predicate P which is the universe of
an elementary substructure M which together with N forms a Vaughtian pair
for (x). The Lwenheim Skolem Theorem applied to T
VP
yields a Vaughtian
pair M
0
N
0
for (x) with M
0
, N
0
countable.
We rst construct an elementary chain
(N
0
, M
0
) (N
1
, M
1
) . . .
of countable Vaughtian pairs, with the aim that both components of the union
pair
(N
,
M)
are homogeneous and realise the same ntypes. If (N
i
, M
i
) is given, we rst
choose a countable elementary extension (N
, M
) such that M
realises all n
types which are realised in N
i
. Then we choose as in the proof of Lemma 19.3
(1) a countable elementary extension (N
i+1
, M
i+1
) of (N
, M
) for which N
i+1
and M
i+1
are homogeneous.
It follows from Lemma 19.3 (3) that M and N are isomorphic.
Next we construct a continuous elementary chain
M
0
M
1
. . . M
. . . ( <
1
)
with (M
+1
, M
)

= (N, M) for all . We start with M
0
= M. If M
is con-
structed, we choose an isomorphism MM
and extend it to an isomorphism

CHAPTER 5.
1
N M
+1
(see Lemma 1.8). For a countable limit ordinal , M
is the union
of the M
( < ). M
is isomorphic to M by Lemma 19.3.2 and 19.3.3.

Finally we set
M=
_
<1
M
.
Since the M
are growing, Mhas cardinality

1
while (M) = (M
) = (M
0
)
is countable.
Corollary 19.4. If T is categorical in an uncountable cardinality, it does not
have a Vaughtian pair.
Proof. If T has a Vaughtian pair, then by Theorem 19.2 it has a model M of
cardinality
1
such that for some (x) L(M) the set (M) is countable. On
the other hand, if T is categorical in an uncountable cardinal, it is
1
-categorical
by Corollary 18.2 and by Theorem 16.10, all models of T of cardinality
1
are
saturated. In particular, each formula is either satised by a nite number or
by
1
many elements, a contradiction.
Corollary 19.5. Let T be categorical in an uncountable cardinal, M a model,
(M) innite and denable over A M. Then M is the unique prime
extension of A (M).
Proof. By Corollary 19.4, T does not have a Vaughtian pair, so M is minimal
over A (M). If N is a prime extension over A (M), which exists by
Theorem 17.3, N is isomorphic to M over A (M) by Lemma 17.8.
Denition 19.6. We say that T eliminates the quantier
x, there are in-

nitely many x, if for every Lformula (x, y) there is a nite bound n
such
that in all models M of T and for all parameters a M,
(M, a)
is either innite or has or at most n
elements
Remark. This means that for all (x, y) there is a (y) such that in all models
M of T and for all a M
M[=
x (x, a) M[= (a).

We denote this by
T y
_
x (x, y) (y)
_
.
If n
exists, we can use (y) =

>n
x (x, y) (there are more than n
many
x such that (x, y)). If, conversely, (y) is a formula which is implied by
x (x, y), a compactness argument shows that there must be a bound n
such that
T
>n
x(x, y) (y).
CHAPTER 5.
1
Lemma 19.7. A theory T without Vaughtian pair eliminates the quantier
x.
Proof. Let P be a new unary predicate and c
1
, . . . , c
n
new constants. Let T
be the theory of all L P, c

1
, . . . , c
n
structures
(M, N, a
1
, . . . , a
n
),
where M is a model of T, N the universe of a proper elementary substructure,
a
1
, . . . , a
n
elements of N and (M, a) N. Suppose that the bound n
does
not exist. Then, for any n, there is a model N of T and a N such that (N, a)
is nite, but has more than n elements. Let Mbe a proper elementary extension
of N. Then (M, a) = (N, a) and the pair (M, N, a) is a model of T
. This
shows that the theory
T

>n
x(x, c) [ n = 1, 2, . . .
is nitely satisable. A model of this theory gives a Vaughtian pair for T.
Exercise 19.1.
If T is totally transcendental and has a Vaughtian pair for (x), T has, for all
uncountable , a model of cardinality with countable (M). Prove Corol-
lary 19.4 from this. (Use Theorem 18.1.)
Exercise 19.2.
Show directly (without using Lemma 16.8) that a theory T which is categorical
in some uncountable cardinality, has a model Mof cardinality
1
in which each
L(M)-formula is either satised by a nite number or by
1
many elements.
Exercise 19.3.
Show that the theory T
RG
of the random graph has a Vaughtian pair.
Exercise 19.4.
Let T be a theory, Ma model of T and a M a nite tuple of parameters. Let
q( x) be the type of a in M. Then for new constants c, the L( c)-theory
T(q) = Th(M, a) = T ( c)[ ( x) q( x)
is complete. Show that T is -stable (or without Vaughtian pair etc.) if and
only if T(q) is. For countable languages, this implies that T is categorical in
some uncountable cardinal if and only if T(q) is.
Exercise 19.5.
Assume that T eliminates the quantier
. Then for every formula (x

1
, . . . , x
n
, y)
there is a formula ( y) such that in all models Mof T a tuple

b satises if and
only if M has an elementary extension M
with elements a
1
, . . . , a
n
M
M
such that M
[= (a
1
. . . . , a
n
,
b).
Exercise 19.6.
Let T
1
and T
2
be two model complete theories in disjoint languages L
1
and L
2
.
Assume that both theories eliminate
. Then T
1
T
2
has a model companion.
CHAPTER 5.
1
20 Algebraic formulas
Formulas dening a nite set are called algebraic. In this section we collect a
few facts and a bit of terminology around this concept which will be crucial in
the following sections.
Denition 20.1. Let M be a structure and A a subset of M. A formula
(x) L(A) is called algebraic if (M) is nite. An element a M is algebraic
over A if there is a nite Adenable set which contains a. We call an element
algebraic if it is algebraic over the empty set. The algebraic closure of A, acl(A),
is the set of all elements of M algebraic over A, and A is called algebraically
closed if it equals its algebraic closure.
Remark. Note that the algebraic closure of A does not grow in elementary
extensions of Mbecause an L(A)formula which denes a nite set in Mdenes
the same set in every elementary extension. As a special case we have that
elementary substructures are algebraically closed.
It is easy to see that
(3) [ acl(A)[ max[T[, [A[
(cf. Theorem 6.1).
In algebraically closed elds, an element a is algebraic over A precisely if a
is algebraic (in the eld theoretical sense) over the eld generated by A. This
follows easily from quantier elimination in T
ACF
.
We call a type p(x) S(A) algebraic if (and only if) p contains an algebraic
formula. Any algebraic type p is isolated by an algebraic formula (x) L(A),
namely by any p having the minimal number of solutions in M. This
number is called the degree deg(p) (or multiplicity) of p. As isolated types are
realised in every model, the algebraic types over A are exactly those of the form
tp(a/A) where a is algebraic over A. The degree of a over A deg(a/A) is the
degree of tp(a/A).
Lemma 20.2. Let p S(A) be nonalgebraic and A B. Then p has a
nonalgebraic extension q S(B).
Proof.
q
0
(x) = p(x) (x) [ (x) algebraic L(B)formula
is nitely satisable. For otherwise there are (x) p(x) and algebraic L(B)
formulas
1
(x) . . .
n
(x) with
M[= x((x)
1
(x) . . .
n
(x)).
But then (x) and hence p(x) is algebraic. So we can take for q any type
containing q
0
.
CHAPTER 5.
1
Remark 20.3. Since algebraic types are isolated by algebraic formulas, an easy
compactness argument shows that a type p S(A) is algebraic if and only if
p has only nitely many realisations (namely deg(p) many) in all elementary
extensions of M.
Lemma 20.4. Let M and N be two structures and f : A B an elementary
bijection between two subsets. Then f extends to an elementary bijection between
acl(A) and acl(B).
Proof. Let g : A
a maximal extension of f to two subsets of acl(A) and

acl(B). Let a be an element of acl(A). Since a is algebraic over A
, a is atomic
over A
. We can therefore realise the type g(tp(a/A
)) in N by an element
b acl(B) and obtain an extension g a, b of g. It follows a A
. So
g is dened on the whole of acl(A). Interchanging A and B shows that g is
surjective. (See Lemma 23.9 for an alternative proof.)
The algebraic closure operation will be used to study models of
1
-categorical
theories in further detail.
Denition 20.5. A pregeometry
5
(or matroid) (X, cl) is a set X with a closure
operator cl : P(X) P(X) such that for all A X and a, b X
a) A cl(A)
b) cl(A) is the union of all cl(A
), where the A
range over all nite subsets

of A.
c) cl(cl(A)) = cl(A)
d) exchange property
a cl(Ab) cl(A) b cl(Aa)
A set A is called closed (or cl-closed) if A = cl(A). Note that the closure
cl(A) of A is the smallest clclosed set containing A. So a pregeometry is given
by the system of clclosed subsets.
The operator cl(A) = A for all A X is a trivial example of a pregeometry.
The three standard examples from algebra are vector spaces with the linear
closure operator, for a eld K with prime eld F, the relative algebraic closure
cl(A) = F(A)
alg
K,
6
and for a eld K of characteristic p, the p-closure given
by cl(A) = K
p
(A) (see Appendix C, Remark 8.1).
Lemma 20.6. If X is the universe of a structure, acl has properties a), b) and
c) of a pregeometry.
5
Pregeometries were introduced by van der Waerden (1930) and Whitney (1934).
6
L
alg
denotes the algebraic closure of the eld L.
CHAPTER 5.
1
Proof. Properties a) and b) are clear. For c) assume that c is algebraic over
b
1
, . . . , b
n
and the b
i
are algebraic over A. We have to show that c is algebraic
over A. Choose an algebraic formula (x, b
1
, . . . , b
n
) statised by c and algebraic
L(A)formulas
i
(y) satised by the b
i
. Let (x, b
1
, . . . , b
n
) be satised by
exactly k elements. Then the L(A)formula
y
1
. . . y
n
(
1
(y
1
) . . .
n
(y
n
)
k
z(z, y
1
, . . . , y
n
) (x, y
1
, . . . , y
n
))
is algebraic and realised by c.
Exercise 20.1.
If A M and M is [A[
+
saturated, then p S(A) is algebraic if and only if
p(M) is nite.
Exercise 20.2 (P.M. Neumann).
Let A and B be subsets of M and (c
0
. . . c
n
) a sequence of elements which are
not algebraic over A. If M is suciently saturated, the type tp(c
0
, . . . c
n
/A)
has a realisation which is disjoint from B. (Hint: Use induction on n. Distinguish
between whether or not ci is algebraic over Acn for some i < n.)
CHAPTER 5.
1
21 Strongly minimal sets
Strongly minimal theories dened below turn out to be uncountably categorical:
the isomorphism type of the model is determined by the dimension of an asso-
ciated geometry. While this appears to be a very special case of such theories,
we will see in the next section that we can always essentially reduce to this
situation.
Throughout this section we x a complete theory T with innite
models. For Corollary 21.9 we have to assume T countable.
Denition 21.1. Let M be a model of T and (x) a nonalgebraic
L(M)formula, not necessarily in one variable.
1. The set (M) is called minimal in M if for all L(M)formulas (x) the
intersection (M) (M) is either nite or conite in (M).
2. The formula (x) is strongly minimal if (x) denes a minimal set in
all elementary extensions of M. In this case, we also call the denable
set (M) strongly minimal. A non-algebraic type containing a strongly
minimal formula is called strongly minimal.
3. A theory T is strongly minimal if the formula x
.
= x is strongly minimal.
Clearly, strong minimality is preserved under denable bijections; i.e., if A
and B are denable subsets of M
k
, M
m
dened by and , respectively, such
that there is a denable bijection between A and B, then if is strongly minimal
so is .
Examples:
1. The following theories are strongly minimal, which is easily seen in each
case using quantier elimination.
T
, the theory of innite sets. The sets which are denable over a
parameter set A in a model M are the nite subsets S of A and their
complements M S.
For a eld K, the theory of innite Kvector spaces. The sets den-
able over a set A are the nite subsets of the subspace spanned by A
and their complements.
The theories T
ACFp
of algebraically closed elds of xed character-
istic. The denable sets of any model K are Boolean combinations
of zerosets
a K [ f(a) = 0
of polynomials f(X) K[X]. Zerosets are nite or, if f = 0, all of
K.
CHAPTER 5.
1
2. If K is a model T
ACFp
, for any a, b K, the formula ax
1
+b = x
2
dening
an ane line A in K
2
is strongly minimal as there is a denable bijection
between A and K.
3. For any strongly minimal formula (x), the induced theory T is strongly
minimal. Here, for any model M of T the induced theory is the theory of
(M) with the structure given by all intersections of 0denable subsets
of M
n
with (M)
n
for all n . This theory depends only on T and ,
not on M.
Whether (x, a) is strongly minimal depends only on the type of the parameter
tuple a and not on the actual model: observe that (x, a) is strongly minimal
if and only if for all Lformulas (x, z) the set
(z, a) =
_
>n
x((x, a) (x, z))
>n
x((x, a) (x, z))
n = 1, 2, . . .
_
cannot be realised in any elementary extension. This means that for all (x, z)
there is a bound n
such that
M[= z
_
x((x, a) (x, z))

n
x((x, a) (x, z))

_
.
This is an elementary property of a, i.e. expressible by a rst-order formula. So
it makes sense to call (x, a) a strongly minimal formula without specifying a
model.
Lemma 21.2. If M is saturated, or if T eliminates the quantier
, any
minimal formula is strongly minimal. If T is totally transcendental, every in-
nite denable subset of M contains a minimal set (M).
Proof. If M is -saturated and (x, a) not strongly minimal, then for some
Lformula (x, z) the set
(z, a) is realised in M, so is not minimal.

If on the other hand (x, a) is minimal and T eliminates the quantier
,
then for all Lformulas (x, z)
x((x, a) (x, z))
x((x, a) (x, z))

_
is an elementary property of z.
If
0
(M) does not contain a minimal set, one can construct from
0
(x) a
binary tree of L(M)formulas dening innite subsets of M. This contradicts
stability.
Lemma 21.3. (M) is minimal if and only if there is a unique non-algebraic
type p S(M) containing (x).
Proof. If (M) is minimal, then clearly
p = [ (x) L(M) such that is algebraic
CHAPTER 5.
1
is the unique non-algebraic type in S(M) containing it. If (M) is not minimal,
there is some Lformula with both and non-algebraic. By
Lemma 20.2 there are at least two non-algebraic types in S(M) containing
.
Corollary 21.4. A strongly minimal type p S(A) has a unique non-algebraic
extension to all supersets B of A in elementary extensions of M. Consequently,
the type of n realisations a
1
, . . . a
n
of p with a
i
/ acl(a
1
, . . . a
i1
A), i = 1, . . . n,
is uniquely determined.
Proof. Existence of non-algebraic extensions follows from Lemma 20.2, which
also allows us to assume that B is a model. Uniqueness now follows from
Lemma 21.3 applied to any strongly minimal formula of p. The last sentence
follows by induction.
Theorem 21.5. If (x) is a strongly minimal formula in Mwithout parameters,
the operation
cl : P((M)) P((M))
dened by
cl(A) = acl
M
(A) (M)
is a pregeometry ((M), cl).
Proof. We have to verify the exchange property. For notational simplicity we
assume A = . Let a (M) be not algebraic over and b (M) not algebraic
over a. By Corollary 21.4, all such pairs a, b have the same type p(x, y). Let
A
be an innite set of nonalgebraic elements realising (which exists in an

elementary extension of M) and b
nonalgebraic over A
. Since all a
have the same type p(x, b
) over b
, no a
is algebraic over b
. Thus also a is not

algebraic over b.
The same proof shows that algebraic closure denes a pregeometry on the set
of realizations of a minimal type, i.e. a non-algebraic type p S
1
(A) having a
unique non-algebraic extension to all supersets B of A in elementary extensions
of M. Here is an example to show that a minimal type need not be strongly
minimal:
Let T be the theory of M = (M, P
i
)
i<
in which the P
i
form a proper
descending sequence of subsets. The type p = x P
i
: i < S
1
() is
minimal. If all P
i+1
are coinnite in P
i
, then p does not contain a minimal
formula and is not strongly minimal.
In pregeometries there is a natural notion of independence and dimension
(see p. 233, Def. 8.2), so in light of Theorem 21.5 we may dene the following:
CHAPTER 5.
1
If (x) is strongly minimal without parameters, the dimension of a model
M of T is the dimension of the pregeometry ((M), cl)
dim
(M).
If M is the model of a strongly minimal theory, we just write dim(M).
If (x) is dened over A
0
M, the closure operator of the pregeometry
(M
A0
) is given by
cl(A) = acl
M
(A
0
A) (M)
and
dim
(M/A
0
) := dim
(M
A0
)
is called the dimension of M over A
0
.
Lemma 21.6. Let (x) be dened over A
0
and strongly minimal, and let Mand
N be models containing A
0
. Then there exists an A
0
elementary map between
(M) and (N) if and only if M and N have the same dimension over A
0
.
Proof. An A
0
elementary map between (M) and (N) maps bases to bases,
so one direction is clear.
For the other direction we use Corollary 21.4: if (M) and (N) have the
same dimension over A
0
, let U and V be bases of (M) and (N), respectively,
and let f : U V be a bijection. By Corollary 21.4, f is A
0
elementary and
by Lemma 20.4 f extends to an elementary bijection g : acl(A
0
U) acl(A
0
V ).
Thus, g (M) is an A
0
elementary map from (M) to (N).
We now turn to showing that strongly minimal theories are categorical in
all uncountable cardinals. For reference we rst note the following special cases
of the preceding lemmas:
Corollary 21.7. 1. A theory T is strongly minimal if and only if over every
parameter set there is exactly one nonalgebraic type.
2. In models of a strongly minimal theory the algebraic closure denes a
pregeometry.
3. Bijections between independent subsets of models of a strongly minimal
theory are elementary. In particular, the type of n independent elements
is uniquely determined.
If T is strongly minimal, by the preceding we have
[ S(A)[ [ acl(A)[ + 1.
CHAPTER 5.
1
Strongly minimal theories are therefore stable for all [T[. It follows that
every restriction to a countable language is stable, so T is totally transcenden-
tal (cf. Exercise 16.4). If (M) is conite and N a proper elementary extension
of M, then (N) is a proper extension of (M). Thus strongly minimal theories
have no Vaughtian pairs.
Theorem 21.8. Let T be strongly minimal. Models of T are uniquely deter-
mined by their dimensions. The set of possible dimensions is an end segment
of the cardinals. A model M is saturated if and only if dim(M)
0
. All
models are homogeneous.
Proof. Let M
0
, M
1
be models of the same dimension, and let B
0
, B
1
be bases for
M
0
and M
1
, respectively. Then any bijection f : B
0
B
1
is an elementary map
by Corollary 21.7, which extends to an isomorphism of the algebraic closures
M
0
and M
1
by Lemma 20.4.
The next claim implies that the possible dimensions form an end segment of
the cardinals.
Claim: Every innite algebraically closed subset S of M is the universe of an
elementary substructure.
Proof: By Theorem 4.2 it suces to show that every consistent L(S)formula
(x) can be realised in S. If (M) is nite, all realisations are algebraic over S
and belong therefore to S. If (M) is conite, (M) meets all innite sets.
Let A be a nite subset of M and p the nonalgebraic type in S(A). Thus,
p is realised in M exactly if M ,= acl(A), i.e. if and only if dim(M) > dim(A).
Since all algebraic types over A are always realised in M, this shows that M is
saturated if and only if M has innite dimension.
It remains to show that all models are homogeneous. Let f : A B be an
elementary bijection between two nite subsets of M. By Lemma 20.4, f extends
to an elementary bijection between acl(A) and acl(B). If a M acl(A), then
p = tp(a/A) is the unique nonalgebraic type over A and f(p) is the unique non
algebraic type over B. Since dim(A) = dim(B), the argument in the previous
paragraph shows that f(p) is realised in M.
Corollary 21.9. If T is countable and strongly minimal, it is categorical in all
uncountable cardinalities.
Proof. Let M
1
and M
2
be two models of cardinality >
0
. Choose two bases
B
1
and B
2
of M
1
and M
2
respectively. By p. 94, equation (3), B
1
and B
2
both
have cardinality . Then any bijection f : B
1
B
2
is an elementary map by
Corollary 21.7, which extends to an isomorphism of the algebraic closures M
1
and M
2
by Lemma 20.4.
Exercise 21.1.
If M is minimal and saturated, then Th(M) is strongly minimal.
CHAPTER 5.
1
Exercise 21.2.
Show directly that strongly minimal theories eliminate
(without using Corol-

laries 19.4 and 21.9.)
Exercise 21.3.
A type is minimal if and only if its set of realisations in any model is minimal
(i.e. has no innite and coinnite relatively denable subsets).
Exercise 21.4.
Show that acl denes a pregeometry on (M) if (M) is minimal. In fact the
following is true: If b (M), a acl(Ab), b , acl(Aa), then a acl(A).
Furthermore we have deg(a/A) = deg(a/Ab).
CHAPTER 5.
1
22 The Baldwin-Lachlan theorem
In this section, we present the characterisation of uncountably categorical theo-
ries due to Baldwin and Lachlan [1]. Since this characterisation is independent
of the uncountable cardinal, it implies Morleys Theorem. The crucial point is
the existence of a strongly minimal formula in a totally transcendental the-
ory. By Corollary 19.5, each model Mis prime over the set of realisations (M)
whose dimension determines the isomorphism type of the model.
Theorem 22.1 (BaldwinLachlan). Let be an uncountable cardinal. A count-
able theory T is categorical if and only if T is stable and has no Vaughtian
pair.
Proof. If T is categorical in some uncountable cardinal, then T is -stable by
Theorem 16.3 and has no Vaughtian pair by Corollary 19.4.
For the other direction we rst obtain a strongly minimal formula: since T
is totally transcendental, it has a prime model M
0
. (This follows from Theo-
rems 14.7 and 14.9 or from Theorem 17.3.) Let (x) be a minimal formula in
L(M
0
), which exists by Lemma 21.2. Since T has no Vaughtian pairs,
can be
eliminated by Lemma 19.7 and hence (x) is strongly minimal by Lemma 21.2.
Let M
1
, M
2
be models of cardinality . We may assume that M
0
is an
elementary submodel of both M
1
and M
2
. Since T has no Vaughtian pair, M
i
is a minimal extension of M
0
(M
i
), i = 1, 2. Therefore, (M
i
) has cardinality
and hence (since is uncountable) we conclude that dim
(M
1
/M
0
) = =
dim
(M
2
/M
0
). By Lemma 21.6 there exists an M
0
-elementary map from (M
0
)
to (M
1
), which by Lemma 17.8 extends to an isomorphism fromM
1
to M
2
.
Corollary 22.2 (Morley). Let be an uncountable cardinal. T is
1
categorical
if and only if T is -categorical.
Notice that the proof of Theorem 22.1 shows in fact the following:
Corollary 22.3. Suppose T is
1
categorical, M
1
, M
2
are models of T, a
i
M
i
and (x, a
i
) strongly minimal, i = 1, 2, with tp(a
1
) = tp(a
2
). If M
1
and M
2
have the same respective -dimension, then they are isomorphic.
For uncountable models, the -dimension equals the cardinality of the model,
so clearly does not depend on the realisation of tp(a
i
). We will show in Section 25
that the converse to Corollary 22.3 holds also for countable models, i.e., if
(x, a
0
) is strongly minimal, then the -dimension of M is the same for all a
realizing tp(a
0
). Thus, also the countable models of an uncountably categorical
theory are in one-to-one correspondence with the possible -dimensions.
Chapter 6
Morley Rank
In this chapter we collect a number of further results about totally transcen-
dental theories, in particular we will introduce Morley rank. For convenience,
we will introduce the monster model (for arbitrary theories), a very large, very
saturated, very homogeneous model. From now on, all models we consider will
be elementary submodels of this monster model. We then nish the analysis of
the countable models of uncountably categorical theories.
23 Saturated models and the Monster
The importance of saturated structures was already visible in Section 16 where
we showed that saturated structures of xed cardinality are unique up to iso-
morphism. Saturated structures need not exist (think about why not), but
by considering special models instead, we can preserve many of the important
properties - and prove their existence.
Denition 23.1. A structure M of cardinality is special if M is the
union of an elementary chain M
where runs over all cardinals less than

and each M
is
+
saturated.
We call (M
) a specialising chain.
Remark. Saturated structures are special. If [M[ is regular, the converse is
true.
Lemma 23.2. Let be an innite cardinal [L[. Then every Lstructure M
of cardinality 2
has a
+
saturated elementary extension of cardinality 2
.
Proof. Every set of cardinality 2
has 2
many subsets of cardinality at most

104
CHAPTER 6. MORLEY RANK 105
. This allows us to construct a continuous elementary chain
M= M
0
M
1
. . . M
. . . ( <
+
)
of structures of cardinality 2
such that all p S(A), for A M
, [A[ , are
realised in M
+1
. The union of this chain has the desired properties.
Corollary 23.3. Let > [L[ be an uncountable cardinal. Assume that
(1) < 2

Then every innite Lstructure M of cardinality smaller than has a special
extension of cardinality .
Let be a limit ordinal. Then for any cardinal , =
() satises (1)
and we have cf() = cf() (see p. 214).
The following is a generalisation of Lemma 16.7
Theorem 23.4. Two elementarily equivalent special structures of the same
cardinality are isomorphic.
Proof. The proof is a rened version of the proof of Lemma 16.7. Let A and B
be two elementarily equivalent special structures of cardinality . Let A be the
union of a chain of elementary substructures A
, B the union of B
, where
runs over all cardinals less than and A
and B
are
+
saturated. Lemma 3.2
of the appendix shows that A and B have enumerations (a
)
<
and (b
)
<
such that a
A
||
and b
B
||
. We construct an increasing sequence of
elementary maps f
: A
such that [A
[ [[, A
A
||
, [B
[ [[,
B
B
||
.
Denition 23.5. A structure M is
universal if every structure of cardinality < which is elementarily
equivalent to M can be elementarily embedded into M.
homogeneous if for every subset A of M of cardinality smaller than
and for every a M, every elementary map A M can be extended to
an elementary map A a M.
strongly homogeneous if for every subset A of M of cardinality less than
, every elementary map A M can be extended to an automorphism of
M.
Theorem 23.6. Special structures of cardinality are
+
universal and strongly
cf()homogeneous.
Proof. Let M be a special structure of cardinality . The
+
universality of
M can be proved in the same way as Theorem 23.4. Let A be a subset of M
of cardinality less than cf() and let f : A M an elementary map. Fix a
specialising sequence (M
). For
0
suciently large, M
0
contains A. The
sequence
M
=
_
(M
, a)
aA
, if
0

(M
0
, a)
aA
, if <
0
is then a specialising sequence of (M, a)
aA
. For the same reason (M, f(a))
aA
is special. By Theorem 23.4 these two structures are isomorphic under an au-
tomorphism of M which extends f.
The monster model
Let T be a complete theory with innite models. For convenience, we would
like to work in a very large saturated structure, large enough so that any model
of T can be considered as an elementary substructure. If T is totally transcen-
dental, by Remark 16.9 we can choose such a monster model as a saturated
model of cardinality where is a regular cardinal greater than all the mod-
els we ever consider otherwise. Using Exercise 33.7, this also works for stable
theories and regular with
|T|
= . = (
|T|
)
+
has this properties for any
innite .
In order to construct the monster model C for an arbitrary theory T we
will work in BGC (BernaysGdel + Global Choice). This is a conservative
extension of ZFC (cf. Appendix, Section A) which adds classes to ZFC. Then
being a model of T is interpreted as being the union of an elementary chain of
(set-size) models of T. The universe of our monster model C will be a proper
class:
Theorem 23.7 (BGC). There is a class-size model C of T such that all types
over all subsets of C are realised in C. C is uniquely determined up to isomor-
phism.
Proof. Global choice allows us to construct a long continuous elementary chain
(M
)
On
of models of T such that all types over M
are realised in M
+1
. Let
C be the union of this chain. The uniqueness is proved as in Lemma 16.7.
We call C the monster model of T. Note that Global Choice implies that C
can be well-ordered.
Corollary 23.8.
C is -saturated for all cardinals .
Any model of T is elementarily embeddable in C
Any elementary bijection between two subsets of C can be extended to an
automorphism of C.
We say that two elements are conjugate over some parameter set A if there
is an automorphism of C xing A elementwise and taking one to the other. Note
that a, b C are conjugate over A if and only if they have the same type over A.
We call types p S(A), q S(B) conjugate over D if there is an automorphism f
of C xing D and taking A to B and such that q = (x, f(a)) [ (x, a) p.
Note that strictly speaking Aut(C) is not an object in Bernays-Gdel Set theory
but we will nevertheless use this term as a way of talking about automorphisms.
Readers who mistrust set theory can x a regular cardinal bigger than the
cardinality of all models and parameter sets they want to consider. For C they
may then use a special model of cardinality =
(
0
). This is
+
universal
and strongly homogeneous.
We will use the following convention throughout the rest of this
book:
Any model of T is an elementary substructure of C. We identify models
with their universes and denote them by M, N,. . . .
Parameter sets A, B, . . . are subsets of C
Formulas (x) with parameters dene a subclass F = (C) of C. Two
formulas are equivalent if they dene the same class.
We write [= for C [= .
A set of formulas with parameters from C is consistent if it is realised in
C.
A global type is a type over C.
This convention changes the avour of quite a number of proofs. As an
example look at the following
Lemma 23.9. An elementary bijection f : A B extends to an elementary
bijection between acl(A) and acl(B).
Proof. Extend f to an automorphism f
of C. Clearly f
maps acl(A) to acl(B).

This implies Lemma 20.4 and the second claim in the proof of Theorem 21.6.
Note that by the remark on p. 94 for any model M and any A M the algebraic
closure of A in the sense of M equals the algebraic closure in the sense of C.
Lemma 23.10. Let D be a denable class and A a set of parameters. Then the
following are equivalent:
a) D is denable over A.
b) D is invariant under all automorphisms of C which x A pointwise.
Proof. Let D be dened by , dened over B A. Consider the maps
C

S(B)

S(A),
where (c) = tp(c/B) and is the restriction map. Let Y be the image of D in
S(A). Since Y = [], Y is closed.
Assume that D is invariant under all automorphisms of C which x A point-
wise. Since elements which have the same type over A are conjugate by an
automorphism of C, this means that Dmembership depends only on the type
over A i.e. D = ()
1
(Y ).
This implies that [] =
1
(Y ), or S(A) Y = [], hence S(A) Y is also
closed and we conclude that Y is clopen. By Lemma 11.3 Y = [] for some
L(A)formula . This denes D.
Denition 23.11. The denable closure dcl(A) of A is the set of elements c for
which there is an L(A)-formula (x) such that c is the unique element satisfying
. Elements or tuples a and b are said to be interdenable if a dcl(b) and
b dcl(a).
Corollary 23.12.
1. a dcl(A) if and only if a has only one conjugate over A.
2. a acl(A) if and only if a has nitely many conjugates over A.
Proof. 1. is clear, since a dcl(A) means that a is Adenable. 2. follows
from Remark 20.3, since the realisations of tp(a/A) are exactly the conjugates
of a over A.
Exercise 23.1.
Finite structures are saturated.
Exercise 23.2.
acl(A) is the intersection of all models which contain A.
Exercise 23.3.
Prove Robinsons Joint Consistency Lemma: Extend the complete Ltheory T
to an L
1
theory T
1
and an L
2
theory T
2
such that L = L
1
L
2
. If T
1
and T
2
are both consistent, show that T
1
T
2
is consistent.
Exercise 23.4.
Prove Beths Interpolation Theorem: If
1

2
for L
i
sentences
i
, there
is an L = L
1
L
2
sentence such that
1
and
2
.
Exercise 23.5.
If M is saturated, then over every set of cardinality smaller than every type
in many variables is realised in M.
Exercise 23.6.
Let be an innite cardinal, not smaller than the cardinality of L and M an
Lstructure. Show that the following are equivalent
a) M is saturated
b) M is
+
universal and homogeneous
If max([L[,
0
) < this is also equivalent to
c) M is universal and homogeneous.
Exercise 23.7.
Let be a an uncountable regular cardinal > [L[. Show that
1. Every Lstructure M of cardinality 2
<
has a saturated elementary
extension of cardinality 2
<
.
2. Assume that < implies 2
. Then every Lstructure M of cardi-

nality has a saturated elementary extension of cardinality .
Exercise 23.8.
Let P
(N) be the set of all nite subsets of N. Show that the theory of
(P
(N), ) has a saturated model of cardinality if and only if is regular

and < implies 2
.
Exercise 23.9.
A type-denable class is the class of all realisations of a set of formulas. Show
that a type-denable class is invariant under all automorphisms of C which x
A pointwise if and only if it can be dened by a set of L(A)formulas. (Use
Exercise 11.2.a and the proof of Lemma 23.10.)
Exercise 23.10.
Let A be contained in B. Show that the following are equivalent:
1. B dcl(A).
2. Every type over A extends uniquely to B.
Exercise 23.11.
1. b is in the denable closure of a if and only if there is a 0denable class
D containing a and a 0denable map D C which maps a to b.
2. a and b are interdenable if and only if a and b are contained in 0denable
classes D and E and there is a 0denable bijection between D and E which
maps a to b.
Exercise 23.12.
Let K be a model of T
ACF0
, T
ACFp
for p > 0, of T
RCF
or of T
DCF0
and let A
be a (dierential) subeld of K. Prove that the denable closure of A is
1. (T
ACF0
) A itself,
2. (T
ACFp
) the perfect hull (see Denition B6.8) of A,
3. (T
RCF
) the relative algebraic closure of A,
4. (T
DCF0
) A itself.
Exercise 23.13.
Use Exercise 23.12 to show the following:
1. Let K be a model of T
ACF0
and f : K
n
K a denable function. Then
K
n
can be decomposed into a nite number of denable subsets X
i
such
that, on each X
i
, f is given by a rational function.
2. The same is true for models of T
ACFp
, p > 0. But on each X
i
, f is of the
form h
p
m
, for some rational function h.
3. In models of T
DCF0
, f is given on each X
i
by a dierential rational func-
tion.
Exercise 23.14 (P. Neumann).
Let X be an innite set, G Sym(X), and B X nite. Suppose that the
orbit of each of the elements c
0
, . . . c
n
X under G is innite. Then there is
some g G with g(c
i
) / B for i = 0, . . . n. (Hint: Proceed as in Exercise 20.2. In
fact, Exercise 20.2 follows from this by compactness.)
Exercise 23.15 (B. Neumann).
Let G be a group (not necessarily abelian), H
0
, . . . , H
n
, subgroups of innite in-
dex. Show that G is not a nite union of cosets of the H
i
(but see Exercise 23.16,
which is easy). Deduce Lemma 9.9 from this.
Exercise 23.16 (B. & P. Neumann).
Deduce Exercise 23.14 from 23.15 and conversely.
24 Morley Rank
The Morley Rank is a rather natural notion of dimension on the formulas of a
theory T or, equivalently, on the denable subsets of the monster model, dened
inductively very much like the dimension of algebraic sets. It is ordinal valued
for all consistent formulas if and only if T is totally transcendental. In this
section, we let T be a complete (possibly uncountable) theory.
We now dene the Morley rank MR for formulas (x) with parameters in
the monster model. We begin by dening
1
the relation MR by induction
on the ordinal .
MR 0 if is consistent;
MR + 1 if there is an innite family (
i
(x)) of formulas (in the
same variables x) which imply , are pairwise incon-
sistent and such that MR
i
for all i;
MR (for a limit ordinal ) if MR for all < .
Denition 24.1. For a given formula there are three cases
1. If there is no with MR , we put MR = .
2. MR for all , we put MR = .
3. Otherwise, by the denition of MR for limit ordinals , there is a
maximal with MR , and we set MR = max [ MR .
It is easy to see by induction on that the relation MR implies the
relation MR for . It follows from this that indeed the Morley rank
of is at least if and only if the relation MR holds.
2
Note that
MR = is inconsistent
MR = 0 is consistent and algebraic.
If a formula has ordinal valued Morley rank, we also say that this formula has
Morley rank. The Morley rank MR(T) of T is the Morley rank of the formula
x
.
= x. The Morley rank of a formula (x, a) only depends on (x, y) and the
type of a. It follows that if a formula has Morley rank, then it is less than
1
Set-theoretically we dene a function which maps each to a class of formulas. In BG
one cannot in general dene functions from ordinals to classes by a recursion scheme. The
more conscienteous reader should therefore use a dierent denition: Dene for each set A
the relation MR
A
, where only formulas with parameters from A are used. And say
MR if MR
A
for some (large enough) A. By Exercise 24.3, if is dened over M
and M is saturated, we have MR = MR
M
.
2
Here, of course is considered as being smaller than all ordinals and to be bigger
than all ordinals.
(2
|T|
)
+
. We will see in Exercise 24.6 that in fact all ordinal ranks are smaller
that [T[
+
.
Remark. Clearly, if implies , then MR MR. It is also clear from
the denition that if has rank < , then for every < there is a formula
which implies and has rank .
The next lemma expresses the fact that the formulas of rank less than
form an ideal in the Boolean algebra of equivalence classes of formulas.
Lemma 24.2.
MR( ) = max(MR, MR)
Proof. By the previous remark, we have MR( ) max(MR, MR). For
the other inequality we show by induction on that
MR( ) + 1 implies max(MR, MR) + 1.
Let MR( ) + 1. Then there is an innite family of formulas (
i
)
which imply , are pairwise inconsistent and such that MR
i
. By
induction hypothesis we have for all i MR(
i
) or MR(
i
) .
If the rst case holds for innitely many i, then MR + 1. Otherwise
MR + 1.
We call and equivalent,

,
if their symmetric dierence . has rank less than . By our previous
considerations it is clear that -equivalence is in fact an equivalence relation.
We call a formula strongly minimal if it has rank and for any formula
implying either or , has rank less than , (equivalently, if every
is equivalent to or to ). In particular, is 0strongly minimal if
and only if is realised by a single element and is 1strongly minimal if and
only if is strongly minimal.
Lemma 24.3. Each formula of rank < is equivalent to a disjunction
of nitely many pairwise disjoint strongly minimal formulas
1
, . . . ,
n
, the
strongly minimal components (or just components) of . The components
are uniquely determined up to equivalence.
Proof. Suppose is a formula of rank without such a decomposition. Then
can be written as the disjoint disjunction of a formula
1
of rank and another
formula
1
of rank not having such a decomposition. Inductively there are
formulas =
0
,
1
. . . of rank and
1
,
2
, . . . so that
i
is the disjoint union
of
i+1
and
i+1
. But then the rank of would be greater than .
To see the uniqueness of this decomposition, let be an strongly minimal
formula implying and let
1
, . . . ,
n
be the strongly minimal components.
Then can be decomposed as
i
, with, say,
i0
, equivalent to . So
up to equivalence the components of are exactly the strongly minimal
formulas implying .
Denition 24.4. For a formula of Morley rank , the Morley degree MD()
is the number of its strongly minimal components.
The Morley degree is not dened for inconsistent formulas or formulas not
having Morley rank. The Morley degree of an algebraic formula equals its
degree as dened on p. 94. Strongly minimal formulas are exactly the formulas
of Morley rank and Morley degree 1. As with strongly minimal formulas it is
easy to see that Morley rank and degree are preserved under denable bijections.
Dening MD
() as the Morley degree for formulas of rank , as 0 for

formulas of smaller rank and as for formulas of higher rank, we obtain the
following:
Lemma 24.5. If is the disjoint union of
1
and
2
, then
MD
() = MD
(
1
) + MD
(
2
).
Theorem 24.6. T is totally transcendental if and only if each formula has
Morley rank.
Proof. Since there are not arbitrarily large ordinal Morley ranks, each formula
without Morley rank can be decomposed into two disjoint formulas without
Morley rank, yielding a binary tree of consistent formulas, a contradiction.
For the other direction let (
s
)
s2
< be a binary tree of consistent formulas.
Choose
s
so that its rank is minimal and (among the formulas of rank )
of minimal degree. Then both
s0
and
s1
have rank and therefore smaller
degree than , a contradiction.
A group is said to have the the descending chain condition (dcc) on denable
subgroups, if there is no innite properly descending chain H
0
H
1
H
2
. . .
of denable subgroups.
Remark 24.7. A totally transcendental group has the descending chain con-
dition on denable subgroups.
Proof. If H is a denable proper subgroup of a totally transcendental group G,
then either the Morley rank or the Morley degree of H must be less than that of
G since any coset of H has the same Morley rank and degree as H. Therefore
the claim follows from the fact that the ordinals are well-ordered.
The previous remark is a crucial tool in the theory of totally transcendental
groups. For example, it immediately implies that (Z, +) is not totally transcen-
dental.
Corollary 24.8. The theory of separably closed elds of degree of imperfection
e > 0 is not totally transcendental.
Proof. The subelds K K
p
K
p
2
K
p
3
. . . form an innite denable
chain of properly descending (additive) subgroups. In fact we will see later that
the proof shows that that separably closed elds are not even superstable (see
Exercise 37.8.)
Denition 24.9. The Morley rank MR(p) of a type p is the minimal rank of
any formula in p. If MR(p) is an ordinal , then its Morley degree MD(p) of p
is the minimal degree of a formula of p having rank . If p = tp(a/A) we also
write MR(a/A) and MD(a/A).
Algebraic types have Morley rank 0 and
MD(p) = deg(p).
Strongly minimal types are exactly the types of Morley rank and Morley de-
gree 1.
Let p S(A) have Morley rank and Morley degree d. Then by denition
there is some p of rank and degree d. Clearly, is uniquely determined
up to equivalence since for all we have MR( ) < if and only if
p. Thus, p is uniquely determined by :
(2) p = (x) [ L(A)formula, MR( ) < .
Obviously, equivalent formulas determine the same type (cf. Lemma 21.3).
Thus L(A) belongs to a unique type of rank if and only if is
minimal over A; i.e. if has rank and cannot be decomposed as the union of
two L(A)-formulas of rank .
Lemma 24.10. Let be a consistent L(A)formula.
1.
MR = maxMR(p) [ p S(A)
2. Let MR = . Then
MD =
_
MD(p)
p S(A), MR(p) =
_
.
Proof. 1: If MR = , then [ L(A)formula, MR < is
consistent. Any type over A containing this set of formulas has rank .
If MR = , there is a decomposition of in L(A)formulas
1
, . . . ,
k
,
minimal over A. (Note that k is bounded by MD.) By (2), the
i
determine
types p
i
of rank .
2: The p
i
are exactly the types of rank containing . Furthermore,
MD
i
= MD(p
i
).
Corollary 24.11. If p S(A) has Morley rank and A B, then
MD(p) =
_
MD(q)
p q S(B), MR(p) = MR(q)

_
Corollary 24.12. Let p S(A) have Morley rank and A B. Then p S(A)
has at least one and at most MD(p) many extensions to B of the same rank.
We will later show that extensions of the same Morley rank are a special
case of the non-forking extensions studied in Chapters 7 and 8. For types with
Morley rank those of the same Morley rank are exactly the non-forking ones.
Exercise 24.1.
Let be a formula of Morley rank < and
0
,
1
, . . . an innite sequence of
formulas. Assume that there is a number k such that the conjunction any k of
the
i
has Morley rank smaller than . Then MR(
i
) < for almost all i.
Exercise 24.2.
We only dened Morley rank for formulas and types in one variable, but it
generalises to formulas and types in any xed number of variables x
1
, . . . , x
n
in
the obvious way. Prove the following:
1. If T is stable, there are only countably many n-types over any countable
set.
2. If T is totally transcendental, then any formula in n variables has Morley
rank.
Exercise 24.3.
Let be a formula with parameters in the saturated model M. If MR() > ,
show that there is an innite family of formulas with parameters in M which
each imply , are pairwise inconsistent and have Morley rank .
Exercise 24.4.
Show that a totally transcendental group G has a connected component, i.e. a
smallest denable subgroup G
0
of nite index. Show also that any nite normal
subgroup of G
0
lies in the center of G
0
.
Exercise 24.5.
If T is totally transcendental, then all types over saturated models have
Morley degree 1. (We will see later that this is true without assuming
saturation (36.12).)
Exercise 24.6 (Lachlan).
If a type p has Morley rank, then MR(p) < [T[
+
. Hence, if T is totally tran-
scendental, we have MR(T) < [T[
+
.
Exercise 24.7.
For a topological space X we dene recursively on ordinals X
0
:= X, X
+1
:=
X
x: x isolated in X
and X
:=
<
X
if is a limit ordinal. The

Cantor-Bendixson rank of x X is equal to if is maximal with x X
.
Show that on S(C) the Morley rank equals the Cantor-Bendixson rank. Note
that S(C) is not even a class. So, for this exercise we have to ignore settheoretic
subtleties.
Exercise 24.8.
If p is a type over acl(A), then p and p A have the same Morley rank.
25 Countable models of
1
-categorical theories
Uncountable models of
1
-categorical theories are determined up to isomor-
phism by their cardinality. Section 22 showed that this cardinality coincides
with the dimension of a strongly minimal formula. We here extend this analysis
in order to classify also the countable models of
1
-categorical theories and show
in Theorem 25.7 that for each possible dimension there is a unique model of the
theory.
Throughout this section we x a countable
1
categorical theory T.
For models M N of T and (x) L(M) a strongly minimal formula, we write
dim
(N/M) for the dimension of N over M.

Theorem 25.1. Let T be a countable
1
-categorical theory, M N be models
of T, A M and (x) L(A) a strongly minimal formula.
1. If b
1
, . . . b
n
(N) are independent over M and N is prime over M
b
1
, . . . b
n
, then
dim
(N/M) = n, and
2.
dim
(N) = dim
(M) + dim
(N/M)
Proof. For ease of notation we assume A = .
1:
Let c (N). We want to show that c is algebraic over M b
1
, . . . , b
n
.
Assume the contrary. Then p(x) = tp(c/M b
1
, . . . , b
n
) is strongly minimal
and is axiomatised by
(x)
i
(x) [ i I,
where the
i
range over all algebraic formulas dened over M b
1
, . . . , b
n
.
Since (M) is innite, any nite subset of p(x) is realised by an element of M.
Thus, p(x) is not isolated. But all elements of the prime extension N are atomic
over M b
1
, . . . , b
n
by Corollary 17.7, a contradiction.
2:
This follows from Remark C 8.8, if we can show that a basis of (N) over (M)
is also a basis of (N) over M. So the proof is complete once we have established
the following lemma.
Lemma 25.2. Let T be stable, M N models of T, (x) strongly minimal
and b
i
(N). If the b
i
are independent over (M), they are independent over
M.
Proof. Assume that

b = b
1
. . . b
n
are algebraically independent over (M) but
dependent over a M. An argument as in the proof of Theorem 19.2 shows
that we may assume that M is saturated. Let p be the type of

b over M.
We choose a sequence

b
0
,
b
1
, . . . in (M) such that

b
2i
is an n-tuple of elements
algebraically independent over

b
0
. . .
b
2i1
and

b
2i+1
realises p a
b
0
. . .
b
2i
. Let
q be the type of a over the set B of elements of (
b
i
). Since the sequence (
b
i
)
is indiscernible, every permutation of denes a type (q) over B. If i [
(2i) even , = i [
(2i) even, we have (q) ,=
(q). So there are uncountably

many types over B and T is not stable.
Corollary 25.3. The dimension
dim(N/M) = dim
(N/M)
of N over M does not depend on : it is the maximal length of an elementary
chain
M = N
0
N
1
. . . N
n
= N
between M and N.
Proof. This follows from the previous theorem since T has no Vaughtian pairs.
For the remainder of this section, we let M
0
denote the prime
model of T. We also x a strongly minimal formula (x, a
0
) L(M
0
)
and put p
0
(x) = tp(a
0
).
Note that the type p
0
(x) of a
0
is isolated by Theorem 14.2, hence realised
in every model of T. For any model M and realisation a of p
0
in M, let
dim
(x,a)
(M) denote the (x, a)dimension of M over a. To simplify notation
we assume that a
0
is some element a
0
rather than a tuple.
Since M
0
is homogeneous by Corollary 14.4, the dimension
m
0
= dim
(x,a)
(M
0
)
does not depend on the realisation a of p
0
in M
0
. We will show in Lemma 25.6
that the same is true for any model of T.
Lemma 25.4. A countable model M is saturated if and only if its (x, a)
dimension is . Hence in this case, the dimension is independent of the reali-
sation of p
0
(x) in M. In particular, T is
0
categorical if and only if m
0
= .
Proof. In a saturated model the (x, a)dimension is innite. Since there exists
a unique countable saturated model by Lemmas 16.7 and 16.8, the rst claim
follows. This obviously does not depend on the realisation of p
0
. The last
sentence now follows from Theorem 16.10.
We need the following observation:
Lemma 25.5. If M is prime over a nite set and m
0
< , then dim
(x,a)
(M)
is nite.
Proof. Suppose M is prime over the nite set C. Let B be a basis of (M, a)
over M
0
. Since M is the minimal prime extension of M
0
B, C is atomic over
M
0
B. Thus there exists a nite subset B
0
of B such that C is contained in
the prime extension N of M
0
B
0
. As M is prime over Ca, it suces to show
that dim
(x,a)
(N) is nite and this follows by Theorem 25.1 from
dim
(x,a)
(N) = m
0
+[B
0
[ .
The crucial lemma and promised converse to Corollary 22.3 is the following:
Lemma 25.6. The dimension dim
(x,a)
(M) does not depend on the realisation
a of p
0
in M.
Proof. The lemma is clear if M is uncountable and also if M is countable with
innite -dimension by Lemma 25.4. Therefore we may assume that M has
nite -dimension, which implies that m
0
is nite.
For the proof we now introduce the following notion: let a
1
and a
2
realise
p
0
. Choose a model N of nite dimension containing a
1
and a
2
, which exists
by Lemma 25.5, and put
di(a
1
, a
2
) = dim
(x,a1)
(N) dim
(x,a2)
(N).
This denition does not depend on the model N: if N
N is prime over
a
1
, a
2
, then by Theorem 25.1 we have for i = 1, 2
dim
(x,ai)
(N) = dim
(x,ai)
(N
) + dim(N/N
),
so
di(a
1
, a
2
) = dim
(x,a1)
(N
) dim
(x,a2)
(N
).
Clearly we have
di(a
1
, a
3
) = di(a
1
, a
2
) + di(a
2
, a
3
).
This implies that di(a
1
, a
2
) only depends on tp(a
1
, a
2
). We will show that
di(a
1
, a
2
) = 0 for all realisations of p
0
. This implies that dim
(x,a1)
(M) =
dim
(x,a2)
(M) for all models M which contain a
1
and a
2
.
For the proof choose an innite sequence a
1
, a
2
, . . . with
tp(a
i
, a
i+1
) = tp(a
1
, a
2
)
for all i.
Now we use the fact that
1
-categorical theories are -stable, so the type p
0
has Morley rank and an extension q
0
to a
1
, a
2
, . . . of the same rank. Let b be
a realisation of q
0
. Then, by Corollary 24.12, there are at most MD(p
0
) many
dierent types of the form tp(ba
i
). So let i < j be such that tp(ba
i
) = tp(ba
j
).
Then di(a
i
, b) = di(b, a
j
) and
(j i) di(a
1
, a
2
) = di(a
i
, a
j
) = di(a
i
, b) + di(b, a
j
) = 0,
implying di(a
1
, a
2
) = 0.
Thanks to the previous lemma we obtain a complete account of the models of
an uncountably categorical theory.
Theorem 25.7 (BaldwinLachlan). If T is uncountably categorical, then for
any cardinal m m
0
there is a unique model M of T with dim
(x,a)
(M) = m.
These models are pairwise non-isomorphic.
Proof. If m = m
0
+ , choose M prime over M
0
b
i
: i < where the
b
i
(C, a
0
) are independent over M
0
. Uniqueness follows from Corollary 22.3
and non-isomorphy from Lemma 25.6.
Exercise 25.1.
All models of an
1
categorical theory are -homogeneous.
Exercise 25.2.
Let T be strongly minimal and m
0
be the dimension of the prime model. Show
that m
0
is the smallest number n such S
n+1
(T) is innite.
26 Computation of Morley Rank
In this section we show that the Morley rank agrees with the dimension of the
pregeometry on strongly minimal minimal sets and give some examples of how
to compute it in -stable elds.
Lemma 26.1. If b is algebraic over aA, we have MR(b/A) MR(a/A).
Proof. Let MR(a/A) = . We prove MR(b/A) by induction over .
Let d = MD(b/Aa). Choose an L(A)formula (x, y) in tp(ab/A) such that
MR(y(x, y)) = and [(a
, C)[ d for all a
.
We show that the Morley rank of (y) = x(x, y) is bounded by . For
this consider an innite family
i
(C) of disjoint subclasses of (C) dened over
some extension A
of A. Put
i
(x) = y((x, y)
i
(y)). Since any d + 1 of
the
i
have empty intersection, some
i
(x) has Morley rank < . Let b
be
any realisation of
i
(y). Choose a
such that [= (a
, b
). Then b
is algebraic
over a
A and since a
realises
i
(x), we have MR(a
/A
) . So by induction
we conclude MR(b
/A
) , which shows MR
i
. So does not contain
an innite family of disjoint formulas of Morley rank greater or equal to . So
MR .
Theorem 26.2. Let (x) be a strongly minimal formula dened over B and
a
1
, . . . , a
n
a sequence of realisations. Then
MR(a
1
, . . . , a
n
/B) = dim
(a
1
, . . . , a
n
/B)
.
Proof. By the lemma we may assume that a
1
, . . . , a
n
are independent over B.
Let a
1
, . . . , a
n
realise the L(B)formula (x
1
, . . . , x
n
). By induction we have
MR(a
1
, a
2
, . . . , a
n
/Ba
1
) = n 1. So the formula
a1
( x) = (x
1
, . . . , x
n
) x
1
.
= a
1
has rank at least n1. The innitely many conjugates of
a1
over B are disjoint
and have rank n 1 as well. This shows that MR n.
Let B
B be an extension of B. By Corollary 21.7 there is only one type

p S
n
(B
) which is realised by an B
independent sequence of elements of (C).

So by induction, there is only one n-type of elements of (C) of rank n. This
implies that (x
1
) . . . (x
n
) has rank n.
Corollary 26.3. Let (x) be a strongly minimal formula and (x
1
, . . . , x
n
) be
dened over B such that implies (x
i
) for all i. Then
MR = maxdim
( a/B) [ C [= ( a).
On strongly minimal, Morley rank is denable:
Corollary 26.4. For any strongly minimal formula (x) and any formula
(x
1
, . . . , x
n
, y) which implies (x
i
) for all i
b [ MR(x
1
, . . . , x
n
,
b) = k
is a denable class for every k.
Proof. We show that MR(x
1
, . . . , x
n
,
b) k is an elementary property of

b
by induction on n. The case n = 1 follows from the fact that MR(x
1
,
b) 1
is equivalent to
x
1
(x
1
,
b). This is an elementary property of

b since is
strongly minimal (see the discussion on page 98.) For the induction step we
conclude from Corollary 26.3 that MR(x
1
, . . . , x
n
,
b) k if and only if one of

the following is true:
there is an a
1
such that MR(a
1
, x
2
. . . , x
n
,
b) k,
there is an a
1
which is is not algebraic over

b such that
MR(a
1
, x
2
. . . , x
n
,
b) k 1.
The rst part an elementary property of

b by induction. For the second part
note that by induction MR(a
1
, x
2
. . . , x
n
,
b) k 1 can be expressed by a
formula (a
1
,
b). The second condition is then equivalent to
x
1
(x
1
,
b).
For algebraically closed elds, these considerations translate into the follow-
ing statement.
Corollary 26.5. Let K be a subeld of a model of T
ACFp
and a a tuple of
elements. Then the Morley rank of a over K equals the transcendence degree
of K(a) over K.
Note that by quantier elimination denable sets in algebraically closed elds
are exactly the constructible sets in algebraic geometry. The previous corollary
expresses the important fact that for a denable set in an algebraically closed
eld the Morley rank equals the dimension of its Zariski closure in the sense of
algebraic geometry (see e.g. [45]).
We now turn to the theory of dierentially closed elds of characteristic 0,
T
DCF0
.
Let K F be an extension of dierential elds and a an element of F.
The dimension of a over K is dened as the transcendence degree of Ka over
K. There is a unique quantier free type over K of innite dimension.
Remark 26.6. If the dimension of a over K equals n, then the type of a over
K is determined by the dminimal polynomial f of a over K: f is irreducible
in K[x
0
, . . . , x
n
] and f(a, . . . , d
n
a) = 0, (see Remark B 6.7).
Corollary 26.7. T
DCF0
is stable. For a dierential eld K and elements a
we have
MR(a/K) dim(a/K).
If a has innite dimension, then the type of a over K has Morley rank .
Proof. There are at most [K[ many dminimal polynomials over K, so at most
[K[ many 1types. Thus, T
DCF0
is stable.
We may assume that K is
1
saturated (otherwise we take a extension of
tp(a/K) to an
1
saturated eld with the same Morley rank. Then the Mor-
ley rank stays the same and the dimension does not increase.) We show
MR(a/K) dim(a/K) by induction on dim(a/K). If dim(a/K) = 0, then
a is algebraic over K and the Morley rank is 0. Let dim(a/K) = n and let f
be the minimal polynomial of a over K. Apart from tp(a/K), all other types
over K containing f(x, . . . , d
n
x)
.
= 0 have dimension and hence Morley rank
less than n. Since K is suciently saturated, this implies that the Morley rank
of tp(a/K) is at most n.
By the next remark there are types of rank n for every n. This implies that
there must be 1types of rank . Since there is only one type p
of innite
dimension, it follows
3
that p
has rank .
Remark 26.8. If a, . . . , d
n1
a are algebraically independent over K and d
n
a
K, then MR(a/K) = n.
Proof. We prove this by induction on n: Consider the formula (x) = (d
n
(x)
.
=
d
n
(a)). If the claim is true for n 1, all formulas
b
(x) = (d
n1
(x)
.
= b) have
rank n1. For all constants c the
d
n1
(a)+c
(x) are contained in (x). So (x)
has rank n. All a
which realise (x) have either at most dimension n 1 or

have the same type as a over K. This shows that a has Morley rank n over
K.
Dimension in pregeometries is additive, i.e. we have dim(ab/B) = dim(a/B)+
dim(b/aB). This translates into additivity of Morley rank for elements in the
algebraic closure of strongly minimal sets:
Remark 26.9. Let be a strongly minimal formula dened over B and a, b
algebraic over (C) B. Then MR(ab/B) = MR(a/B) + MR(b/aB).
Proof. Assume B = for notational simplicity. Then a is algebraic over some
nite set of elements of (C). We can split this set into a sequence

f of elements
which are independent over a and a tuple a which is algebraic over

fa. By taking
nonforking extensions if necessary, we may assume that

f is independent from
ab. Now a and a are interalgebraic over

f. In the same way we nd tuples g
3
This is immediate if K is saturated, see Exercise 24.3.
and

b in (C) in which b is algebraic with g independent from ab
f and so that
b is interalgebraic with b over F =

f g. The claim now follows from
MR(ab) = MR(ab/F)
= MR( a
b/F) = MR( a/F) + MR(
b/ aF)
= MR(a/F) + MR(b/aF) = MR(a) + MR(b/a)
In fact, Exercise 26.5 shows that if F is any innite Bindependent subset
F of (C) then every element of acl((C) B) is interalgebraic over FB with a
tuple in (C).
Exercise 26.1.
Let T be strongly minimal. Show that a nite tuple a is geometrically indepen-
dent from B over C if and only if MR(a/BC) = MR(a/C). (See page 235 and
Exercise C 8.1.)
Exercise 26.2.
Let be a formula without parameters. Assume that is almost strongly
minimal , i.e. that there is a strongly minimal formula dened over some set
B such that all elements of (C) are algebraic over (C) B. Then for all a, b
in (C) and any set C we have MR(ab/C) = MR(a/C) + MR(b/aC).
The following exercise shows that for arbitrary totally transcendental theo-
ries the Morley rank need not be additive:
Exercise 26.3.
Consider the following theory in a two-sorted language having sorts A and B
and a function f : B A. Assume that sort A is split into innitely many
innite predicates A
1
, A
2
, . . . such that any a A
n
has exactly n preimages
under f. Let a be a generic element of A, i.e., an element such that MR(a) is
maximal, and choose f(b) = a. Show that MR(ab) = MR(a) = 2, MR(b/a) = 1.
However, this is the worst that can happen: Shelah (see [47], Thm. V 7.8)
and Erimbetov (see [14]) proved independently that the Morley rank of B is
bounded by ( + 1) if 1 and MR(A) and MR(f
1
(a)) for all
a A. It can be shown that this bound is optimal.
Exercise 26.4.
Find an example of a theory without the denable multiplicity property (DMP).
Exercise 26.5.
Let be a strongly minimal formula without parameters and F an innite
independent subset of (C). Then every element of acl (C) is interalgebraic
over F with a tuple in (C).
Chapter 7
Simple theories
So far, we mainly studied totally transcendental theories, a small subclass of
the class of stable theories, the totally transcendental theories being the most
stable ones. Before we turn to stable theories in general, we consider simple (but
possibly unstable) theories, a generalisation which after their rst introduction
by Shelah [49] gained new attention after the fundamental work of Kim and
Pillay [30]. Interest in simple theories increased with Hrushovskis results on
pseudo-nite elds, see [25]. Much of the exposition is taken from Casanovas
article [12].
27 Dividing and Forking
We will characterise simple theories by the existence of a well-behaved notion
of independence, a relation on types satisfying certain properties. To this end
we here dene forking and dividing. In the context of totally transcendental
theories, these concepts correspond to type extensions of smaller Morley rank.
We begin with a reformulation of the Standard Lemma 15.3 on indiscernibles.
Lemma 27.1 (The Standard Lemma). Let A be a set of parameters, 1 an
innite sequence of tuples and J a linear order. Then there is a sequence of
indiscernibles over A of order type J realizing EM(1/A).
Denition 27.2. We say (x, b) divides over A (with respect to k) if there
is a sequence (b
i
)
i<
of realisations of tp(b/A) such that ((x, b
i
))
i<
is k-
inconsistent.
1
A set of formulas (x) divides over A if (x) implies some (x, b)
which divides over A.
1
A family (
i
(x))
xI
is kinconsistent if for every kelement subset K of I the set {
i
|
i K} is inconsistent.
125
CHAPTER 7. SIMPLE THEORIES 126
If (x, a) implies (x, a
) and (x, a
) divides over A, then (x, a) divides

over A. Thus divides over A if and only if divides over A. Also a
set divides over A if and only if a conjunction of formulas from divides
over A. Note that it makes sense to say that ( x) divides over A for x an
innite sequence of variables as we may use blind variables without changing
the meaning of dividing.
The following is easy to see:
Remark 27.3. 1. If a , acl(A), then tp(a/Aa) divides over A.
2. If (x) is consistent and dened over acl(A), then (x) does not divide
over A.
Lemma 27.4. (x, b) divides over A if and only if there is a sequence (b
i
)
i<
of
indiscernibles over A with tp(b
0
/A) = tp(b/A) and

i<
(x, b
i
) inconsistent.
We may replace by any innite linear order. Note also that b may be a
tuple of innite length.
Proof. If (b
i
)
i<
is a sequence of indiscernibles over A with tp(b
0
/A) = tp(b/A)
and

i<
(x, b
i
) inconsistent there is a conjunction (x, b) of formulas from
(x, b) for which (x) = (x, b
i
) [ i < is inconsistent. contains some
kelement inconsistent subset. This implies that ((x, b
i
))
i<
is kinconsistent.
Assume conversely that (x, b) divides over A. Then some nite conjunction
(x, b) of formulas from (x, b) divides. Let (b
i
)
i<
be a sequence of realisa-
tions of tp(b/A) such that ((x, b
i
) [ i < ) is k-inconsistent. We may assume
by Lemma 27.1 that (b
i
)
i<
is indiscernible over A. Clearly,

i<
(x, b
i
) is
inconsistent.
Corollary 27.5. The following are equivalent:
1. tp(a/Ab) does not divide over A.
2. For any innite sequence of A-indiscernibles 1 containing b, there exists
some a
with tp(a
/Ab) = tp(a/Ab) and such that 1 is indiscernible over

Aa
.
3. For any innite sequence of A-indiscernibles 1 containing b, there exists
1
with tp(1
/Ab) = tp(1/Ab) and such that 1
is indiscernible over Aa.

Proof. 2) 3): this is clear by considering appropriate automorphisms. It is
also easy to see that the conclusion of 2) and 3) is equivalent to:
()
There exist a
and 1
with tp(a
/Ab) = tp(a/Ab) and tp(1
/Ab) =
tp(1/Ab) such that 1
is indiscernible over Aa
.
1) (): Let 1 = (b
i
)
iI
be an innite sequence of indiscernibles with b
i0
= b.
Let p(x, y) = tp(ab/A). Then

iI
p(x, b
i
) is consistent by Lemma 27.4; let
a
be a realisation. By Lemma 27.1, there is 1
= (b
i
)
iI
indiscernible over
Aa
and realising EM(1/Aa
). Since [= p(a
, b
i0
), there is an automorphism
Aut(C/Aa
) taking b
i0
to b. Put 1
= (1
).
2)1) Let p(x, y) = tp(ab/A) and let (b
i
)
i<
be a sequence of indiscernibles over
A with tp(b
0
/A) = tp(b/A). We have to show that

i<
p(x, b
i
) is consistent.
By assumption there is a
with tp(a
/Ab) = tp(a/Ab) such that 1 is indiscernible

over Aa
. As [= p(a
, b), a
is a realisation of

i<
p(x, b
i
).
The next proposition states a transitivity property of dividing. See Corol-
lary 28.17 and its proof.
Proposition 27.6. If tp(a/B) does not divide over A B and tp(c/Ba) does
not divide over Aa, then tp(ac/B) does not divide over A.
Proof. Let b B be a nite tuple and 1 an innite sequence of Aindiscernibles
containing b. If tp(a/B) does not divide over A, there is some 1
with tp(1
/Ab) =
tp(1, Ab) and indiscernible over Aa. If tp(c/Ba) does not divide over Aa, there
is 1
with tp(1
/Aab) = tp(1
/Aab) and indiscernible over Aac proving the

claim.
Denition 27.7. (x) forks over A if (x) implies a disjunction
_
l<d
l
(x) of
formulas
l
(x) each dividing over A.
Thus, if (x) divides over A, it forks over A. By denition (and compact-
ness), we immediately see the following:
Remark 27.8 (Non-forking is closed). If p S(B) forks over A, there is some
(x) p such that any type in S(B) containing (x) forks over A.
Corollary 27.9 (Finite character). If p S(B) forks over A, there is a nite
subset B
0
B such that p AB
0
forks over A.
Lemma 27.10. If is nitely satisable in A, then does not fork over A.
Proof. If (x) implies the disjunction
_
l<d
l
(x, b), then some
l
has a realisa-
tion a in A. If the b
i
, i < , realise tp(b/A), then
l
(x, b
i
) [ i < is realised
by a. So
l
does not divide over A.
Lemma 27.11. Let A B and let be a partial type over B. If does not
fork over A, it can be extended to some p S(B) which does not fork over A.
Proof. Let p(x) be a maximal set of L(B)formulas containing (x) which does
not fork over A. Clearly, p is consistent. Let (x) L(B). If neither nor
belongs to p, then both p and p fork over A, and hence p forks
over A. Thus p is complete.
Exercise 27.1.
1. Let (x) be a formula over A with Morley rank and let (x) dene a
subclass of (C). If forks over A, it has smaller Morley rank than .
2. Let p be a type with Morley rank and q an extension of p. If q forks over
A, it has smaller Morley rank than p.
We will see in Exercise 36.4 that in both statements the converse is also true.
Exercise 27.2.
Let p be a type over the model M and A M. Assume that M is [A[
+
saturated. Show that p forks over A if and only if p divides over A.

Exercise 27.3.
A global type which is Ainvariant, i.e. invariant under all Aut(C/A), does
not fork over A.
Exercise 27.4.
Let M be a saturated and strongly homogeneous model. If p S(M) forks
over a subset A of cardinality smaller than , p has many conjugates under
Aut(M/A).
Exercise 27.5.
Dene the cyclical order on Q by
cyc(a, b, c) (a < b < c) (b < c < a) (c < a < b).
Show:
1. (Q, cyc) has quantier elimination.
2. cyc(a, x, b) divides over if a ,= b.
3. The unique p S
1
() forks over .
Exercise 27.6.
If tp(a/Ab) does not divide over A and (x, b) divides over A with respect to k,
then (x, b) divides over Aa with respect to k.
28 Simplicity
In this section, we dene simple theories and the notion of forking independence
whose properties characterise such theories. By the absence of binary trees of
consistent formulas, totally transcendental theories are simple. Using the fact
that in simple theories types do not divide over (suitable) small sets, we will see
later that in fact all stable theories are simple.
Denition 28.1. 1. A formula (x, y) has the tree property (with respect
to k) if there is a tree of parameters (a
s
[ , = s
<
) such that:
a) For all s
<
, ((x, a
si
) [ i < ) is kinconsistent.
b) For all

, (x, a
s
) [ , = s is consistent.
2. A theory T is simple if there is no formula (x, y) with the tree property.
Clearly, for simplicity it suces to consider formulas (x, y) without parameters.
Remark 28.2. It is not hard to see that in totally transcendental theories no
formula has the tree property. This is immediate for k = 2. In the general case
this follows from Exercise 24.1.
Denition 28.3. Let be a nite set of formulas (x, y) without parameters.
A kdividing sequence over A is a sequence (
i
(x, a
i
) [ i < ) such that
1.
i
(x, y)
2.
i
(x, a
i
) divides over A a
j
[ j < i with respect to k.
3.
i
(x, a
i
) [ i < is consistent.
Lemma 28.4. 1. If has the tree property with respect to k, then for every
A and there exists a kdividing sequence over A of length .
2. If no has the tree property with respect to k, there is no innite
kdividing sequence over .
Proof. 1): Note rst that we may assume that is a limit ordinal. A compact-
ness argument shows that for every and there is a tree (a
s
[ ,= s
<
)
such that all families ((x, a
si
) [ i < ) are kinconsistent and for all

,
(x, a
s
) [ ,= s is consistent. If is bigger than 2
max(|T|,)
, we nd
a path such that for all s , innitely many a
si
have the same type over
a
t
[ t s. Now (
i
(x, a
i+1
) [ i < ) is a kdividing sequence.
2): Suppose there is an innite kdividing sequence over . If appears
innitely many times in this sequence, there is an innite kdividing sequence
((x, a
i
) [ i < ). For each i we choose a sequence (a
n
i
[ n < ) with tp(a
n
i
/a
j
[
j < i) = tp(a
i
/a
j
[ j < i) such that ((x, a
n
i
) [ n < ) is kinconsistent.
Then we nd parameters b
s
showing that has the tree property with respect
to k as follows: assume s
i+1
and
b = (b
s1
, . . . , b
si
) have been dened such
that tp(a
1
, . . . , a
i1
) = tp(
b). Choose Aut(C) with (a

1
, . . . , a
i1
) =
b and
put b
s
= (a
s(i)
i
).
It is easy to see that in simple theories for every nite set and all k there
exists a nite bound on the possible length of kdividing sequences.
Proposition 28.5. Let T be a complete theory. The following are equivalent.
a) T is simple
b) For all p S(B) there is some A B with [A[ [T[ such that p does not
divide over A.
c) (Local Character) There is some such that for all models M and p S(M)
there is some A M with [A[ such that p does not divide over A.
Proof. a)b): If b) does not hold, there is a sequence (
i
(x, b
i
) [ i < [T[
+
) of
formulas from p(x) such that every
i
(x, b
i
) divides over b
j
[ j < i with re-
spect to k
i
. There is an innite subsequence for which all
i
(x, y) equal (x, y)
and all k
i
= k yielding a kdividing sequence.
b)c): Clear.
c)a): If has the tree property, there are kdividing sequences ((x, b
i
) [
i <
+
). It is easy to construct an ascending sequence of models M
i
, (i <
+
)
such that b
j
M
i
for j < i and (x, b
i
) divides over M
i
. Extend the set of
(x, b
i
) to some type p(x) S(M) where M =
i<
+ M
i
. Then p divides over
each M
i
.
Corollary 28.6. Let T be simple and p S(A). Then p does not fork over A.
Proof. Suppose p forks over A, so p implies some disjunction
_
l<d
l
(x, b) of
formulas all of which divide over A with respect to k. Put =
l
(x, y) [ l < d.
We will show by induction that for all n there is a kdividing sequence
over A of length n. This contradicts the remark after Lemma 28.4. We will
assume also that the dividing sequence is consistent with p(x).
Suppose that (
i
(x, a
i
) [ i < n) is kdividing sequence over A, consis-
tent with p(x). By Exercise 28.4 we can replace b with a conjugate b
over A
such that (
i
(x, a
i
) [ i < n) is a dividing sequence over Ab
. Now one of the

formulas
l
(x, b
), say
0
(x, b
), is consistent with p(x)

i
(x, a
i
) [ i < n. So
0
(x, b
),
0
(x, a
0
), . . . ,
n1
(x, a
n1
) is a kdividing sequence over A and
consistent with p(x).
Let p be a type over A and q an extension of p. We call p a forking extension
if q forks over A.
Corollary 28.7 (Existence). If T is simple, every type over A has a nonforking
extension to any B containing A.
Proof. This follows from Corollary 28.6 and Lemma 27.11.
Denition 28.8. A is independent from B over C,
A [
C
B,
if for every nite tuple a from A, the type tp(a/BC) does not fork over C.
This denition makes sense since forking of tp(a/BC) does not depend on the
enumeration of a and since tp(a/BC) forks over C if the type of a subsequence
of a forks over C. So this is the same as saying that tp(A/BC) does not fork
over C.
Denition 28.9. Let I be a linear order. A sequence (a
i
)
iI
is called
1. independent over A if a
i
[
A
a
j
[ j < i for all i,
2. Morley sequence over A if it is independent and indiscernible over A,
3. Morley sequence in p(x) over A if it is a Morley sequence over A consisting
of realisations of p.
Example 28.10. Let q be a global type invariant over A. Then any sequence
(b
i
)
iI
where each b
i
realises q A b
j
[ j < i is a Morley sequence over A.
Proof. Let us call such sequences good. Clearly a subsequence of a good se-
quence is good again. So for indiscernibility it suces to show that all nite
good sequences b
0
. . . b
n
and b
0
. . . b
n
have the same type over A. Indeed, using
induction, we may assume that b
0
. . . b
n1
and b
0
. . . b
n1
have the same type
and so (b
0
. . . b
n1
) = b
0
. . . b
n1
for some Aut(C/A). Then
(tp(b
n
/Ab
0
. . . b
n1
)) = (q Ab
0
. . . b
n1
) = q Ab
0
. . . b
n1
= tp(b
n
/Ab
0
. . . b
n1
),
which proves our claim.
We call such a sequence (b
i
)
iI
a Morley sequence of q over A. Note that
our proof shows that the type of a Morley sequence of q over A is uniquely
determined by its order type.
Lemma 28.11. If (a
i
)
iI
is independent over A and J < K are subsets of I,
then tp((a
k
)
kK
/Aa
j
[ j J) does not divide over A.
Proof. We may assume that K is nite. The claim now follows from Proposi-
tion 27.6 by induction on [K[.
Lemma 28.12 (Shelah). For all A there is some such that for any linear order
I of cardinality and any family (a
i
)
iI
there exists an Aindiscernible sequence
(b
j
)
j<
such that for all j
1
< . . . < j
n
< there is a sequence i
1
< . . . < i
n
in
I with tp(a
i1
. . . a
in
/A) = tp(b
j1
. . . b
jn
/A).
Proof. We only need that satises the following: let = sup
n<
[ S
n
(A)[.
1. cf() >
2. For all < and all n < there is some
< with
()
n
(see
Appendix C, Denition 10.1).
By Erds-Rado (see Appendix C, Theorem 10.2) we may take =
+.
We now construct a sequence of types p
1
(x
1
) p
2
(x
1
, x
2
) . . . with p
n

S
n
(A) such that for all < there is some I
I with [I
[ = such that
tp(a
i1
. . . a
in
) = p
n
for all i
1
< . . . < i
n
from I
.
Then we can choose the (b
i
)
i<
as a realisation of

i<
p
i
.
If p
n1
has been constructed and we are given < , we choose
<
with
()
n
and some I
I with [I
[ =
such that tp(a

i1
. . . a
in1
/A) =
p
n1
for all i
1
< . . . < i
n1
from I
. Thus there are I
and p
n
with
tp(a
i1
. . . a
in
) = p
n
for all i
1
< . . . < i
n
from I
. Since cf() > , there is some

p
n
with p
n
= p
n
for conally many .
The existence of a Ramsey cardinal > (see p. 239) would directly imply
that any sequence of order type contains a countable indiscernible subsequence
(in fact even an indiscernible subsequence of size ).
Lemma 28.13. If p S(B) does not fork over A, there is an innite Morley
sequence in p over A which is indiscernible over B. In particular, if T is simple,
for every p S(A), there is an innite Morley sequence in p over A.
Proof. Let a
0
be a realisation of p. By Lemma 27.11 there is a nonforking
extension p
of p to Ba
0
. Let a
1
be a realisation of p
. Continuing in this
way we obtain a sequence (a
i
)
i<
with a
i
[
A
B(a
j
)
j<i
for arbitrary . By
Lemma 28.12 we obtain a sequence of length with the same property and
indiscernible over B. The last sentence is immediate by Corollary 28.6.
Proposition 28.14. Let T be simple and (x, y) be a partial type over A.
Let (b
i
)
i<
be an innite Morley sequence over A and

i<
(x, b
i
) consistent.
Then (x, b
0
) does not divide over A.
Proof. By Lemma 27.1, for every linear order I there is a Morley sequence (b
i
)
iI
in tp(b
0
/A) over A such that (x) =
iI
(x, b
i
) is consistent. Choose I having
the inverse order type of [T[
+
. Let c be a realisation of . By Proposition 28.5.b
there is some i
0
such that tp(c/A b
i
[ i I) does not divide over A b
i
[
i > i
0
. This implies that tp(c/A b
i
[ i i
0
) does not divide over A b
i
[
i > i
0
. By Lemma 28.11, tp((b
i
[ i > i
0
)/Ab
i0
) does not divide over A. Hence
tp(c (b
i
[ i > i
0
)/Ab
i0
) does not divide over A by Proposition 27.6. This implies
that (x, b
i0
) does not divide over A.
Proposition 28.15. Let T be simple. Then (x, b) divides over A if and only
if it forks over A.
Proof. By denition, if (x, b) divides over A, it forks over A. For the converse
assume (x, b) does not divide over A. So if (x, b) =
_
l<d
l
(x, b) is implied by
(x, b), it does not divide over A. Let (b
i
)
i<
be a Morley sequence in tp(b/A)
over A, which exists since T is simple. So (x, b
i
) [ i is consistent. By the
pigeon hole principle there must be some l and some innite I such that
l
(x, b
i
) [ i I is consistent. By Proposition 28.14,
l
(x, b) does not divide
over A. Hence (x, b) does not fork over A.
Proposition 28.16 (Symmetry). In simple theories, independence is symmet-
ric.
Proof. Assume A [
C
B and consider nite tuples a A and b B. Since
a [
C
b, Lemma 28.13 gives an innite Morley sequence (a
i
)
i<
in tp(a/Cb)
over C, indiscernible over Cb. Let p(x, y) = tp(ab/C). Then

i<
p(a
i
, y) is
consistent because it is realised by b. Thus, by Proposition 28.14, p(a, y) does
not divide over C. This proves b [
C
a. Since this holds for all a A, b B, it
follows B [
C
A by Finite character.
Corollary 28.17 (Monotony and Transitivity). Let T be simple, B C D.
Then we have A [
B
D if and only if A [
B
C and A [
C
D.
Proof. One direction of this equivalence, Monotony, holds for arbitrary theories
and follows easily from the denition. For Transitivity, the other direction, note
that by Proposition 28.15 we may read Proposition 27.6 after replacing nite
tuples by innite ones as
A
A
B and C [
AA
B CA
A
B.
Swapping the left and the right hand side, this is exactly the transitivity. Hence
the claim follows from Proposition 28.16.
Corollary 28.18. (a
i
)
iI
being independent over A does not depend on the
ordering of I.
Proof. Let i be an element of I and J, K two subsets such that J < i < K. Write
a
J
= a
j
[ j J and a
K
= a
k
[ k K. We have to show that i [
A
a
J
a
K
.
Now by Lemma 28.11 we have a
K
[
A
A
J
i. Monotony yields a
K
[
AA
J
i and
by Symmetry we have i [
AAJ
A
K
. The claim follows now from i [
A
A
J
and
Transitivity.
So we can dene a family (a
i
[ i I) to be independent over A if it is
independent for some ordering of I. Clearly (a
i
)
iI
is independent over A if
and only if a
i
[
A
a
j
[ j ,= i for all i. One calls a set B independent over B if
b [
A
(B b) for all b B.
The following lemma is a generalisation of Proposition 28.14.
Lemma 28.19. Let T be simple and 1 be an innite Morley sequence over A.
If 1 is indiscernible over Ac, then c [
A
1.
Proof. We may assume that 1 = (a
i
)
i<
. Consider any (x, a
0
, . . . , a
n1
)
tp(c/A1). Put b
i
= (a
ni
, . . . , a
ni+n1
). Then by Lemma 28.11 (b
i
)
i<
is again
a Morley sequence over A and (x, b
i
) [ i is consistent since realised by
c. We see from Proposition 28.14 that (x, a
0
, . . . , a
n1
) does not fork over
A.
Exercise 28.1.
If T is simple, there does not exist an ascending chain (p
)
|T|
+ of forking
extensions.
Exercise 28.2 (Diamond Lemma).
Assume T to be simple and p S(A). Let q be a nonforking extension of p and
r any extension of p. Then there is an Aconjugate r
of r with a nonforking
extension s which also extends q.
p
q
r
`
`
`
`
`
`
nf
nf
r
can be chosen in such a way that the domains of r
and q are independent

over A.
Exercise 28.3.
If T is simple and (a
i
)
iI
is an A-independent sequence, then
a
X
[
AaXY
a
Y
for all X, Y I where a
X
= a
i
[ i X.
Exercise 28.4.
If (x, b) divides over A and A B, there is some A-conjugate B
of B such
that (x, b) divides over B
.
Exercise 28.5.
If T is simple, then
ab [
A
B a [
A
B and b [
Aa
B.
Exercise 28.6.
Assume that T simple and b
1
. . . b
n
[
A
C. Then the sequence b
1
, . . . , b
n
is
independent over A if and only if it is independent over AC.
Exercise 28.7.
If T is simple and Aa [
B
C, then a acl(ABC) implies a acl(AB).
29 The Independence Theorem
The core of this section is the characterisation of simple theories in terms of
a suitable notion of independence. This is due to Kim and Pillay and will
be applied to pseudo-nite elds in Section 31. We will later specialise this
characterisation to stable theories. Unless explicitly stated otherwise, we
assume throughout this section that T is a simple theory.
Denition 29.1. For any set A we write nc
A
(a, b) if a and b start an innite
sequence of indiscernibles over A.
A formula (x, y) is called thick if there are no innite antichains, i.e. se-
quences (c
i
)
i<
where (c
i
, c
j
) for all i < j < . By compactness this says
that there is a bound k < on the length of nite antichains. See Exercise 29.2
for an explanation of the terminology.
Lemma 29.2. (T arbitrary.) For any set A and ntuples a, b the following are
equivalent:
a) nc
A
(a, b)
b) [= (a, b) for all thick (x, y) dened over A.
In particular, nc
A
is typedenable.
Proof. Let p(x, y) = tp(ab/A). By 15.3, a and b start an innite sequence of
indiscernibles if and only if there is a sequence (c
i
)
i<
with [= p(c
i
, c
j
) for i < j
if and only if for all p the complement of (C) contains arbitrarily long
antichains, and so
,[= (a, b) is not thick.
Corollary 29.3. ( T arbitrary.) If a and b have the same type over a model
M, there is some c such that nc
M
(a, c) and nc
M
(c, b).
Proof. We have to show [= z((a, z)(z, b)) for every thick formula (x, y)
L(M). We may assume that is symmetric.
2
Since M is a model, there is a
maximal antichain a
0
, . . . , a
k1
of in M. Thus, for some i we have [= (a, a
i
)
and hence [= (b, a
i
).
In Exercise 32.1 below we give a dierent proof of the corollary, independent
from Lemma 29.2.
Lemma 29.4. ( T arbitrary.) Let (b
i
)
i<
be indiscernible over A and (b
i
)
1i<
indiscernible over Aa
0
b
0
. Then there is some a
1
such that nc
A
(a
0
b
0
, a
1
b
1
).
2
I.e. |= x, y ((x, y) (y, x)).
Proof. Choose a
i
with tp(a
i
b
i
b
i+1
. . . /A) = tp(a
0
b
0
b
1
. . . /A). Let a
0
b
0
, a
1
b
1
, . . .
be a sequence of indiscernibles realizing the EM-type of a
0
b
0
, a
1
b
1
, . . . over A.
Since (b
i
)
1i<
is indiscernible over Aa
0
b
0
, we have
tp(a
i1
, b
i1
, b
i2
, . . . , b
in
/A) = tp(a
0
, b
0
, b
1
, . . . , b
n
/A)
for all i
1
< . . . < i
n
. So tp(a
0
, b
0
, b
1
, . . . /A) = tp(a
0
, b
0
, b
1
, . . . /A) and we may
assume that a
0
b
0
, a
1
b
1
, . . . is indiscernible over A.
Lemma 29.5. Let 1 be indiscernible over A and an innite initial segment
without last element.Then 1 is a Morley sequence over A.
Proof. Let 1 = (a
i
)
iI
and = (a
i
)
iJ
. It suces to show a
i
[
AJ
a
X
for all
i I J and all nite X I with X < i. But this follows from Lemma 27.10
as tp(a
X
/Aa
i
) is nitely satisable in A.
Proposition 29.6. If (x, a) does not fork over A and nc
A
(a, b), then (x, a)
(x, b) does not fork over A.
Proof. Let 1 be an innite sequence of indiscernibles over A containing a and
b. We extend 1 by an innite initial segment without last element. Let c be
a realisation of (x, a) independent from a over A. By Corollary 27.5 we may
assume 1 to be indiscernible over Ac.
It follows from Lemma 29.5 that 1 is a Morley sequence over A. So by
Lemma 28.19 we have c [
AJ
1. Transitivity now implies c [
A
1 and hence
the claim.
Lemma 29.7. Let nc
A
(b, b
) and a [
Ab
b
. Then there is some a
with
nc
A
(ab, a
).
Proof. Let (b
i
)
i<
indiscernible over A, b = b
0
and b
= b
1
. By Corollary 27.5
we may assume (b
i
)
1i<
to be indiscernible over Aab. The claim now follows
from Lemma 29.4.
Proposition 29.8. If (x, a) (x, b) does not fork over A, nc
A
(b, b
) and
a [
Ab
b
, then also (x, a) (x, b
) does not fork over A.

Proof. By Lemma 29.7 there is some a
such that nc
A
(ab, a
). Proposition 29.6
implies that (x, a) (x, b) (x, a
) (x, b

Corollary 29.9. Assume that (x, a) (x, b) does not fork over A, a [
A
b
b
and that b and b
have the same type over some model containing A. Then

(x, a) (x, b

Proof. By Corollary 29.3 there is some c such that nc
A
(b, c) and nc
A
(c, b
) By re-
placing a, if necessary, by a realisation of a nonforking extension of tp(a/Abb
)
to Abb
c, which exists by Corollary 28.7, we may assume that a [

A
bb
c. Propo-
sition 29.8 yields now rst that (x, a) (x, c) does not fork over A and then
that (x, a) (x, b

Corollary 29.10. Let a [
M
b, a
[
M
a, b
[
M
b, [= (a
, a) (b
, b) and
assume that a
and b
have the same type over M. Then (x, a) (x, b) does

not fork over M.
Proof. Choose a
such that tp(a
/M) = tp(bb
/M) and a
[
Ma
abb
. Then
by Transitivity we have
a
M
aa
bb
.
It follows that the sequences a, a
, a
and a, b, a
are both independent over A.

This implies
(1) a
M
aa
and
(2) a [
M
a
b
Since [= (a
, a
), (1) implies that (x, a) (x, a
) does not fork over M.

So (x, a) (x, b) does not fork over M by (2) and Corollary 29.9.
Theorem 29.11 (Independence Theorem). Suppose that b and c have the same
type over the model M and suppose that
B [
M
C, b [
M
B and c [
M
C.
Then there exists some d such that d [
M
BC, tp(d/B) = tp(b/B) and tp(d/C) =
tp(c/C).
Proof. By Corollary 29.10, tp(b/B) tp(c/C) does not fork over M. So we nd
a d such that d [
M
BC, tp(d/B) = tp(b/B) and tp(d/C) = tp(c/C).
Corollary 29.12. Let B
i
, i I, be independent over M and b
i
such that
b
i
[
M
B
i
all b
i
having the same type over M. Then there is some d such that
d [
M
B
i
[ i I and tp(d/B
i
) = tp(b
i
/B
i
) for all i.
Proof. Wellorder I and show the existence of d by recursively constructing
p
i
= tp(d/B
i
[ i I). The details are left as an exercise.
Theorem 29.13 (Kim-Pillay [30]). Let T be a complete theory and a [
0
A
B a
relation between nite tuples a and sets A and B invariant under automorphisms
and having the following properties:
a) (Monotony and Transitivity) a [
0
A
BC if and only if a [
0
A
B and
a [
0
AB
C.
b) (Symmetry) a [
0
A
b if and only if b [
0
A
a.
c) (Finite Character) a [
0
A
B if a [
0
A
b for all nite tuples b B.
d) (Local Character) There is a cardinal , such that for all a and B there
exists B
0
B of cardinality less than such that a [
0
B0
B
e) (Existence) For all a, A and C there is a
such that tp(a
/A) = tp(a/A)
and a
0
A
C.
f ) (Independence over models) Let M be a model, a [
0
M
b, a
0
M
a,
b
0
M
b and tp(a
/M) = tp(b
/M). Then there is some c such that c [
0
M
ab,
tp(c/Ma) = tp(a
/Ma) and tp(c/Mb) = tp(b
/Mb).
Then T is simple and [
0
= [
.
We have seen that in simple theories the relation [
satises the properties

of the previous theorem (for = [T[
+
).
Proof. We may assume to be a regular cardinal, otherwise just replace by
+
.
Assume now a [
0
A
b. We will use Lemma 27.4 to show that tp(a/Ab) does not
divide over A. So, let (b
i
)
i<
be a sequence of Aindiscernibles starting with
b = b
0
.
Claim: We can nd a model M containing A such that (b
i
)
i<
is indiscernible
over M and b
i
[
0
M
b
j
[ j < i for all i.
Proof: By Lemma 27.1 we can extend the sequence to (b
i
)
i
. Furthermore,
it is easy to construct an ascending sequence of models A M
0
M
1
. . .
such that for all i < all b
j
, (j < i) is contained in M
i
and (b
j
)
ij
is
indiscernible over M
i
. By Local Character there is some i
0
such that
b
0
Mi
0
b
j
[ i
0
j < . From the Indiscernibility it now follows that
b
i
[
0
Mi
0
b
j
[ i
0
j < i for all i. We can take M = M
i0
and the sequence
b
i0
, b
i0+1
, . . ..
Claim: We may assume a [
0
M
b.
Proof: By Existence we may replace a by a
with tp(a
/Ab) = tp(a/Ab) and

a
0
Ab
M and then apply Monotony and Transitivity.
We now nd elements a = a
0
, a
1
, . . . such that a
i
[
0
M
b
j
[ j i, q
i
(x) =
tp(a
i+1
/Mb
j
[ j i) = tp(a
i
/Mb
j
[ j i) and tp(a
i
b
i
/M) = tp(ab/M):
given a
0
. . . , a
i
, choose a
with tp(a
b
i+1
/M) = tp(ab/M). Now apply Inde-
pendence over Models to
b
j
[ j i [
0
M
b
i+1
, a
i
[
0
M
b
j
[ j i and a
0
M
b
i+1
to nd a
i+1
.
This implies that

i<
q
i
(x) is consistent and contains all p(x, b
i
) where
p(x, y) = tp(ab/M). By Lemma 27.4, tp(a/Ab) does not divide over A. Sim-
plicity of T now follows from Local Character and Proposition 28.5.
It is left to show a [
0
A
b if tp(a/Ab) does not divide over A. Using Ex-
istence we construct for any a sequence (b
i
)
i<
which is [
0
A
-independent
and for which tp(b
i
/A) = tp(b/A). If this sequence is suciently long, the same
argument as for Lemma 28.13 (but now using Monotony and Finite Char-
acter) yields an Aindiscernible sequence (b
i
)
i<
which is [
0
A
-independent as
well and satises tp(b
i
/A) = tp(b/A) and b = b
0
. We now apply Corollary 27.5
and obtain a
such that tp(a
/Ab) = tp(a/Ab) and so that (b
i
)
i<
is indiscernible
over Aa
. Local Character and Monotony yield the existence of i

0
with
a
0
A{b
i
|i<i0}
b
i0
.
Since
b
i0
[
0
A
b
i
[ i < i
0
we get
a
0
A
b
i0
and hence a [
0
A
b
from Symmetry and Transitivity using that tp(a
i0
/A) = tp(a
0
/A) =
tp(ab/A).
Corollary 29.14. The theory of the random graph is simple.
Proof. Dene A [
0
B
C by A C B and apply Theorem 29.13,
Exercise 29.1.
Let T be simple. Assume that the partial type (x, b) does not fork over A and
that 1 is an innite sequence of indiscernibles over A containing b. Show that
there is a realisation c of (x, b) such that c [
A
1 and 1 is indiscernible over
Ac.
Exercise 29.2.
A symmetric formula is thick if and only if there is no innite anti-clique, i.e.
a sequence (c
i
)
i<
, where [= (x, y) for all i ,= j. This explains the notation
nc
A
. Prove the following, without using Lemma 29.2:
1. The conjunction of two thick formulas is thick.
2. If (x, y) is thick, then
(x, y) = (y, x) is thick.

3. A formula is thick if and only if it is implied by a symmetric thick formula.
Exercise 29.3.
Let T be a simple theory and A a set of parameters. Assume that there is an
element b which is algebraic over A but not denable over A. Then the Inde-
pendence Theorem does not hold if in its formulation the model M is replaced
by A.
Exercise 29.4.
Fill in the details of the proof of Corollary 29.12.
Exercise 29.5.
Prove directly from the axioms in Theorem 29.13 that
a [
0
A
B a [
0
A
AB.
Exercise 29.6.
Let T be simple and p be a type over a model. Then p has either exactly one
nonforking extension to C (p is stationary) or arbitrarily many.
30 Lascar strong types
In this section we will prove a version of the Independence Theorem 29.11 over
arbitrary parameter sets A. For this we have to strengthen the assumption that
b and c have the same type over A to having the same Lascar strong type over
A. In what follows T is an arbitrary complete theory.
Denition 30.1. Let A be any set of parameters. The group Aut
f
(C/A) of Las-
car strong automorphisms of C over A is the group generated by all Aut(C/M)
where the M are models containing A. Two tuples a and b have the same Lascar
strong type over A if (a) = b for some Aut
f
(C/A). We denote this by
Lstp(a/A) = Lstp(b/A).
It is easy to see that two elements a and b have the same Lascar strong type
over A if and only if there is a sequence a = b
0
, b
1
, . . . , b
n
= b such that for all
for all i < n, b
i
and b
i+1
have the same type over some model containing A.
Lemma 30.2. Assume that T is simple. If (x, a) (x, b) does not fork over
A, Lstp(b/A) = Lstp(b
/A) and a [
A
b
b, then (x, a) (x, b
) does not fork

over A.
Proof. Choose a sequence b = b
0
, b
1
, . . . , b
n
= b
such that for each i < n, b

i
and
b
i+1
have the same type over some model containing A. By the properties of
forking we may assume that a [
A
b
0
b
1
. . . b
n
. We thus always have a [
A
b
i
b
i+1
and the claim follows by induction from Corollary 29.9.
As in the previous proof we will repeatedly use Existence (Corollary 28.7)
to assume that we have realisations of types which are independent from other
sets. This is also crucial in the following lemma:
Lemma 30.3. Assume T to be simple. For all a, A and B there is some a
such that Lstp(a
/A) = Lstp(a/A) and a
[
A
B.
Proof. By Existence and Symmetry, we can choose a model M A with
a [
A
M. By Existence again we can nd a
such that tp(a
/M) = tp(a/M)
and a
[
M
B, so the claim now follows from Corollary 29.3 and Transitivity.
A stronger statement will be proved in Exercise 30.3.
Corollary 30.4. Let Lstp(a/A) = Lstp(b/A). For all a
, B there exists b
such
that Lstp(aa
/A) = Lstp(bb
/A) and b
[
Ab
B.
Proof. Choose bb
such that Lstp(aa
/A) = Lstp(bb
/A) and b
by the Lemma
with b
[
Ab
B and Lstp(b
/Ab) = Lstp(b
/Ab). It is easy to see that this implies

Lstp(bb
/A) = Lstp(bb
/A).
Corollary 30.5. Let a [
A
b, a
[
A
a, b
[
A
b, Lstp(a
/A) = Lstp(b
/A) and
[= (a
, a) (b
, b). Then (x, a) (x, b) does not fork over A.

Proof. Choose a
by Corollary 30.4 such that Lstp(a
/A) = Lstp(bb
/A)
and a
[
Aa
abb
. Then proceed as in the proof of Corollary 30.5, but use

Lemma 30.2 instead of Corollary 29.9.
Theorem 30.6 (Independence Theorem). Let T be simple and suppose
Lstp(b/A) = Lstp(c/A),
B [
A
C, b [
A
B and c [
A
C.
Then there exists some d such that d [
A
BC, Lstp(d/B) = Lstp(b/B) and
Lstp(d/C) = Lstp(c/C).
Proof. By Corollary 30.5, tp(b/B) tp(c/C) does not fork over A. So we nd
some d such that d [
A
BC, tp(d/B) = tp(b/B) and tp(d/C) = tp(c/C). The
stronger claim about Lascar strong types is left as Exercise 30.2.
Corollary 30.7. Assume T to be simple and let B
i
, i I, be independent over
A and b
i
such that b
i
[
A
B
i
all b
i
having the same Lascar strong type over A.
Then there is some d such that d [
A
B
i
[ i I and Lstp(d/B
i
) = Lstp(b
i
/B
i
)
for all i.
Proof. Assume the B
i
are models containing A. The proof goes then as the
proof of Corollary 29.12.
Lemma 30.8. Let T be simple, a [
A
b and Lstp(a/A) = Lstp(b/A), there is
an innite Morley sequence over A containing a and b.
Proof. Consider p(x, y) = tp(ab/A). Starting from a
0
= a and a
1
= b we
recursively construct a long independent sequence (a
i
) of elements all having
the same Lascar type over A. If (a
i
[ i < ) is given, the p(a
i
, y) are realised by
elements b
i
with b
i
[
A
a
i
and Lstp(b
i
/A) = Lstp(a/A). By Corollary 30.7 there
is some a
with a
[
A
a
i
[ i < , [= p(a
i
, a
) for all i < and Lstp(a
/A) =
Lstp(a/A). If the sequence is suciently long, by Lemma 28.12 there is an A
indiscernible sequence (a
i
)
i<
such that [= p(a
i
, a
j
) for all i < j and furthermore
the sequence is independent over A because all types (a
j
/Aa
i
[ i < j) appear
in (a
i
). Since tp(a
1
a
0
/A) = tp(ba/A), we may assume a
0
= a and a
1
= b.
Corollary 30.9. Lstp(a/A) = Lstp(b/A) if and only if nc
2
A
(a, b). In particular,
the relation E
L
A
(a, b) Lstp(a/A) = Lstp(b/A) is typedenable.
Proof. Choose c with c [
A
ab and Lstp(c/A) = Lstp(a/A) = Lstp(b/A). By
Lemma 30.8, we have nc
A
(c, a) and nc
A
(c, b). Hence
E
L
A
(x, y) z (nc
A
(x, z) nc
A
(z, y))
and this is typedenable by Lemma 29.2.
It is an open problem whether in simple theories Lascar strong types are the
same as strong types (see Exercise 35.9).
Exercise 30.1.
Show that nc
A
(a, b) implies nc
acl(A)
(a, b).
Exercise 30.2.
Deduce Theorem 30.6 from the weaker version which claims only the equalities
tp(d/B) = tp(b/B) and tp(d/C) = tp(c/C).
Exercise 30.3.
Show that in arbitrary theories nc
A
(x, a) does not divide over A. Use this to
prove a stronger version of Lemma 30.3: Assume that T is simple. For all
a, A, B there is some a
such that nc
A
(a, a
) and a
[
A
B.
Exercise 30.4.
If nc
A
(a, b), there is some model M containing A such that tp(a/M) = tp(b/M).
Conclude that E
L
A
is the transitive closure of nc
A
.
Exercise 30.5.
A relation R C
n
C
n
is called bounded if there are no arbitrarily long an-
tichains, i.e. sequences (c
[ < ) with R(c
, c
) for all < < . Show

that the intersection of a family R
i
, (i I), of bounded relations is again
bounded.
Exercise 30.6.
We call a relation Ainvariant if it is invariant under all automorphisms in
Aut(C/A). Show that nc
A
is the smallest bounded Ainvariant relation.
Exercise 30.7.
Show that E
L
A
is the smallest bounded Ainvariant equivalence relation.
31 Example: Pseudo-nite elds
We now turn to an important example of simple theories, namely those of
pseudo-nite elds. A perfect eld K is called pseudo-nite if it is pseudo-
algebraically closed, i.e. if every absolutely irreducible ane variety dened over
K has a K-rational point and if its absolute Galois group is

Z, i.e., K has a
unique extension of degree n for each n 1. For background on pseudo-nite
elds and pronite groups see Appendix B 7.
Proposition 31.1. Let L
1
and L
2
be regular procyclic extensions of a eld K
and let L
2
be pseudo-nite. Then L
1
can be regularly embedded over K into an
elementary extension of L
2
.
Proof. By Lemma B 7.16, we may assume that N is a common regular procyclic
extension of the L
i
. As a regular procyclic extension of L
2
, N is 1free (by
Appendix B 7.14). Since L
2
is existentially closed in N, N is embeddable over
L
2
into an elementary extension L
2
of L
2
. Let N
denote the image of this

embedding. By Appendix B 7.14, L
2
/N
is regular.
Theorem 31.2. Let L
1
and L
2
be regular pseudo-nite extensions of K. Then
L
1
and L
2
are elementarily equivalent over K.
Proof. By Proposition 31.1 we obtain an alternating elementary chain. Its union
is an elementary extension of both L
1
and L
2
.
Corollary 31.3. Let L be pseudo-nite and Abs(L) the relative algebraic closure
of the prime eld in L, the absolute part of L. The elementary theory of L is
determined by the isomorphism type of Abs(L). A eld K algebraic over its
prime eld is the absolute part of some pseudo-nite eld if and only if it is
procyclic (this is always true in nite characteristic).
We now x the complete theory of a pseudo-nite eld and work in its
monster model C.
Corollary 31.4. Let K be a subeld of C, a and b tuples of elements C. Then
a and b have the same type over K if and only if the relative algebraic closures
of K(a) and K(b) in C are isomorphic over K via some isomorphism taking a
to b.
Proof. Let A and B be the relative algebraic closures of K(a) and K(b), respec-
tively. If a and b have the same type over K, then A and B are isomorphic in the
required way by Lemma B 6.13. Conversely, if A and B are isomorphic over K
by such an isomorphism, the claim follows immediately from Theorem 31.2.
Theorem 31.5. In pseudo-nite elds, algebraic independence has all the prop-
erties of forking listed in Theorem 29.13.
Proof. We keep working in C. All properties are clear except (Existence) and
(Independence over models).
For (Existence): Let K be a subeld of C and L and H two extensions
of K. We may assume that all three elds are relatively algebraically closed in
C. By Appendix B 7.16, there is a procyclic extension C of H (not necessarily
contained in C) containing a copy L
of L/K independent from H over K and

such that C/L
is regular. By Proposition 31.1, C can be regularly over H

embedded into C. Let L
denote the image of L
in C. Then L
and H are
independent over K and L
and L have the same type over K (see p. 32).

For (Independence over models): Let M be an elementary submodel
of C and let K and L be eld extensions independent over M. Assume further
that we are given extensions K
and L
so that K and K
as well as L and L
are independent over M and such that K
and L
have the same type over M.

We may assume that all these elds are relatively algebraically closed in
C. Then, if is a generator of G(C), the relative algebraic closures of if KK
and LL
in C are the xed elds of
= (KK
)
alg
in (KK
)
alg
and of
= (LL
)
alg
, respectively.
We now take another eld extension H/M (possibly outside C) isomorphic
to K
/M and L
/M and independent of KL over M. Then KK
and KH, and

LL
and LH, respectively, are isomorphic over M and the isomorphisms are
compatible with the given isomorphism between K
and L
. We transport
and
to (KH)
alg
and (LH)
alg
via these isomorphism and call the transported
automorphisms and . Clearly , and = (KL)
alg
agree on K
alg
and
L
alg
. They also agree on H
alg
, since H
alg
and H
alg
are both the unique
extension of M
alg
to G(H).
Since (KH)
alg
and (LH)
alg
are independent over H
alg
, and extend to
an automorphism
of (KH)
alg
(LH)
alg
which agrees with on K
alg
L
alg
. We
will see in Corollary 32.8 that
_
(KH)
alg
(LH)
alg
_
(KL)
alg
= KL.
So
and have a common extension to some automorphism of

(KH)
alg
(LH)
alg
(KL)
alg
which again can be extended to some automorphism
of (KLH)
alg
. Let C be the xed eld of , so C is procyclic. The relative
algebraic closures of KL, KH and LH in C are isomorphic to the relative
algebraic closures of KL, KK
and LL
in C. Let N be relative algebraic

closure of KL in C. C is a regular extension of N. So, by Proposition 31.1 we
nd a regular embedding of C into C over N. The image of H has the required
properties.
Note that we did not make use of the fact that M is a model, but only that
M is 1-free and C/M is regular.
We have now proved:
Corollary 31.6. Pseudo-nite elds are simple. Forking independence agrees
with algebraic independence.
Chapter 8
Stable Theories
Recall from Section 16, that a theory is -stable if [ S(A)[ for all sets A of
size at most . We call a theory stable if it is -stable for some . Another
equivalent denition will be given in Section 33: a theory is stable if no formula
has the order property. In order to apply the results of the previous chapter to
stable theories, we will eventually show that stable theories are simple and then
specialise the characterisation given in Theorem 29.13 to stable theories. But
before that we will introduce some of the classical notions of stability theory,
all essentially describing forking in stable theories.
32 Heirs and Coheirs
In this section we x an arbitrary complete theory T. For types over
models we here dene some special extensions to supersets, viz. heirs, coheirs,
and denable type extensions. All these extensions have in common that they
do not add too much new information to the given type. For stable theories,
we will see in Section 34 that these extensions coincide with the nonforking
extension (and this is in fact unique).
Denition 32.1. Let p be a type over a model M of T and q S(B) an
extension of p to B M.
1. We call q an heir of p if for every L(M)-formula (x, y) such that (x, b)
q for some b B there is some m m with (x, m) p.
2. We call q a coheir of p if q is nitely satisable in M.
It is easy to see that tp(a/Mb) is an heir of tp(a/M) if and only if tp(b/Ma)
is a coheir of tp(b/M).
148
CHAPTER 8. STABLE THEORIES 149
The following observation is trivial, but used frequently:
Remark 32.2. Suppose q is an heir of p S(M). If (x, b) q and [= (b),
then there is some m M with [= (m) and (x, m) p.
Lemma 32.3. Let q S(B) be a (co)heir of p S(M) and C an extension of
B. Then q can be extended to a type r S(C) which is again a (co)heir of p.
Proof. Suppose q is an heir of p. We have to show that
s(x) = q(x) (x, c) [ c C, (x, y) L(M), (x, m) p for all m M
is consistent: if there are formulas (x, b),
1
(x, c
1
), . . . ,
n
(x, c
n
) s(x) with
(x, b) q(x) whose conjunction is inconsistent, then as M is a model and q
is an heir of p there would be m, m
1
, . . . , m
n
M with (x, m) p and its
conjunction with
1
(x, m
1
), . . . ,
n
(x, m
n
) inconsistent. Since
i
(x, m
i
) p,
this is impossible. Any type r(x) S(C) containing s(x) is then an heir of p(x).
If q is a coheir of p, let r be a maximal set of L(C)formulas containing q
which is nitely satisable in M. Clearly, r is consistent. Let (x) L(C). If
neither nor belongs to r, then both r and r are not nitely
satised in M and so neither is r (cf. the proof of Lemma 5.2).
Denition 32.4. A type p(x) S
n
(B) is denable over C if the following
holds: for any Lformula (x, y) there is an L(C) formula (y) such that for
all b B
(x, b) p if and only if [= (b).
p is denable if it is denable over its domain B.
We write (y) as d
p
x(x, y) to indicate the dependence on p, (x, y) and the
choice of the variable tuple x. (So d
p
has the syntax of a generalised quantier,
cf. [52].) Thus, we have
(x, b) p if and only if [= d
p
x(x, b).
Note that d
p
x(x, y) is also meaningful for formulas with parameters in B.
Example:
In strongly minimal theories all types p S(A) are denable. To see this x
0
p of minimal Morley rank k and minimal degree and consider a formula
(x, y) without parameters. The discussion on page 114 shows that (x, a) p
if and only if MR(
0
(x)(x, a)) < k. By Corollary 26.4 this is an Adenable
property of a.
Lemma 32.5. A denable type p S(M) has a unique extension q S(B)
denable over M for any set B M, namely
(x, b) [ (x, y) L, b B, C [= d
p
x(x, b)
and q is the only heir of p.
Proof. The fact that the d
p
x(x, y) dene a type is a rst order property ex-
pressible in M and is hence true in any elementary extension of M. This proves
existence. On the other hand, if q is a denable extension of p, d
q
x(x, y)
and d
p
x(x, y) agree on M and hence in all elementary extensions proving
uniqueness. Clearly, q is an heir of p. If q
S(B) is dierent from q,

then for some (x, b) q
we have ,[= d
p
x(x, b). But there is no m with
(x, m) d
p
x(x, m) p, so q
is not an heir of p.
Lemma 32.6. A global type which is a coheir of its restriction to a model M
is invariant over M.
Proof. Let q S(C) be nitely satiable in M and Aut(C/M). Consider
a formula (x, c). Since c and (c) have the same type over M, (x, c)
(x, (c)) is not satisable in M. So (x, c) q implies (x, (c)) q.
We conclude by exhibiting coheirs in strongly minimal theories:
Proposition 32.7. Let T be strongly minimal, M a model and B an exten-
sion of M. Then tp(a/B) is an heir of tp(a/M) if and only if MR(a/B) =
MR(a/M).
Note that in strongly minimal theories MR(a/B) = MR(a/M) is equivalent to
a and B being geometrically independent over M (see Exercise 26.1). This is
a symmetric notion, which implies that in strongly minimal theories heirs and
coheirs coincide. We will later see in Corollary 34.7 (see also Corollary 36.11)
that this is actually true for all stable theories. Note also that this implies that
in strongly minimal theories types over models have a unique extension of the
same Morley rank, i.e. they have Morley degree 1. This is true in all totally
transcendental theories (Corollary 36.12, see also Corollary 36.4.)
Proof. Let k be the Morley rank of p = tp(a/M). Choose a formula
0
p of
same rank and degree as p. We saw in the Example on page 32 that the unique
heir q of p on B is dened by
(x) L(B)formula [ MR(
0
(x) (x, a)) < k.
On the other hand this set of formulas must be contained in all extensions of p
to B having rank k. So q is also the unique extension of p of rank k.
Corollary 32.8 (Hrushovski-Chatzidakis). Let K, L, H be algebraically closed
extensions of an algebraically closed eld M. If H algebraically independent
from KL over M, then
_
(KH)
alg
(LH)
alg
_
(KL)
alg
= KL.
Proof. We work in the monster model C. Let c be an element of the left hand
side. So there are tuples a (KH)
alg
and b (LH)
alg
such that c dcl(a, b),
witnessed by, say, [= (a, b, c). Furthermore, there are tuples a
K and h
1
H
such that a is algebraic over a
h
1
witnessed by, say, [=
1
(a, a
, h
1
). Similarly,
we nd [=
2
(b, b
, h
2
) for b. By independence and the heir property there are
h
1
, h
2
M such that
[= x, y (a, x, y)
1
(x, a
, h
1
)
2
(y, b
, h
2
).
Since K and L are algebraically closed, this implies c KL.
Exercise 32.1.
Use Example 28.10 and Lemma 32.6 to give an alternative proof of Corol-
lary 29.3.
Exercise 32.2.
Let T be a complete theory, M an saturated model.
1. Let (x) be a formula over M with Morley rank, and a formula with
arbitrary parameters which implies and has the same Morley rank. Then
is realized in M.
2. Let B an extension of M and MR(a/B) = MR(a/M) < . Show that
tp(a/B) is a coheir of tp(a/M).
It will follow from Corollaries 34.7 and 36.11 that in totally transcendental
theories this is true for arbitrary M. In fact this holds for arbitrary theories,
see Exercise 36.4.
Exercise 32.3 (Hrushovski-Pillay).
Let p(x) and q(y) be global types, and suppose that p(x) is A-invariant. We
dene a global type p(x) q(y) by setting (p q) B = tp(ab/B) for any
B A where b realizes q(y) B and a realizes p Bb. Show that p(x) q(y) is
well-dened, and Ainvariant if both p(x) and q(y) are.
33 Stability
In analogy with simple theories we here dene stable theories via (several equiv-
alent) properties of their formulas and note that this denition ts well with
the denition of -stability given in Section 16.
In this section, let T be a complete (possibly uncountable) theory.
Let S
(B) denote the set of all types over B; i.e. maximal consistent sets of
formulas of the form (x, b) or (x, b) where b B.
Denition 33.1. Let (x, y) be a formula in the language of T.
1. The formula is stable if there is an innite cardinal such that [ S
(B)[
whenever [B[ .
The theory T is stable if all its formulas are stable.
2. has the order property if there are elements a
0
, a
1
, . . . and b
0
, b
1
, . . . such
that for all i, j
[= (a
i
, b
j
) if and only if i < j.
3. (x, y) has the binary tree property if there is a binary tree (b
s
[ s
<
2) of parameters such that for all

2,
(n)
(x, b
n
) [ n < is
consistent. (We use the notation
0
= and
1
= .)
It is important to note that T is stable if and only if it is -stable for some
, see Exercise 33.7.
Remark 33.2. The notion of (x, y) having the order property is symmetrical
in x and y. This means that if (x, y) has the order property, then there are
elements a
0
, a
1
, . . . and b
0
, b
1
, . . . such that [= (a
i
, b
j
) if and only if j < i.
Proof. Apply Lemma 27.1 to 1 = (a
i
b
i
)
i<
and J = (, >).
Theorem 33.3. For a formula (x, y) the following are equivalent:
a) is stable.
b) [ S
(B)[ [B[ for any innite set B.

c) does not have the order property.
d) does not have the binary tree property.
Proof. a) d): Let be minimal such that 2
> . Then the tree I =

<
2 has
cardinality at most . If (x, y) has the binary tree property, by compactness
there are parameters b
s
, (s I), such that for all

2, q
=
()
(x, b
) [
< is consistent. Complete every q
to a type p
over B = b
s
[ s I.
Since the p
are pairwise dierent, we have [B[ < 2
[ S
(B)[.
d)c): Choose a linear ordering of I =

2 such that for all

2 and
n <
< n (n) = 1
If (x, y) has the order property, then by Lemma 27.1 one can nd a
i
and b
i
indexed by I such that
[= (a
i
, b
j
) if and only if i < j
Now the tree (x, b
s
), s
<
2, shows that has the binary tree property.
c)b): Let B be an innite set of parameters and [ S
(B)[ > [B[. For any a

the type of a over B is given by S
a
= b B
n
[ [= (a, b). Since [B[ = [B
n
[
we may assume for simplicity that n = 1 and so S
a
B. Applying the Erds-
Makkai Theorem (see Appendix C Theorem 9.1) to B and o = S
a
[ a C,
we obtain a sequence (b
i
[ i < ) of elements of B and a sequence (a
i
)
i<
such
that either b
i
S
aj
j < i or b
i
S
aj
i < j for all i, j. In the rst case
has the order property by denition. In the second case (x, y) has the order
property by Remark 33.2.
b) a): Clear.
If has the binary tree property witnessed, say, by (b
s
[ s
<
2), then
the family
s
=
_
n<|s|
s(n)
(x, b
sn
), s
<
2, is a binary tree of consistent
formulas. This shows that totally transcendental theories are stable. We will
see below (Corollary 34.6, also Exercise 33.11)) that stable theories are simple.
Remark 33.4. By Example 37.6 and Exercise 33.7 the theory of any R-module
is stable (but not necessarily totally transcendental) providing a rich class of
examples for stable theories. Note that the theory of the random graph is
simple by Corollary 29.14 but not stable (see Exercise 33.3).
Exercise 33.1.
T is unstable if and only if there is an Lformula (x, y) and elements a
0
, a
1
, . . . ,
ordered by ; i.e., such that
[= (a
i
, a
j
) i < j.
may contain parameters.
Exercise 33.2.
A formula (x, y) is said to have the independence property (IP) if there are
a
i
, i , such that for each A the set (x, a
i
): i A (x, a
i
): i ,
A is consistent. Show that T is unstable if it contains a formula with the
independence property.
Exercise 33.3.
Show that the theory of the random graph is not stable.
Exercise 33.4.
A formula (x, y) is said to have the strict order property (SOP) if there is a
sequence (a
i
)
i<
such that
[= y((a
i
, y) (a
j
, y)) i j.
T has the strict order property if there is a formula in T with the strict order
property. Show: T has the SOP if and only if there is a partial ordering with
innite chains denable in T
eq
. (For the denition of T
eq
see p. 160.)
Exercise 33.5.
Show that a theory with SOP is not simple.
Exercise 33.6.
(Shelah) If T is unstable, either there is a formula having the IP or a formula
having the SOP.
Exercise 33.7.
The following are equivalent:
a) T is stable.
b) T is stable for all such that
|T|
=
c) T is stable for some .
Exercise 33.8.
Show that for any innite there is a linear order of cardinality greater than
with a dense subset of size .
Exercise 33.9.
Show that the set of stable formulas is closed under Boolean combinations, i.e.
conjunction, disjunction, and negation. Use this to show that the theory T
Tree
,
dened on page 74, is stable.
Exercise 33.10.
Fix an L-formula (x, y). Let denote the class of Boolean combinations of
formulas of the form (x, b). Dene the rank R
as the smallest function

from formulas (x) to On such that
R
0 if is consistent;
R
+ 1 if there are innitely many

i
which are pairwise
inconsistent and such that R
(
i
) for all i.
Prove:
1. R
< if and only if (x) (x, y) is stable.

2. If R
< , then R
< .
Exercise 33.11.
If has the tree property, it is unstable.
This shows that stable theories are simple. We will give a dierent proof in
Corollary 34.6.
34 Denable types
Denability of types turns out to be a crucial feature of stable theories. We
show here that in stable theories the extensions of a type over a model given by
its denition agree with the nonforking extensions (and with heirs and coheirs).
Theorem 34.1. (x, y) is stable if and only if all types are denable.
Proof. Let A be a set of parameters of size [T[. If all types over A are
denable, there exists no more types over A than there are dening formulas,
i.e. at most [A[ many. So is stable.
For the converse assume that (x, y) is stable. Dene for any formula (x)
the degree D
() to be the largest n for which there is a nite tree (b

s
[ s
<n
2)
of parameters such that for every
n
2 the set (x)
(i)
(x, b
i
) [ i < n
is consistent. This is welldened since does not have the binary tree property.
Now, let p be a type over B. Let be a conjunction of formulas in p with n =
D
() minimal. Then (x, b) belongs to p if and only if n = D
((x) (x, b)).

This shows that p is denable.
Corollary 34.2. T is stable if and only if all types are denable.
Observe that the proof of Theorem 34.1 applies also to a proper class of
parameters. From this we obtain the following important corollary:
Corollary 34.3 (Separation of variables). Let T be stable and let F be a 0
denable class. Then any denable subclass of F
n
is denable using parameters
from F.
Proof. Let (a, C) be a denable subclass of F
n
. The type q = tp(a/F) is
denable over a subset of F by Corollary 34.2. Thus,
(a, C) = f F
n
[ [= d
q
x(x, f).
If the property in the conclusion of Corollary 34.3 holds for a 0denable
class F (in a not necessarily stable theory T), then F is called stably embedded.
At rst glance, the next lemma looks mysterious. In essence it states that
in stable theories heirs and coheirs coincide. We need it in the proof of Corol-
lary 36.3.
Lemma 34.4 (Harrington). Let T be stable and let p(x) and q(y) be global
types. Then for every formula (x, y) with parameters
d
p
x(x, y) q(y) d
q
y(x, y) p(x).
Proof. Let p, q and be denable over A. We recursively dene sequences a
i
and b
i
, i : if a
0
, . . . , a
n1
and b
0
, . . . , b
n1
have been dened, let b
n
be a
realisation of q Aa
0
, . . . , a
n1
and a
n
a realisation of p Ab
0
, . . . , b
n
. Then we
have for i < j
[= (a
i
, b
j
) [= d
q
y(a
i
, y) d
q
y(x, y) p(x)
and for j i
[= (a
i
, b
j
) [= d
p
x(x, b
j
) d
p
x(x, y) q(y).
Since does not have the order property, the claim follows.
Lemma 34.5. Let p S(C) be a global type.
1. If p is denable over A, then p does not divide over A.
2. If T is stable and p does not divide over the model M, p is denable over
M.
Note that for global types dividing and forking coincide (Exercise 27.2).
Proof. 1): Consider a formula (x, m) p and an innite sequence of indis-
cernibles m = m
0
, m
1
, . . . over A. If p is denable over A, all (x, m
i
) belong
to p. So (x, m) does not divide over A by Lemma 27.4.
2): Now assume that T is stable and p does not divide over the model M.
We will show that p is an heir of p M. By Corollary 34.2 and Lemma 32.5
this implies that p is denable over M. So assume that (x, b) p, we want to
show that (x, b
) p for some b
M.
Let 1 = (b
i
)
i<
be a Morley sequence of a global coheir extension of tp(b/M)
over M starting with b
0
= b (see Example 28.10 and Lemma 32.6). Since
tp(a/Mb) does not divide over M, Lemma 27.5 implies that we may assume that
1 is indiscernible over Ma. So we have [= (a, b
i
) for all i. By Corollary 34.2,
the type q = tp(a/Mb
i
[ i < ) is denable. Assume that the parameters of
d
q
x(x, y) are in Mb
0
, . . . , b
n1
. Since tp(b
n
/Mb
0
, . . . , b
n1
) is a coheir of
tp(b/M), and since [= d
q
x(x, b
n
), there is a b
M with [= d
q
x(x, b
). This
implies [= (a, b
) and so (x, b
) tp(a/M) = p M.
Corollary 34.6. Stable theories are simple.
Proof. Let p be a type over a model M. Then p is denable over some A M
of cardinality [T[. Let p
be the global extension of p given by the deni-

tion over A. By Lemma 34.5.1, p
and hence also p does not divide over A.

Proposition 28.5 implies that T is simple.
This implies in particular that forking and dividing coincides in stable the-
ories (see Proposition 28.15).
Corollary 34.7. Let T be a stable theory, p a type over a model M and A an
extension of M. Then p has a unique extension q S(A) with the following
equivalent properties.
a) q does not fork over M.
b) q is denable over M.
c) q is an heir of p.
d) q is a coheir of p.
Proof. By Lemma 34.5, q does not fork over M if and only if it is denable
over M. Since p is denable, we know by Lemma 32.5 that there is a unique
extension q which is denable over M, and which is also the unique heir of p.
To prove the equivalence with d) we may assume that A = M a for
a nite tuple a. Fix a realisation b of q. Then q = tp(b/Ma) is a coheir of
p = tp(b/M) if and only if tp(a/Mb) is an heir and hence, by the rst part of
the proof, a nonforking extension of tp(a/M). Now forking symmetry and the
rst part of the proof imply the desired.
Exercise 34.1.
Find a theory T and a type p over the empty set such that no denition of p
denes a global type. (A denition which denes a global type is called a good
denition of p, see Theorem 36.1).
Exercise 34.2.
Let T be an arbitrary complete theory and M be a model. Consider the following
four properties of a global type p :
(D) p is denable over M.
(C) p is a coheir of p M.
(I) p is Minvariant.
(H) p is an heir of p M.
Use the example T = T
DLO
and M = Q to show that DI, CI, DH are
the only logical relations between these notions.
Exercise 34.3.
Let p(x) S(M) be denable over B. Then, for any n, the map
r(y
1
, . . . , y
n
) (x, y
1
, . . . , y
n
) [ d
p
x r
denes a continuous section
n
: S
n
(B) S
n+1
(B). Show that this denes a
bijection between all types denable over B and all coherent families (
n
) of
continuous sections S
n
(B) S
n+1
(B).
Exercise 34.4.
Let (x) be a formula without parameters and let M be a model of T. Show
that (M) is stably embedded in M (i.e. every M-denable relation of (M) is
denable over (M)) if and only if for all n, every p(x
1
, . . . , x
n
) S
n
((M))
which contains (x
1
) . . . (x
n
) has a unique extension p
S
n
(M). If is
(absolutely) stably embedded and p is denable, show that p
is denable over
(M).
Exercise 34.5.
Call a formula
0
(x) stable if
0
(x) (x, y) is stable for all (x, y). We call a
type stable if it contains a stable formula. Prove:
1. Types with Morley rank are stable.
2. Stable types are denable.
Exercise 34.6.
Let T be stable, and p S(A). Show that p is denable over C if p is nitely
satisable in C. Furthermore for every (x, y), d
p
x(x, y) is a positive Boolean
combination of formulas (c, y), c C.
Exercise 34.7.
We call q a weak heir of p S(M) if the heir property holds for all (x, y)
without parameters. Show that in stable theories, weak heirs are in fact heirs.
Exercise 34.8.
In Corollary 34.7 prove the equivalence of c) and d) directly from Lemma 34.4.
35 Elimination of imaginaries and T
eq
This section is an excursion outside the realm of stable theories: for a model M
of an arbitrary theory T and any 0denable equivalence relation E(x, y) on n
tuples, we now consider the equivalence classes of M
n
/E as elements of a new
sort of socalled imaginary elements. Adding these imaginaries makes many
arguments more convenient. For certain theories, these imaginaries are already
coded in the original structure. However, if this is not already the case, then
adding imaginaries leads to a new theory T
eq
which does have this property so
that we do not run into an innite regression.
Let D be a denable class in C
n
. A nite tuple d C is called a canonical
parameter if d is xed by the same automorphisms of C which leave D invariant.
Lemma 23.10 implies that D is denable over d, and by Corollary 23.12 (1) d
is determined by D up to interdenability. We write d = D, or d = (x)
if D = (C). Note that the empty tuple is a canonical parameter for every
0denable class.
Denition 35.1. T eliminates imaginaries if any class e/E of a 0denable
equivalence relation E on C
n
has a canonical parameter d C.
Theorem 35.2. If T eliminates imaginaries, then the following holds:
1. Every denable class D C
n
has a canonical parameter c.
2. Every denable type p S(C) has a canonical base. This is a set B which
is pointwise xed by the same automorphisms which leave p invariant.
Proof. Write D = (C, e). Dene the equivalence relation E by
y
1
Ey
2
x (x, y
1
) (x, y
2
)
and let d be a canonical parameter of e/E. Then d is a canonical parameter of
D.
If d
p
is a denition of p, the set B = d
p
x(x, y) [ (x, y) Lformula is
a canonical base of p.
Lemma 35.3. Assume that T eliminates imaginaries. Let A be a set of para-
meters and D a denable class. Then the following are equivalent:
a) D is acl(A)denable.
b) D has only nitely many conjugates over A.
c) D is the union of equivalence classes of an Adenable equivalence relation
with nitely many classes (a nite equivalence relation).
Proof. Let d be a canonical parameter of D. Then D is denable over acl(A) if
and only if d belongs to acl(A). On the other hand D has as many conjugates
over A as d. So a) and b) are equivalent.
For the equivalence of b) and c), rst notice that any class of an Adenable
nite equivalence relation has only nitely many conjugates over A, which yields
c) b). For the converse, let D = D
1
, . . . D
n
be the conjugates of D over A.
Consider the nite equivalence relation E(c, c
) dened by
c D
i
if and only if c
D
i
for all i.
Clearly D is a union of Eclasses. E is denable and since it is invariant under
all A-automorphisms of C, it is in fact denable over A.
Note that elimination of imaginaries was only used for b) a).
The previous results show why it is convenient to work in a theory elim-
inating imaginaries. It is easy to see that a theory eliminates imaginaries if
every 0-denable equivalence relation arises from bers of a 0-denable func-
tion. While not all theories have this property (e.g. the theory of an equivalence
relation with innitely many innite classes), we now show how to extend any
complete theory T to a theory T
eq
(in a corresponding language L
eq
) which
does:
Let E
i
(x
1
, x
2
), (i I), be a list of all 0denable equivalence relations on
n
i
tuples. For any model M of T we consider the manysorted structure
M
eq
= (M, M
ni
/E
i
)
iI
,
which carries the home sort M and for every i the natural projection
i
: M
ni
M
ni
/E
i
The elements of the sorts S
i
= M
ni
/E
i
are called imaginary elements, the
elements of the home sort are real elements.
The M
eq
form an elementary class axiomatised by the (complete) theory
T
eq
which, in in the appropriate many-sorted language L
eq
, is axiomatized by
the axioms of T and for each i I by
y x
i
( x)
.
= y (y a variable of sort S
i
)
and
x
1
, x
2
(
i
( x
1
)
.
=
i
( x
2
) E
i
( x
1
, x
2
))
The algebraic (denable, respectively) closure of in M
eq
is denoted by acl
eq
(dcl
eq
, respectively).
The rst two statements of the following proposition explain why we consider
T
eq
as an inessential expansion of T.
Proposition 35.4. 1. Elements of C
eq
are denable over C in a uniform
way.
2. The 0denable relations on the home sort of C
eq
are exactly the same as
those in C.
3. T
eq
eliminates imaginaries,
Proof. 1: Every element of sort S
i
has the form
i
( a) for an n
i
tuple a from C.
2: We show that every L
eq
formula ( x) with free variables from the homesort
is equivalent to an Lformula
( x) by induction on the complexity of . If is

atomic, it is either an Lformula or of the form
i
( x
1
)
.
=
i
( x
2
), in which case
we set
= E
i
( x
1
, x
2
). We let commute with negations, conjunctions and
quantication over homesort variables. Finally, if y is a variable of sort S
i
, we
set
_
y( x, y)
_
= x
( x,
i
( x
))
.
3: We observe rst that C
eq
is the monster model of T
eq
, i.e. that every type
p(y) over a set A is realized in C
eq
: By part 1 we may assume that A C.
If y is of sort S
i
, the set ( x) = p(
i
( x)) is nitely satisiable. By part 2 is
equivalent to a set
of Lformulas.
has a realisation

b, which gives us a
realisation
i
(
b) of p.
It is now clear that
i
(e) is a canonical parameter of the class e/E
i
. By the
proof of Theorem 35.2 this implies that every relation in C
eq
which is denable
with parameters from C has a canonical parameter in C
eq
. On the other hand,
by part 1, every denable relation in C
eq
is denable in C
Corollary 35.5. T eliminates imaginaries if and only if in T
eq
every imaginary
is interdenable with a real tuple.
Proof. Since every automorphism of C extends (uniquely) to an automorphism
of C
eq
, a real tuple d is a canonical parameter of e/E
i
in the sense of T if and
only if it is a canonical paramter in the sense of T
eq
. But this is equivalent to
d being interdenable with
i
(e).
The proof of the following criterion for elimination of imaginaries shows how
T
eq
can be used:
Lemma 35.6. The following are equivalent:
a) T eliminates imaginaries and has at least two 0denable elements.
b) Every 0denable equivalence relation on C
n
is the bration of a 0denable
function f : C
n
C
m
.
Proof. b) a): If E is the bration of a 0denable function f : C
n
C
m
, we
have e/E = f(e). To see that there are at least two 0denable elements look
at the following equivalence relation on C
2
:
x
1
x
2
E y
1
y
2
(x
1
= x
2
y
1
= y
2
).
This has two classes, which are both 0denable. If E is the bration of a
0denable : C
2
C
m
the two images of are two dierent 0denable m
tuples.
a) b): Let E be a 0denable equivalence relation on C
n
. Every e/E is
interdenable with an element of some power C
me
. So by Exercise 23.11, e/E
belongs to a 0denable D C
n
/E with a 0denable injection f : D C
me
.
A compactness argument shows that we can cover C
n
/E by nitely many 0
denable classes D
1
, . . . , D
k
with 0denable injections f
i
: D
i
C
mi
. We may
assume that the D
i
are pairwise disjoint, otherwise we replace D
i
by D
i
(D
0
. . . D
i1
). Now, using the two 0denable elements, we can nd, for some
big m, 0denable injections g
i
: C
mi
C
m
with pairwise disjoint images. The
union of the g
i
f
i
is a 0denable injection from C
n
/E into C
m
.
Using 1. and 2. of the previous proposition, one can see that in general all
properties of T which concern us here are preserved when going from T to T
eq
.
Here are some examples:
Lemma 35.7. 1. T is
1
categorical if and only if T
eq
is
1
categorical.
2. T is stable if and only if T
eq
is stable.
3. T is stable if and only if T
eq
is stable.
Proof. 1. is clear.
For 2: Let A be a set of parameters in T
eq
of cardinality . A is contained
in the denable closure of some set B of cardinality of the home sort. For
any p S(B) we may rst take the unique extension of p to dcl
eq
(B) and then
its restriction to A. This denes a surjection S(B) S(A). Notice that we
now have to specify not only the number of variables but also the sorts for the
variables in the types.
If S
(A) consists of types of elements of the sort C

n
/E, S
n
(A) denotes the
ntypes of the home sort and the projection C
n
C
n
/E, then tp(
b/A)
tp((
b)/A) denes a surjection S

n
(A) S
(A). This shows that T

eq
is stable if
T is.
Of course 3. follows from 2. and Exercise 33.7. We still give a direct proof as
an example of how to translate between T and T
eq
. Let (y
1
, y
2
) be a formula
in T
eq
with the order property. If y
1
and y
2
belong to C
n1
/E
1
and C
n2
/E
2
,
respectively, there are tuples a
i
of the home sort such that
[= (
1
( a
1
),
2
( a
2
)) if and only if i < j.
By Proposition 35.4 the formula (
1
( x
1
),
2
( x
2
)) is equivalent to some L
formula ( x
1
, x
2
), which has the order property in T.
For applications the following special cases are often useful:
Denition 35.8.
1. T eliminates nite imaginaries if every nite set of ntuples has a canon-
ical parameter.
2. T has weak elimination of imaginaries if for every imaginary e there is a
real tuple c such that e dcl
eq
(c) and c acl(e).
Lemma 35.9. T eliminates imaginaries if and only if it has weak elimination
of imaginaries and eliminates nite imaginaries
Proof. This follows from the observation that T has weak elimination of imag-
inaries if and only if every imaginary e is interdenable with the canonical
parameter of a nite set of real ntuples: Indeed, if e dcl
eq
(c) and c acl(e),
and c
1
, . . . , c
m
are the conjugates of c over e, then e is interdenable with
c
1
, . . . , c
m
. If conversely e is interdenable with c
1
, . . . , c
m
, then e
dcl
eq
(c
1
. . . c
m
) and c
1
. . . c
m
x acl(e).
Lemma 35.10 (Lascar-Pillay). Let T be strongly minimal and acl() innite.
Then T has weak elimination of imaginaries.
Proof. Let e = c/E be an imaginary. It suces to show that c/E contains
an element algebraic over e or, more general, that every non-empty denable
X C
n
contains an element of A = acl(X). We proceed by induction on n.
For n = 1 there are two cases: if X is nite, it is a subset of A. If X is innite,
almost all elements of acl() belong to X. If n > 1, consider the projection
Y of X to the rst coordinate. Y contains an element a of acl(Y ), which is
a subset of A. By induction the bre X
a
contains an element b of acl(X
a
),
which is also a subset of A. So (a, b) is in X A.
Corollary 35.11. The theory T
ACFp
of algebraically closed elds of character-
istic p eliminates imaginaries.
Proof. By the preceding lemmas it suces to show that every theory of elds
eliminates nite imaginaries. Let S = c
0
, . . . , c
k1
be a set of ntuples c
i
=
(c
i,j
)
j<n
. Consider the polynomial
p(X, Y
0
, . . . , Y
n1
) =
i<k
(X
j<n
c
i,j
Y
j
).
An automorphism leaves p xed if and only if it permutes S. So the coecients
of p serve as a canonical parameter of S.
Lemma 35.12. A totally transcendental theory in which every global type has
a canonical base in C has weak elimination of imaginaries.
Proof. Let e = c/E be an imaginary and the Morley rank of the class c/E.
Let p be a global type of Morley rank which contains E(x, c). By assumption
p has a canonical base d C. Since there are only nitely many such p, d is
algebraic over e. Also e is denable from d since for an automorphism xing
p, c/E and c/E cannot be disjoint, so they must be equal. Clearly we may
assume that d is a nite tuple (see also Exercises 35.1 and 35.7).
Corollary 35.13. T
DCF0
eliminates imaginaries.
Proof. By quantier elimination every global type p(x) is axiomatised by its
quantier free part p
qf
(x), which is equivalent to a union of q
i
(x, dx, . . . , d
i
),
i = 0, 1, . . ., where the q
i
(x
0
, . . . , x
i
) are quantier free pure eld-theoretic types.
If C
i
is the canonical base of q
i
in the sense of T
ACF0
, C
0
C
1
. . . is a canonical
base of p.
Exercise 35.1.
Let D be a denable class. Assume that there is a set D which is xed by the
same automorphisms which leave D invariant. Show that D contains a canonical
parameter of D.
Exercise 35.2.
A theory T has weak elimination of imaginaries if and only if for every denable
class D there is a smallest algebraically closed set over which D is denable.
Exercise 35.3.
Use Exercise 35.2 to prove that the theories T
and T
DLO
(not easy) have weak
elimination of imaginaries. Show also that T
DLO
has elimination of imaginaries,
but T
does not.
Exercise 35.4.
Show that all extensions of p S(A) to acl(A) are conjugate over A. More
generally this is true for every normal extension B of A. These are sets which
are invariant under all Aut(C/A). Note that normal extensions must be
subsets of acl(A).
Exercise 35.5.
An algebraic type over A has a good denition (see Exercise 34.1 or p. 166) over
B A if and only if it is realised in dcl(B).
Exercise 35.6.
Let d be a canonical parameter of D. Then d is 0denable in the L P
structure (C, D).
Exercise 35.7.
Let T be totally transcendental and p a global type.
1. Show that p has a nite canonical base in C
eq
.
2. If p has a canonical base D C, then it has a nite base d C.
Exercise 35.8.
Show that Lemma 35.12 is true for stable theories. (Hint: In the proof of 35.12
replace p by a suitable E(x, y)type.)
Exercise 35.9.
Dene the strong type of a over A as stp(a/A) = tp(a/ acl
eq
(A)). Show that
stp(a/A) is axiomatised by
E(x, a): E(x, y) Adenable nite equivalence relation.
36 Properties of forking in stable theories
Except in Theorem 36.10 we assume throughout this section that T
is stable. We now collect the crucial properties of forking in stable theories.
As with simple theories, we will see in Theorem 36.10 that these properties
characterise stable theories and forking. Since we already know that stable
theories are simple, some of these properties are immediate. For completeness
and reference we restate them in the context of stable theories.
Let p S(B) be dened by L(B)formulas d
p
x. We call the denition
good if it denes a global type (or, equivalently, if it denes a type over some
model containing B).
Theorem 36.1. Let T be stable. A type p S(B) does not fork over A B if
and only if p has a good denition over acl
eq
(A).
Proof. If p does not fork over A, p has a global extension p
which does not

fork over A. Let M be any model which contains A. By Lemma 34.5, p
is
denable over M, so the canonical base of p
belongs to M
eq
. By Exercise 23.2
the canonical base belongs to acl
eq
(A).
If conversely p has a good denition over acl
eq
(A), p does not fork over
acl
eq
(A) and therefore does not fork over A.
Denition 36.2. A type is stationary if and only if it has a unique nonforking
extension to any superset.
Corollary 36.3 (Uniqueness). If T is stable and eliminates imaginaries, then
any type over an algebraically closed set is stationary.
Proof. Let A = acl
eq
(A). Let p
and p
two global nonforking extensions of

p S(A). Consider any formula (x, b), and let q(y) be a global nonforking
extension of tp(b/A). By Theorem 36.1, p
, p
and q are denable over A. Now

we apply Harringtons Lemma 34.4:
(x, b) p
d
p
x(x, y) q d
q
y(y, x) p
d
p
x(x, y) q (x, b) p
Corollary 36.4. In a stable theory, types over models are stationary.

Proof. This is immediate by the above proof since we can replace acl
eq
(A) every-
where by M. It follows also formally from Corollary 36.3 since M
eq
= dcl
eq
(M)
is an elementary substructure of C
eq
and so algebraically closed.
For the remainder of this section we also assume that T eliminates
imaginaries. In view of T
eq
(cp. Section 35) this is a harmless assumption.
As stable theories are simple we may rst collect some of the properties of
forking established in Section 28.
We keep using
a [
A
B
to express that tp(a/AB) does not fork over A.
Theorem 36.5. If T is stable, forking independence has the following proper-
ties:
1. (Monotony and Transitivity) Let A B C and q S(C). Then q does
not fork over A if and only if q does not fork over B and q B does not
fork over A.
2. (Symmetry)
a [
A
b = b [
A
a
3. (Finite character) If p S(B) forks over A, there is a nite subset B
0
B
such that p AB
0
forks over A.
4. (Local character) For p S(A) there is some A
0
A of cardinality at
most [T[ such that p does not fork over A
0
5. (Existence) Every type p S(A) has a nonforking extension to any set
containing A.
6. (Algebraic Closure)
(a) p S(acl(A)) does not fork over A.
(b) If tp(a/Aa) does not fork over A, then a is algebraic over A.
Proof. This is contained in 28.17, 28.16, 27.9, 28.5, 28.7 and 27.3.
The following properties do not hold in arbitrary simple theories.
Theorem 36.6. Assume T is stable.
1. (Conjugacy) If A M and M is strongly homogeneous for some >
max[T[, [A[, then all nonforking extensions of p S(A) to M are con-
jugate over A.
2. (Boundedness) Any p S(A) has at most 2
|T|
nonforking extensions for
every B A.
Proof. For 1. let q
1
and q
2
be nonforking extensions of p to M. Any A
automorphism of M which takes q
1
acl(A) to q
2
acl(A) (see Exercise 35.4)
takes q
1
to q
2
. Since types over algebraically closed sets are stationary by
Corollary 36.3, the claim now follows.
To prove 2. let A
0
be a subset of A of cardinality at most [T[ such that p
does not fork over A
0
. Then p has at most as many nonforking extensions as
p A
0
has extensions to acl(A
0
).
Corollary 36.7. Let T be stable and p S(A). Then p is stationary if and
only if it has a good denition over A.
Proof. Assume rst that p is stationary and let q be the global nonforking
extension. q is denable and invariant under all automorphisms over A, so q is
denable over A by Lemma 23.10. This shows that p has a good denition over
A. For the converse assume that p has a good denition over A. So p has a
nonforking global extension p
, denable over A by 36.1. Since all global non

forking extensions of p are conjugate over A, and p
is xed by all automorphisms

over A, p
is the only global nonforking extension of p.

Let p S(A) be a stationary type. We call p based on B if p is parallel
to some stationary type q dened over B, i.e. if p and q have the same global
nonforking extension (see Exercise 38.4). The canonical base Cb(p) of p is the
canonical base of the nonforking global extension of p (see Theorem 35.2).
Lemma 36.8. A stationary type p S(A) is based on B if and only if
Cb(p) dcl(B). So p does not fork over B A if and only if Cb(p) acl(B).
Proof. Let r be the global nonforking extension of p and q = r B. Assume
that p is based on B. Then q is stationary and r the unique nonforking exten-
sion of q. By Corollary 36.7 q has a good denition over B, which also denes
r. So r is denable over B, which means Cb(p) dcl(B).
If, conversely, r is denable over B, we know by Theorem 36.1 that r does
not fork over B and that q is stationary by Corollary 36.7.
The last statement follows from the easy fact that p does not fork over B if
and only if p is based on acl(B).
For A B let N(B/A) be the set of all types over B that do not fork over
A. By Remark 27.8, N(B/A) is closed in S(B). For future reference we record
the following useful fact:
Theorem 36.9. (Open mapping) The restriction map : N(B/A) S(A) is
open.
Proof. It is easy to see that we may replace B by C. If (q) = (q
) for some
q, q
N(C/A), then q and q
are conjugate. So if O is a (relative) open subset

of N(C/A), then
O
=
1
((O)) =
_
(O) [ Aut(C/A)
is again open. So
S(A) (O) = (N(C/A) O
)
is closed since it is the image of a closed set.
Theorem 36.10 (Characterisation of Forking). Let T be a complete theory and
n > 0. T is stable if and only if there is a special class of extensions of ntypes,
which we denote by p q, with the following properties.
a) (Local character) There is a cardinal such that for q S
n
(C) there is
C
0
C of cardinality at most such that q C
0
q.
b) (Weak Boundedness) For all p S
n
(A) there is a cardinal such that p
has, for any B A, at most extensions q S
n
(B) with p q.
If satises in addition
c) (Invariance) is invariant under Aut(C),
d) (Existence) For all p S
n
(A) and A B, there is q S
n
(B) such that
p q,
e) (Transitivity) p q r implies p r,
f ) (Weak Monotony) p r and p q r implies p q,
then coincides with the nonforking relation.
Proof. Assume properties a) and b). First, choose
big enough so that for all

A
0
of cardinality at most all ntypes over A
0
have at most
extensions
to any supserset.
Let A be a set of parameters. Then the number of ntypes over A is bounded
by the product of the number of subsets A
0
of A of cardinality at most , times
a bound for the number of types p over A
0
, times a bound for the number of
extensions of p S
n
(A
0
) to A. So we have
[ S
n
(A)[ [A[
2
max(,|T|)

and it follows that T is stable if
= and max([T[,
), hence stable
by Exercise 33.7.
Now let have the properties a) to f). Consider a type p S
n
(A) with an
extension q S
n
(B).
Assume rst that p q. Let be the cardinal given Boundedness applied to
p. By Exercise 27.4 there is an extension M of B such that every r S
n
(M)
which forks over A has more than conjugates over A. By Existence and
Transitivity q has an extension r to M such that p r. By Invariance we
have p r
for all conjugates r
. So r has no more than conjugates, which

implies that r does not fork over A and that q is a nonforking extension of p.
Now assume that q is a nonforking extension of p. Choose an extension M of B
which is suciently saturated in the sense of Theorem 36.6.1. Let r S
n
(M) be
a nonforking extension of q and r
S
n
(M) such that p r
. By the above r
is a nonforking extension of p. So r and r
are conjugate over A. This implies

p r and p q by Weak Monotony.
Corollary 36.11. Let T be totally transcendental, p S(A) and q an extension
of p to some superset of A. Then q is a nonforking extension if and only if
MR(p) = MR(q). Hence p is stationary if and only if it has Morley degree 1.
Proof. In a totally transcendental theory, extensions having the same Morley
rank satisfy the conditions of Theorem 36.10 (see Section 24).
The same is true for types with Morley rank in stable theories (see Exercise
36.4). It follows in particular that in totally transcendental theories for any type
p S(A) there is a nite set A
0
A such that p does not fork over A
0
. Stable
theories with this property are called superstable, see section 37.
Corollary 36.12. In a totally transcendental theory, types over models have
Morley degree 1.
Proof. This follows from Corollaries 36.4 and 36.11.
Corollary 36.13. If T is strongly minimal, we have A [
B
C if and only if
A and C are algebraically independent over B in the pregeometry sense, i.e. if
dim(a/B) = dim(a/BC) for all nite tuples a A.
Proof. This follows from Theorem 26.2 and Corollary 36.11.
Corollary 36.14. Let K F
1
, F
2
be dierential elds contained in a model of
T
DCF0
. Then F
1
[
K
F
2
if and only if F
1
and F
2
are algebraically independent
over K.
Proof. By Exercises 9.2 and 28.7, F
1
[
K
F
2
implies the algebraic independence.
For the converse we may assume that K is algebraically closed. So let F
1
and
F
2
be algebraically independent over K. By existence of nonforking extensions
choose a copy F
of F
1
satisfying the same type over K and forking independent
of F
2
over K. Then F
is algebraically independent of F
2
over K. Since K is
algebraically closed, F
and F
1
satisfy the same type over F
2
in the sense of
eld theory. Since F
, F
1
and F
2
are d-closed and F
and F
1
are isomorphic
as d-elds, we conclude that F
and F
1
have the same type over F
2
. Thus,
F
1
[
K
F
2
.
Exercise 36.1 (Finite equivalence relation theorem).
Let A B and let tp(a/B) ,= tp(b/B) be types which do not fork over A. Then
there is an A-denable nite equivalence relation E with q
1
(x)q
2
(y) E(x, y).
Exercise 36.2.
If a and b are independent realisations of the same type over acl(A), then
tp(a/Ab) is stationary.
Exercise 36.3.
Prove:
1. Let p S(A) be stable and q an extension of p. Then q does not fork over
A if and only if q has a good denition over acl
eq
(A).
2. Stable types over sets which are algebraically closed in C
eq
are stationary.
Exercise 36.4.
1. Let T be stable, p S(A) a type with Morley rank and q an extension of
p to some superset of A. Then q is a nonforking extension if and only if
MR(p) = MR(q). It follows that a type with Morley rank is stationary if
and only if it has Morley degree 1.
2. Show that the same is true for an arbitrary theory T.
Exercise 36.5.
Assume that T is stable. For any p S(A) there is some A
0
A of cardinality
at most [T[ such that p is the unique nonforking extension of p A
0
to A. If p
has Morley rank, A
0
can be chosen as a nite set.
Exercise 36.6 (Forking multiplicity).
(T stable) Dene the multiplicity of a type p as the number mult(p) of its global
nonforking extensions. Show:
1. If p is algebraic, mult(p) is the number of realisations of p. (This was our
previous denition of the multiplicity of algebraic types. See page 94 and
Remark 20.3.)
2. If T is countable, then mult(p) is either nite or 2
0
.
3. If p has Morley rank, show that mult(p) = MD(p).
Exercise 36.7.
Let G be a totally transcendental group. Show:
1. If a and b are independent elements, then MR(a b) MR(a). Equality
holds if and only if a b and b are independent.
2. Assume that G is saturated. Then an element a is generic, i.e. MR(a) =
MR(G), if and only if for all b G
a [
b a b [
b.
It can be shown that all saturated stable groups contain generic elements,
i.e., elements satisfying property 2), see Poizat [41], or Wagner [54].
Exercise 36.8 (Group conguration).
Let G be a totally transcendental group, and let a
1
, a
2
, a
3
G be independent
generic elements, i.e. elements of maximal rank in Th(G). Put b
1
= a
1
a
2
,
b
2
= a
1
a
2
a
3
and b
3
= a
2
a
3
. We consider these six elements as the points
of a geometry with lines A
0
= a
1
, b
1
, a
2
, A
1
= a
2
, b
3
, a
3
, A
2
= a
1
, b
2
, b
3
and A
3
= b
1
, b
2
, a
3
. It is easy to see that every permutation of the four lines
gives rise to an automorphism of this geometry.
a
1
a
3
b
1
b
2
b
3
a
2
`
`
`

`
`
`

`
`
``
``
Show:
1. Each point an a line is algebraic over the other two points on the line.
2. Any three non-collinear points are independent.
Any family of points a
1
, a
2
, a, a
3
, b
1
, b
2
, b
2
, b
3
with these properties
1
is called
a group conguration. Hrushovski proved that whenever a totally transcenden-
tal structure contains a group conguration, there is a group denable in this
structure whose Morley rank equals the Morley rank of any of the points. For
more details see Bouscaren [10], Wagner [54] or Pillay [39].
Exercise 36.9.
Let T be an arbitrary complete theory, not necessarily stable. For any set of
parameters A the map S(acl(A)) S(A) is open. (For stable theories, this is
just the Open Mapping Theorem.)
1
It is easy to see that 2) can equivalently be replaced by: 2a) Any two points on a line are
independent and 2b) Any two lines are independent over their intersection.
37 SU-rank and the stability spectrum
We saw that in totally transcendental theories forking is governed by the Mor-
ley rank. The SU-rank, which we dene here, generalizes this to superstable
theories. We use it to show that the stability spectrum of countable theories is
rather restrictive: there are only four possibilities for the class of cardinals in
which a countable theory is stable.
Denition 37.1. Let T be a simple theory. We dene SU(p) for a type p
by recursion on :
SU(p) 0 for all types p;
SU(p) + 1 if p has a forking extension q with SU(q) ;
SU(p) (for a limit ordinal ) if SU(p) for all < .
and the SU-rank SU(p) of p as the maximal such that SU(p) . If there is
no maximum, we set SU(p) = .
Lemma 37.2. Assume T to be simple. Let p have ordinal valued SU-rank and
let q be an extension of p. Then q is a nonforking extension of p if and only if
q has the same SU-rank as p. If p has SU-rank , then so does any nonforking
extension.
Proof. It is clear that the SU-rank of an extension cannot increase. So it is
enough to show for all that SU(p) implies SU(q) whenever q is a
nonforking extension of p S(A). The interesting case is where = + 1
is a successor ordinal. Then p has a forking extension r with SU(r) . By
the Diamond Lemma (Exercise 28.2) there is an Aconjugate r
of r with a
nonforking extension s which also extends q. By induction SU(s) . But s
is a forking extension of q, so SU(q) .
Since every type does not fork over a set of cardinality at most [T[, there
are at most 2
|T|
dierent SU-ranks. Since they form an initial segment of the
ordinals, all ordinal ranks are smaller than (2
|T|
)
+
. (Actually one can prove
that they are smaller than [T[
+
.) It follows that every type of SU-rank has
a forking extension of SU-rank .
Denition 37.3. A simple theory is supersimple if every type does not fork over
a nite subset of its domain. A stable, supersimple theory is called superstable.
Note that totally transcendental theories are superstable.
Lemma 37.4. T is supersimple if and only if every type has SU-rank < .
Proof. If SU(p) = , there is an innite sequence p = p
0
p
1
. . . of forking
extensions of SUrank . The union of the p
i
forks over every nite subset of
its domain. If p S(A) forks over every nite subset of A, there is an innite
sequence A
0
A
1
. . . of nite subsets of A such that p A
i+1
forks over A
i
.
This shows that p has SU-rank .
Let T be a complete theory. The stability spectrum Spec(T) of T is the class
of all innite cardinals in which T is stable.
Theorem 37.5. Let T be a countable complete theory. There are four cases:
1. T is totally transcendental. Then Spec(T) = [
0
.
2. T is superstable but not totally transcendental. Then Spec(T) = [
2
0
.
3. T is stable but not superstable. Then Spec(T) = [
0
= .
4. T is unstable. Then Spec(T) is empty.
Proof. 1: This follows from Theorem 16.5.
2: Let T be superstable and [A[ = . Since every type over A does not fork
over a nite subset of A, an upper bound for the size of S(A) can be computed
as the product of
the number of nite subsets E of A,
the number of types p S(E),
the number of nonforking extensions of p to A.
So we have [ S(A)[ 2
0
2
0
= max(2
0
, ). If T is not totally transcendental,
the proof of Theorem 16.5 shows that T cannot be stable in cardinals smaller
than 2
0
.
3: If T is stable, then T is stable whenever
0
= by Exercise 33.7. If T is
not superstable, the proof of Lemma 37.4 shows that there is a type p over the
empty set of innite SU-rank with a forking extension p
of innite SU-rank.
Let q be a nonforking global extension of p
and let
0
. By Exercise 27.4 q
has many dierent conjugates q
, ( < ). Choose A
0
of size such that all
p
= q
A
0
are dierent. By Lemma 37.2 the p
have again innite SU-rank.

Continuing in this manner we get a sequence A
0
A
1
. . . of parameter sets
and a tree of types p
0,...,n
S(A
n+1
), (n < ,
i
< ). We may assume
that all A
i
have cardinality . Each path through this tree denes a type over
A =
n<
A
n
. So we have [ S(A)[
0
.
4: This follows from Theorem 33.3.
The spectrum of uncountable theories is more dicult to describe, see [47,
Chapter III].
Example 37.6 (Modules). For any R-module M, the L
M(R)
-theory of M is
-stable if =
|R|+0
.
See [42] or [56] for more on the model theory of modules.
Proof. Let B be a subset of some model N of Th(M), [B[ . Let S
N
(B)
denote the set of all complete types over B which are realized in N. Every type
tp(a/B) is axiomatised by
tp
(a/B) = tp
+
(a/B) tp
(a/B),
where
tp
+
(a/B) = (x, b) [ ppformula, b B, N [= (a, b)
and
tp
(a/B) = (x, b) [ ppformula, b B, N [= (a, b).

Clearly, tp
is determined by tp
+
.
By Corollary 9.8, tp
+
(a/B) contains - up to equivalence - at most one for-
mula (x, b) for any ppformula (x, y). Hence tp
+
(a/B) is determined by a
partial map f from the set of ppformulas to the set of nite tuples in B in the
sense that it is axiomatised by (x, f()): ppformula. Hence we have
[ S
N
(B)[ ([B[ +
0
)
|R|+0
.
Thus T is -stable in every with =
|R|+0
.
Example 37.7 (Separably closed elds). The theory SCF
p,e
of separably closed
elds is stable for all with =
0
-
Proof. Let L be a model of SCF
p,e
. Fix a pbasis b
1
, . . . , b
e
and consider the
corresponding -functions
. Now let K be a subeld of cardinality . By

Theorem 9.20 every type tp(a/K) is axiomatized by Boolean combinations of
equations t(x)
.
= 0 where the t(x) are L(c
1
, . . . , c
e
,
)
p
eterms with parame-
ters from K. To compute an upper bound for the number of types over K we
may assume that K contains the pbasis and is closed under the -functions. It
is now easy to see that every t(x) is equivalent to a term p(
1
(x), . . . ,

k
(x))
where p(X
1
, . . . , X
k
) K[X
1
, . . . , X
k
] and the
i
are iterated -functions:
1
...
m =
1 . . .
m
So, if
0
,
1
, . . . is a list of all iterated -functions, the type of a over K
is determined by the sequence of the quantier free Ltypes of the tuples
(
0
(a), . . . ,
n
(a)) over K. By Example 16.2 for each n there are only many
such types. So we we can bound the number of types over K by
0
.
Exercise 37.1.
Show that the types of SU-rank 0 are exactly the algebraic types. Show also
that a type is minimal if and only if it is stationary and has SUrank 1.
Exercise 37.2.
Let T be an arbitrary theory. Dene the U-rank (or Lascar rank) U(p) of a
type p S(A) as its SUrank, except for the clause
U(p) + 1 if for any there is a set B A to which p has at
least many extensions q with U(q) .
Show that in stable theories Urank and SUrank coincide.
Exercise 37.3.
Show that in simple theories SU(p) MR(p).
Exercise 37.4.
Let (x, y) be a formula without parameters and k natural number. Dene the
rank D(p, , k) by
D(p, , k) + 1 if p has an extension q with D(q, , k) and which
contains a formula (x, b) which divides over the do-
main of p with respect to k.
Show for simple T:
1. D(p, , k) is bounded by a natural number which depends only on and
k.
2. Let q be an extension of p. Then p is a nonforking extension of p if and
only if D(p, , k) = D(q, , k) for all and k.
Exercise 37.5.
A countable theory is stable if and only if it is superstable, small and if every
type (over a nite set) has nite multiplicity.
Exercise 37.6 (Lachlan).
Show that in an
0
categorical theory there are only nitely many strong 1-
types over a nite set. Conclude that an
0
categorical superstable theory is
-stable.
Note that there are stable
0
categorical theories which are not -stable
(see [22]).
Exercise 37.7.
Let T be a stable theory. Assume that there is a sequence of denable equiva-
lence relations E
0
E
1
. . . such that every E
i
class contains innitely many
E
i+1
classes. Show that T is not superstable.
Exercise 37.8.
Use Exercise 37.7 to show that a superstable group has no innite descending
sequence of denable subgroups G = G
0
G
1
G
2
G
3
. . . each of innite
index in the previous one. Conclude from this that SCF
p,e
is not superstable if
e > 0.
Exercise 37.9.
Prove that a module M is totally transcendental if and only if it has the dcc on
ppdenable subgroups of M. M is superstable if and only if there is no innite
descending sequence of ppdenable subgroups each of which is of innite index
in its predecessor.
Chapter 9
Prime extensions
In this chapter we return to questions around the uniqueness of prime extensions.
We will now prove their uniqueness for totally transcendental theories and for
countable stable theories having prime extensions.
38 Indiscernibles in stable theories
We assume throughout this section that T is complete, stable and
eliminates imaginaries. Indiscernibles in stable theories are in fact indis-
cernible for every ordering of the underlying set. More importantly, we show
that they form a Morley sequence in some appropriate stationary type.
A family 1 = (a
i
)
iI
of tuples is totally indiscernible over A, if
C [= (a
i1
, . . . a
i
k
) (a
j1
, . . . a
j
k
)
for all L(A)formulas and sequences i
1
. . . i
k
, j
1
. . . j
k
of pairwise distinct
indices.
Lemma 38.1. If T is stable, indiscernibles are totally indiscernible.
Proof. Assume that 1 = (a
i
)
iI
is indiscernible over A, but not totally in-
discernible over A. By Lemma 27.1 we may assume (I, <) = (Q, <). Be-
cause any permutation on 1, . . . n is a product of transpositions of neigh-
bouring elements, there are some L(A)formula (x
1
, . . . , x
n
), rational numbers
r
1
< . . . < r
n
and some j Q such that
[= (a
r1
, . . . , a
rj
, a
rj+1
, . . . , a
rn
)
and
[= (a
r1
, . . . , a
rj+1
, a
rj
, . . . , a
rn
).
178
CHAPTER 9. PRIME EXTENSIONS 179
The formula (x, y) = (a
r1
, . . . , x, y, . . . , a
rn
) orders the elements (a
r
), (r
j
<
r < r
j+1
). By Exercise 33.1, this contradicts stability.
Let p S(A) be a stationary type. Recall that a Morley sequence in p is a
set of realisations of p independent over A. Morley sequences (a
)
<
are easy
to construct as follows: choose a
realising the unique nonforking extension

of p to A a
[ < . By Corollary 28.18, a
[ < is independent
over A. Since any Morley sequence arises in this way (independently of the
enumeration), the Morley sequences of p are uniquely determined by up to
isomorphism over A. If p is not an algebraic type, the a

and hence indiscernible over A.
Theorem 38.2. If T is stable, then any innite sequence of indiscernibles over
A is a Morley sequence for some stationary type dened over some extension of
A.
Proof. Let 1 = (a
i
)
iI
be indiscernible over A. Notice that for all formulas
(x, b) the set
J
= i I[ [= (a
i
, b)
is nite or conite in I: otherwise for all J I the set of formulas
(a
i
, y) [ i J (a
i
, y) [ i , J
would be consistent. So there would be 2
|I|
many types over 1, contradicting
stability of T.
This shows that for every either J
or I J
is bounded by some k
(which
depends only on ).
The average type of 1 is a global type dened by
Av(1) = (x, b) [ b C, [= (a
i
, b) for all but nitely many i I.
By the preceding remarks, this is a complete type. Let I
0
be an innite
subset of I. Since (x, b) Av(1) if and only if i I
0
[ [= (a
i
, b) contains
more than k
many (and hence innitely many) elements, Av(1) is denable

over 1
0
. Hence Av(1) does not fork over 1
0
and its restriction to 1
0
is stationary
(cf. Theorem 36.1 and Corollary 36.7.)
It is easy to see that all a
i
, i I I
0
, realise the type
p = Av(1) A1
0
.
As this is also true for all I
0
I
0
, we see that a
i
, i I I
0
, forms a Morley
sequence for p.
At the beginning of the proof we can now replace 1 by an innite set of
indiscernibles 1
containing 1 as a coinnite subset which shows 1 to be a

Morley sequence for p
= Av(1
) A(1
1).
Exercise 38.1.
If p is stationary and q a nonforking extension of p, then any Morley sequence
of q is also a Morley sequence for p.
Exercise 38.2.
Let p S(A) be stationary and 1 a Morley sequence of p.
a) Let B A and 1
0
1 such that B [
AI0
1. Then 11
0
is a Morley sequence
of the nonforking extension of p to B.
b) Av(1) is the nonforking global extension of p.
Exercise 38.3.
We call indiscernibles 1
0
and 1
1
parallel if there is some innite set such that
1
0
and 1
1
are indiscernible. Show that 1
0
and 1
1
are parallel if and only if
they have the same average type.
Exercise 38.4.
Stationary types are called parallel if they have the same global nonforking
extension. Show that two types are parallel if and only if two (or all) of their
innite Morley sequences are parallel.
Exercise 38.5.
Show the converse of Lemma 38.1: if all indiscernibles sequences are totally
indiscernible, then T is stable.
39 Totally transcendental theories
Let T be a totally transcendental theory. In this section we will prove the
following theorem:
Theorem 39.1 (Shelah [46]). Let T be totally transcendental.
1. A model M is a prime extension of A if and only if M is atomic over A
and does not contain an uncountable set of indiscernibles over A.
2. Prime extensions are unique.
We rst aim to show that a constructible set does not contain an uncountable
set of indiscernibles.
Lemma 39.2. Let 1 be indiscernible over A and B a countable set. Then 1
contains a countable subset 1
0
such that 1 1
0
is indiscernible over AB1
0
.
Proof. By Theorem 38.2, 1 is a Morley sequence of some stationary types over
some extension A
of A. Since T is totally transcendental, we only need to extend

A by nitely many elements (this follows using Exercise 36.5 and Exercise 38.1).
So we may assume A
= A. For every nite tuple b in B there is some nite set

1
0
such that b [
AI0
1. In this way we nd a countable set 1
0
with
B1
0
[
AI0
1.
It now follows from Exercise 38.2 that 11
0
is a Morley sequence over AB1
0
.
We need a bit of set theory: a club D
1
is a closed and unbounded subset
where closed means that sup( D) D for all
1
.
Theorem 39.3 (Fodor). Let D
1
be a club and f : D
1
a regressive
function, i.e., f() < for all D. Then f is constant on an unbounded
subset of
1
.
Proof. Suppose not, then for each
1
, the bre D
= x D: f(x) =
is bounded. Construct a sequence
0
<
1
< . . . of elements of D as follows.
Let
0
be arbitrary. If
n
is constructed, choose for
n+1
an upper bound of
<n
D
. So we have for all D

n+1
f()
n
.
For = sup
n<

n
, this implies f() . A contradiction.
Recall from Section 17 that a set B = b
<
is a construction over A if all
tp(b
/AB
) are isolated where as above B
= b
[ < ). A subset C B
is called construction closed if for all b
C the type tp(b
/AB
) is isolated by
some formula over A (B
C).
The following lemma holds for arbitrary T.
Lemma 39.4. Let B = b
<
be a construction over A.
1. Any union of construction closed sets is construction closed.
2. Any b B is contained in a nite construction closed subset of B.
3. If C B is construction closed, then B is constructible over AC.
Proof. The rst part is clear. For the second part, let b = b
B. Since the
type of b = b
0
is isolated over A, we can do induction on . As b
<
is
a construction over A, tp(b
/AB
) is isolated by some formula (x, c) where

c = b
1
. . . b
n
with
i
< . By induction, each b
i
is contained in a nite
construction closed set C
i
. Thus C
1
. . . C
n
b
is nite and construction

closed.
For 3. we assume A = to simplify notation. We will show that the type
tp(b
/CB
) is isolated for all . This is clear if b
C. So assume b
, C.
From the assumption it is easy to see that C is isolated over B
+1
where the
isolating formulas only contain parameters from B
+1
C B
.
We thus have
tp(C/B
) tp(C/B
).
If (x) isolates the type tp(b
/B
), then (x) also isolates the type tp(b
/CB
).
Lemma 39.5. If B is constructible over A, then B does not contain an un-
countable set of indiscernibles over A.
Proof. We assume A = . Let 1 = c
<1
be indiscernible. By Lemma 39.4
we can build a continuous sequence C
of countable construction closed subsets

of B such that c
C
+1
. By Lemma 39.2 there is a club D consisting of limit
ordinals such that for all D the set c
[ is indiscernible over C
.
Each c
is isolated over C
by a formula with parameters from C
with
< .
By Fodors Theorem there is some
0
such that for some conal set of s from
D the parameters can be chosen in C
0
. Assume that
1
<
2
are such elements.
Then c
1
and c
2
have the same type over C
0
. But this is impossible since
tp(c
2
/C
0
) tp(c
2
/C
0
c
1
).
Let M be a model and A M. We call a subset B of M normal in M over
A if for every element b B all realisations of tp(b/A) in M are contained in B.
Lemma 39.6. Let T be a (not necessarily totally transcendental) theory which
has prime extensions. If M is atomic over A and B is a normal subset of M
containing A, then M is atomic over B.
Proof. Let c be a tuple from M, so tp(c/A) is isolated. Since the isolated types
are dense over B by Theorem 14.7, there is some d M atomic over B and
realizing the type tp(c/A). Let tp(d/B) be isolated by (x, d
0
) for some tuple
d
0
B. Since c and d satisfy the same type over A there is some c
0
M such
that tp(cc
0
/A) = tp(dd
0
/A). Then c satises (x, c
0
) and as B is normal, we
have c
0
B. It follows easily that (x, c
0
) is complete over B as well.
Proof of 39.1(1). If M is a prime extension of A, then M is atomic over A by
Corollary 17.7 and since M can be embedded over A into some constructible
prime extension, M does not contain an uncountable set of indiscernibles over
A by Lemma 39.5.
For the converse assume again A = , i.e. suppose M is atomic over without
uncountable set of indiscernibles. In order to prove that M can be embedded
into any model N we enumerate all types over realised in M as (p
)
<
and
recursively extend the empty map to the normal sets C
<
p
(M). That
this is possible follows from the following
Claim: Let M be atomic over B, p S(B) and B C M normal over B.
Then any elementary map C N can be extended to C p(M).
We proof the claim by induction on the Morley rank of p and note that the
claim is clear if p is algebraic.
Assume inductively that the claim is proved for all types of Morley rank less
than (over arbitrary sets B). Then any given elementary map f : C N
with B C M and C normal over B can be extended to C a M [
MR(a/B) < .
Let now p S(B) with MR(p) = . Let c
i
i<
be a maximal set of
realisations of p in M independent over B. By Exercise 39.1, c
i
i<
splits into
a nite number of indiscernible sequences, which implies that is countable,
and we can assume that . Let B
i
= B c
0
, . . . , c
i1
and C
i
= C a
M [ MR(a/B
i
) < , so B = B
0
. By maximality we have p(M)
i<
C
i
. As
M is atomic over B, M is also atomic over B
i
and since C
i
is normal over B
i
even atomic over C
i
. If f has been extended to C
i
, we may extend f to C
i
B
i+1
by the atomicity. By induction hypothesis applied to B
i+1
and C
i
B
i+1
there is
an extension to C
i+1
.
The proof of 39.1 (1) can easily be symmetrised to yield 39.1 (2).
Example:
[48] Let L contain a binary relation symbol E
for every ordinal <

1
and let
T be the theory stating that each E
is an equivalence relation such that E

0
consists of only one class and for any < <
1
each E
equivalence class is
the union of innitely many E
equivalence classes. T is complete, stable (but

not totally transcendental) and admits quantier elimination. Every consistent
L(A)formula (x) can be completed over A. Therefore there exists a model M
which is constructible over the empty set.
In M any chain (K
)
<1
of E
equivalence classes has non-empty inter-

section: otherwise there would be a countable subset A of M and some limit
ordinal <
1
such that
(i) A is construction closed (with respect to a xed construction);
(ii) A
<
K
= ;
(iii) A K
,= for all < .

M/A is atomic. Let c K
; then tp(c/A) is isolated by some formula (x; a);

by (ii) there exists some < such that a K
= . By (iii) there is some

d A K
; since a K
= , it would follow [= (d; a); but this is impossible

as isolates tp(c/A). Therefore we have (K
)
<1
,= .
Now let a M and let N be the set of all b from M for which there is some
ordinal <
1
with [= aE
b; N is also prime model, but M and N are not

isomorphic.
Exercise 39.1.
Let T be totally transcendental, and 1 be an independent set of realisations
of p S(A). Then 1 can be decomposed into a nite number of 1
1
,. . . ,1
n
of
indiscernible sets over A.
40 Countable stable theories
We assume throughout this section that T is countable and stable.
For such T we will show that prime extensions, if they exist, are unique. The
main point is Shelahs result that in this situation subsets of constructible sets
are again constructible.
The proof of Theorem 39.1(1) showed that atomic extensions of A with-
out uncountable sets of indiscernibles are constructible and hence prime. The
uniqueness of such prime extensions then also follows directly from the following
theorem which holds for arbitrary theories.
Theorem 40.1 (Ressayre). Constructible prime extensions are unique.
Proof. It suces to prove the theorem for constructible prime extensions M and
M
over the empty set. Let f

0
: E
0
E
0
be a maximal elementary map between
construction closed subsets E
0
M and E
0
M
. If E
0
,= M, there is some
proper construction closed nite extension E
1
of E
0
. Since E
1
is atomic over
E
0
by Lemma 39.4.3 and Corollary 17.6 there is an extension f
1
: E
1
E
1
of
f
0
. Then E
1
need not be construction closed, but there is a construction closed
nite extension E
2
of E
1
. Similarly there exists a (not necessarily construction
closed) set E
2
M and some extension f
1
to an isomorphism f
2
: E
2
E
2
.
Continuing in this way we obtain an ascending chain of elementary isomorphisms
f
i
: E
i
E
i
. Then E
:= E
i
and E
:= E
i
are construction closed and
f
:= f
i
is an elementary isomorphism from E
to E
, contradicting the
maximality of f
0
.
Theorem 40.2 (Shelah [48]). If T is stable and countable, then any subset of
a set constructible over A is again constructible over A.
We immediately obtain the following corollary:
Corollary 40.3. If T is a countable stable theory with prime extensions (see
Exercise 17.1), then all prime extensions are unique.
For the proof of the theorem we need the following lemma:
Lemma 40.4. Let T be countable and stable; if A and B are independent over
C and B
is countable, then there is a countable subset C
of A with A and BB
independent over CC
.
Proof. Using the properties of forking, we nd a countable subset C
A with
ABC [
BCC
; then A [
BCC
, and since A [
CC
B we have A [
CC
BB
.
Proof of 40.2. Let B be constructible over A and D a subset of B. We may
assume that B is innite since nite sets are constructible. If E is an arbitrary
construction closed subset of B and E
B a countable extension of E (i.e.

E
E is countable), then there is some countable construction closed extension

E
of E
. Similarly, for any E with D [

A(DE)
E and every countable extension
E
of E there is a countable extension E
with D [
A(DE
)
E
(40.4).
Applying these closure procedures alternatingly countably many times, one
sees that for any construction closed subset E of B with
D [
A(DE)
E
and for any countable extension E
of E there exists a construction closed count-

able extension E
of E
such that
D [
A(DE
)
E
.
In this way we obtain a continuous chain (C
)
<
of construction closed sets
with C
0
= ,

<
C
= B, countable dierences C
+1
C
and
D [
A(DC)
C
.
For each we can choose an enumeration of D (C
+1
C
). These
enumerations can be composed to an enumeration of D. In order to show that
this enumeration is a construction of D it suces to show that every initial
segment d of D (C
+1
C
) is atomic over A(D C
). This follows from the

Open Mapping Theorem (36.9) as d is atomic over AC
.
Chapter 10
The ne structure of
1
41 Internal types
By the results in Sections 22 and 25 we know that models of
1
categorical
theories are (minimal) prime extensions of strongly minimal sets. We will see
in the next section that in this case the prime extensions are obtained in a par-
ticularly simple way. Unless stated otherwise, we assume in this section
that T is totally transcendental.
We need the concept of an internal type. We x a 0denable innite
subclass F =
0
(C) of C.
Denition 41.1. A partial type (x) is called Finternal if for some set B, the
class (C) is contained in dcl(FB).
Note that by compactness is F-internal if and only if there is some nite
conjunction of formulas in which is F-internal.
Example:
Let G be a group. Let
M = (G, A)
be a twosorted structure where A is a copy of G without the group structure.
Instead, the structure M contains the map
: GA A
dened as (g, a) = ga. Then M is the prime extension of G and clearly the
type p of an element in A is Ginternal, in fact, p(M) dcl(G, a) for any a A.
We will see in Corollary 41.6 that this is the typical picture for internal types.
187
CHAPTER 10. STRUCTURE OF
1
Lemma 41.2. A type p S(A) is Finternal if and only if there is some set of
parameters B and some realisation e of p such that e dcl(FB) and e [
A
B.
Proof. If p(C) dcl(FB), we just choose e as a realisation of p independent of
B over A.
Conversely, given a realisation e of p with e [
A
b and e dcl(Fb), we choose
a Morley sequence b
0
, b
1
, . . . of tp(b/ acl
eq
(A)) of length [T[
+
. For e
p(C) there
is some i such that e
[
A
b
i
as otherwise the p
i
= tp(e
/ acl
eq
(A)b
1
. . . b
i1
)
form a chain of forking extensions of length [T[
+
contradicting Exercise 28.1.
If p is stationary, then eb and e
b
i
realise the same type over A and hence
e
dcl(Fb
i
). So p(C) dcl(FB) for B = b
0
, b
1
. . .. If p is not stationary,
then by the previous argument all extensions of p to acl
eq
(A) are Finternal.
By taking unions we obtain a set B such that the realisations of all extensions
of p to acl
eq
(A) are contained in dcl(FB).
Lemma 41.3. A consistent formula is F-internal if and only if there is a
denable surjection from some F
n
onto (C).
Proof. If there is a Bdenable surjection F
n
(C), (C) is contained in
dcl (FB). For the converse assume that (C) dcl (FB). Then, by Exer-
cise 23.11, for every e (C) there is a Bdenable class D
e
F
ne
and a
Bdenable map f
e
: D
e
C such that e is in the image of f
e
. A compactness ar-
gument shows that there is a nite number of denable classes D
1
, . . . , D
m
F
n
and denable maps f
i
: D
i
C such that (C) is contained in the union of the
f
i
(D
i
). Fix a sequence of distinct elements a
1
, . . . , a
m
F and an element b of
(C). Dene f : F
n+1
(C) by setting f( x, y) = f
i
( x) if y = a
i
and x D
i
and f( x, y) = b otherwise. Then f is a surjection from F
n+1
onto (C).
Lemma 41.4. Let T be an arbitrary theory and F a stably embedded 0denable
class. If a and b have the same type over F, they are conjugate under an element
of Aut(C/F).
Proof. We construct the automorphism as the union of a long ascending se-
quence of elementary maps : A F B F which are the identity on F.
Assume that is constructed and consider an element c C. Since F is stably
embedded, the type of cA over F is denable over some subset C of F. Choose
some d C with tp(dB/C) = tp(cA/C). We can then extend to an elemen-
tary map
: c A F d B F. To see this assume that [= (c, a, f),

where a A and f F. Then (x, y, f) belongs to the type of cA over F, so
(x, y, f) belongs also to the type of dB over F which shows [= (d, (a), f).
A groupoid is a category where all morphisms are isomorphism. A groupoid
is connected if there are morphisms between any two objects. A denable
groupoid ( is a groupoid whose objects are given by a denable family (O
i
)
iI
1
of classes and whose morphisms by a denable family (M
i,j
)
i,jI
of bijections
O
i
O
j
. We use the notation Hom
G
(O
i
, O
j
) and Aut
G
(O
i
) for M
i,j
and M
i,i
.
We denote by F
eq
C
eq
the collection of those 0denable equivalence
classes having representatives in F.
Theorem 41.5 (Hrushovskis Binding Groupoid). Let T be totally transcen-
dental, E and F be 0denable and assume that E is Finternal and non-empty.
Then E is an object of a 0denable connected groupoid ( with the following
properties
a) E is not the only object of (. The objects other than E are subsets of F
eq
.
b) Aut
G
(E) = Aut(E/F).
Aut
G
(E) as a denable group is Poizats Groupe de Liaison ([41]).
Proof. By Lemma 41.3 there is a denable surjection from some power F
n
to E.
This induces a bijection f : O E for some denable class O of F
eq
. Let f = f
c
be dened from a parameter c and O = O
d
from a parameter d. By Lemma 34.3
we nd d in F. Let M
0
be an atomic model of T, so we may assume d FM
0
.
Since, by Lemma 17.4, the isolated types are dense over any parameter set, we
may assume that the type of c over FM
0
is isolated, say by a formula (x, a).
It is easy to see that (x, a) isolates a type over F. By extending d if necessary
we may assume that d = a. Now let (y) isolate the type of d.
The objects of G are E and the O
e
where e realises (y). Hom
G
(O
e
, E) is
the set of all f
c
where c
realises (x, a). We claim that Hom

G
(O
e
, E) is a right
coset of Aut(E/F). We can then set
Hom
G
(E, O
e
) = f
1
[ f Hom
G
(O
e
, E)
Hom
G
(E, E) = f g
1
[ f, g Hom
G
(O
e
, E)
Hom
G
(O
e
, O
d
) = f
1
g [ f Hom
G
(O
d
, E), g Hom
G
(O
e
, E).
To see that Hom
G
(O
e
, E) is a right coset of Aut(E/F) we have to show that for
some c
we have
Hom
G
(O
e
, E) = Aut(E/F) f
c
.
This follows easily from the fact that, by Lemma 41.4, the elements of Hom
G
(O
e
, E)
have the form f
(c) for automorphisms Aut(C/F) and from the formula

f
(c
) = f
c
.
Recall that a group G acts regularly on a set A if for all a, b A there exists
a unique g G with ga = b.
1
Corollary 41.6 (Binding group). Let T be a totally transcendental theory, E
and F be 0denable and assume that E is Finternal. Then the following holds.
1. There is a denable group G F
eq
, the binding group, and a denable
class A on which G acts regularly and such that E dcl(Fa) for all a A.
G, A, the group operation and the action of G on A are denable with
parameters from F.
2. Aut(E/F) is a 0denable permutation group, Aut(E/F) acts regularly on
A and is denably isomorphic to G.
Proof. Let ( be the groupoid of Theorem 41.5. Fix any object O
i
dierent
from E. Set G = Aut
G
(O
i
) and A = Hom
G
(O
i
, E). Now replace the denable
bijections in G and A by their canonical parameters in order to obtain elements
of F
eq
and C
eq
respectively.
Exercise 41.1.
Prove that every element of dcl
eq
(F) is interdenable with an element of F
eq
.
Exercise 41.2.
Let T be arbitrary, F 0denable and C a subset of F. Show that tp(a/F) is
denable over C if and only if tp(a/C) tp(a/F).
Exercise 41.3 (Chatzidakis-Hrushovski, [13]).
Let T be arbitrary and F 0denable. Show that the following are equivalent:
a) F is stably embedded.
b) Every type tp(a/F) is denable over a subset C of F.
c) For every a there is a subset C of F such that tp(a/C) tp(a/F).
d) Every automorphism of F extends to an automorphism of C.
1
42 Analysable types
Throughout this section we assume that T is a stable theory elimi-
nating imaginaries.
Denition 42.1. Let F be a 0denable class. A type p S() is called F
analysable if for every realisation a of p there is a sequence of tuples a
0
, . . . , a
n
=
a in dcl(a) such that tp(a
i
/a
0
. . . a
i1
) is Finternal for i = 0, . . . n.
Theorem 42.2 (Hrushovski [23]). Let T be
1
categorical and F a 0denable
strongly minimal set. Then every type p S() is Fanalysable.
We need some preparation in order to prove this theorem:
Theorem 42.3. Let p S(A) be a stationary type and 1 an innite Morley
sequence for p. Then Cb(p) dcl(1).
Proof. By Exercise 38.2 b) the average type Av(1) is the nonforking extension
of p to the monster model. The proof of Theorem 38.2 implies that Av(1) is
based on 1.
Denition 42.4. 1. We call types p, q S(A) almost orthogonal if any
realisation of p is independent over A from any realisation of q. The types
are orthogonal if for any set B A all nonforking extensions of p to B
are almost orthogonal to all nonforking extensions of q to B. Note that
algebraic types are orthogonal to all types.
2. A theory T is called unidimensional if all stationary non-algebraic types
are pairwise non-orthogonal.
Let F be a strongly minimal set. We call a type p S(A) orthogonal to F if
for every realisation b of p, any c F and any extension B of A over which F is
dened we have
b [
A
B = b [
B
c.
If q is any type of Morley rank 1 containing F(x), this is equivalent to p being
orthogonal to q.
Lemma 42.5. Let T be
1
categorical and F a strongly minimal set. Then no
non-algebraic type is orthogonal to F.
Proof. Let p S(A) be orthogonal to F. We choose a model M containing
A over which F is dened, a realisation b of p independent from M over A,
and a model N containing M and b. It follows from the assumptions that b is
independent from F(N) over M.
1
On the other hand N is the (minimal) prime extension of M F(N). Hence
b is atomic over Mc for some tuple c F(N) and isolated by, say, (x, c). By
Corollary 34.7 the type tp(c/Mb) is an heir of tp(c/M). Therefore there must
be some m M realizing (x, c). But this m can only be b itself. So we see
that b [
A
b, hence b acl(A) by Theorem 36.5.6. and p is algebraic.
Proof of 42.2. By induction on = MR(p). If = 0, p is algebraic and hence
trivially internal (in any denable class). If > 0, we apply Lemma 42.5 and
nd a realisation b of p, some c F and some set of parameters B such that
b [
B and b ,[
B
c. By nite character of forking we may assume that B is
nite.
Let D be the canonical base of tp(cB/ acl(b)). As cB [
D
b, but cB ,[
b, we
have b ,[
D.
Let (c
i
B
i
) be an innite Morley sequence for tp(cB/D). By Theorem 42.3, we
have
D dcl(c
0
B
0
c
1
B
1
. . .).
It follows from b [
B (and D acl(b)) that we must have D [
B and
hence D [
B
0
B
1
. . .. This shows that the type of any tuple d D is Finternal
by Lemma 41.2. We choose a nite tuple d D with b ,[
d. Then we have
MR(b/d) < . We may absorb the parameter d into the language and apply
the induction hypothesis to T(d) = T (d): tp(d) to nd a sequence
b
1
, . . . , b
n
= b such that b
i
dcl(db) and the types tp(b
i
/db
1
. . . b
i1
) are F
internal. Setting b
0
= d, we would be done if we knew that d dcl(b). For this
we replace d by the canonical parameter d
of the nite set d

1
, . . . , d
k
of conju-
gates of d over b. We have thus achieved d
dcl(b). Because d
dcl(d
1
. . . d
k
)
the type of d
is Finternal and since d acl(d
), we have MR(b/d
) MR(b/d) <
. We can use this d
to nish the proof.

Theorem 42.6 (Baldwin [2]).
1
categorical theories have nite Morley rank.
To prove Theorem 42.6 we need the following denition which allows us to
extend additivity of Morley rank beyond strongly minimal sets (see also Remark
26.9 and Exercise 26.3):
Denition 42.7. Let f : B A be a denable surjection. We say that the
bers of f have denable Morley rank if there is a nite bound for the Morley
rank of the bers f
1
(a) and if for every denable B
B and every k
the class
a A [ MR(f
1
(a) B
) = k
is denable.
Remark 42.8. If B is a power of a strongly minimal set, the bers of f have
denable Morley rank by Corollary 26.4.
For the next statement remember that the Morley rank of the empty set is
dened as .
1
Lemma 42.9. If the bers of f : B A have denable Morley rank and MR(A)
is nite, we have
MR(B) = max
k<
(MR(A
k
) +k),
where A
k
= a A [ MR(f
1
(a)) = k.
Proof. We leave it as an exercise (Exercise 42.1) to show that MRB MRA
k
+k
for all k. For the converse we may assume that all bers have Morley rank k
and that A has Morley degree 1 and Morley rank . We show MRB + k
by induction on .
Let B
i
be a family of disjoint denable subsets of B. We want to show that
one of the B
i
has smaller Morley rank than +k. Consider any a A. Then for
one i the ber f
1
(a) B
i
must have rank smaller than k. So the intersection
of all A
i
k
= a [ MR(f
1
(a) B
i
) = k is empty, which implies that one of the
A
i
k
has smaller rank that . Induction yields MRB
i
< +k.
Proof. (of Theorem 42.6) It is enough to prove that every element a has nite
Morley rank over . Each a has an analysing sequence a
0
, . . . , a
n
= a where
all types tp(a
i
/a
0
. . . a
i1
) are Finternal. We prove by induction on n that the
tuple a
0
. . . a
n
has nite Morley rank. By Lemma 26.1 this implies that a
n
has
nite Morley rank.
By induction hypothesis, a
0
. . . a
n1
is contained in a 0denable set A of
nite Morley rank. Since tp(a
n
/a
0
. . . a
n1
) is Finternal, a
n
is contained in an
(a
0
. . . a
n1
)denable set which is an image of some power of F by a denable
map. So we may assume that a
0
. . . a
n
belongs to a 0denable set B which
projects onto A by the restriction map : B A and such that the bers
1
(a) are denable images of some power of F. By Corollary 42.8 the bers
of have denable Morley rank. If the rank of the bers is bounded by k,
Lemma 42.9 bounds the rank of B by MRA +k.
We end this section with a dierent characterisation of
1
-categorical theo-
ries due to Erimbetov [14].
Theorem 42.10. A countable theory T is
1
categorical if and only if it is
stable and unidimensional.
Proof. Assume rst that T is
1
-categorical. Let F be strongly minimal, dened
over A, p and q be two stationary types over A. By Lemma 42.5 there is an
extension A B, realisations a, b, c
1
, c
2
of p, q and F such that a [
A
B,
b [
A
B, a ,[
B
c
1
,b ,[
B
c
2
. That means that c
1
acl(aB)acl(B), c
2
acl(bB)
acl(B). So c
1
and c
2
have the same type over B and we may assume that c
1
= c
2
.
But then c
1
,[
B
c
1
implies a ,[
B
b and p and q are not orthogonal.
For the converse assume that T is stable and unidimensional. The proof
of the BaldwinLachlan Theorem shows that it is enough to prove that there
1
are no Vaughtian pairs M N for strongly minimal formulas (y) dened over
M. Let a be any element in N M and p(x) = tp(a/M). By assumption there
is an extension M
of M and an element c (C) M
such that a [
M
M
and
c acl(aM
). Let (a, m
, y) be a formula which isolates the type of c over aM
.
Then the following sentences are true in M
d
p
xy(x, m
, y)
d
p
xy((x, m
, y) (y))
y d
p
x(x, m
, y)
So we nd an m M for which the corresponding sentences are true in M.
This implies that there is an b B such that N [= (a, m, b) and that all such
b lie in (N) M. So M N is not a Vaughtian pair for .
Exercise 42.1.
Let f : B A denable and all bers of Morley rank . Then MRB
+ MRA.
1
43 Locally modular strongly minimal sets
In this section, we let T denote a complete stable theory and (x) a strongly
minimal formula without parameters.
We call modular if its pregeometry is modular in the sense of Denition 8.9
in Appendix C, i.e. if for all relatively algebraically closed A, B in (C)
(1) dim(A B) + dim(A B) = dim(A) + dim(B).
T is locally modular if (1) holds whenever A B contains an element not in
acl().
It is easy to see that is modular if and only if any two relatively alge-
braically closed subsets A and B of (C) are independent over their intersection:
A [
AB
B.
(cf. Appendix C, Lemma 8.10). In fact this holds for arbitrary sets B, not
necessarily contained in (C):
Lemma 43.1. If is modular, then
A [
AB
B
for algebraically closed B and any A which is relatively algebraically closed in
(C).
Proof. Let C be the intersection of B and (C). It is enough to show that A
is independent from B over C. For this we may assume that B is the algebraic
closure of a nite set and the elements of A are algebraically independent over
C. We have to show that the elements of A remain algebraically independent
over B. Choose a B-independent sequence A
0
, A
1
, . . . of sets realising the same
type as A over B. For any i the intersection of acl(A
0
. . . A
i
) and acl(A
i+1
) is
contained in B. So by local modularity A
0
. . . A
i
and A
i+1
are independent over
C. This implies that the elements of A
0
A
1
. . . are algebraically independent
over C. By Exercise 38.2 for some i the elements of A
i
A
i+1
. . . are algebraically
independent over B. Hence also the elements of A are algebraically independent
over B.
Denition 43.2. A formula (x) without parameters is 1-based if
A [
acl
eq
(A) acl
eq
(B)
B
for all B and all subsets A of (C).
1
For a set B and a tuple a, the strong type stp(a/B) = tp(a/ acl
eq
(B)) is
stationary (cf. Exercise 35.9). We denote by Cb(a/B) the canonical basis of
stp(a/B). Note that Cb(a/B) is a subset of C
eq
.
Lemma 43.3. is 1based if and only if
Cb(a/B) acl
eq
(a)
for all sets B and nite tuples a in (C).
Proof. Let C be the intersection acl
eq
(a) acl
eq
(B). If T is 1based, the strong
type of a over B does not fork over C. So by Lemma 36.8 Cb(a/B) is con-
tained in C acl
eq
(a). If conversely Cb(a/B) is contained in acl
eq
(a), it is also
contained in C and a [
Cb(a/B)
B implies a [
C
B.
Corollary 43.4. 1. 1basedness is preserved under adding and removing pa-
rameters, i.e., is 1based if and only if is 1based in C
A
for any set
A of parameters.
2. If is 1based and if every element of
(C) is algebraic over (C), then
(C) is 1based.
Proof. 1. If is 1based, then Cb
A
(a/B) = Cb(a/AB) acl
eq
(a) acl
eq
A
(a).
If conversely is 1based in C
A
and a (C) and B are given, we may as-
sume that a, B is independent from A. We have than Cb(a/B) = Cb(a/AB) =
Cb
A
(a/B) acl
eq
A
(a). Since Cb(a/B) is also contained in acl
eq
(B), we con-
clude Cb(a/B) acl
eq
(a).
2. First note that if c is algebraic over a, then c and B are independent over
Cb(a/B), so we have Cb(c/B) acl
eq
Cb(a/B). Now let a
be a nite tuple
from
(C) and B any set. Choose a tuple a

1
from (C) over which a
is alge-
braic. Then choose a
2
which realises the type of a
1
over a
and is independent
from a
1
over a
. We have then Cb(a
/B) acl
eq
Cb(a
1
/B) acl
eq
Cb(a
2
/B)
acl
eq
(a
1
) acl
eq
(a
2
) acl
eq
(a
).
Theorem 43.5. Let T be totally transcendental and a strongly minimal for-
mula without parameters. Then the following are equivalent.
a) is locally modular.
b) is 1based.
c) Every family of plane curves in has dimension at most 1. This means that
for all B and elements a, b of (C), if tp(ab/B) has Morley rank 1, then
Cb(ab/B) has Morley rank at most 1 over the empty set.
1
Proof. ab: If is locally modular, becomes modular if we add a name for
any element x (C) acl
eq
() to the language. If B and a (C) are given,
we choose x independent of a, B. It follows from Lemma 43.1 that a and Bx
are independent over acl
eq
(ax)acl
eq
(Bx). This implies Cb(a/Bx) acl
eq
(ax).
On the other hand, we have Cb(a/Bx) = Cb(a/B) acl
eq
(B). Since x and B
are independent over a, we have acl
eq
(ax) acl
eq
(B) acl
eq
(a). This implies
Cb(a/B) acl
eq
(a).
bc: Write d = Cb(ab/B). Then MR(ab/d) = 1. If the Morley rank of d is
not zero, ab and d are dependend over the empty set. By our assumption d
acl
eq
(ab), so we have MR(abd) 2. Since MR(ab/d) = 1, we have MR(d) 1
using Remark 26.9.
ca: Let x be a non-algebraic element of (C). By Lemma 8.11 of Appendix C
we have to show the following: For all elements a, b and sets B in (C), if
MR(ab/x) = 2, MR(ab/Bx) = 1, there is a c acl(Bx) such that MR(ab/cx) =
1. We may assume that a , acl(Bx). Consider the imaginary element d =
Cb(ab/Bx). Since d is contained in acl
eq
(Bx), a is not algebraic over d. Also,
since d is algebraic over ab by assumption, x is not algebraic over d. So a and
x have the same type over d and we can nd an element c such that xc and ab
have the same type over d. Now b acl(ad) implies c acl(xd) acl(Bx). We
have also d acl
eq
(xc), which implies MR(ab/cx) = 1.
Proposition 43.6. A totally transcendental theory T which contains an innite
denable eld is not one-based.
This is true for stable theories, see [38].
Proof. Let K be an innite denable set with a denable eld structure. We
may assume that everything is denable over the empty set. Let be the Morley
rank of K. We call an element x of K generic if MR(x) = . We note rst that
if p = (x, y) is an element of the line g
a,b
= (x, y) [ ax + b = y which is not
algebraic over a, b, then a, b dcl Cb(p/a, b). This follows from the fact, that
two lines intersect in at most one point. So it is enough to nd such p and g
a,b
with (a, b) not algebraic over p.
For this we choose four independent generic elements a, b, a
, b
and let p =
(x, y) be the intersection of g
a,b
and g
a
,b
. Since x and b
are interdenable
over a,b,a, we can conclude that x, a, b, a
are generic and independent. So p is

not algebraic over a, b. Since y and b are interdenable over x, a we have that
x, y, a, a
are generic and independent. This implies that a, b is not algebraic

over p.
The converse of the previous proposition for strongly minimal theories was
known as Zilbers conjecture. This conjecture, namely that for any non-locally
modular strongly minimal theory T an innite eld is denable in T
eq
, was
refuted by Hrushovski in [24]. A variant of his construction of a new strongly
1
minimal set will be given in the next section. However, Hrushovski and Zilber
proved in their fundamental work [26] that the conjecture holds for so-called
Zariski structures.
1
44 Hrushovskis Examples
To end the book, we present a modication of Hrushovskis ab initio example of
a new strongly minimal set [24]. This counterexample to Zilbers conjecture has
been the starting point of a whole new industry constructing new uncountably
categorical groups [6], elds [5], [7], and geometries [4], [51]. The dimension
function dened below also reappears in Zilbers work around Shanuels Con-
jecture [57].
Following Baldwin [4] (see also [51]), we construct an almost strongly mini-
mal projective plane as a modied Frass limit: instead of considering structures
with all their substructures we restrict the amalgamation to so-called strong sub-
structures and embeddings. This will allow us to keep control over the algebraic
closure of sets so that the resulting structure is uncountably categorical.
Recall that a projective plane is a point-line geometry such that any two
lines meet in a unique point, any two points determine a unique line through
them and there are four points no three of which are collinear. For convenience,
we consider a projective plane as a bipartite graph where the bipartition divides
the vertices into points and lines, and incidence between a point and a line is
represented by an edge. In these terms a projective plane is a bipartite graph
with the property that the distance between any two vertices is at most 3,
the smallest cycles in the graph have length 6, and any element has at least 3
neighbours.
We x a language L = P, E for bipartite graphs where P is a predicate
denoting the partition and E denotes the edges. For a nite graph A, let e(A)
denote the number of edges in A and [A[ the number of vertices. We dene
(A) = 2[A[ e(A) and more generally for nite subgraphs A, B M we put
(A/B) = (AB) (B). We call a nite set B M strong in M, B M, if
(A/B) 0 for all nite A M containing B.
Note that if A and B are disjoint subsets of a graph M, then
(AB) = (A) +(B) e(A, B)
where e(A, B) denotes the number of edges connecting a vertex in A to one in
B.
Let / be the class of graphs A, bipartite with respect to P, not containing
any 4-cycles, and such that (A
) 4 for any nite subgraph A
of A with
[A
[ 3. Note that this implies (A) 0 for all A /.

From now on all nite graphs considered are in /.
Denition 44.1. We call a proper strong extension B over A minimal if it
cannot be split into two proper strong extensions A C B. For A, B nite
subgraphs of a graph M, we say that B is i-simple over A if A and B are
disjoint, (AB/A) = i and for every proper nonempty subset C of B we have
1
(AC/C) > i. A graph B is minimally i-simple over A if it is i-simple over A
but not i-simple over any proper nonempty subset of A.
Remark 44.2. It is easy to see that a 0-simple extension B of A is minimally
0-simple over the set of elements of A which are incident with an element of B.
Note also that if A B and C B, then A C C (see Exercise 44.1).
A minimal extension B over A is i-simple for i = 0, 1 or 2. More precisely,
either B is 0-simple over A, or it is of the form B = Ab where b is incident
to at most one element of A. If b is incident with exactly two vertices a
1
, a
2
of A, then b is 0-simple over A and minimally 0-simple over a
1
, a
2
. Clearly
(b/A) equals 1 if b is incident to an element of A and 2 otherwise. Any 1-simple
extension is of this form: if B is 1-simple over A with [B[ > 1, then for any
b B we would have (bA/A) = 2, i.e. no element of B is connected to any
element of A. Then (AB) = (A) +(B) > (A) + 1, a contradiction.
We next x a function from pairs (A, B) of nite graphs in / with B
minimally 0-simple over A into the natural numbers satisfying the following
properties:
1. The function depends only on the isomorphism type of (A, B).
2. If A = a
1
, a
2
, B = b and b is connected to a
1
and a
2
, then (A, B) = 1;
otherwise (A, B) max(A), 3.
For any graph N and any minimally 0-simple pair (A, B) with A N we
dene
N
(A, B) to be the number of pairwise disjoint graphs B
N such that
B and B
are isomorphic over A.

Let now /
be the subclass of / consisting of those N / for which for

every minimally 0-simple pair (A, B) with A N we have
N
(A, B) (A, B).
We need the following lemma:
Lemma 44.3. Let C
0
, C
1
, C
2
/
, C
0
C
1
, C
2
with

C
1
= C
1
C
0
minimally
0-simple over A
0
C
0
. If D = C
1

C0
C
2
/ /
, one of the following happens:

1.
C2
(A
0
,

C
1
) = (A
0
,

C
1
) and there is an isomorphic copy of

C
1
over C
0
inside C
2
, or
2. there is some minimally 0-simple pair (A, B) with
a)
D
(A, B) > (A, B),
b) A A
0

C
1
, and
c) at least one copy of B is properly contained in

C
1
.
Proof. Without loss of generality we may assume that C
0
= A
0
. Suppose that
D = C
1

C0
C
2
/ /
. If D contains a 4-cycle (a
0
, a
1
, . . . , a
4
= a
0
), then since
1
this cannot be entirely contained in either C
1
or C
2
, without loss of generality
we have a
1

C
1
, a
3

C
2
, a
0
, a
2
C
0
. Since C
1
is 0-simple over C
0
, it follows
that

C
1
= a
1
and a
3
is the isomorphic copy of a
1
over C
0
.
Now suppose that there is some minimally 0-simple pair (A, B) with
D
(A, B) >
(A, B). By the previous paragraph, we may assume that [
C
1
[ > 1 and hence
D
(A, B) > (A, B) 3. Then for every copy B
of B over A, every element of

A is connected to some element of B
because B is minimally 0-simple over A.

First consider the case A C
2
. Since C
2
/
there is some copy B
of B
over A with B
C
1
,= . If B
,

C
1
, then (B
C
2
/A) > 0 and thus (B
/C
2
) <
0, contradicting C
2
D. So B

C
1
and since (B
/A) = (B
/C
2
) = 0, by
0-simplicity of

C
1
over A
0
we see that B
=

C
1
and A = A
0
.
Next consider the case A , C
2
, i.e. A

C
1
,= . Then since B is minimally
0-algebraic over A, no copy of B is contained in

C
2
= C
2
C
0
. Now suppose that
there are k copies B
1
, . . . , B
k
of B over A contained in C
2
and that the copies
B
k+1
, . . . , B
k+l
intersect both B
2
and

C
1
. Since each B
i
contains vertices which
are connected to the vertices of A

C
1
, it follows immediately that (A/C
2
)
(A/C
2
A) k (A) k.
Since the B
i
are 0-simple over A, we have for each i = k + 1, . . . , k +l:
(B
i
/C
2
A B
k+1
. . . B
i1
) < 0
This implies
(
k+l
_
i=k+1
B
i
/C
2
A) l.
Hence
0 (
k+l
_
i=k+1
B
i
A/C
2
) (A/C
2
) l (A) (k +l).
Thus at most (A) many copies of B over A are not contained in

C
1
, leaving
at least one copy of B over A inside

C
1
. Since each element of A is connected to
some element of this copy, we see that A must be contained in A
0

C
1
, nishing
the proof.
As in Section 13 we say that M /
is /
-saturated if for all nite B M

and strong extensions C of B with C /
there is a strong embedding of C

into M xing B elementwise.
Theorem 44.4. The class /
is closed under substructures, and has the joint

embedding and the amalgamation property with respect to strong embeddings.
There exists a countable /
-saturated structure M
which is unique up to iso-

morphism. In particular, in M
any partial isomorphism f : A A
with
A, A
extends to an automorphism of M
.
1
Proof. Clearly, /
is closed under substructures. We will show that /
has
the joint embedding and the amalgamation property with respect to strong
embeddings. Then exactly as in the proof of Theorem 13.4 we obtain a count-
able /
which is unique up to isomorphism. In par-

ticular, in M
any partial isomorphism f : A A
with A, A
extends
to an automorphism of M
. Since the empty graph is in /
and strong in
any nonempty A /
, it suces to prove the amalgamation property. Let

C
0
, C
1
, C
2
/
, C
0
C
1
, C
2
,

C
i
= C
i
C
0
. We have to nd some D /
such
that the natural embeddings C
0
C
1
, C
2
D are all strong. We prove the
amalgamation property by induction on [
C
1
[.
Given graphs A B, C we denote by B
A
C the trivial amalgamation of B
and C over A, obtained as the graph whose set of vertices is the disjoint union
B A) (C A) A with incidence and predicate P induced by B and C. Note
that B, C B
A
C. It is easy to verify that if (F) 4 for all nite subgraphs
F of B, C with [F[ 3, then the same is true for D.
Case 1: If there is some proper subset X of

C
1
with (XC
0
) = (C
0
), then
C
0
C
0
X C
1
and by induction hypothesis we may amalgamate C
0
X with
C
2
over C
0
to get D
, and then amalgamate D
with C
1
over XC
0
to obtain D.
Case 2: There is no proper subset X of

C
1
with (XC
0
) = (C
0
).
Case 2.i): If (C
1
/C
0
) = 0, then

C
1
is 0-simple over C
0
. If D = C
1

C0
C
2
, /
and C
2
does not contain an isomorphic copy of

C
1
over C
0
, then by
the Lemma 44.3 there is some minimally 0-simple pair (A, B) with B properly
contained in

C
1
and A C
1
, but A , C
0
. Then 0 = (B/A) > (B/A C
0
).
But since C
1
is a strong extension of C
0
, this is not possible.
Case 2.ii): Suppose (C
1
/C
0
) = 1. If there is a proper subset X of

C
1
with
(X/C
0
) = 1, then we have C
0
XC
0
C
1
by the failure of Case 1 and we can
use induction. Otherwise,

C
1
= b is 1-simple over C
0
. Then D = C
1

C0
C
2
contains no k-cycle for k < 6. To see that D /
note rst that there is no

0-simple extension (A, B) inside D with b B. If C D is minimally 0-simple
over F D with b F, then 1 =
D
(C, F) (C, F) as b is connected to a
unique element of C
0
. So we have D /
.
Case 2.iii): If (C
1
/C
0
) > 1, then by the failure of Case 1 we must have
(b/C
0
) 1 for all b

C
1
. First suppose that there is some b

C
1
with
(bC
0
/C
0
) = 1. Then we can use Case 2.ii) to amalgamate b and C
2
over C
0
into some D
. Again by the failure of Case 1 it follows that bC

0
is strong in C
1
and we can use the induction hypothesis to amalgamate C
1
with D
over bC
0
.
Otherwise we have (bC
0
/C
0
) = 2 for all b

C
1
, and so no element of

C
1
is
connected to any element of C
0
. Then D = C
1
C0
C
2
is in /
since D contains
no k-cycle for k < 6 and any minimally 0-simple pair (A, B) is either entirely
contained in

C
1
or in C
2
.
1
Remark 44.5. It is easy to see that the /
is a projec-
tive plane: since any two vertices of the same colour form a strong substructure
of M
, for any two pairs of such vertices with the same colouring there is an
automorphism taking one pair to the other. Since there are pairs of vertices
of the same partition at distance 2 in the graph, the same is true for any such
pair. Thus any two points lie on a common line and any two lines intersect in
a point. Uniqueness is immediate since there are no 4-cycles. The existence
of four points in M
no three of which are collinear follows from the fact that

the corresponding graph, an 8-cycle of pairwise distinct elements is contained
in /
.
We now turn to the model theoretic properties of M
.
Theorem 44.6. Let T
be the theory (in the language of bipartite graphs)

axiomatising the class of models M such that:
1. M is a projective plane;
2. M /
;
3. M C , /
for each C which is 0-simple over M.

Then T
= Th(M
).
Proof. Note rst that this forms an elementary class whose theory T
is con-
tained in Th(M
): clearly, 1. is a rst-order property, which holds in M
by
the previous remark. For each minimally 0-simple pair (A, B) we can express
that
M
(A, B) (A, B), so 2. is rst-order expressible and holds in M
by
construction. For 3. notice that if C is 0-simple over M, then by Lemma 44.3
the fact that M C , /
is witnessed by some minimally 0-simple pair (A, B)

with B C and A CC
0
where C is minimally 0-simple over C
0
. Since there
are only nitely many isomorphism types of such (A, B), this can be expressed
in a rst order way. It again holds in M
by construction.
Since M
is a /
-saturated model of this theory, for the proof it now suces

to show the following:
Claim: M is an -saturated model of T
if and only if it is /
-saturated.
Let M [= T
be -saturated. To show that M is /
-saturated, let A M
and A B /
nite. By induction we may assume that B is minimal over A.

If B is a 0-simple extension of A minimally 0-simple over, say, A
0
M,
then since M
A0
B , /
, by Lemma 44.3 there is an isomorphic copy of B

over A
0
inside M and since there are no connections between A A
0
and B,
this isomorphism is in fact over A. Otherwise B = Ab where b is connected
to at most one element of A. By w-saturation there is another vertex b
M
1
either not connected to any element of A or connected to the same element of
A as b.
Conversely, suppose M is /
-saturated, so M is in /
. Suppose that for

some C which is 0-simple over M we have MC /
. If A
0
M is such that C
is minimally 0-simple over A
0
, then by /
-saturation we may embed C over A

0
into M. Iterating this ad libitum, we contradict the fact that M /
. Hence
M is a model of T
, which is -saturated because M is partially isomorphic to

M
.
Since every saturated model of T
is partially isomorphic to M
, it now
follows that T
is the complete theory of M
Denition 44.7 (Coordinatisation). Let = (T, L, 1) be a projective plane,

let L be a line and let x
1
, x
2
T be points not on . Let D
denote the set

of points on . Then every point of is in the denable closure of D
x
1
, x
2
:
for y D
x
1
, x
2
the lines
1
= (y, x
1
),
2
= (y, x
2
) meet in two
distinct elements y
1
, y
2
D
. Then y dcl(x
1
, x
2
, y
1
, y
2
). This process is called
coordinatisation.
Let cl(B) = cl
M
(B) be the smallest strong subgraph of M containing B.
Note that cl(B) is nite if B is. We also dene
d(A) = mind(B) [ A B M = (cl(A)).
Similarly, we put d(A/B) = d(AB) d(B).
We will show that for any vertex x M
the set D
x
of neighbours of x is
strongly minimal with dimension function given by d. To this end we start with
the following easy lemma:
Lemma 44.8. Let M and M
be models of T
. Then tuples a M and a
have the same type if and only if the map a a
extends to an isomorphism of
cl(a) to cl(a
). In particular, d(a) depends only on the type of a.

Lemma 44.9. Suppose M /
, A strong in M, and (B/A) = 0. Then there

exist only nitely many copies of B over A in M, so cl(A) acl(A).
Proof. By decomposing into minimal extensions, we may assume that B is 0-
simple over A. We claim that any two copies of B over A in M are disjoint, and
hence there are only nitely many of them.
Let B
1
, B
2
M be distinct copies of B over A. Since A is strong in M and
B
1
is 0-simple over A, we have B
1
A M. So by the last part of Remark 44.2,
B
2
A B
1
A is strong in B
2
A. By 0-simplicity we must have B
2
A B
1
A = A
proving the claim.
Proposition 44.10. For any model M of T
and any a M, the set D

a
is
strongly minimal (with dimension function given by d).
1
Proof. We work over the parameter a. We may assume that M = M
. Since
for x
1
,= x
2
D
a
, the path (x
1
, a, x
2
) is strong in M, there is a unique type of
pairs of distinct elements. Let X D
a
and x D
a
. Note that if [X[ 2, then
a cl(X), so d(x/X) 1.
If d(x/X) = 0, then by the previous lemma, x is algebraic over cl(X) and
hence over X.
If d(x/X) = 1, then cl(X) x is strong in M, so any other element
x
D
a
with d(x
/X) = 1 is conjugate to x over X, showing that

there is a unique non-algebraic type over X. The claim now follows from
Lemma 21.3.
Together with coordinatisation, the strong minimality of D
a
now implies
that T
is almost strongly minimal : a complete theory is called almost strongly

minimal if there is a strongly minimal formula (possibly containing parame-
ters) such that every model M is in the algebraic closure of (M) a for some
nite tuple a M.
Theorem 44.11. The theory T
is almost strongly minimal, not 1based and

of Morley rank 2.
Proof. By Proposition 44.10 and coordinatisation, T
is almost strongly mini-

mal. By Theorem 26.2 the Morley rank is given by the dimension function of
the strongly minimal set D
a
, say, which again coincides with d. Hence for any
x M we have MR(x) = d(x) = 2.
To see that T
is not 1based, let be a line of M and p be a point on

, so clearly Cb(p/). Then MR(p/) = 1 and MR() = MR(p) = 2. By
Remark 26.9 (and Exercise 26.2), we have
3 = MR(p) = MR(/p) + MR(p),
so , acl(p).
To show that this yields indeed a counterexample to Zilbers Conjecture, we
nish by showing that
Proposition 44.12. There is no innite group denable in T
eq
.
Proof. Again it suces to work in M = M
. We work over the parameters

, p
1
, p
2
, p
3
, p
4
M where is a line, p
1
, p
2
are points not on and p
3
, p
4
are points on . By Exercise 44.3, acl(, p
1
, . . . , p
4
) is innite. Hence, by
Lemma 35.10 the theory Th(M, , p
1
, . . . p
4
) has weak elimination of imaginaries.
1
Suppose now that G is an innite group denable in M
eq
. Then G is totally
transcendental (even uncountably categorical) and so every type has Morley
rank. Let a
1
G be an element realizing a generic type of G, i.e. a type of
maximal Morley rank. Then by weak elimination of imaginaries and coordi-
natisation 44.7, there are x
1
, . . . x
m
D
such that a
1
dcl(x
1
, . . . x
m
) and
x acl(a). By Theorem 26.2, the Morley rank of G, which is equal to the
Morley rank of a
1
, is at most m, say MR(a) = n. Let a
2
be another realisation
of tp(a
1
) independent from a
1
, and put a
3
= a
1
a
2
. Let b
1
realise tp(a
1
) and
independent from a
1
, a
2
. Put b
2
= a
1
b
1
, b
3
= a
1
3
b
2
= a
1
2
b
1
. Now put
A
0
= a
1
, a
2
, a
3
, A
1
= a
3
, b
2
, b
3
, A
2
= a
1
, b
1
, b
2
and A
3
= a
2
, b
1
, b
3
. Let
A
ij
= A
i
A
j
for i, j = 0, . . . 3 (see Exercise 36.8). Then
A
i
[
Aij
A
j
and hence also acl(A
i
) [
Aij
acl(A
j
).
We move back to working in D
: by coordinatisation and weak elimination

of imaginaries, we replace each of a
i
, b
j
, i, j = 1, 2, 3 by a tuple x
i
, y
j
D
with a
i
dcl( x
i
), b
j
dcl( y
j
) and x
i
acl(a
i
), y
j
acl(b
j
) to obtain nite sets
E
0
= x
1
, x
2
, x
3
, . . . , E
3
= x
2
, y
1
, y
3
D
.
Since cl(A) acl(A), we have cl(E
i
) [
Eij
cl(E
j
) for i, j = 0, . . . 3. Let
E =
3
i=0
cl(E
i
). Since x
1
, x
2
, y
1
are algebraically independent, Theorem 26.2
implies
(E) d(E) RM(E) = 3n.
On the other hand, note that by Proposition 44.10, for the strong subsets
cl(E
i
), Morley rank agrees with the d- and the -value. So cl(E
i
) [
cl(Eij)
cl(E
j
)
implies that e(cl(E
i
) cl(E
ij
), cl(E
j
) cl(E
ij
)) = 0.
Noting that (cl(E
i
)) = 2n, i = 0, . . . 3, and (cl(E
ij
)) = n for the six choices
of i ,= j 3, we can now use Exercise 44.2 to calculate (E) as
(E) = 4 2n 6 n = 2n.
Thus n = 0 and G is nite, a contradiction.
We now have the promised counterexample to Zilbers Conjecture:
Corollary 44.13. Let T
a
be the induced theory of T
on the strongly minimal

set D
a
(after adding the parameter a to the language). Then T
a
is strongly
minimal, not locally modular and does not interpret an innite eld.
Proof. By Corollary 43.4 since T
is almost strongly minimal over D

a
and not
1based, the strongly minimal set D
a
itself cannot be locally modular. Hence
1
the induced theory T
a
is strongly minimal and not 1-based. If T
a
would interpret
an innite eld, then so would T
, which it doesnt.
Exercise 44.1.
Prove that if A B and C B, then A C C.
Exercise 44.2.
If A, B are subgraphs of M, then (AB) = (A)+(B)(AB)e(AB, BA).
Exercise 44.3.
Let M = M
, and , p
1
, . . . , p
4
M where is a line, p
1
, p
2
are points not on ,
and p
3
, p
4
are points on . Show that acl(, p
1
, . . . , p
4
) is innite. (Hint: For any
k 3, a 2k-cycle (x0, x1, . . . , x
2k1
, x
2k
= x0) is a 0-simple extension of a graph A if
every xi has a unique neighbour in A.)
Exercise 44.4.
Show that for any a Aut(M
) acts 2-transitively on D
a
: for any two pairs
of elements x
1
,= x
2
and y
1
,= y
2
in D
a
, there is some g Aut(B
) such that
g(x
1
) = y
1
and g(x
2
) = y
2
.
Appendix A
Set Theory
In this appendix we collect some facts from set theory presented from the naive
point of view and refer the reader to [27] for more details. In order to talk about
classes as well as sets we begin with a brief axiomatic treatment.
1 Sets and classes
In modern mathematics, the underlying axioms are mostly taken to be ZFC,
i.e. the Zermelo-Fraenkel axioms (ZF) including the axiom of choice (AC) (see
e.g. [27], p.3). In order to talk about the monster model we may work in Bernays-
Gdel set theory (BG) which is formulated in a twosorted language, one type
of objects being sets and the other type of objects being classes, the element-
relation being dened between sets and sets and between sets and classes only.
Since the axioms are less commonly known, we give them here following [27],
p.70. We use lower case letters as variables for sets and capital letters for
referring to classes. BG has the following axioms:
1. (a) Extensionality: u(u x u y) x = y.
(b) Pairing: For any sets x and y, there is a set x, y.
(c) Union: For every set x, the set

x exists.
(d) Power Set: For every set x, the power set P(x) exists.
(e) Innity: There is an innite set.
2. (a) Class extensionality: u(u X u Y ) X = Y .
(b) Comprehension: Y
1
. . . Y
n
X X = x [ (x, Y
1
, . . . Y
n
) where is
a formula in which only set-variables are quantied.
(c) Replacement: If a class F is a function and x is a set, then
F(z): z x is a set.
208
APPENDIX A. SET THEORY 209
3. Regularity: Every nonempty set has an -minimal element.
For BGC we add:
5. Global Choice: There is a function F such that F(x) x for every non-
empty set x.
A model M of ZF becomes a model of BG by taking the denable subsets
of M as classes. This shows that BG is a conservative extension of ZF: any set-
theoretical statement provable in BG is also provable in ZF. Similarly, BGC is a
conservative extension of ZFC, see [16]. For a historical discussion see also [8].
2 Cardinals
Two sets are said to have the same cardinality if there is a bijection between
them. To each set x we can associate a cardinal [x[ (its cardinality) in such
a way that two sets have the same cardinal if and only if they have the same
cardinality.
1
The cardinality of the empty set is denoted by 0, a one-element set has
cardinality 1 etc.
Denition 2.1. For cardinals and we dene
,
if is the cardinality of x and the cardinality of a subset of x.
It is easy to see that is reexive and transitive. The CantorBernstein
Theorem (see [27] 3.2) states that two sets which can be injectively mapped
to each other have the same cardinality. Thus, is antisymmetric and hence
denes a partial ordering on the class of cardinals.
For any two sets x, y, by Zorns lemma (see [27] 5.4) there is a maximal
bijection f between subsets x
0
of x and a subset y
0
of y. If both x
0
and y
0
were
proper subsets, we could extend f. So either f is a bijection between x
0
and y
or between x and y
0
, showing that is in fact a linear ordering.
Denition 2.2. A wellordering of X is a linear ordering such that any non
empty subset of X contains a smallest element or, equivalently, such that X does
not contain innite properly descending chains. If X is a proper class rather
than a set, we also ask that for all x X the set y [ y < x of predecessors is
a set.
Lemma 2.3. The class of cardinals is wellordered by .
Proof. By the wellordering theorem (which, like Zorns Lemma is equivalent
to the Axiom of Choice, see [27] 5.1) any set can be wellordered. Using Zorns
lemma one can see as above that for any two wellorderings one them of is
isomorphic to an initial segment
2
of the other one.
If x is a wellordering with cardinality , then every smaller cardinal is the
cardinality of an initial segment of x. Since the initial segments of a well
ordering are wellordered by inclusion, the claim follows.
1
It is tempting to dene |x| as the class of all sets having the same cardinality as x, but
then a cardinal would not be a set.
2
An initial segment of a partial order X is a subset which contains for each of its elements
all smaller elements of X.
0
denotes the cardinality of the natural numbers (which of course are well
ordered). Sets of cardinality
0
are called countable. All proper initial segments
are nite, so
0
is the smallest innite cardinal.
Any family (
i
)
iI
of cardinals has an upper bound (e.g. the cardinality of
iI
x
i
, where
i
= [x
i
[). Thus there is a least upper bound sup
iI

i
.
Sums, products, and powers of cardinals are dened by disjoint union, carte-
sian power and sets of maps, respectively. Thus
[x[ +[y[ = [x y[ (1)
[x[ [y[ = [x y[ (2)
[x[
|y|
= [
y
x[ (3)
where we assume in (1) that x and y are disjoint. In (3), the set
y
x denotes the
set of all functions from y to x.
It is easy to see that these operations satisfy the same rules as the corre-
sponding operations on the natural numbers. E.g.,
_
.
The following is known as Cantors Theorem, see [27, 3.1, 3.5]) for a proof.
The rst part also follow from Theorem A 3.2.
Theorem 2.4.
1. If is innite, then = .
2. 2
> .
Note that 2.4.2. implies that there is no largest cardinal. Since every set of
cardinals has an upper bound, the class of cardinals is not a set.
Corollary 2.5.
1. If is innite, then + = max(, ).
2. If > 0 and are innite, then = max(, ).
3. If is innite, then
= 2
.
Proof. Let = max(, ). Then + + 2 = and if
> 0, then = .
Finally,
2
(2
= 2
= 2
.
Corollary 2.6. The set of all nite sequences of elements of a nonempty set
x has cardinality max([x[,
0
).
Proof. Let be the cardinality of all nite sequences in x. Clearly, [x[ and
0
. On the other hand
=
nN
[x[
n
_
sup
nN
[x[
n
_

0
= max([x[,
0
),
because
sup
nN
[x[
n
=
_
_
1, if [x[ = 1
0
, if 2 [x[
0
[x[, if
0
[x[.
For every cardinal ,
+
denotes the smallest cardinal greater than , the
successor cardinal of . By 2.4.2, we have
+
2
. The Generalised Continuum

Hypothesis (GCH) states that
+
= 2
for all innite .

GCH is independent of ZFC (assuming these axioms are consistent, see e.g. [27]
14.32).
3 Ordinals
The class On of ordinals is characterised up to isomorphism by properties shared
by the class of cardinals:
a) On is wellordered;
b) On is a proper class.
The wellordering of On yields the following principle of transnite induction:
Let c be a property of ordinals. Assume that whenever all ordinals
less than have property c, then itself has this property. Then
all ordinals have property c.
Let (X, <) be a wellordering and f a maximal isomorphism between an initial
segment of X and an initial segment of On. It is easy to see that f is dened on
all of X. If X is a proper class, then f is an isomorphism between X and On.
If X is a set, then f is an isomorphism between X and a proper initial segment,
which is of the form
[ < .
It is easy to see that f, and hence are uniquely determined by (X, <). (If
g : X On is another isomorphism to an initial segment, just consider the
smallest x such that f(x) ,= g(x).) This is called the order type of (X, <)
= otp(X, <).
To simplify notation one often identies an ordinal with its set of predecessors
3
= [ < .
In this way, is its own order type.
By the wellordering theorem any cardinal is the cardinality of an ordinal.
In axiomatic set theory, a cardinal is identied with the smallest ordinal having
this cardinality. In this way the class of cardinals is a subclass of On: an ordinal
is a cardinal if and only if all ordinals < have smaller cardinality.
For any ordinal its successor is dened as ; it is the smallest
ordinal greater than . Starting from the smallest ordinal 0 = , its successor
is 1 = 0, then 2 = 0, 1 and so on, yielding the natural numbers. The order
type of the natural numbers is denoted by = 0, 1, . . ., the next ordinal is
+ 1 = 0, 1, . . . . , , et cetera. Notice that the natural numbers and =
0
are cardinals.
3
In axiomatic set theory ordinals are dened in this way.
By denition, a successor ordinal contains a maximal element (so is
the successor of ) and we write = + 1. For natural numbers n, we put
+n = + 1 +. . . + 1
. .
n times
.
Ordinals > 0 which are not successor ordinals are called limit ordinals. When-
ever
i
[ i I is a non-empty set without biggest element, sup
iI

i
is a limit
ordinal. Any ordinal can be uniquely written as
+n,
with = 0 or a limit ordinal.
Theorem 3.1. Let G be an assignment such that for every function f dened
on an ordinal, G(f) is a set. Then there is a unique function F dened on all
of On satisfying the recursion formula
F() = G(F ).
Proof. It is easy to see by induction that for all there is a unique function
F
dened on and satisfying the recursion formula for < . Put F =
On
F
.
Example:
The aleph-function assigns ordinals to all innite cardinals. It is dened as
= the smallest innite cardinal greater than all
( < ).
We may thus talk about successor and limit cardinals. Similarly, we write
to denote the ordinal corresponding to
.
Example:
For every cardinal the beth function is dened as
() =
_
_
, if = 0,
2
()
, if = + 1,
sup
<
(), if is a limit ordinal.

A nite sequence of length n of elements from A is a function s: n A.
<
A =
_
n<
n
A
denotes the set of all nite sequences of elements from A.
Lemma 3.2 (The Gdel wellordering, see [27, 3.5]). There is a bijection On
On On, which induces a bijection for all innite cardinals . This
is a wellordering of On On, which has the property that for each pair (, )
the class of predecessors is actually a set. This implies that there is a (unique)
Proof. Dene
(, ) < (
)
_
max(, ), ,
_
<
lex
_
max(
),
_
where <
lex
is the lexicographical ordering on triples. This is a wellordering
with the additional property that for all , the class X
,
of predecessors of
(, ) is a set. This implies that there is a (unique) order-preserving bijection
: On On On. We show by induction that maps to for every
innite cardinal , which in turn implies = (see A 2.4.1). Since the
image of is an initial segment, it suces to show that X
,
has smaller
cardinality than for every , < . We note rst that X
,
is contained in
with = max(, ) + 1. Since is innite, we have that the cardinality
of is smaller that . Hence by induction [X
,
[ [[ [[ < .
For any linear order (X, <) we can easily construct a well-ordered conal
subset, i.e. a subset Y such that for any x X there is some y Y with x y.
Denition 3.3. The conality cf(X) is the smallest order type of a well ordered
conal subset of X.
It is easy to see that cf(X) is a regular cardinal where an innite cardinal
is regular if cf() = . Successor cardinals and are regular. The existence
of weakly inaccessible cardinals, i.e., uncountable regular limit cardinals cannot
be proven in ZFC.
Appendix B
Fields
4 Ordered elds
Let R be an integral domain. A linear < ordering on R is compatible with the
ring structure if for all x, y, z R
x < y x +z < y +z
x < y 0 < z xz < yz
A eld (K, <) together with a compatible ordering is an ordered eld.
Lemma 4.1. Let R be an integral domain and < a compatible ordering of R.
Then the ordering < can be uniquely extended to an ordering of the quotient
eld of R.
Proof. Put
a
b
> 0 ab > 0.
It is easy to see that in an ordered eld sums of squares can never be negative.
In particular, 1, 2, . . . are always positive and so the characteristic of an ordered
eld is 0. A eld K in which 1 is not a sum of squares is called formally real.
Lemma 4.2. A eld has an ordering if and only if it is formally real.
Proof. A eld with an ordering is formally real by the previous remark. For
the converse rst notice that , the set of all sums of squares in K, is a
216
APPENDIX B. FIELDS 217
semi-positive cone, i.e., a set P such that
P (1)
P +P P (2)
P P P (3)
1 , P (4)
The rst and third condition easily imply
x P 0
1
x
P.
Therefore, condition (4) is equivalent to P (P) = 0. It is also easy to see that
for all b, the set P + bP has all the properties of a semi-positive cone, except
possibly (4). Condition (4) holds if and only if b = 0 or b , P. We now choose
P as a maximal semi-positive cone. Then
x y y x P
and we obtain a compatible ordering of K.
Corollary 4.3 (of the proof). Let K be formally real
1
and let a be an element
of K. There is an ordering of K making a negative if and only if a is not the
sum of squares.
Proof. If a , , then a is a semi-positive cone.
Denition 4.4. An ordered eld (R, <) is real closed if
a) every positive element is a square,
b) every polynomial of odd degree has a zero.
(R, <) is called a real closure of the subeld (K, <) if R is real closed and
algebraic over K.
The eld of real numbers is real closed. Similarly, the eld of real algebraic
numbers is the real closure of Q. More generally, any eld which is relatively
closed in a real closed eld is itself real closed.
Theorem 4.5. Every ordered eld (K, <) has a real closure, and this is uniquely
determined up to isomorphism over K.
Proof (Sketch). Existence: Let K
0
be the positive cone of (K, <) and let L be a eld
extension of K. The ordering of K can be extended to L if and only if
P = {x1y
2
1
+ +xny
2
n
| xi K
0
, yi L}
1
It suces to assume that the characteristic of K is dierent from 2.
is a semi-positive cone of L, i.e. if 1 P. Therefore we may apply Zorns Lemma to
obtain a maximal algebraic extension R of K with an ordering extending the ordering
of K. We claim that R is real closed.
Let r be a positive element of R and assume that r is not a square. Since the ordering
cannot be extended to L = R(
r), there are ri R

0
and si, ti R such that
1 =
ri(si
r +ti)
2
.
Then 1 =
ri(s
2
i
r +t
2
i
), contradicting the fact that the right hand side is positive.
Thus every element of R is a square (and the ordering of R is unique).
Let f R[X] be a polynomial of minimal odd degree n without zero (in R). Clearly,
f is irreducible. Let be a zero of f (in the algebraic closure of R) and put L = R().
Since L cannot be ordered, 1 is a sum of squares in L. So there are polynomials
gi R[X] of degree less than n such that f divides h = 1 +
g
2
i
. The leading
coecients of the g
2
i
are squares in R and so cannot cancel out. Hence the degree of
h is even and less than 2n. But then the polynomial hf
1
has odd degree less than n
and no zero in R since h does not have a zero. A contradiction.
Uniqueness: Let R and S be real closures of (K, <). It suces to show that R and
S are isomorphic over K as elds. By (the easy characteristic 0 case of) Lemma 6.13
below it is enough to show that an irreducible polynomial f K[X] has a zero in R
if and only if it has one in S. As pointed out by Knebusch this follows from the next
lemma which we prove at the end of the section.
Lemma 4.6 (J.J. Sylvester). Let f be irreducible
2
in K[X] and (R, <) a real closed
extension of (K, <). The number of zeros of f in R equals the signature of the trace
form of the K-algebra K[X]/(f).
Notice that the uniqueness depends on the given ordering. For a formally real
eld there may be dierent orderings leading to non-isomorphic real closures.
The Fundamental Theorem of Algebra holds for arbitrary real closed elds
3
:
Theorem 4.7. Let R be real closed. Then C = R(
1) is algebraically closed.
Proof. Notice that all elements of C are squares: One square root of a+b
1 is given
by
a
2
+b
2
+a
2

a
2
+b
2
a
2
1,
where we choose according to the sign of b.
Let F be a nite extension of C. We claim that F = C. We may assume that
F is a Galois extension of R. Let G be a 2Sylow subgroup of Aut(F/R) and L
the xed eld of G. Then the degree L/R is odd. The minimal polynomial of a
generating element of this extension has the same degree and is irreducible. But since
2
It suces that f is nonconstant and all zeros in acl(K) have multiplicity 1.
3
Since in real closed elds the ordering is uniquely determined by the eld structure, it
makes sense to say that a eld is real closed without specifying its ordering.
all irreducible polynomials over R of odd degree are linear, we have L = R. Therefore
G = Aut(F/R) and hence also H = Aut(F/C) are 2groups. 2-groups are soluble
(even nilpotent). So if H is nontrivial, it has a subgroup of index 2 and then C has
a eld extension of degree 2; but this is impossible since every element is a square.
Hence H = 1 and F = C.
Corollary 4.8. If R is real closed, the only monic irreducible polynomials are:
Linear polynomials
X a,
(a R)
Quadratic polynomials
(X b)
2
+c,
(b, c R, c > 0)
Proof. Since all nonconstant polynomials f R[X] have a zero in R(
1), all irre-

ducible polynomials must be linear or quadratic. Any monic polynomial of degree 2 is
of the form (X b)
2
+c. It is reducible if and only if it has a zero x in R if and only
if c 0 (namely x = b
c).
Finally we prove Sylvesters Lemma (4.6). Let K be an ordered eld, R a real
closure of K and f K[X] irreducible. Consider the nite dimensional Kalgebra
A = K[X]/(f). The trace TrK(a) of a A is the trace of left multiplication by a,
considered as a vector space endomorphism. The trace form is a symmetric bilinear
form given by
(a, b)K = TrK(ab).
Let a1, . . . , an be a basis of A diagonalising ( , )K, i.e., (ai, aj)K = iij. The signature
of such a form is dened as the number of positive i minus the number of negative
i. Sylvesters Theorem (from linear algebra) states that the signature is independent
of the diagonalising basis. By tensoring A with R, we obtain the Ralgebra
AR = AR

= R[X]/(f).
The basis a1, . . . , an is also a diagonalising Rbasis for the trace form of AR with the
same i and hence the same signature as the trace form of A. We now split f in R[X]
into irreducible polynomials g1, . . . , gm. Since f does not have multiple zeros, the gi
are pairwise distinct. By the Chinese Remainder Theorem we have
R[X]/(f)

= R[X]/(g1) R[X]/(gm).
This shows that the trace form of R[X]/(f) is the direct sum of the trace forms of
the R[X]/(gi), and hence its signature is the sum of the corresponding signatures.
Sylvesters Lemma follows once we show that the signature of the trace form is equal
to 1 for a linear polynomial, while the signature is 0 for an irreducible polynomial of
degree 2.
If g is linear, then R[X]/(g) = R. The trace form (x, y)R = xy has signature 1. If
g is irreducible of degree 2, then R[X]/(g) = R(
1) and TrR(x +y
1) = 2x. The
trace form is diagonalised by the basis 1,
1. With respect to this basis we have

1 = 2 and 2 = 2 and so its signature is zero.
5 Dierential Fields
In this section, all rings considered have characteristic 0. Let R be a
commutative ring and S an R-module. (We mainly consider the case that S is
a ring containing R.) An additive map d: R S is called a derivation if
d(rs) = (dr)s +r(ds)
For any ring S the ring of dual numbers is dened as
S[] = a +b [ a, b S,
where
2
= 0. The following is an easy observation:
Lemma 5.1. Let S be a ring containing R and let : S[] S; a + b a.
Any derivation d: R S denes a homomorphism
t
d
: R S[]; r r + (dr)
inverting . Conversely, any such homomorphism arises from a derivation.
Lemma 5.2. Let S be a ring containing the polynomial ring R
= R[x
1
, . . . , x
n
]
and let d: R S be a derivation. For any sequence s
1
, . . . , s
n
of elements of S
there is a unique extension of d to a derivation d
: R
S taking x
i
to s
i
for
i = 1, . . . n.
Proof. Extend t
d
: R S[] via t
d
(x
i
) = x
i
+s
i
to a homomorphism t
d
: R

S[].
Lemma 5.3. Let R be a subring of the eld S and let R
be an intermediate eld
algebraic over R. Then every derivation d: R S can be uniquely extended to
R
.
Proof. It suces to consider the following two cases:
1. R
is the quotient eld of R: if a is a unit, then a +b is a unit in S[e] (with

inverse a
1
ba
2
.) Therefore, t
d
takes elements ,= 0 of R to units in S[] and
thus can be uniquely extended to R
.
2. R is a eld and R
= R[a] a simple algebraic extension: Let f(x) be the

minimal polynomial of a over R. Then t
d
can be extended to R
if and only if
(t
d
f)(x) has a zero of the form a + b in S[]. But (t
d
f)(x) = f(x) + (f
d
)(x),
hence
(5) (t
d
f)(a +b) = f(a +b) +f
d
(a) =
_
f
x
(a)b +f
d
(a)
_
.
Since char(S) = 0, f is separable, so
f
x
(a) ,= 0 and d can be extended to R
by
d
(a) = a f
d
(a)
_
f
x
(a)
_
1
.
Let C = c K [ dc = 0 denote the set of constants of K. Clearly, C is a
subring containing 1. The previous lemma implies that C is a subeld which is
relatively algebraically closed in K.
Corollary 5.4. Let R be a subring of the eld K. Any derivation d: R K
can be extended to a derivation of K.
Remark 5.5. Let (K, d) be a dierential eld and F a eld extension of K,
a, b F
n
. Assume that the ideal of all f in K[x] with f(a) = 0 is generated by
I
0
. Then the following are equivalent:
a) There is an extension of d to F with da = b.
b) For all f I
0
we have
f
x
(a)b +f
d
(a) = 0.
Proof. It is clear that a) implies b) because
d(f(a)) =
f
x
(a)b +f
d
(a).
In order to show the converse we have to extend the homomorphism t
d
: K
F[] to K[a] in such a way that a is mapped to a+b. This is possible if and only
if (t
d
f)(a +b) = 0 for all f I
0
. By (5) this implies
f
x
(a)b +f
d
(a) = 0.
Let (F, d) be a saturated model of T
DCF0
and V F
n
an irreducible ane
variety (see [45, Chapter 1] for basic algebraic geometry). The torsor T (V )
F
2n
is dened by the equations
f(x) = 0 and
f
x
(x)y +f
d
(x) = 0
for all f I(V )
4
. Clearly (a, da) T (V ) for all a V .
Remark 5.5 states that for any small subeld K over which V is dened,
and any (a
0
, b
0
) T (V ) such that a
0
V is generic over K there is a pair
(a, da) T (V ) satisfying the same eldtype over K as (a, b). This already
proves one direction of the following equivalence:
Remark 5.6 (Pillay). An algebraically closed dierential eld (K, d) is a model
of T
DCF0
if and only if for every irreducible ane variety V dened over K and
every regular section s: V T (V ) of the projection T (V ) V there is some
a V (K) with da = s(a).
Proof. We show that an algebraically closed dierential eld (K, d) with this
property is a model of the axioms of T
DCF0
.
4
I(V ) is called the vanishing ideal of V .
Let f and g be given with f irreducible. As in the proof of Theorem 9.22
we nd a eld extension F = K(, . . . , d
n
) in which , . . . , d
n1
are alge-
braically independent over K and f(, . . . , d
n
) = 0. Let be the inverse of
g(, . . . , d
n1
)
f
xn
(, . . . , d
n
). Note that K[, . . . , d
n
, ] is closed under d.
Putting c = (, . . . , d
n
, ), let (c, dc) = s(c) for some tuple s of polynomials
in K[x
0
, . . . , x
n+1
]. Then s denes a section V T (V ) where V K
n+1
is
the variety dened by f = 0. By assumption there is some (a, b) V (K) with
(a, b, da, db) = s(a, b). Hence f(a, . . . , d
n
a) = 0 and g(a, . . . , d
n1
a) ,= 0.
Remark 5.7 (Linear dierential equations). Let K be a model of T
DCF0
and
A an n nmatrix with coecients from K. The solution set of the system of
dierential equations
(6) dy = Ay
is an ndimensional Cvector space.
Proof. Choose an nnmatrix Y over an extension of K whose coecients are
algebraically independent over K. Lemma 5.2 shows that d can be extended to
K(Y ) by dY = AY . Since K is existentially closed, there is a regular matrix
over K, which we again denote by Y , such that dY = AY . The columns of Y
are n linearly independent solutions of (6). Such a matrix is called fundamental
system of the dierential equation. We show that any solution y is a C-linear
combination of the columns of Y : let z be a column over K with y = Y z. Then
AY z = d(Y z) = (dY )z +Y dz = AY z +Y dz.
Hence Y dz = 0, so dz = 0, i.e. the elements of z are constants.
Remark 5.8. Let K be a dierential eld and K
an extension of K with
elds of constants C and C
, respectively. Then K and C
are linearly disjoint

over C, i.e. any set of C-linearly independent elements of K remains linearly
independent over C
(see Section B.6).

Proof. Let a
0
, . . . , a
n
be in K and linearly dependent over C
. Then the columns

of
B =
_
_
_
_
_
a
0
. . . a
n
da
0
. . . da
n
.
.
.
.
.
.
d
n
a
0
. . . d
n
a
n
_
_
_
_
_
are linearly dependent. Let m < n be maximal with
Y =
_
_
_
_
_
a
0
. . . a
m
da
0
. . . da
m
.
.
.
.
.
.
d
m
a
0
. . . d
m
a
m
_
_
_
_
_
regular. We want to conclude that a
0
, . . . , a
m+1
is linearly dependent over C.
So we may assume n = m+1. Then the last row of B is a Klinear combination
of the rst m+ 1 rows. Thus for some mm matrix A all columns of Y and
z =
_
_
_
a
n
.
.
.
d
m
a
n
_
_
_
are solutions of the dierential equation dy = Ay. The proof of Remark 5.7
shows that z is a Clinear combination of the columns of Y .
6 Separable and regular eld extensions
Denition 6.1. Two rings R, S contained in a common eld extension are
said to be linearly disjoint over a common subeld k if any set of k-linearly
independent elements of R remains linearly independent over S. Note that then
R is also linearly disjoint over K from the quotient eld of S.
Remark 6.2. It is easy to see that this is equivalent to saying that the ring
generated by R and S is canonically isomorphic to the tensor product R
k
S.
So linear disjointness is a symmetric notion.
If S is a normal algebraic extension of K, so invariant under automorphims
of K
alg
/K, then the linear disjointness of R and S over K does not depend on
the choice of a common extension of R and S (see Exercise 35.4 for a similar
phenomenon).
Recall that an algebraic eld extension L over K is separable if every element
of L is a zero of a separable polynomial f K[X] and galois if L is normal and
separable over K.
Lemma 6.3. If F and L are eld extensions of K with L galois over K, then
F and L are linearly disjoint over K if and only if F L = K.
Proof. One direction is clear (see also the remark belowy). For the other direc-
tion assume F L = K. We may assume that L/K is nitely generated. Since
L/K is separable, it is nitely generated, say L = K(a). Let f be the minimal
polynomial of a over F. Since alle roots of f belong to L, f is in L[X] and
therefore in K[X]. It follows that [K(a) : K] = [F(a) : F] and F and L are
linearly disjoint over K.
That h
1
, . . . , h
n
are algebraically independent over L means that the mono-
mials in h
1
, . . . , h
n
are linearly independent over L. This observation implies
that L and H are algebraically independent over K if they are linearly disjoint
over K. The converse holds if L/K is regular:
Denition 6.4. A eld extension L over K is regular if L and K
alg
are linearly
disjoint over K in some common extension.
Lemma 6.5. If L is a regular extension of K and H/K is algebraically inde-
pendent from L/K, then L and H are linearly disjoint over K.
The reader may note that in the special case where H = K
alg
this is just
the denition of regularity.
Proof. Let l
i
L, h
i
H,

i<n
l
i
h
i
= 0, but not all h
i
= 0. Since L and H
are independent over K
alg
, the type tp(L/K
alg
H) is an heir of tp(L/K
alg
) by
Corollaries 26.5 or 36.13. Thus, there is a non-trivial ntuple

h
K
alg
such
that

l
= 0. Since L/K is regular, there is a nontrivial

h
K such that
= 0. This proves the claim.

Lemma 6.6. Let K be a eld. There is a natural bijection between isomorphism
types of eld extensions K(a
1
, . . . , a
n
) of K with n generators
5
and prime ideals
P in K[X
1
, . . . , X
n
]. Regular extensions correspond to absolutely prime ideals,
i.e. ideals which generate a prime ideal in K
alg
[X
1
, . . . , X
n
].
Proof. We associate with L = K(a
1
, . . . , a
n
) the vanishing ideal P = f
K[X
1
, . . . , X
n
] [ f(a
1
, . . . , a
n
) = 0. Conversely if P is a prime ideal the quo-
tient eld of K[X
1
, . . . , X
n
]/P is an extension of K generated by the cosets of
X
1
, . . . , X
n
.
Let I be the ideal generated by P in K
alg
[X
1
, . . . , X
n
]. Then K[a
1
, . . . , a
n
]
K
K
alg
and K
alg
[X
1
, . . . , X
n
]/I are isomorphic as Kalgebras. Now L/K is regu-
lar if and only if K[a
1
, . . . , a
n
]
K
K
alg
is an integral domain, which means that
I is prime.
Remark 6.7. It is easy to see that the case tr. deg(a
1
, . . . , a
n
) = n 1 cor-
responds exactly to the case where P is a principal ideal, generated by an
irreducible polynomial f K[X
1
, . . . , X
n
]. f is uniquely determined up to a
constant factor. ([31, VII, Ex.26])
Denition 6.8. A eld extension L over K is separable if, in some common
extension, L and the perfect hull of K are linearly disjoint over K. The perfect
hull is the smallest perfect eld containing K, so in characteristic 0 equals K,
and in characteristic p is the union of all K
p
n
= a K
alg
[ a
p
n
K. This
extends the denition in the algebraic case.
Again this does not depend on the choice of the common extension L and the
perfect hull of K. Note that regular extensions are separable.
Let K be a eld of characteristic p > 0. For any subset A we call the eld
K
p
(A) the pclosure of A. This denes a pregeometry on K (see Section C8)
whose associated dimension is the degree of imperfection. A basis of K in the
sense of this pregeometry is called a pbasis. If b = (b
i
[ i I) is a pbasis of
K, then the products b
iI
b
i
i
dened for multiindices = (
i
[ i I)
where 0
i
< p and almost all
i
equal to zero, form a linear basis of K
over K
p
. If the degree of imperfection of K is a nite number e, it follows that
[K : K
p
] = p
e
.
Remark 6.9. L is a separable extension of K if and only if L and K
p
1
are
linearly disjoint over K. It follows that L/K is separable if and only if a pbasis
of K stays pindependent in L.
5
By quantier elimination this is just Sn(K) in the theory of K
alg
.
Proof. Consider a sequence (a
i
) of elements of L which is linearly independent
over K. Then (a
i
) is independent over K
p
1
by assumption which implies that
(a
p
i
) is independent over K. Then (a
p
i
p
1
implying that
(a
i
p
2
. In this way one can show inductively that (a
i
)
is independent over K
p
n
for all n.
Lemma 6.10. If b
1
, . . . , b
n
are pindependent in K, they are algebraically in-
dependent over F
p
.
Proof. b
1
, . . . , b
n1
form a pbasis of F = F
p
(b
1
, . . . , b
n1
). So K/F is separable
by Remark 6.9. If an element c K is algebraic over F, it is separably algebraic
over F. So c is also separably algebraic over K
p
F, which is only possible if c
belongs to K
p
F. Since b
n
does not belong to K
p
F, it follows that b
n
is not
algebraic over F.
Lemma 6.11. 1. Let R be an integral domain of characteristic p and let b be
a pbasis of R in the sense that every element of R is a unique R
p
linear
combination of the b
. Then b is also a pbasis of the quotient eld of R.

2. Let b be pbasis of K and L is a separable algebraic extension of K, then
b is also a pbasis of L.
Proof. 1): Let K be the quotient eld of R. Clearly b is pindependent in
K. Since the b are algebraic over K
p
, K
p
[b] is a subeld of K which contains
R = R
p
[b]. So K = K
p
[b].
2): We may assume that L/K is nite. Since L
p
and K are linearly disjoint over
K
p
, we have [L
p
K : K] = [L
p
: K
p
] = [L : K]. It follows that L
p
K = L.
Lemma 6.12. The eld L = K(a
1
, . . . , a
n
) is separable over K if and only if
there is a transcendence basis a
a
1
, . . . , a
n
of L/K such that L is separably
algebraic over K( a
).
Proof. This follows from the fact that every irreducible polynomial in K[X] is of
the form f(X
p
k
) for some k 0 and some separable polynomial f K[X].
Lemma 6.13. Algebraic eld extensions of K are isomorphic over K if and
only if the same polynomials in K[X] have a zero in these extensions.
Proof. Consider two algebraic extensions L and L
of K. A compactness argu-
ment shows that L and L
are isomorphic over K if and only if they contain, up

to isomorphy over K, the same nitely generated subextensions. The condition
that the same polynomials in K[X] have zeros in L and L
is equivalent to L
and L
having the same simple subextensions. So the lemma is clear if L and

L
are separable over K.

For the general case it suces to prove the following: let L be a nite
algebraic extension of K and L
an extension of K such that for every a L

there is an a
such that K(a) is isomorphic to K(a
) under an automorphism
xing K and taking a to a
. Then L
contains a copy of L over K.

By the above we may assume that K has characteristic p and is innite.
The extension L/K contains a separable extension F/K such that L is purely
inseparable over F. Let us rst remark that for every a L, the eld F(a) is
simply generated over K. To see this choose a generator c of F/K and a power
q of p with a
q
F. For all d K the element a + dc also generates L over
F. We claim that for some d K, (a + dc)
q
generates F/K. If not, innitely
many of the (a +dc)
q
are contained in the same intermediate eld K F
F
as there are only nitely many of these. So there are d
1
,= d
2
K such that
(a + d
1
c)
q
, (a + d
2
c)
q
F
. This implies c
q
F
. But since F/K is separable,

c
q
is also a generator of F/K, a contradiction.
By assumption, for every a L there is an embedding f
a
: F(a) L
over K. Let f
1
, . . . , f
n
be a list of all embeddings of F into L
. We construct a
sequence a
0
, a
1
, . . . as follows. Assume that a
0
. . . , a
n1
are already constructed.
Consider for each i the eld L
i
/F generated by all a
j
with f
aj
F = f
i
. We are
nished once one of the L
i
equals L. As long this is not the case, by Neumanns
Lemma 9.9, we nd an a
n
L, which does not yet belong to any of the L
i
.
Adding a
n
to the appropriate L
i
this construction stops after nitely many
steps yielding a nite set A of generators of L over F such that all f
a
for a A
agree on F. These f
a
combine to an embedding of L = F(A) into L
.
Exercise 6.1.
Show that elementary extensions are regular.
7 Pseudo-nite elds and pronite groups
Denition 7.1. A eld K is pseudo algebraically closed (PAC) if every ab-
solutely irreducible ane variety dened over K has a Krational point.
See [45, Chapter 1] for basic algebraic geometry.
Lemma 7.2. Let K be a eld. The following are equivalent:
a) K is PAC,
b) Let f(X
1
, . . . , X
n
, T) K[X
1
, . . . , X
n
, T] be absolutely irreducible and of
degree greater than 1 in T and let g K[X
1
, . . . , X
n
]. Then there exists
(a, b) K
n
K such that f(a, b) = 0 and g(a) ,= 0.
c) K is existentially closed in every regular eld extension.
Proof. a)b): Let C be an algebraically closed eld containing K. The zero
set of f denes an absolutely irreducible variety X C
n
C over K. Then the
set V = (a, b, c) [ f(a, b) = 0, g(a)c = 1 is absolutely irreducible as well and
dened over K.
b)c): Let L/K be a nitely generated regular extension. Since regular ex-
tensions are separable, by Lemma 6.12 we nd a transcendence basis a of L/K
such that L is separable algebraic over K( a). Then there is b separable algebraic
over K( a) such that L = K(a
1
, . . . , a
n
, b). Let now c
1
, . . . , c
m
be elements of
L satisfying certain equations over K.
6
We need c
i
K satisfying the same
equations. Write c
i
=
hi( a,b)
g(a)
for polynomials h
i
( x, y) and g( x) over K. The
vanishing ideal of ( a, b) over K is generated by some f( x, y), which is absolutely
irreducible since L/K is regular. There are a
, b
K such that f( a
, b
) = 0
and g( a
, b
) ,= 0. Now the c
i
=
hi( a
,b
)
g(a
)
satisfy all equations satised by the c
i
.
c)a): Let V be an absolutely irreducible variety over K. If c V is a point
in C generic over K, then K( c)/K is regular by Lemma 6.6.
Corollary 7.3. PAC is an elementary property.
Proof. Being absolutely irreducible is an elementary property of the coecients
in K because the theory of algebraically closed elds has quantier elimination.
Denition 7.4. A pronite group is a compact
7
topological group with a neigh-
bourhood basis for 1 consisting of subgroups.
6
As we are working in an innite eld, we dont need inequalities.
7
Note that for us compact spaces are Hausdor.
Pronite groups can be presented as inverse limits of nite groups, see [55]
for more details.
Denition 7.5. A pronite group is called procyclic if all nite (continuous)
quotients are cyclic.
Lemma 7.6 (see [55]). Let G be a pronite group. The following are equivalent:
a) G is procyclic;
b) G is an inverse limit of nite cyclic groups;
c) G is (topologically) generated by a single element a, i.e. the group abstractly
generated by a is dense in G.
By a generator for a pronite group we always mean a topological generator.
Denition 7.7.

Z is the inverse limit of all Z/nZ, with respect to the canonical
projections.
Lemma 7.8. A pronite group G is procyclic if and only if for every n there
is at most one closed subgroup of index n. If G has exactly one closed subgroup
of index n for each n > 0, then G is isomorphic to

Z.
Proof. It suces to show that a nite group G of order n is cyclic if and only if
for each k dividing n there is a unique subgroup of order k. If a is a generator
for G and H a subgroup of G of order k, then a
n/k
is a generator for H, proving
uniqueness. For the other direction, note that by assumption any two elements
having the same order generate the same subgroup. Thus, if G has an element
of order k, G contains exactly (k) such elements. The claim now follows from
the equality
n =
k|n
(k).
Lemma 7.9. Let G H be an epimorphism of procyclic groups. Then any
generator of H lifts to a generator of G.
Proof. As it suces to prove the claim for nite cyclic groups G and H, we may
again assume that G and H are pgroups. If H = 0, we choose any generator
as the preimage. If H ,= 0, any preimage of a generator of H is a generator
of G.
For a eld K we write G(K) for its absolute Galois group Aut(K
alg
/K).
Denition 7.10. A perfect eld K is called procyclic if G(K) is procyclic, and
1free if G(K)

=

Z.
By Lemma 7.8 a perfect eld is procyclic (1free, respectively) if and only if
it has at most one (exactly one, respectively) extension of degree n in K
alg
for
every n. Thus, being procyclic or 1free is an elementary property of K.
Denition 7.11. A perfect 1free PACeld is called pseudo-nite.
Remark 7.12. Being pseudo-nite is an elementary property. A theorem of
Lang and Weil (see [32]) on the number of points on varieties over nite elds
shows that innite ultraproducts of nite elds are PAC and hence pseudo-nite.
Proposition 7.13. Let K be perfect and L/K a eld extension. Then the
following are equivalent:
a) L/K regular,
b) K is relatively algebraically closed in L,
c) the natural map G(L) G(K) is surjective.
Proof. Using remark 6.2, it is easy to see that a) implies c). Clearly, c) implies
b). To see that b) implies a) assume that K is relatively algebraically closed
in L. Since every nite extension between K and K
alg
is generated by a single
element, it suces to show for a K
alg
that the minimal polynomial f of a
over L has coecients in K.
But this follows as the coecients of f can be expressed by conjugates of
a K
alg
L.
If L is perfect and N/L and L/K are regular extensions, then also N/K is
regular.
Corollary 7.14. Let L/K be regular, L procyclic and K 1free. Then L is
1free. If N/L is an extension of L such that N/K is regular, then N/L is
regular.
Proof. If G(L) is procyclic and : G(L)

Z = G(K) is surjective, then is
an isomorphism. Let : G(N) G(L) be the natural map. If is surjective,
then so is .
Any pronite group G acts faithfully and with nite orbits on the set S
consisting of all left cosets of nite index subgroups. Since point stabilisers
are open subgroups of G, this action is continuous with respect to the discrete
topology on S. We use this easy observation to prove the following:
Lemma 7.15. Any procyclic eld has a regular pseudo-nite eld extension.
Proof. Let K be a procyclic eld. Let S be as in the previous remark for G =

Z
and let L
= K(X
s
)
sS
with the action of G on L
given by the action of G

on S. Then L
is a Galois extension of L = Fix(G) with Galois group

Z and
L
and L are regular extensions of K. Let Aut(K

alg
L
/L) be a common
extension of a generator of G(K) and a generator of Gal(L
/L). Extend to
some
Aut(L
alg
/L) and consider the xed eld L
of
. Then L
is a 1free
regular extension of K. It is now easy to construct, by a long chain, a regular
procyclic extension N of L
which is existentially closed in all regular procyclic

extensions. N is again 1free. In order to see that N is PAC, we consider a
regular extension N
of N. Let G(N
) be a lift of a generator of G(N) and

N
the xed eld of in N
alg
. Then N
is a regular procyclic extension of N.

By construction, N is existentially closed in N
and hence also in N
.
Procyclic elds have the amalgamation property for regular extensions:
Lemma 7.16. If L
1
, L
2
are regular procyclic extensions of a eld K, there is
a common procyclic regular extension H of L
1
and L
2
in which L
1
and L
2
are
linearly disjoint over K.
Proof. Let L
1
and L
2
be regular procyclic extensions of a eld K (so K is also
procyclic). We may assume that the L
i
are algebraically independent over K in
some common eld extension. Let be a (topological) generator of G(K). By
Lemma 7.9, we can lift to generators
i
of G(L
i
). By Lemma 6.5, L
1
and L
2
are linearly disjoint. As L
alg
1
L
alg
2
is the quotient eld of the tensor product of
L
alg
1
and L
alg
2
over K
alg
, the
i
generate an automorphism of L
alg
1
L
alg
2
which can
be extended to an automorphism of (L
1
L
2
)
alg
. The xed eld of is procyclic
and a regular extension of L
1
and L
2
.
By Lemma 7.15 this also implies that pseudo-nite elds have the amalga-
mation property for regular extensions.
Appendix C
Combinatorics
8 Pregeometries
In this section, we collect the necessary facts and notions about pregeometries,
existence of bases and hence a well-dened dimension, modularity laws etc. First
recall Denition 20.5:
Denition. A pregeometry (X, cl) is a set X with a closure operator cl : P(X)
P(X) such that for all A X and a, b X
a) A cl(A)
b) cl(A) is the union of all cl(A
), where the A
range over all nite subsets

of A.
c) cl(cl(A)) = cl(A)
d) (exchange property) a cl(Ab) cl(A) b cl(Aa)
Remark 8.1. The following structures are pregeometries:
1. A vector space V with the linear closure operator.
2. For a eld K with prime eld F, the relative algebraic closure cl(A) =
F(A)
alg
3. The pclosure in a eld K of characteristic p > 0, i.e. cl(A) = K
p
(A).
Proof. We prove the exchange property for algebraic dependence in a eld K as
an example.
1
We may assume that A is a subeld of K. Let a, b be such that
1
For the pclosure see [9, 13].
232
APPENDIX C. COMBINATORICS 233
a A(a)
alg
. So there is a nonzero polynomial F A[X, Y ] with F(a, b) = 0.
If we also assume that b / A(b)
alg
, it follows that F(a, Y ) = 0. This implies
a A
alg
.
A pregeometry in which points and the empty set are closed, i.e., in which
cl
() = and cl
(x) = x for all x X

is called geometry. For any pregeometry (X, cl), there is an associated geometry
(X
, cl
) obtained by setting X
= X
/ , and cl
(A/ ) = cl(A)
/ where
is the equivalence relation on X
= X cl() dened by cl(x) = cl(y). Starting

from a vector space V , the geometry obtained in this way is the associated
projective space P(V ). The important properties of a pregeometry are mostly
in fact properties of the associated geometry.
Denition 8.2. Let (X, cl) be pregeometry. A subset A of X is called
1. independent if a , cl(A a) for all a A;
2. a generating set if X = cl(A);
3. a basis if A is an independent generating set.
Lemma 8.3. Let (X, cl) be a pregeometry with generating set E. Any indepen-
dent subset of E can be extended to a basis contained in E. In particular every
pregeometry has a basis.
Proof. Let B be an independent set. If x is any element in Xcl(B), Bx is
again independent. To see this consider an arbitrary b B. Then b , cl(Bb),
whence b , cl(B b x) by exchange.
This implies that for a maximal independent subset B of E, we have E
cl(B) and therefore X = cl(B).
Denition 8.4. Let (X, cl) be a pregeometry. Any subset S gives rise to two
new pregeometries, the restriction (S, cl
S
) and the relativisation (X, cl
S
), where
cl
S
(A) = cl(A) S,
cl
S
(A) = cl(A S).
Remark 8.5. Let A be a basis of (S, cl
S
) and B a basis of (X, cl
S
). Then the
(disjoint) union A B is a basis of (X, cl).
Proof. Clearly A B is a generating set. Since B is independent over S, we
have b , cl
S
(B b) = cl(A B b) for all b B. Consider an a A. We
have to show that a , cl(A
B), where A
= A a. As a , cl(A
), we let
B
be a maximal subset of B with a , cl(A
). B
,= B would imply that

a cl(A
b) for any b B B
which again would imply b cl(AB
),
a contradiction.
A is called a basis of S and B a basis over or relative to S.
Lemma 8.6. All bases of a pregeometry have the same cardinality.
Proof. Let A be independent and B a generating subset of X. We show that
[A[ [B[.
Assume rst that A is innite. Then we extend A to a basis A
. Choose for
every b B a nite subset A
b
of A
with b cl(A
b
). Since the union of the A
b
is a generating set, we have A
bB
A
b
. This implies that B is innite and
[A[ [A
[ [B[.
Now assume that A is nite. [A[ [B[ follows immediately from the fol-
lowing exchange principle: For any a A B there is b B A such that
A
= b A a is independent. Proof: Since a cl(B), B cannot be con-

tained in cl(A a). Choose b in B but not in cl(A a). It follows from the
exchange property that A
is independent.
Denition 8.7. The dimension dim(X) of a pregeometry (X, cl) is the cardi-
nality of a basis. For a subset S of X let dim(S) be the dimension of (S, cl
S
)
and dim(X/S) the dimension of (X, cl
S
).
By Remark 8.5 we have
Lemma 8.8.
dim(X) = dim(S) + dim(X/S).
The dimensions in our three standard examples are:
The dimension of a vector space.
The transcendence degree of a eld.
The degree of imperfection of a eld of nite characteristic (see [9, 13,
Ex.1]).
Let (X, cl) be a pregeometry. For arbitrary subsets A and B one sees easily
that
dim(A B) + dim(A B) dim(A) + dim(B).
One may hope for equality to hold if A and B are closed:
Denition 8.9. We call a pregeometry (X, cl) modular, if
(1) dim(A B) + dim(A B) = dim(A) + dim(B)
for all clclosed A and B.
The main examples are
trivial pregeometries where cl(A B) = cl(A) cl(B) for all A, B.
vector spaces with the linear closure operator.
Let K be a eld of transcendence degree at least four over its prime eld F.
The following argument shows that the pregeometry of algebraic dependence
on K is not modular: Choose x, y, x
, y
K algebraically independent over

F. From these elements we can compute a, b K such that ax + b = y and
ax
+b = y
. Since the elements x, x
, a, b generate the same subeld as x, y, x
, y
,
they are also algebraically independent. This implies that F(x) and F(x
) are
isomorphic over F(a, b)
cl
, where cl denotes the relative algebraic closure in K.
This isomorphism maps y to y
and therefore we have

F(x, y)
cl
F(a, b)
cl
F(x, y)
cl
F(x
, y
)
cl
= F
cl
.
So K is not modular since
tr. deg F(x, y, a, b) + tr. deg F = 3 + 0 < 2 + 2 = tr. deg F(x, y) + tr. deg F(a, b).
Let us call two sets A and B (geometrically) independent over C if all
subsets A
0
A and B
0
B which are both independent over C are disjoint
and their union is again independent over C. The following is then easy to see:
Lemma 8.10. For a pregeometry (X, cl) the following are equivalent:
1. (X, cl) is modular.
2. Any two closed A and B are independent over their intersection.
3. For any two closed sets A and B we have dim(A/B) = dim(A/A B).
Lemma 8.11. A pregeometry (X, cl) is modular if and only if for all a, b, B
with dim(ab) = 2, dim(ab/B) = 1, there is c cl(B) such that dim(ab/c) = 1.
Proof. If (X, cl) is modular and a, b, B are as in the lemma, then ab and B are
dependent, but independent over the intersection of cl(ab) and cl(B). Let c be
an element of the intersection which is not in cl(). Then dim(ab/c) = 1.
Assume that the property of the lemma holds. We show that the third condi-
tion of Lemma 8.10 is satised. For this we may assume that n = dim(A/AB)
is nite and proceed by induction on n. The cases n = 0, 1 are trivial. So as-
sume n 2. Let a
1
, . . . a
n
be a basis of A over A B. We have to show that
dim(a
1
, . . . , a
n
/B) = n. By induction we know that dim(a
1
. . . a
n2
/B) = n2.
So it is enough to show that dim(a
n1
a
n
/B
) = 2 where B
be the closure
of a
1
, . . . , a
n2
B. If not, by our assumption there is c B
such that
dim(a
n1
a
n
/c) = 1.
Denition 8.12. (X, cl) is locally modular if (1) holds for all closed sets A, B
with dim(A B) > 0.
Remark 8.13. Clearly, (X, cl) is locally modular if and only if for all x
X cl() the relativised pregeometry (X, cl
x
) is modular.
An ane subspace of a vector space V is a coset of a subvector space. The
ane subspaces of V dene a locally modular geometry, which is is not modular,
because of the existence of parallel lines. Note that in this example, points have
dimension one, lines dimension two, etc.
The arguments on page 235 show also that a eld of transcendence degree at
least ve is not locally modular. Just replace F by a subeld of transcendence
degree 1.
Exercise 8.1.
Consider a pregeometry (X, cl). Let A [
cl
C
B be the relation of A and B being
independent over C and A [
0
C
B the relation cl(AC) cl(BC) = cl(C). Show
the following:
1. [
cl
has the following properties as listed in Theorem 29.13: Monotony,
Transitivity, Symmetry, Finite Character and Local Charac-
ter.
2. [
0
has Weak Monotony (cf. Theorem 36.10), Symmetry, Finite
Character and Local Character. Monotony holds only if [
0
=
[
cl
, i.e. if X is modular.
Exercise 8.2.
Prove that trivial pregeometries are modular.
Exercise 8.3 (P. Kowalski).
Let K be a eld of characteristik p > 0 with degree of imperfection at least 4.
Prove that K with pdependence is not locally modular.
Exercise 8.4.
(X, cl) is modular if and only if for all c cl(A B) there are a cl(A) and
b cl(B) such that c cl(a, b).
Exercise 8.5.
The set of all closed subsets of a pregeometry forms a lattice, where the inmum
is intersection and the supremum of X and Y is X.Y = cl(XY ). Show that
a pregeometry is modular if and only if the lattice of closed sets is modular, i.e.
if for all closed A, B, C
A C A. (B C) = (A. B) C.
Exercise 8.6.
Let (X, cl) be a pregeometry of uncountable dimension. Suppose that for all
closed B of countable dimension the automorphism group Aut(X/B) acts tran-
sitively on X B. Then (X, cl) is locally modular if and only if for all closed B
of countable dimension and all a, b with dim(a, b) = 2 and dim(a, b/B) = 1 and
every Aut(X/B) the two pairs (a, b) and ((a), (b)) are not independent
over . Conclude that for any nite A, X is locally modular if and only if X
A
is locally modular.
9 The ErdsMakkai Theorem
Theorem 9.1 (ErdsMakkai). Let B be an innite set and o a set of subsets
of B with [B[ < [o[. Then there are sequences (b
i
[ i < ) of elements of B and
(S
i
[ i < ) of elements of o such that either
b
i
S
j
j < i (2)
or
b
i
S
j
i < j (3)
for all i, j .
Proof. Choose a subset o
be a subset of o of the same cardinality as B such

that any two nite subsets of B which can be separated by an element of o can
be separated by an element of o
. The hypothesis implies that there must be

an element S
of o which is not a Boolean

2
combination of elements of o
.
Assume that for some n, three sequences (b
i
[ i < n) in S
, (b
i
[ i < n) in
B S
and (S
i
[ i < n) in o
have already been constructed. Since S
is not a
Boolean combination of S
0
, . . . , S
n1
, there are b
n
S
and b
n
B S
such
that for all i < n
b
n
S
i
b
n
S
i
.
Choose S
n
as any set in o
, which separates b
0
, . . . , b
n
and b
0
, . . . , b
n
.
Now, an application of Ramseys theorem shows that we may assume that
either b
n
S
i
or b
n
, S
i
for all i < n. In the rst case we set b
i
= b
i
and get
(2), in the second case we set b
i
= b
i+1
and get (3).
2
It actually suces to consider positive Boolean combinations.
10 The ErdsRado Theorem
Denition 10.1. For cardinals , , we write ()
n
( arrows ) to
express that for any function f : []
n
there is some A with [A[ = ,
such that f is constant on [A]
n
. In other words, every partition of []
n
into
pieces has a homogeneous set of size .
With this notation Ramseys Theorem 15.5 states that
()
n
k
for all n, k < .
In an analogous manner one can dene a cardinal to be a Ramsey cardinal
if ()
<
2
. In other words, if for any n a partition of []
n
into two classes
is given, there is a set of size simultaneously homogeneous for all partitions.
A Ramsey cardinal satises ()
<
for all < (see [29] 7.14). More

generally, an uncountable cardinal is called weakly compact if it satises
()
2
2
. Such cardinals are weakly inaccessible and their existence cannot be
proven from ZFC.
Theorem 10.2 (Erds-Rado).
+
n
() (
+
)
n+1
.
Proof. This follows from
+
(
+
)
1
and the following lemma:

Lemma 10.3. If
+
(
+
)
n
, then (2
)
+
(
+
)
n+1
.
Proof. We note rst that the hypothesis implies . Now let B be a set of
cardinality (2
)
+
and f : [B]
n+1
be a colouring. If A is a subset of B, we
call a function p: [A]
n
a type over A. If b B A, the type tp(b/A) is the
function which maps each nelement subset s of A to f(s b). If [A[ ,
there are at most 2
many types over A. Thus an argument as in the proof of

Lemma 23.2 shows that there is some B
0
B of cardinality 2
such that for

every A B
0
of cardinality at most every type over A which is realised in B
is already realised in B
0
.
Fix an element b B B
0
. We can easily construct a sequence (a
)
<
+ in
B
0
such that every a
has the same type over a
[ < as b. By assumption
a
[ <
+
contains a subset A of cardinality
+
such that tp(b/A) is
constant on [A]
n
. Then f is constant on [A]
n+1
.
Appendix D
Solutions of Exercises
Exercise 2.3
We consider formulas which are built using , , , and we move the quantiers
outside using
x x
x x
( x(x, y)) z ( (z, y))
( x(x, y)) z ( (z, y))
In the last equivalence we have replaced the bound variable x by a variable z,
which does not occur freely in .
Exercise 3.1
The theory T
DLO
of dense linear orders without endpoints is axiomatised in L
O
by
x x < x
x, y, z (x < y y < z x < z)
x, y (x < y x
.
= y y < x)
x, z (x < z y (x < y y < z))
xy x < y
yx x < y
The class of all algebraically closed elds can be axiomatised by the theory
T
ACF
:
240
APPENDIX D. SOLUTIONS OF EXERCISES 241
T
F
(eld axioms)
For all n > 0:
x
0
, . . . x
n1
y x
0
+x
1
y +. . . +x
n1
y
n1
+y
n
.
= 0
Exercise 3.3
Let A and B be two elementarily equivalent Lstructures. It is easy to see
that A and B are isomorphic if L is nite. Let L be arbitrary and assume
for contradiction that A and B are not isomorphic. Then for every bijection
f : A B there is a Z
f
L which is not respected by f. If L
0
is the set of
all the Z
f
, A L
0
and B L
0
are not isomorphic, which contradicts our rst
observation. One should note that the isomorphy follows also from Exercise 23.1
and Lemma 12.3.
Exercise 4.2
Hint: Let T = : M[= for all M c. For M[= T construct an ultraprod-
uct of Lstructure in c having the same theory.)
Exercise 6.2
For every prime p, T
ACFp
has a model which is the union of a chain of nite
elds.
Exercise 7.1
We imitate the proof of Lemma 7.1. That a) implies b) is clear. For the converse
consider an element y
1
of Y
1
and 1
y1
, the set of all elements of 1 which contain
y
1
. b) implies that the intersection of the sets in 1
y1
is disjoint from Y
2
. So a
nite intersection h
y1
of elements of 1
y1
is disjoint from Y
2
. The h
yi
, y
1
Y
1
,
cover Y
1
. So Y
1
is contained in the union H of nitely many of the h
yi
. H
separates Y
1
from Y
2
.
b) a):
For any model A
1
of T
1
let 1
A1
be the set of all sentences from 1 which are
true in A
1
. b) implies that 1
A1
and T
2
cannot have a common model. By the
Compactness Theorem there is a nite conjunction
A1
of sentences from 1
A1
inconsistent with T
2
. Clearly,
T
1

A1
[ A
1
[= T
1
is inconsistent since any model A

1
of T
1
satises
A1
. Again by compactness T
1
implies a disjunction of nitely many of the
A1
. This is in 1 and separates
T
1
from T
2
.
Exercise 8.1
Hint: For any simple existential formula write down the equivalence to a quan-
tier free formula. Show that this is equivalent to an sentence and that T
is axiomatised by these.
Exercise 9.1
Like the proof of 9.2 with order preserving automorphisms replaced by edge
preserving ones.
Exercise 9.2
The cases T
ACF
, T
RCF
are easy. So we concentrate on T
DCF0
. Clearly, the
algebraic closure of the dierential eld generated by A is contained in the
model theoretic algebraic closure. For the converse, let K
0
be an algebraically
closed dierential eld and a
0
an element not contained in K
0
. If dim(a
0
/K
0
) is
innite, then a
0
and all its derivatives have the same type over K, so a
0
is not
not modeltheoretically algebraic over K
0
. If dim(a
0
/K
0
) = n > 0, consider
the minimal polynomial f of a
0
over K
0
. Let K
1
be a d-closed extension of K
0
containing a
0
. Then f remains irreducible over K
1
and there is some a
1
whose
minimal polynomial over K
1
is f. Now extend K
1
to some eld K
2
containing
a
1
etc.. In this way we obtain an innite sequence of distinct elements having
the same type over K
0
as a
0
, showing that a
0
is not algebraic over K
0
in the
sense of model theory.
Exercise 9.3
Let K be algebraically closed. X = a [ / [= (a, b) a denable subset of K
n
and f : X X is given by an ntuple of polynomials f(x, b). We may assume
that (x, z) is quantier free. We want to show that K satises
y
_
x
_
(x, y) (f(x, y), y)
_
x, x
_
(x, y) (x
, y) f(x, y)
.
= f(x
, y) x
.
= x
_

x
_
(x
, y) x((x, y) f(x, y)
.
= x
)
_
_
.
This is obviously true in nite elds (even in all nite L
R
structures) and logi-
cally equivalent to an sentence. So the claim follows from Exercise 6.2.
For the second part use Exercise 23.13 and proceed like before.
Exercise 11.1
[] is a singleton if and only if [] is nonempty and cannot be divided into
two non-empty clopen subsets [ ] and [ ]. This means that for all
either or follows from modulo T. So [] is a singleton if and only if
generates the type
= (x) [ T x ((x) (x)),
which of course must be the only element of [].
This shows that [] = p implies that isolates p. If, conversely, isolates
p, this means that is consistent with T and contains p. Since p is a type, it
follows p = .
Exercise 11.2
a): The sets [] are a basis for the closed subsets of S
n
(T). So the closed sets
of S
n
(T) are exactly the intersections

[] = p S
n
(T) [ p.
b): X is the union of a sequence of countable nowhere dense sets X
i
. We may
assume that the X
i
are closed, i.e. of the form p S
n
(T) [
i
p. That
X
i
has no interior means that
i
is not isolated. The claim follows now from
Corollary 10.3.
Exercise 11.3
Let X = tp(a
0
, a
2
, . . .) [ the a
i
enumerate a model of T. Consider for every
formula ( v, y) the set X
= p S
[ (y( v, y) ( v, v
i
)) p for some i.
The X
are open and dense and X is the intersection of the X
.
Exercise 11.5
The homeomorphism from S
m
(aB) to the ber above tp(a/B) is given by
tp(c/aB) tp(ca/B).
Exercise 14.1
If T is countable, not small, S
n
(T) uncountable, say. Prove that there is a
consistent formula such that both [] and [] are uncountable. Inductively
we obtain a binary tree of consistent formulas (cf. proof of Theorem 16.5.2).
Exercise 15.2
To ease notation we replace the partition by a function : [A]
n
1, . . . , k.
Fix a non-principal ultralter | on A, i.e. an ultralter which does not contain
any nite set. For each s [A]
n1
chose a c(s) such that a A [ (s a) =
c(s) belongs to |. Construct a sequence a
0
, a
1
, . . . of distinct elements such
that (s a
n
) = c(s) for any s [a
0
, . . . , a
n1
]
n1
. Apply induction to c
restricted to [a
0
, a
1
, . . .]
n1
.
Exercise 16.1
Let T dene a linear ordering of the universe. By Exercise 33.8 there is a linear
ordering J of bigger cardinality than which has a dense subset of cardinality
. A compactness argument shows that T has a model with a subset B which
is order-isomorphic to J. Let A be a dense subset of B of cardinality . Then
all elements of B have dierent types over A and so [ S(A)[ [B[ > .
Exercise 17.1
a b: Let A a countable subset of the model M. That A has a prime extension
means that T
A
= Th(M
A
) has a prime model. Now apply Theorem 14.7.
b c: Assume that A is contained in the model M and that the isolated
types are not dense over A. So there is a consistent L(A)formula , which
does not contain a complete L(A)formula. Add a predicate P for the set A
and consider the L Pstructure (M, A). Choose a countable elementary
substructure (M
0
, A
0
) which contains the parameters of . Then does not
contain a complete L(A
0
)formula.
c a: Like the proof of 17.3.
Exercise 19.4
Clearly, T(q) is complete -stable if and only if T is. It is also clear that if T(q)
has a Vaughtian pair then so does T. For the converse use a construction as in
Theorem 19.2 to nd a Vaughtian pair M N such that q is realized in N.
Exercise 19.6
Hint: Use Exercise 19.5.
Exercise 20.1
If p S(A) is not algebraic, then all n-types
q(x
1
, . . . x
n
) = x
i
, i = 1, . . . n, satises p and the x
i
are consistent and hence realised in M.
Exercise 20.2
If c
i
acl(Ac
n
) for some i < n, we realise tp(c
0
. . . c
n
/A) such that the c
0
. . . c
n1
is disjoint from acl(AB). Then c
n
cannot belong to B. If c
i
, acl(Ac
n
) for all
i < n, we realise the tp(c
n
/A) and then tp(c
0
. . . c
n1
/Ac
n
), both disjoint from
B.
Exercise 23.3
Choose special models A
i
of T
i
of the same cardinality.
Exercise 23.4
If does not exist, the set T
of all Lsentences such that

1
or

2
is consistent. Choose a complete Ltheory T which contains T
and
apply Exercise 23.3 to T
1
=
1
T and T
2
=
2
T
.
Exercise 23.5
Let [A[ < and p = p(x
i
)
i<
) be a type over p. Denote by p
the restriction
of p to the variables (x
i
)
i<
. Construct a realisation (a
i
)
i<
of p inductively: If
(a
i
)
i<
realises p
, choose a
as a realisation of p
+1
((a
i
)
i<
, x
).
Exercise 23.10
If B dcl(A), every formula with parameters in A is equivalent to a formula
with parameters in A. So every type over A axiomatises a type over B. This
proves 12. For the converse show that b dcl(A) if tp(b/A) has a unique
extension to Ab.
Exercise 23.11
If (a, b) is a formula witnessing b dcl(a), let D equal the set of elements
x for which there is a unique y with (x, y) and let f : D E denote the
corresponding map, which we may assume to be surjective. If also a dcl(b)
witnessed by (y, x) and function g : E
1
D
1
, then we get a 0-denable bijec-
tion (x, y) E D
1
[f(x) = y and g(y) = x.
Exercise 23.12
Use Exercise 9.2.
Exercise 23.13
Use Exercise 23.12 and compactness.
Exercise 23.15
By induction on n. We distinguish two cases. First assume that for some i < 0,
H
n
H
i
has nite index in H
n
. We can then cover every coset of H
n
by nitely
many cosets of H
i
. Since G is not a nite union of cosets of H
0
, . . . , H
n1
, we
are done. Now assume that all H
i
= H
n
H
i
have innite index in H
n
. Assume
for contradiction that G is a nite union of cosets of the H
0
, . . . , H
n
. Since H
n
has innite index in G, there must be a coset of H
n
which is covered by a nite
number of cosets of H
0
, . . . , H
n1
. This implies that H
n
is a nite union of
cosets of the H
i
, i < n, which is imposible by induction.
Exercise 23.16
To see that Exercise 23.15 implies Exercise 23.14 consider the subgroups H
i
:=
G
ci
, i n which have innite index in G. The nitely many cosets a
j
H
i
with
a
j
(c
i
) B, i n do not cover G, so there is some g G such that for all i n
we have g(c
i
) / B.
For the converse, let H
1
, . . . H
n
G be subgroups of innite index, and
consider the action of G on the disjoint union of the G/H
i
by left translation.
By Exercise 23.14, for any a
1
, . . . a
n
G there is some g G such that for all
i n we have g(1 H
i
) / a
i
H
i
, proving Exercise 23.15.
Exercise 24.1
Let I be that set of all i for which
i
has rank . The hypothesis implies that
all kelement subsets of I contain two indices i, j such that
i
,

j
.
So [I[ (k 1) MD.
Exercise 24.2
Let T be stable and A countable. Choose elements b
i
and c
ij
with S
1
(A) =
tp(b
i
/A) [ i S
1
(Ab
i
) = tp(c
ij
/Ab
i
) [ j . Then S
2
(A) = tp(b
i
c
ij
/A) [
j is countable. Similarly, one can show that all the S
n
(A) are countable.
If there is a binary tree of formulas in n variables, there is a countable sub-
language L
0
L for which the countable theory T
0
= [ L
0
sentence, T
= T L
0
is not stable for ntypes. Hence T
0
is not stable and T
0
and
T are not totally transcendental.
Exercise 24.4
Let G
0
be the intersection of all denable subgroups of nite index. G
0
is
denable by Remark 24.7. If N is a nite subgoup which is normalised by G,
the centraliser of N in G is a denable group of nite index in G
0
.
Exercise 24.5
Let M be an saturated model and let p be a type over M. Let (x, m) p
be of minimal rank and degree n. If n > 1, there is a formula (x, b) such
that (x, m) (x, b) and (x, m) (x, b) both have Morley rank . Choose
a M with tp(a/m) = tp(b/m). Then both formulas (x, m) (x, a) and
(x, m) (x, a) have rank and one of these formulas belongs to p, contra-
dicting the maximal choice of (x, m).
Exercise 24.6
Assume that there is a formula (x,
b) of rank [T[
+
. Construct a binary
tree of formulas
1
,
s
(x, y
s
), (s
<
2), below (x,
b) so that for all k and all

< [T[
+
there are parameters a
s
such that MR
s
(x, a
s
) for all s with
[s[ = k. Conclude that MR(x,
b) = ,
Exercise 24.8
Let a be in acl(A) and a
1
, . . . , a
n
the conjugates of a over A. Then (x, a) and
(x, a
1
) . . . (x, a
n
) have the same Morley rank.
Exercise 26.1
In a pregeometry a nite set A is independent from B over C if and only if
dim(A/BC) = dim(A/C). Now use Theorem 26.2.
Exercise 26.2
Assume that abC is independent from B and apply Remark 26.9.
Exercise 27.2
If p forks over A, there is a (x, m) p which implies a conjunction
_
l<d
l
(x, b)
of formulas which fork over A. Choose a tuple b
in M which realises the type

of b over Am. The formulas
l
(x, b
) fork over A and one of them belongs to p.

Exercise 27.3
Let q be Ainvariant. Use Lemma 27.4 to show that q does not divide over A:
If (x, b) belongs to q, then all the (x, b
i
) also belong to q. That q does not
fork over A follows from Exercise 27.2.
Exercise 27.4
p containes a formula which divides over A. So there are many
i

Aut(M/A) such that the system of all
i
() is kinconsistent. This implies
that many of the
i
(p) must be distinct.
Exercise 27.5
p(x) forks over the empty set since it implies the disjunction of cyc(0, x, 3) and
cyc(2, x, 1).
Exercise 27.6
This is just a variant of Proposition 27.6. Let 1 = (b
i
[ i < ) a sequence of
A-indiscernibles containing b such that ((xb
i
) [ i < ) is kinconsistent. Since
tp(a/Ab) does not divide over A, we can assume that 1 is indiscernible over Aa.
Exercise 28.1
Let (p
)
|T|
+ be a chain of types with p
S(A
). Their union

|T|
+ p
does not fork over a subset A
of

|T|
+ A
of cardinality at most [T[. This

follows from Proposition 28.5 (Local Character) and Proposition 28.15. Since
A
for some suciently large [T[

+
, from that index on the chain does
not fork anymore.
1
in the sense of Denition 14.8
Note that this property is just a reformulation of Local Character for =
[T[
+
.
Exercise 28.2
Let q = tp(b/B) and r = tp(c/C). Find an Aautomorphism, which maps c to
b and C to C
such that B [
Ab
C
. Then r
= tp(b/C
) and s = tp(b/BC
) are
as required: we have B [
A
b, which together with B [
Ab
C
yields B [
A
C
b
from which b [
C
B.
Exercise 28.3
By monotony and symmetry it suces to show that a
X
and a
Y \X
are inde-
pendent over A. So we can assume that X and Y are disjoint and, by nite
character, that X and Y are nite. We proceed by induction on [X Y [. Let
z be the maximum of X Y . By symmetry we may assume that z Y . Then
a
z
is independent from a
XY \{z}
. The claim follows now from the induction
hypothesis and transitivity.
Exercise 28.4
This follows from: If (b
i
[ i < ) is indiscernible over A, there is an Aconjugate
B
of B such that (b
i
[ i < ) is indiscernible over B
.
Exercise 28.5
Up to symmetry this is only a reformulation of Monotony and Transitivity.
Exercise 28.6
For ease of notation we restrict to the following special case:
If b
1
b
2
is independent from C, then we have
b
1
[
b
2
b
1
[
C
b
2
.
Proof: By Corollary 28.18 both sides of this equivalence are equivalent to the
triple (b
1
, b
2
, C) being independent. (For the direction from right to left it
suces in fact to assume that b
1
and b
2
are independent from C individually.)
Exercise 28.7
The hypothesis implies a [
AB
C. Now the claim follows from Remark 27.3.
Exercise 29.1
Let 1 be the sequence b = b
0
, b
1
. . .. Use Proposition 29.6 and induction on n
to show that

(x, b
i
) [ i < 2
n
does not fork over A. So we nd a realisation
of

(x, b
i
) [ i < which is independent from 1 over A. That one can nd
c in such a way that 1 is indiscernible over Ac follows from Lemma 15.3 and
Finite Character.
Exercise 29.2
1. Let (c
i
[ i < ) be an antichain for
1
2
. Then by Ramseys theorem there
is an innite A such that (c
i
[ i A) is an antichain for
1
or for
2
.
2. If
(x, y) is not thick, it has, by compactness, antichains (c

i
[ i I) indexed
by arbitrary linear orders I. If I
is the inverse order, (c

i
[ i I
) is an
antichain for .
3. Consider
.
Exercise 29.3
Choose an Aconjugate c of b, dierent from b, and set B = b and C = c.
Exercise 29.5
Show that a [
0
A
A with Existence and use Monotony and Transitivity.
Exercise 29.6
Let p S(M) with two dierent nonforking extensions to B M. Let
B
0
, B
1
, . . . be an Mindependent family of conjugates of B. Then on each
B
i
there are two dierent extensions q
0
i
, q
i
of p. Now by the independence theo-
rem for any function : 2 there is a nonforking extension q
of p to
i<
B
i
,
which extends every each q
(i)
i
.
Exercise 30.1
If is thick and dened over A, the conjunction of the Aconjugates of is
thick and dened over A.
Exercise 30.2
Extend B and C to models M
B
and M
C
such that M
B
[
A
M
C
, b [
A
M
B
and
c [
A
M
C
. Now it suces to nd a d such d [
A
M
B
M
C
, tp(d/M
B
) = tp(b/M
B
)
and tp(d/M
C
) = tp(c/M
C
).
Exercise 30.3
We have to show that for every thick (x, y) the formula (x, a) does not divide
over A. So let a = a
0
, a
1
, . . . be indiscernible over A. Then [= (a
i
, a
j
) for all
i, j. This shows that (x, a
i
) [ i < is nitely satisable. If T is simple and
B is any set, chose a
independent from B over A such that nc

A
(a
, a). Finally
choose a
such that tp(a
/Aa) = tp(a
/Aa) and a
[
Aa
B.
Exercise 30.4
If 1 is an innite sequence of indiscernibles over A, 1 is indiscernible over some
model which contains A.
Exercise 30.5
Use the Erds-Rado-Theorem C10.2.
Exercise 30.6
Let R be bounded and Ainvariant and a
0
, a
1
, . . . indiscernible over A. Show
that R(a
i
, a
j
) for all i < j.
Exercise 30.7
This follows from Exercise 30.6.c
Exercise 32.1
Consider a Morley sequence of a global coheir extension of tp(a/M) = tp(b/M)
over M
Exercise 32.2
We may assume that has Morley degree 1. Then the MR( ) = is
smaller than the Morley rank of . Choose a family of Mdenable formulas
i
of rank which dene disjoint subsets of (C). The for some i the rank of
i
must be smaller than . is realised in M by induction on the
rank of .
The second part follows from the rst.
Exercise 32.3
If b
is another realisation of q(y) B and a
realizes p(x) Bb
there is a B
automorphism taking b
to b. If p(x) is Ainvariant, (a
) realizes tp(a/Bb)
and so p(x) q(y) is well-dened. The same proof shows p(x) q(y) to be
Ainvariant if p, q both are A-invariant.
Exercise 33.1
If (x, y) has the order property witnessed by a
0
, a
1
, . . . and b
0
, b
1
, . . ., then the
sequence a
0
b
0
, a
1
b
1
, . . . is ordered by the formula
(xy, x
) = (x, y
). The
converse is obvious.
Exercise 33.3
The formula xRy has the binary tree property.
Exercise 33.5
If (x, y) has SOP, the formula (x, y
1
, y
2
) = (y
1
, x) (y
2
, x) has the tree
property with respect to k = 2.
Exercise 33.6
Hint: if T is unstable, there are a formula (x, y) and indiscernibles (a
i
b
i
[ i Q)
with [= (a
i
, b
j
) i < j. If does not have the independence property,
there are nite disjoint subsets J, K of Q such that
J,K
(y) = (a
i
, y) [ i
J (a
i
, y) [ i K is inconsistent. Not all of J can be less than all
elements of K. Choose J and K minimising the number of inversions F =
(j, k) J K [ k < j. Choose (j, k) F so that the interval (k, j) does not
contain any elements of J K. Write J = J
0
j and K = K
0
k. Then
J0{k},K0{j}
(y) is consistent and the formula (with parameters)
J0,K0
(y) (x, y)
has the strict order property.
Exercise 33.7
a) b): A type p S(A) is determined by the family of all p
, the parts of
p.
Hence
[S(A)[
[S
(A)[
[A[ = [A[
|T|
.
So if [A[ = and
|T|
= , then [S(A)[ .
b) c): Clear.
c a): Follows directly from 33.3, a b.
Exercise 33.8
See the proof of 33.3, a) d). Let A =
_
0,
1
2
, 1
_
. We embed I into

A by
extending sequences from I
0
to a sequence of length with the constant value
1
2
.
We order I by the ordering induced from the lexicographic ordering of

A.
Exercise 33.9
Clearly, S
(B) = S
(B). If has the order property witnessed by

a
i
, b
j
, i, j , use Ramseys Theorem to nd an innite subset J such
that a
i
, b
j
, i, j J witness the order property for either of . The same
argument works for .
Exercise 33.10
1. R
= implies as in Theorem 24.6 that there is a binary tree of consistent

formulas of the form , . Now we follow the proof of Theorem 16.5 and
conclude rst that over some countable A there are incountable many types
which contain . This implies then that (x, y) has the binary tree property.
Again by the proof of 24.6 this implies that R
= .
2. If < R
there is a formula (x, a) such that (x) (x, a) and (x)

(x, a) have at least rank . So if R
, there would have binary trees

of arbitrary nite height and had the binary tree property.
Exercise 33.11
Assume that the tree property of is witnessed by the parameters A = a
s
[
s
<
. If has the tree property with respect to k = 2, it is easy to see
that has the binary tree property.
For the general case we make use of Exercise 33.10: if is stable, all
ranks are less than . It follows that there is a sequence

such that
the ranks of the formulas
_
1i<n
(x, a
i
) are strictly decreasing, which is
impossible.
Exercise 34.1
Let T the theory of all equivalence relations with three, innite classes. There
is only one 1type p over the empty set. p has no good denition.
Exercise 34.2
Six of the eight possible cases are realised by 1types. For the cases (D, C, I,
H), (D, C, I, H) use 2types and for the case (D, C, I, H) a 3type.
Exercise 34.4
Consider the Boolean algebra of all Mdenable subsets of (M)
n
and the
subalgebra of (M)denable subsets. The two algebras coincide if and only if
they have the same Stone spaces (see p. 61). For the second part note that if
(C) has a least two elements then for every (x, y) there is a formula (x, z)
such that every class (x, b) which is a subclass of (C)
n
has the form (C, c)
for some c (C).
Exercise 34.5
1. Prove that formulas with Morley rank are stable. The proof that totally
transcendental theories are stable on page 153 is similar.
2. It is enough to nd d
p
x(x, y) for stable (x, y). Use the proof of Theo-
rem 34.1.
Exercise 34.6
The proof follows the pattern of the proof of the ErdsMakkaiTheorem, 9.1.
Assume that B
= b B [ (x, b) p is not a positive Boolean combination

of sets of the form b B : [= (c, b), c C. Construct three sequences
(b
i
[ i < n) in B
, (b
i
[ i < n) in B B
and (c
i
[ i < n) in C such that for all
i < n
[= (c
i
, b
n
) [= (c
i
, b
n
)
and for all i n
[= (c
n
, b
i
) and [= (c
n
, b
i
).
Exercise 34.7
Hint: If q is a weak heir of p, then D
(q) = D
(p) where D
(p) is dened as the

minimum of D
() for p. The argument in Theorem 34.1 now shows that q

is denable over M.
Exercise 34.8
Let p be the global extension of tp(a/M) which is denable over M. By
Lemma 32.5 tp(a/Mb) is an heir of tp(a/M) if and only if tp(a/Mb) q,
i.e. if and only if (x, b) tp(a/Mb) [= d
p
x(x, b). So if q is the global M
denable extension of tp(b/M), we have that tp(a/Mb) is an heir of tp(a/M)
if and only if (x, b) tp(a/Mb) [= d
p
x(x, y) q(y). Lemma 34.4 implies
now that tp(a/Mb) is a heir of tp(a/M) if and only if tp(b/Ma) is an heir of
tp(b/M).
Exercise 35.1
D is denable from some some tuple d D. Each such d is a canonical parameter
of D.
Exercise 35.2
Let e be an imaginary and A the smallest algebraically closed set in the home
sort over which e is denable. Then e is denable from a nite tuple a A.
Since every automorphism which xes e leaves A invariant, all elements of A
are algebraic over e.
For the converse let a acl(e) be a real tuple over which e is denable. Then
A = acl(a) is the smallest algebraically closed set over which e is denable.
Exercise 35.3
T
and T
DLO
have the following property: If A, B are nite sets and if the
tupels a and b have the same type over A B, then there is a sequence a =
a
0
, b
0
, . . . , a
n
, b
n
= b such that a
i
and b
i
have the same type over A and b
i
and
a
i+1
have the same type over B. This implies that for every denable class D
there is a smallest set over which D is denable.
T
does not eliminate imaginaries since no nite set with at least two ele-
ments has a canonical parameter.
Exercise 35.4
Let q
1
and q
2
be extensions of p to B. Choose realisations a
1
, a
2
of q
1
, q
2
,
respectively. There is an Aut(C/A) taking a
1
to a
2
. Since (B) = B, we
have (q
1
) = q
2
.
Exercise 35.5
Let p S(A) be algebraic. If p is realised by b dcl(B) d
p
x(x, y) = (b, y)
is a good denition of p over B. Conversely, if q is an extension of p to M,
then q is realised by some b in M and d
q
x(x
.
= y) denes the set b. So if q is
denable over B, then b dcl(B).
Exercise 35.6
Let d ba a canonical parameter of D = (C, d). If d
has the same type as d, we

have
(C, d
) = D d
= d.
By compactness this is true for all d
which satisy some (x) tp(d). Consider

the L Pformula (x, P) = (x) y ((y, x) P(y)).
Exercise 35.7
1. Let (x, a) p have the same Morley rank as p and be of degree 1. Then
d
x
(x, y) is a canonical base of p.
2. Use 1. and Exercise 35.1.
Exercise 35.9
If stp(a/A) ,= stp(b/A) there is an acl(A)-denable class D = (x, a) with
(a, a) and ,= (b, a). By Lemma 35.3, D is the union of equivalence classes of
an A-denable nite equivalence relation E
a
, proving the claim.
Exercise 36.1
Let P be the set of all strong types over A which are consistent with p, and Q
be the set of all strong types which are consistent with q. P and Q are closed
subsets of S(acl
eq
(A)) and disjoint since strong types are stationary. So they
can be separated by a formula (x) over acl
eq
(A). By Lemma 35.3 (C) is a
union of classes of a nite Adenable equivalence relation E(x, y).
Exercise 36.3
Use the second part of Exercise 34.5. The rst claim can now be proved like
Theorem 36.1.
The second claim is proved like Corollary 36.3, but we must be more careful
and use the rank introduced in Exercise 33.10. We call the the minimal
rank of a formula in a type p the rank of p. Let A = acl
eq
(A), and p S(A)
a stable type and p
and p
two nonforking global extensions. p
and p
are
denable over A. We want to show that (x, b) p
(x, b) p
. We
may assume that (x, y) is stable (containing parameters from A). Let q(y) be
a global extension of tp(b/A) which has the same
rank, where (x, y)
=
(y, x). Since there are only nitely many possibilites for the part of q, the
part is denable over A. Now the claim follows from an adapted version of
34.4.
We have still to show that p has a nonforking global extension, i.e. an exten-
sion which is denable over A. Choose for every stable (x, y) a global extension
of p whith the same rank. By the above the part p
of this extension is
denable over A and as such is uniquely determined. It remains to show that
the union of all p
is consistent. Consider a nite sequence

1
(x, y), . . . ,
n
(x, y)
of stable formulas. Choose a stable formula (x, y, z) such that every instance
i
(x, b) has the form (x, b, c) for some choice of c. Then p
contains all p
i
for
i = 1, . . . , n.
Exercise 36.4
1. Argue as in the second part of the proof of Theorem 36.10. Replace p q
by MR(p) = MR(q).
2. By the rst part of Exercise 34.5 p is stable Let q be global extension of p.
If MR(p) = MR(q), then q has only nitely many conjugates over A. Since q is
denable, this implies that q is denable over acl
eq
(A). So q does not fork over A
by Exercise 36.3. Now assume that q does not fork over A. Using Exercise 24.8
we see that we can assume that A = acl
eq
(A). Let q
be an extension of p with
the same Morley rank. Then q
does not fork over A. So be Exercise 36.3 q = q
.
Exercise 36.5
Choose A
1
A of cardinality at most [T[ over which p does not fork. Let (p
i
)
be the nonforking extensions of p A
1
to A. For each Lformula (y, y) there
are only nitely many dierent p
i
. Hence there is a nite subset A
of A such
that for all i
(p
i
A
= (p A
= p
i
= p
.
Now put A
0
= A
1

.
If p has Morley rank, choose p having the same Morley rank and degree
as p. Any set A
0
containing the parameters of does the job.
Exercise 36.6
1: Easy.
2: Exercise 36.5 shows that it is enough to consider types p over a countable set
A. The multiplicity is the number of extensions of p to acl(A). These extensions
form a separable compact space. By Exercise 35.4, either all or none of them are
isolated. In the rst case the space is nite, in the other case it has cardinality
2
0
.
3: Use Exercise 36.4.
Exercise 36.7
1) a and a b are interalgebraic over b. This implies that MR(a) = MR(a/b) =
MR(a b/b) MR(a b). If MR(a) = MR(a b), we have MR(a b/b) = MR(a b)
and a b and b are independent.
2) Let b be independent from a. If a is generic, MR(a b) cannot be bigger than
MR(a), so a b and b are independent by part 1). For the converse we choose
b generic. Part 1) (with sides reversed) implies that a b is also generic. If a b
and b are independent, it follows that MR(a) = MR(a b) and a is generic.
Exercise 36.8
For 1) notice that each line A
i
consists of two elements and their product.
For 2) note that by Exercise 36.7, if a, b, c are independent generics, then a, b, bc
are again independent generics. If one applies this rule repeatedly starting with
a
1
, a
2
, a
3
, one obtains every noncollinear triple of our diagram.
Exercise 37.1
It follows from Remark 27.3 and Symmetry that a type is algebraic if and only if
it has no forking extensions. A type has SUrank 1 if and only if if the algebraic
and the forking extensions coincide. So a type is minimial if and only if it has
SUrank 1 and has only one nonforking extension to every set of parameters.
Exercise 37.2
Use Exercise 27.4.
Exercise 37.3
This follows from Exercise 27.1.
Exercise 37.4
The rst claim follows from Lemma 28.4(2). and the remark thereafter. The
second claim is easily proved using the Diamond Lemma (Exercise 28.2) and
Exercise 27.6.
Exercise 37.5
Totally transcendental theories are superstable by Corollary 36.11. It follows
also that the multiplicity of a type over arbitrary sets is nite, namely equal to
its Morley degree. If T is superstable, one can compute an upper bound for the
number of types over a set A of cardinality as in the proof of Theorem 37.5 (2).
If T is small, there are only countably many types over a nite set E. If we know
also that all p S(E) have nite multiplicity, we have [ S(A)[
0

0
= .
Exercise 37.7
We can assume that all E
i
are 0denable. Choose a sequence a
0
, a
1
, . . . such
that [= E
i
(a
i
, a
i+1
) and a
i
/E
i
is not algebraic over a
0
. . . a
i1
. Let b be an
element in the intersection of all a
i
/E
i
, A = a
0
, a
1
, . . . and p = tp(b/A). Then
for all i we have b ,[
a0,...,ai1
a
i
/E
i
by Remark 27.3. This shows that p forks
over each nite subset of A.
Exercise 37.8
Dene E
i
(x, y) as xy
1
G
i
. For any imperfect eld K of nite characteristik
p set G
i
= K
p
i
. x y K
p
i+1
.
Exercise 37.9
Half of the claim follows from Remark 24.7 and Exercise 37.8. Assume that M
has the dcc on pp-denable subgroups. Then for every element a and every set B
of parameters the positive type tp
+
(a/B) contains a smallest element
0
(x, b).
So there are at most max([T[, [A[) many types over A. This shows that M R
0
is stable for every countable subring, so M is totally transcendental (see Ex-
ercise 16.4). Now assume that there is no innite sequence of pp-subgroups
with innite index in each other. Then tp
+
(a/B) contains a formula
0
(x, b
0
)
such that tp
+
(a/B) is axiomatised by formulas (x, b) where (M, 0) is a sub-
group of nite index in
0
(M, 0). There are max([T[, [A[) many possibilities for
0
(x, b
0
) and for each (x, y) nitely many possibilites. So the number of types
over A is bounded by max(2
|T|
, [A[) and M must be superstable. Indeed, the
proof of Theorem 37.5.3 shows that otherwise for every there would be a set
A of cardinality with [ S(A)[
0
.
Exercise 38.1
Let p S(A) and q a nonforking extension to B. Let 1 be a Morley sequence
of q so 1 is independent over B. Since every element of 1 is independent from
B over A, 1 is independent over A as well (cp. Exercise 28.6), hence a Morley
sequence of p.
Exercise 38.2
a): 1 1
0
is independent from B over A, hence independent over B. The
elements of 1 realise the nonforking extension of p to B.
b): Let B A and q the nonforking extension of p. We extend 1 to a very
long sequence 1
indiscernible over A. Then 1
is still a Morley sequence of p.

So if we choose 1
0
1
with [1
0
[ [T[ +[B[ and B [
AI0
1
, then 1
1
0
is an
innite Morley sequence of q having the same average type as 1, so q Av(1).
Exercise 38.3
If 1
0
and 1
1
are parallel, hence 1
0
and 1
1
indiscernible, then
Av(1
0
) = Av(1
0
) = Av(1
1
) = Av(I
1
).
If conversely p = Av(1
0
) = Av(1
1
), note that by the proof of Theorem 38.2
there are sets B
0
and B
1
over which p does not fork and such that 1
0
and 1
1
are Morley sequences of the stationary types p B
0
and p B
1
. Let be a
Morley sequence of p B
0
1
0
B
1
1
1
. Then 1
0
and 1
1
are Morley sequences of
p B
0
and p B
1
, respectively.
Exercise 38.4
Let p and q be stationary types with innite Morley sequences 1 and . Then
the average types Av(1) and Av() are the global nonforking extensions of p
and q, respectively. Now p and q are parallel if and only if Av 1 = Av , i.e., if
and only if 1 and are parallel.
Exercise 39.1
Let p
1
,. . . ,p
n
be the extensions of p to acl
eq
(A) and let 1
i
be the elements of 1
which realise p
i
.
Exercise 41.1
If d is in dcl
eq
(F) the set d is denable over F. The proof of Theorem 35.2
shows that d has a canonical parameter in F
eq
.
Exercise 41.3
a b was implicitly used in 41.4.
b a: this is the proof of 34.3.
b c: this is Exercise 41.2.
b d: same as the proof of 41.4
d a: Assume that (a, F) is not denable with parameters from F. Let a
i
be
an enumeration of C. Construct a sequence
i
of partial automorphisms of F so
that the domain of
i
contains some f with [= (a, f) [= (a
i
,
i
(f)).
Exercise 42.1
Induction on MRA.
Exercise 44.4
This follows because any path (x
1
, a, x
2
) is strong in M
.
Appendix
Exercise 6.1
Let L be an elementary extension of K. Show rst that if S is an integral domain
which contains K, then L
K
S is again an integral domain. So L
K
K
alg
is
an integral domain.
Exercise 8.1
1. Symmetry is clear from the denition. For the other properties show rst
that a nite set A is independent from B over C i dim(A/BC) = dim(A/C).
This implies Monotony and Transitivity. Finite Character and Local
Character follow from the fact that for nite A and any D there is a nite
D
0
D such that dim(A/D) = dim(A/D
0
).
2. Symmetry, Finite Character and Weak Monotony are clear. If A
is nite and D is an set, let D
0
be a basis of cl(A) D. D
0
is nite and
A [
0
D0
D. This shows Local Character. A [
cl
C
B always implies A [
0
C
B.
If [
0
satises Monotony, the converse is true: We may assume that B =
b
1
, . . . , b
n
is nite. Then A [
0
C
B implies A [
0
Cb1...bi
b
i+1
for all i. But this
is the same as A [
cl
Cb1...bi
b
i+1
, from which follows that A [
cl
C
B.
Exercise 8.3
Choose a, b, x, c K pindependent. Set F
0
= K
p
(c), F
1
= K
p
(c, a, b) and
F
2
= K
p
(c, x, ax+b). Then F
0
has pdimension 1, F
1
and F
2
have pdimension 3
and F
1
F
2
has pdimension 4. To show that F
0
= F
1
F
2
, prove that dim
F0
F
1
=
dim
F0
F
2
= p
2
and dim
F0
(F
1
+F
2
) = 2p
2
1.
Exercise 8.5
First show that in any pregeometry for any closed A B: If dim(A/B) is nite,
then it is the longest length n of a proper chain B = C
0
C
1
. . . C
n
= B
of closed sets C
i
.
Bibliography
[1] J. T. Baldwin and A. H. Lachlan. On strongly minimal sets. J. Symbolic
Logic, 36:7996, 1971.
[2] John T. Baldwin.
T
is nite for
1
categorical T. Trans. Amer. Math.
Soc., 181:3751, 1973.
[3] John T. Baldwin. Fundamentals of Stability Theory. Perspectives in Math-
ematical Logic. Springer Verlag; Berlin, Heidelberg, New York, London,
Paris, Tokyo, 1988.
[4] John T. Baldwin. An almost strongly minimal non-Desarguesian projective
plane. Trans. Amer. Math. Soc., 342(2):695711, 1994.
[5] A. Baudisch, A. Martin-Pizarro, and M. Ziegler. Red elds. J. Symbolic
Logic, 72(1):207225, 2007.
[6] Andreas Baudisch. A new uncountably categorical group. Trans. Amer.
Math. Soc., 348(10):38893940, 1996.
[7] Andreas Baudisch, Martin Hils, Amador Martin Pizarro, and Frank O.
Wagner. Die bse farbe. Journal de lInstitut de Mathmatiques de Jussieu.
[8] Paul Bernays. Axiomatic set theory. Dover Publications Inc., New York,
1991. With a historical introduction by Abraham A. Fraenkel, Reprint of
the 1968 edition.
[9] N. Bourbaki XI Algbre, chapitre 5. Corps commutatifs. Hermann, Paris,
1959.
[10] Elisabeth Bouscaren. The group congurationafter E. Hrushovski. In
The model theory of groups (Notre Dame, IN, 19851987), volume 11 of
Notre Dame Math. Lectures, pages 199209. Univ. Notre Dame Press, Notre
Dame, IN, 1989.
[11] Steven Buechler. Essential stability theory. Perspectives in Mathematical
Logic. Springer-Verlag, Berlin, 1996.
[12] Enrique Casanovas. Simplicity.
258
BIBLIOGRAPHY 259
[13] Zo Chatzidakis and Ehud Hrushovski. Model theory of dierence elds.
Trans. Amer. Math. Soc., 351(8):29973071, 1999.
[14] M. M. Erimbetov. Complete theories with 1-cardinal formulas. Algebra i
Logika, 14(3):245257, 368, 1975.
[15] Ju. L. Erov. Fields with a solvable theory. Dokl. Akad. Nauk SSSR,
174:1920, 1967. English transl., Soviet Math. Dokl., 8:575-576, 1967.
[16] Ulrich Felgner. Comparison of the axioms of local and universal choice.
Fund. Math., 71(1):4362. (errata insert), 1971.
[17] Steven Givant and Paul Halmos. Introduction to Boolean algebras. Under-
graduate Texts in Mathematics. Springer, New York, 2009.
[18] Victor Harnik. On the existence of saturated models of stable theories.
Proc. Amer. Math. Soc., 52:361367, 1975.
[19] Wilfrid Hodges. Model theory, volume 42 of Encyclopedia of Mathematics
and its Applications. Cambridge University Press, Cambridge, 1993.
[20] Wilfried Hodges. Model Theory. Encyclopedia of Mathematics and its
Applications. Cambridge University Press, 1993.
[21] Wilfried Hodges. A Shorter Model Theory. Cambridge University Press,
1997.
[22] Ehud Hrushovski. A stable
0
-categorical pseudoplane. Preprint, 1988.
[23] Ehud Hrushovski. Unidimensional theories are superstable. Ann. Pure
Appl. Logic, 50:117138, 1990.
[24] Ehud Hrushovski. A new strongly minimal set. Ann. Pure Appl. Logic,
62(2):147166, 1993. Stability in model theory, III (Trento, 1991).
[25] Ehud Hrushovski. A non-PAC eld whose maximal purely inseparable
extension is PAC. Israel J. Math., 85(1-3):199202, 1994.
[26] Ehud Hrushovski and Boris Zilber. Zariski geometries. J. Amer. Math.
Soc., 9(1):156, 1996.
[27] Thomas Jech. Set theory. Springer Monographs in Mathematics. Springer-
Verlag, Berlin, 2003. The third millennium edition, revised and expanded.
[28] Klaus Kaiser. ber eine Verallgemeinerung der Robinsonschen Modellver-
vollstndigung. Z. Math. Logik Grundlag. Math., 15, 1969.
[29] Akihiro Kanamori. The higher innite. Springer Monographs in Mathe-
matics. Springer-Verlag, Berlin, second edition, 2003. Large cardinals in
set theory from their beginnings.
BIBLIOGRAPHY 260
[30] Byunghan Kim and Anand Pillay. From stability to simplicity. Bull. Sym-
bolic Logic, 4(1):1736, 1998.
[31] Serge Lang. Algebra. AddisonWesley Publishing Company, second edition,
1984.
[32] Serge Lang and Andr Weil. Number of points of varieties in nite elds.
Amer. J. Math., 76:819827, 1954.
[33] Daniel Lascar. Stability in Model Theory. Longman, New York, 1987.
[34] Angus Macintyre. On
1
-categorical theories of elds. Fund. Math.,
71(1):125. (errata insert), 1971.
[35] David Marker. Model theory, volume 217 of Graduate Texts in Mathematics.
Springer-Verlag, New York, 2002. An introduction.
[36] M. Morley. Categoricity in power. Trans. Amer. Math. Soc., 114:514538,
1965.
[37] Anand Pillay. An introduction to stability theory, volume 8 of Oxford Logic
Guides. The Clarendon Press Oxford University Press, New York, 1983.
[38] Anand Pillay. The geometry of forking and groups of nite Morley rank.
J. Symbolic Logic, 1995.
[39] Anand Pillay. Geometric stability theory, volume 32 of Oxford Logic Guides.
The Clarendon Press Oxford University Press, New York, 1996. Oxford
Science Publications.
[40] Bruno Poizat. Cours de Thorie des Modles. Nur Al-Mantiq Wal-Marifah,
Villeurbanne, 1985.
[41] Bruno Poizat. Groupes Stables. Nur Al-Mantiq Wal-Marifah, Villeurbanne,
1987.
[42] Mike Prest. Model theory and modules, volume 130 of London Mathematical
Society Lecture Note Series. Cambridge University Press, Cambridge, 1988.
[43] V. A. Puninskaya. Vaughts conjecture. J. Math. Sci. (New York),
109(3):16491668, 2002. Algebra, 16.
[44] Gerald E. Sacks. Saturated model theory. W. A. Benjamin, Inc., Reading,
Mass., 1972. Mathematics Lecture Note Series.
[45] Igor R. Shafarevich. Basic algebraic geometry. 1. Springer-Verlag, Berlin,
second edition, 1994. Varieties in projective space, Translated from the
1988 Russian edition and with notes by Miles Reid.
[46] Saharon Shelah. Uniqueness and characterization of prime models over sets
for totally transcendental rst-order-theories. J. Symbolic Logic, 37:107
113, 1972.
BIBLIOGRAPHY 261
[47] Saharon Shelah. Classication Theory. North Holland, Amsterdam, 1978.
[48] Saharon Shelah. On uniqueness of prime models. J. Symbolic Logic, 43:215
220, 1979.
[49] Saharon Shelah. Simple unstable theories. Ann. Math. Logic, 19(3):177
203, 1980.
[50] Joseph R. Shoeneld. Mathematical logic. Association for Symbolic Logic,
Urbana, IL, 2001. Reprint of the 1973 second printing.
[51] K. Tent. Very homogeneous generalized n-gons of nite Morley rank. J.
London Math. Soc. (2), 62(1):115, 2000.
[52] Jouko Vnnen. Barwise: abstract model theory and generalized quanti-
ers. Bull. Symbolic Logic, 10(1):3753, 2004.
[53] Frank Wagner. Simple Theories. Kluwer Adacemic Publishers, Dordrecht,
2000.
[54] Frank O. Wagner. Stable groups, volume 240 of London Mathematical So-
ciety Lecture Note Series. Cambridge University Press, Cambridge, 1997.
[55] John S. Wilson. Pronite groups, volume 19 of London Mathematical So-
ciety Monographs. New Series. The Clarendon Press Oxford University
Press, New York, 1998.
[56] Martin Ziegler. Model theory of modules. Ann. Pure Appl. Logic,
26(2):149213, 1984.
[57] Boris Zilber. Analytic and pseudo-analytic structures. In Logic Colloquium
2000, volume 19 of Lect. Notes Log., pages 392408. Assoc. Symbol. Logic,
Urbana, IL, 2005.
Index
0, 213
0denable set, 16
1, 213
2, 213
minimal formula, 114
strongly minimal formula, 112
Av(1), 179
_
i<m
, 17
, 112
, 14
, 14
(A, R), 9
(A, a
1
, . . . , a
n
), 9
A
B, 37
A
B, 38
A
= B, 7
A B, 26
A K, 9
A B, 8
A
B
, 10
formula, 39
[A[, 7
a [
B
C, 167
A
1
B, 45
Abs(L), 145
A [
C
B, 131
acl(A), 94
acl
eq
(A), 160
A B, 23
0
, 211
, 214
+ 1, 214
+n, 214
A [= [
b], 15
[A]
n
, 77
Aut(A), 8
Aut(A/B), 10
b
a
x
, 15
Bdenable set, 16
, 214
Cb(a/B), 196
Cb(p), 168
cf, 215
cl
S
, 233
cl
S
, 233
Diag(A), 18
dim(N/M), 118
dim
(M), 100
D, 159
, 159
dcl(A), 108
dcl
eq
(A), 160
deg(p), 94
dim
(N/M), 117
dim
S
(X/S), 234
dim(X), 234
d
p
x(x, y), 149
D(p, , k), 176
.
=, 14
, 14
, 14
formula, 39
>n
, 92
EM(1/A), 77
B
, 8
, 92
(), 19
(t
1
, . . . , t
n
), 17
(x
1
, . . . , x
n
), 16
f : A
B, 37
[], 60
(A), 16
F
eq
, 189, 190
dimension, 100
262
INDEX 263
over parameters, 100
f(p), 61
type, 152
G(K), 229
G(k), 229
h : A

B, 25
h : A

B, 7
h: A B, 7
, 14
categorical theory, 35
, 210
homogeneous structure, 105
+
, 212
saturated, 82
stable, 81, 152, 154
universal structure, 105
kinconsistent Familie, 125
/saturated, 69
L
AG
, 6
L
alg
, 95
, 55
L(B), 10
L(C), 10
L
eq
, 160
L
G
, 6
L
sets
, 6
L
M(R)
, 49
L
N
, 6
L
, 6
L
O
, 6
L
OR
, 6
L
R
, 6
L
sep
, 55
L
Sk
, 79
Lstp(a/A), 142
M
eq
, 160
MD(a/A), 114
MD(p), 114
MD, 113
MR(T), 111
MR(a/A), 114
MR(p), 114
MR, 111
M
B
, see A
B
MD
, 113
M
eq
, 160
[=, 107
mult(p), 171
N(B/A), 168
, 14
nc
A
, 136, 141
ntype, 31
_
i<m
, 17
, 14
otp(X, <), 213
0dimensional space, 60
, 213
<
A, 214
homogeneous, 65
stable, 81
On, 213
saturated, 64
pbasis, 225, 226
pclosure, 225
p
, 249
rank, 154, 253
p A, 61
P(V ), 233
R
, 154
S
B
, 8
sup
iI

i
, 214
sup
iI

i
, 211
S
A
(B), 31
S
A
n
(B), 31
S(B), 31
S
n
(B), 31
S
(B), 152
S(T), 60
S
n
(T), 60
stp(a/A), 165
SU(p), 173
T
eq
, 160
t(t
1
, . . . , t
n
), 13
t(x
1
, . . . , x
n
), 13
t
A
[
b], 12
t
A
[a
1
, . . . , a
n
], 13
[T[, 22
T
, 38
T
ACF
, 23, 51, 240
T
ACF0
, 52
INDEX 264
T
ACFp
, 52
T
, 40
T
AG
, 21
T
DLO
, 23, 240
T-e.c, 45
T
eq
, 160
T S, 22
Th(A), 22
T
KH
, 46
T
F
, 21
T
M(R)
, 49
tp(a), 73
tp(a), 31
tp(a/B), 31
tp
A
(a/B), 31
T , 22
tp
qf
(a), 72
T S, 22
T
R
, 21
T
RCF
, 52
T
Tree
, 74
T
RG
, 57
, 17
, 17
SCF
p,e
, 54
SCF
p
(c
1
, , c
e
), 54
SCF
p,
, 54
T
Sk
(L), 79
T
, 48
iI
A
i
, 9
U(p), 176
v
0
, v
1
, . . ., 12
y
x, 211
Z, 229
nonforking extension
existence, 131, 139, 167
17. Hilberts Problem, 53
abelian group, 21
absolute Galois group, 229
absolute part of a pseudo-nite eld,
145
absolutely prime, 225
additivity of Morley rank, 123, 192
ane subspace, 236
Aleph function, 214
algebraic
closure, 94
element, 94
formula, 94
type, 94
algebraically closed eld, 35, 51, 52, 94,
97, 150
almost orthogonal type, 191
almost strongly minimal, 124, 205
Amalgamation Property, 69
analysable type, 191
antichain, 136
assignment, 12
atom, 62
atomic
diagram, 18
extension, 87
formula, 14
structure, 73
tuple, 73
automorphism, 8
automorphism group, 8
average type, 179
axiom, 21
1based formula, 195
basic formula, 17
basis, 233, 234
Bernays-Gdel Set theory, 208
Beth function, 214
Beths Interpolation Theorem, 108
BG, 111, 208
binary tree, 75, 82
binary tree property, 152
binding group, 190
binding strength, 15
Boolean algebra, 61
bounded relation, 144
canonical
base, 159, 168, 196
parameter, 159
Cantor-Bendixson rank, 116
cardinal, 210, 213
arithmetic, 211
Ramsey, 239
INDEX 265
regular, 215
successor, 212
weakly compact, 239
weakly inaccessible, 215, 239
cardinality, 210
cardinality of a structure, 7
categoricity, 35
chain
continuous, 32
elementary, 27
Lemma, 27
of structures, 9
Charakterisierungssatz, 139
class
elementary, 23, 47
closed
sets of types, 62
closedness of the set of nonforking ex-
tensions, 127
closure
denable, 108
club, 181
conal, 215
conality, 215
commutative ring, 21
Compactness Theorem, 28
complete
formula, 62
theory, 22
components of a formula, 112
conjugacy, 167
conjugate over, 107
conjunction, 14
conjunctive normal form, 20
conservative extension, 62
consistent
set of formulas, 21, 107
theory, 21
constant, 6
constant term, 14
construction, 182
continuous chain, 32
Continuum Hypothesis, 212
contructible set, 85
countable, 211
theory, 58
cyclical order, 128
denable
bijection, 97
class, 107
closure, 108
set, 16
type, 149
denable bijection, 97
degree
of imperfection, 54, 234
degree of a type, 94
degree of imperfection, 225
density, 74
derivation, 220
descending chain condition (dcc), 113
diagram
atomic, 18
elementary, 25
Diamond-lemma, 134
dierential eld, 220
dierentially
closed eld, 56
dierentially closed eld, 122, 170
dimension
dimension, 100
over parameters, 100
of a pregeometry, 234
of extensions of dierential elds,
122
directed
family, 8
elementary, 27
partial order, 8
disjoint formulas, 75
disjunction, 14
disjunctive normal form, 20
dividing
formula, 125
sequence, 129
set of formulas, 125
domain, 7
of a type, 31
dual numbers, 220
Ehrenfeucht-Mostowski-type, 77
INDEX 266
elementary
class, 23, 47, 160
diagram, 25
directed family, 27
embedding, 25
equivalence, 23
extension, 26
map, 25, 61
property, 98
substructure, 25
elimination
of
, 92
of nite imaginaries, 163
of imaginaries, 159, 161
weak, 163
of quantiers, 42
embedding, 7
elementary, 25
equality symbol, 14
equivalence, 14
equivalent
formulas, 16, 38, 107
modulo T, 38
theories, 22
exchange property, 95
existence of nonforking extensions, 131,
139, 167
existential formula, 18
primitive, 42
simple, 42
existential quantier, 14
existentially closed
structure, 45
substructure, 45
expansion, 9
extension, see eld extension
atomic, 87
conservative, 62
elementary, 26
forking, 131
minimal, 87
of a structure, 8
of types, 61
prime, 85
eld, 21
algebraically closed, 35, 51, 52, 94,
97, 150
dierential, 220
dierentially closed, 56, 122, 170
formally real, 216
1free, 230
ordered, 216
procyclic, 230
pseudo algebraically closed, 228
pseudo-nite, 230
real closed, 217, 218
separably closed, 54, 176
eld extension
galois, 224
normal, 224
regular, 224
separable, 224, 225
lter, 20
nite character of forking, 127, 139,
167
nite equivalence relation, 159
Finite equivalence relation theorem, 171
nitely axiomatisable, 32, 33
nitely generated substructure, 8
nitely satisable
set of formulas, 31
theory, 28
forking
extension, 131
independence, 131
multiplicity, 171
set of formulas, 127
symmetry, 133, 139, 167
formally real eld, 216
formula
1based, 195
strongly minimal, 112
minimal over A, 114
, 39
, 39
algebraic, 94
atomic, 14
basic, 17
complete, 62
dividing, 125
existential, 18
INDEX 267
primitive, 42
simple, 42
has Morley rank, 111
isolating, 58
locally modular, 195
positive primitive, 49
quantier free, 17
stable, 152, 158
stably embedded, 155, 190
symmetric, 136
thick, 136, 141
universal, 18
with the binary tree property, 152
with the order property, 152
with the tree property, 129
formulas
disjoint, 75
equivalent, 16, 38, 107
Frass limit, 70
1free eld, 230
free occurrence, 15
function symbol, 6
Fundamental Theorem of Algebra, 218
galois eld extension, 224
GCH, 212
generated
substructure, 8
type, 242
generating set, 233
generic, 124, 221
type, 206
generic element, 172
generic point, 228
generic type, 206
geometry, 233
global type, 107
good denition of a type, 166
graph, 32
connected, 32
random, 57, 66, 72, 140, 153
group conguration, 172
Groupe de Liaison, 189
groupoid, 188
connected, 188
denable, 188
heir, 148, 149
Henkinconstants, 28
Henkintheory, 28
Hereditary Property, 69
Hilbert Basis Theorem, 81
Hilberts 17. Problem, 53
Hilberts Nullstellensatz, 52
home sort, 160
homomorphism, 7
imaginary elements, 160
implication, 14
independence
forking, 131
geometric, 235
independence property, 153
Independence Theorem, 138, 143
independent
family, 134
sequence, 131
set, 134, 233
indiscernibles, 76
parallel, 180
total, 178
induced theory, 98
induction, 213
inductive theory, 39
innitesimal, 33
initial segment, 210
interdenable, 108
internal type, 187
Interpolation Theorem, 108
interpretation, 7
invariant
class, 108, 109, 144
relation, 67
type, 128
isolated
set of formulas, 58
isolating formula, 58
isomorphic over a set, 87
isomorphic structures, 7
isomorphism, 7
Joint Consistency, 108
INDEX 268
Joint Embedding Property, 69
Kaiser hull, 46
-functions, 55
language, 6
Lascar rank, 176
Lascar strong type, 142
limit ordinal, 214
linearly disjoint, 222, 224
local character of forking, 130, 139, 167
locally nite, 67
locally modular
pregeometry, 236
formula, 195
logical symbols, 14
manysorted structure, 10
map
elementary, 25, 61
matroid, 95
meager set, 62
minimal
prime extension, 87
set, 97
type, 99, 175
model, 17, 107
consisting of constants, 29
prime, 73
model companion, 45
model complete theory, 44
modular
lattice, 236
pregeometry, 234
theory, 195
module, 49
monotony, 133, 139, 167
weak, 169
monster model, 106110
Morley degree
of a formula, 113
of a type, 114
Morley rank
of T, 111
of a formula, 111
of a type, 114
Morley sequence, 131, 179
in p, 131
of q over A, 131
Morleyisation, 42
multiplicity, 94, 171
negation normal form, 18
negation symbol, 14
normal eld extension, 224
normal form
conjunctive, 20
disjunctive, 20
negation, 18
prenex, 20
normal subset, 164, 183
nowhere dense set, 62
Nullstellensatz, 52
omitting a set of formulas, 58
Omitting Types, 58
order property, 152
order type, 213
ordered eld, 216
ordinal, 213
limit, 214
orthogonal type, 191
PAC eld, 228
parallel
indiscernibles, 180
type, 168, 180
parameter set, 107
parentheses, 14, 15
partial type, 58
perfect hull, 110, 225
plane
projective, 199
Poizat, Bruno, 189, 268
positive primitive formula, 49
ppdenable subgroup, 50
ppformula, 49
predicate, 6
pregeometry, 95
modular, 234
prenex normal form, 20
preservation theorems, 36
INDEX 269
prime
absolutely, 225
extension, 85
minimal, 87
model, 73
structure, 42
primitive existential formula, 42
procyclic
eld, 230
group, 229
pronite group, 228
projective plane, 199
projective space, 233
property
independence, 153
order, 152
pseudo algebraically
closed eld, 228
pseudo-nite eld, 230
quantier
existential, 14
universal, 14
quantier elimination, 42
quantier free formula, 17
Ramsey cardinal, 239
Ramseys Theorem, 77
random graph, 57, 66, 72, 140, 153
rank
SU, 173
U, 176
phirank, 154, 253
Cantor-Bendixson, 116
Lascar, 176
Morley, 111
real closed eld, 217, 218
real closure, 217
real elements, 160
realisation of a set of formulas, 31
realisation set, 16
recursion theorem, 214
regular action, 189
regular cardinal, 215
regular eld extension, 224
relation
bounded, 144
relation symbol, 6
relativisation of a pregeometry, 233
restriction
of a pregeometry, 233
of a structure, 9
of a type, 61, 62
ring, 21
Robinsons
Joint Consistency Lemma, 108
Test, 44
satisfaction, 15
saturated structure, 64, 82, 104106
sentence, 17
separable eld extension, 224, 225
separably closed eld, 54, 176
separating sentence, 36
Separation Lemma, 36
set
minimal, 97
set of formulas
dividing, 125
Shelahs Lemma about indiscernibles,
132
simple existential formula, 42
simple theory, 129
skeleton, 69
Skolem function, 79
Skolem theory, 79
small theory, 66
SOP, 154
sort, 160
home, 160
special structure, 104
stability spectrum, 174
stable
stable, 81, 152, 154
stable, 81
formula, 152, 158
theory, 152
type, 158
stably embedded formula, 155, 158, 190
Standard Lemma on indiscernibles, 77,
125
stationary type, 141, 166
INDEX 270
Stone duality, 61
Stone space, 61
strict order property, 154
strong type, 165
strongly homogeneous structure, 105
strongly minimal
almost, 124, 205
formula, 97
theory, 97
type, 97
structure, 7
homogeneous, 105
saturated, 82
universal, 105
saturated, 64
atomic, 73
existentially closed, 45
manysorted, 10, 160
saturated, 82
special, 104
strongly homogeneous, 105
sublanguage, 9
substitution, 13
lemma, 13, 17
substructure, 8
elementary, 25
existentially closed, 45
nitely generated, 8
generated, 8
substructure complete theory, 44
successor cardinal, 212
supersimple theory, 173
superstable theory, 173
SU-rank, 173
Sylvester, J.J., 218
symmetric formula, 136
symmetry of forking, 133, 139, 167
Tarskis
Chain Lemma, 27
Test, 26
term, 12
constant, 14
Theorem
Beths Interpolation, 108
Cantors, 211
CantorBernstein, 210
of Cantor, 35
of LwenheimSkolem, 34
of Lachlan, 88
of Morley, 103
downwards, 89
of RyllNardzewski, 64
of Vaught, 66
Ramseys, 77
Ressayres, 185
Vaughts Two-cardinal Theorem, 90
theorem
Finite equivalence relation, 171
theory, 21
stable, 81, 152, 154
stable, 81
complete, 22
consistent, 21
countable, 58
equivalent, 22
nitely satisable, 28
induced, 98
inductive, 39
categorical, 35
model complete, 44
modular, 195
of abelian groups, 21
of commutative rings, 21
of elds, 21
simple, 129
small, 66
stable, 152
substructure complete, 44
supersimple, 173
superstable, 173
totally transcendental, 82
unidimensional, 191
universal, 38
with prime extensions, 87
thick formula, 136, 141
totally transcendental theory, 82
trace, 219
transcendence degree, 234
transitivity, 133, 139, 167
tree, 75
INDEX 271
tree property, 129
trivial pregeometry, 235
tuple
atomic, 73
type, 31, 60
algebraic, 94
almost orthogonal, 191
analysable, 191
based on, 168
denable over C, 149
generated, 242
global, 107
internal, 187
invariant, 128
minimal, 99, 175
of a set, 32
of an element, 31
orthogonal, 191
parallel, 168, 180
partial, 58
quantier free, 72
stable, 158
stationary, 141, 166
strong, 165
type-denable class, 109
ultra-homogeneous, 69
ultralter, 20
unidimensional theory, 191
unique decomposition, 12, 15
uniqueness of nonforking extensions,
166
universal
formula, 18
quantier, 14
theory, 38
universe, 7
U-rank, 176
valid sentence, 22
vanishing ideal, 225
variable, 12
Vaughts conjecture, 67
Vaughts Test, 34
Vaughts Two-cardinal Theorem, 90
Vaughtian pair, 90
vector space, 49
weak elimination of imaginaries, 163
weak monotony, 169
weakly compact cardinal, 239
weakly inaccessible cardinal, 215, 239
wellordering, 210
theorem, 210

Buch

Caricato da

Informazioni sul documento

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Buch

Caricato da

Copyright:

Formati disponibili

A Course in Model Theory

= The empty language.

: A B is another homomorphism, then C = b [ h(b) = h

coincide on S, then S is a subset of C, and

be an isomorphism and Ban extension of A. Then

Proof. There are at most max[L[,

b. Here is the formal denition:

b we dene the relation

b] there exists a A such that A [=

b] holds we say holds in A for

b satises (in A).

b depends only on the free variables of :

b and c agree on all variables which are free in . Then

b] for some (all)

Proof. The proof is an easy induction on .

b) for all . is consistent with T if T is consistent.

B for every L(S

. It is clear that A is the universe of

would not be nitely satisable. This proves the lemma.

has a model A (unique

is nitely complete, every sentence which

. Otherwise the negation of that

, but would also be inconsistent together with a

. But clearly R(d

is a Henkintheory, there exists c

. This shows that everything is well dened.

be the L(C)structure (A, a

. Other atomic sentences contain function symbols f or

holds and by the induction

contains all sentences which follow from a nite subset of T

[= (c) for some c C (c) T

, we choose c satisfying (c) T

is an elementary extension of A, then

. Therefore we can realise p

is the model we were looking for.

~ A innitesimal if r < x < r for all r A

takes innitesimal elements to innitesimal elements.

B of cardinality and apply Corollary 4.3.

are isomorphic, and follows that

B[= T. If is true in A it is also true in Band therefore

where is existential (see p. 18). The following is clear:

The same procedure applied to A and B

for every Lformula (x

can be extended to a model of T.

and conversely. Let A

are the substruc-

e.c. This is clear for condition a) since every

can be embedded in a model M of T. For b) this follows from the fact

. We choose an enumeration ()< of all existential

to be the union of all A, ( < ). Note that A

AN, there is an embedding of M in

. Let M be a model of T. As in the proof of 8.9 we extend M

. It follows from 8.11 that T

is contained in the Kaiser hull.

be the model companion of T. As a model complete theory, T

is Te.c. Let A be a model of T which extends M. A can be embedded in a model N

. Criterion 8.7 implies, that T

there is an element b with

|= z(R(z, a) R(z, b)).

of innite sets are all innite sets without addi-

is empty, the axioms are (for n = 1, 2, . . .)

of innite sets has quantier elimination and

/F has a transcendence basis

with the following properties: