An Invitation to Mathematical Logic

David Marker
2
Preface

My goal was to write a text for a one semester graduate level introduction
to mathematical logic, one that I would have liked to learn from when I was a
student, and one I would like to teach from as a professor Two thirds of the book
evolved from lecture notes for courses given over 30 years teaching introductory
logic at the University of Illinois Chicago.
The heroes of most introductory logic texts are Gödel and Turing.1 Cer-
tainly the Gödel’s Completeness and Incompleteness Theorems and Turing’s
formalization of computability, universal machines and undecidable problems
must be at the center of any course in mathematical logic, but I think focusing
only on Gödel and Turing gives an unbalanced view. Gödel’s results on incom-
pleteness and undecidability in arithmetic become even more interesting when
contrasted with Tarski’s tameness results for the real and complex fields. One
of my goals is to raise Tarski to the podium alongside Gödel and Turing.
I titled this book an invitation to mathematical logic as I hope that it
will excite readers and make them hungry for further study in logic. One can
easily get bogged down at the beginning of a logic course when confronted with
a myriad of new definitions, formalism and syntax. I have tried to streamline
this part of the text in order to arrive as quickly as possible to the meat. From
my experience in the classroom this seems to work well. This book definitely
contains more material than could possibly covered in one semester. My hope is
that a student would be excited by what they have seen and will be motivated to
work through some of the more advanced material in Chapters 8, 12, 14, 15 and
16. Another goal is to put in one place some topics that often fall through the
gaps in one’s logic education such as cut elimination and models of arithmetic.
Mathematical logic grew out of the study of questions on the foundations of
mathematics. Foundational questions are the focus of Part’s I and IV and Part
III develops the foundations of computability. I have also tried, particularly in
Chapters 5, 7, 8, 14 and 16, to illustrate the interplay between logic and other
areas of mathematics, notably algebra, number theory and combinatorics. To
me this is one of the most fascinating aspects of modern logic.
One difficult decision was to exclude set theory–except for Appendix A which
contains a brief introduction to some useful fundamentals. While I view set
1 Gödel and Turing are even embedded in popular culture in books like Gödel, Escher Bach

and films like The Immitation Game and Oppenheimer. Indeed, “Pharma bro” Martin Shkreli
named two of his companies after Turing and Gödel.

i
ii Preface

theory as a central, highly important part of logic, I think it belongs in a separate

companion course. Including the material I would want for a proper introduction
to set theory would have probably added another 200 pages (and several years
of writing) to this volume, defeating the purpose of a one semester introductory
text. While perhaps there is a niche for a future An Invitation to Set Theory,
for now there are excellent introductory texts by Kunen [57], Schindler [86] and
Jech [38].
Another decision that some will disagree with is postponing the Incomplete-
ness Theorem until after an introduction to computability theory. Historically,
the Incompleteness Theorem preceded and inspired the development of com-
putability and it is both possible and reasonable to take a faster path toward
these results. I, however, believe that incompleteness phenomena are best un-
derstood having first encountered computability, particularly, the arithmetic
hierarchy and computably inseparable computably enumerable sets.

Detailed Overview
In part I we begin by introducing the basic concepts of logic: structures, truth,
proofs and Gödel’s Completeness Theorem. Chapter 1 begins with the basic con-
cepts of logic–the syntactic notions of languages, terms, formulas and theories
and the semantic notions of structures, truth in a structure, logical consequences
and definability. It is easy to get bogged down in some of the technical formal-
ism so I try to go though this material as quickly as possible. To this end, I cut
some corners on issues such as unique readability of formulas; for completeness,
I return to these issues in Appendix B. Induction on the complexity of formulas
is a basic proof technique in logic that does not have a counterpart in other
areas of mathematics. I include a number of results on equivalent normal forms
in this chapter, in part to give the reader more examples of this method.
A fundamental lesson of category theory is that when one studies any type
of mathematical objects it is imperative to study structure preserving maps
between them. Chapter 2 is devoted to the study of embeddings of structures,
isomorphisms of structures and how the truth of formulas is preserved under
mappings. This chapter provides the foundation for Part II Elements of Model
Theory, and will also be needed in Chapters 13 and 16 of Part IV. Chapter
2 concludes with the Tarski–Vaught characterization of elementary submodels
and the Downward Löwenheim–Skolem Theorem. This chapter also has the
pedagogical goal of reinforcing the student’s proficiency with proofs by induction
on complexity of formulas.
Chapters 3 and 4 culminate in Gödel’s Completeness Theorem, one of the
intellectual gems of logic. I still find it remarkable that the semantic notion of
logical consequence, which a priori requires quantification over all models of a
theory, is completely captured by a finitistic syntactic notion of proof. Chapter
3 introduces and studies a system of formal proof. I have chosen a variant of
sequent calculus. There are simpler systems, but this one has the advantage that
it is relatively easy to formalize proofs in this framework. Chapter 4 is devoted
 iii

to a proof, in the style of Henkin, of the Completeness Theorem. We need to

show that every consistent theory has a model. How do we build a structure
from scratch? In a surprising twist, the syntactic elements of the language can
be used to build the desired structure.
Part II is an introduction to model theory and algebraic applications. Chap-
ter 5 begins with the Compactness Theorem. Though the Compactness The-
orem is a simple consequence of the Completeness Theorem, it has many sur-
prising and intriguing consequences. I introduce the notions of complete and
-categorical theories, and Vaught’s test for completeness, and use these to show
that we can completely axiomatize the theory of the field of complex numbers
by saying it is an algebraically closed field of characteristic zero. This is ap-
plied in Ax’s astonishing proof that an injective polynomial map f : Cn ! Cn
must be surjective, where he reduces the problem to polynomial maps on finite
fields. I also introduce the back-and-forth method to prove several countable
categoricity results. One interesting combinatorial consequence is the 0-1 law
for random graphs.
Chapter 6 introduces ultraproducts, a model theoretic method for averag-
ing structures and building rich elementary extensions. Ultraproducts can be
used to give an alternative proof of the Compactness Theorem.2 In my earlier
book [63], I chose to deemphasize ultraproducts, a decision I came to regret.
Over the last two decades ultraproducts have come to play an important role in
continuous model theory and in connections between model theory and combi-
natorics. The usefulness of ultraproducts in set theory is also a strong argument
for inclusion in any first course in logic.
Chapters 7 and 8 are devoted to Tarski’s results that, in contrast to the Gödel
phenomena of incompleteness and undecidability in arithmetic, the theories of
the complex and real field are quite tame. Chapter 7 develops a model theoretic
test for quantifier elimination. As a warm up we apply these results to divisible
abelian groups and ordered divisible abelian groups. We then prove quantifier
elimination for algebraically closed fields, apply this to characterize the definable
sets as the constructible sets of algebraic geometry, and give a model theoretic
proof of Hilbert’s Nullstellensatz.
In Chapter 8 we begin be “reviewing” some of the basic algebra of ordered
fields. As I expect most students will not have seen this before, I include a
detailed survey of real algebra with proofs in Appendix C. We then prove quan-
tifier elimination for the theory of real closed fields. Some of the applications we
include are a characterization of definable sets as the semialgebraic sets, discus-
sion of tameness of definable functions and sets, and Abraham Robinson’s model
theoretic version of Artin’s solution to Hilbert’s 17th Problem.3 The chapter
concludes with a brief survey of more recent results on o-minimal expansions of
the real field and exponentiation.
2 Indeed, if one wanted to get to the model theoretic material as quickly as possible, one

could do Chapter 6 directly after Chapter 2 and then cover Chapters 5, 6 and 7, postponing
formal proofs to a later part of the course.
3 I have adopted the style of referring to most people only by their last name–this is im-

possible with Abraham, Julia and Raphael Robinson.

iv Preface

Part III is devoted to computability. Chapter 9 begins by introducing reg-

ister machines as a simple model of a programmable computing device. While
Turing machines are even simpler, register machines are much easier to program.
Although they are simple, the Church–Turing thesis asserts that they capture
completely our intuitive notion of potentially computable by algorithm. As ex-
amples of things computable by register machines we introduce the primitive
and partial recursive functions of Gödel and Herband. We first show that they
are all register machine computable and that indeed they describe the same
class of computable functions. As evidence for the Church–Turing thesis we in-
troduce random access machines and show that even though they at first seem
more powerful, anything they can do can be done on a register machine. For
completeness, we conclude with a description of Turing machines and provide
some examples.
Chapter 10 discusses Turing’s celebrated results that there are universal ma-
chines and that the halting problem is undecidable. We use this to give several
other examples of undecidability, including Church’s theorem on the undecid-
ability of validity in first order logic and Rice’s theorem on index sets. We
conclude with Kleene’s Recursion Theorem, one of the more intriguing conse-
quences of the existence of universal machines.
Chapter 11 introduces the computably enumerable sets and the arithmetic
sets and discusses some of their properties.4 This material will be crucial in our
approach to the First Incompleteness Theorem in Chapter 13. As an example
of an interesting ⇧01 -set, we study the Kolmogorov random numbers and arising
incompleteness phenomena.
Chapter 12 is a brief introduction to some further topics in computability
theory. We start by introducing Turing reducibility, the Turing jump, and giv-
ing Post’s characterization of the arithmetic hierarchy. I then have chosen to
survey some of my favorite results in computability theory: the Kleene–Post
construction of incomparable degrees, Spector’s construction of a minimal de-
gree, a finite injury priority argument to prove the existence of incomplete non-
computable computably enumerable sets, and the Jockusch–Soare Low Basis
Theorem. These may seem like random choices but I’ve included these results
because a reasonable number of logic students will never take a more advanced
course in computability, and I think it is important that they see these results.
In set theory, the Kleene–Post construction is a precursor to Cohen forcing, and
the minimal degree construction led to Sacks forcing. I think it is important
for a student in set theory to see these results in their original context. The
Low Basis Theorem gives new insight on problems like finding the completion
of a theory. Finally, every student of logic should understand at least one basic
priority argument.
Part IV is devoted to various manifestations of the incompleteness and un-
decidability of arithmetic. Chapter 13 is devoted to Gödel’s Incompleteness
Theorems and refinements. We first prove the first Incompleteness Theorem by
4 Computably enumerable sets used to be called recursively enumerable. While I have

adopted the new terminology, I still can not bring myself to use the common abbreviation c.e.
sets.
 v

showing that we can define the graph of every primitive recursive function in
N, then concluding that the sets definable in N are exactly the arithmetic sets.
The results of Part III can then be used to conclude, in a very strong way, that
Th(N) is undecidable and can not be recursively axiomatized. We then turn
to Gödel’s original proof, where we code formulas by numbers and use diago-
nalization to produce a sentence asserting its own unprovability. A sketch is
given of a proof of the Second Incompleteness Theorem, that PA does not prove
its own consistency, though we leave out some tedious, but necessary, details
showing that some of the basic properties of proof systems can be formalized in
PA. Finally, we sketch the proof by Hilbert and Bernays that the Completeness
Theorem can be formalized in PA and Kreisel’s model theoretic proof of the
Second Incompleteness Theorem.
Chapters 14–16 explore di↵erent aspects of incompleteness phenomena. One
of the most astonishing manifestations of the incompleteness phenomena is the
undecidability of Hilbert’s 10th problem on the solvability of Diophantine equa-
tions. In Chapter 14 we discuss some aspects of the proof. In particular, fol-
lowing the early work of Davis, Putnam and Julia Robinson [14], we will show
that if we assume that y = 2x is Diophantine definable, then the Diophantine
definable sets are exactly the computably enumerable sets. We then discuss
Pell equations, one of the key ideas in Matiyasevich’s proof that y = 2x is Dio-
phantine. I will not say much about the rest of the proof. It is long, clever
and detailed, relying mostly on very elementary number theory but there are
no further logical aspects. As I feel I have nothing new to add, I refer the reader
to clear, elegant published treatments such as Murty and Fodden’s book [72].
Goodstein found a surprising number theoretic statement whose proof makes
essential use of ordinals below ✏0 . Chapter 15 begins with Goodstein’s proof,
and then proves the independence of Goodstein’s result by showing that the
use of ✏0 is essential and beyond Peano Arithmetic. The bulk of the chapter
is devoted to a theorem of Wainer, building on work of Gentzen and Kreisel,
calibrating the growth rates of computable functions provably total in PA. My
treatment of this material follows closely unpublished notes of Henry Towsner
[100]. and I thank him for letting me adapt his presentation.
Chapter 16 centers on a model theoretic proof due to Paris and Harrington of
the independence from PA of a combinatorial statement that is a minor variant
of the fundamental result of Ramsey Theory. As a warm up we give model
theoretic proofs of two results in the spirit of Chapter 15 characterizing the
growth rates of computable functions provably total in weak fragments of Peano
Arithmetic. Having introduced the study of nonstandard models of PA as a tool
in independence results, we turn to studying these models as interesting objects
in their own right and prove several fundamental results about extensions and
embeddings of models of PA.
vi Preface

Using this book as a text

This book contains significantly more material than could be covered in a one
semester course. This was done intentionally. First, I wanted to give instructors
some flexibility in shaping the content of their course. At the end of this intro-
duction I have included a graph of the essential dependencies between chapters.
Secondly, I wanted to provide additional material with the hope that interested
students would be tempted to explore material beyond what can be covered in
one semester.
At a bare minimum I think a course should cover:

• Chapter 1: omitting the material on normal forms;

• Chapter 2: through Proposition 2.16;

• Chapters 3 and 4: all;

• Chapter 5: through the Upward Löwenheim–Skolem Theorem;

• Chapter 9: through the proof of the equivalence of register machine com-

putable and general recursive;

• Chapter 10: omitting index sets and the Recursion Theorem;

• Chapter 11: through the analysis of the arithmetic hierarchy;

• Chapter 13: through the Second Incompleteness Theorem.5

When I taught the course at the University of Illinois Chicago, I would

usually also cover: the rest of Chapters 2 and 5, all of Chapter 6, the quantifier
elimination tests with applications to algebraically closed fields in Chapter 7,
quantifier elimination for real closed fields and some consequences in Chapter 8,
Kolmogorov randomness from Chapter 11, Turing reducibility, Post’s theorem,
the existence of incomparable degrees and the finite injury priority argument
from Chapter 12 and, if time permitted, a briefly discuss Goodstein’s Theorem
from Chapter 15–though the exact content changed from year to year.
Chapter 14, 15 and 16 will probably not be covered in most one semester
courses. They, like some of the material in Chapters 8 and 12, would make
excellent subject matter for a reading course, a graduate student working group,
or self-study by motivated students.
Each chapter ends with a section of exercises. The exercises range from
quite easy to quite challenging. Some of the exercises develop important ideas
that I would have included in a longer text. I have left some important results
as exercises because I think students will benefit by working them out. Some
exercises are embedded in the chapters. These tend to be ones where I think
5 If one needs a quicker path to the Incompleteness Theorem, it would be possible to do

Chapter 13 immediately after proving the undecidability of the Halting Problem in Chapter
10, adapting slightly some of the arguments in Chapter 13.
 vii

it would be good for the student to work on the exercise immediately to fully
understand what is going on. Some exercises will require more comfort with
algebra, computability, or set theory than I assume in the rest of the book. I
mark those exercises with a dagger †.

Prerequisites
For most of the text the only prerequisite is “mathematical maturity”. It should
be suitable for first year graduate students or advanced undergraduates in math-
ematics, philosophy graduate students with a solid math background or students
in computer science who want a mathematical introduction to logic. While some
prior exposure to logic would be helpful, it is not assumed. We assume some
familiarity with basic set theory–countability, cardinality and Zorn’s Lemma–
more or less at the same level one would expect for a first graduate analysis
course. Appendix A covers most of this material. In Chapter 15 we assume
some familiarity with ordinals. This is also covered in Appendix A.
In Chapters 5, 7 and 8 we assume some familiarity with algebra, particularly
algebraically closed fields. A student simultaneously taking a graduate algebra
course should be well prepared. Chapter 8 uses algebraic results on ordered
fields. While this material is included in many graduate algebra texts–Lang’s
Algebra [58] is a good reference for the material we need–I suspect most students
will not have seen it, so I develop the necessary results in Appendix C.
viii Preface

Chapter dependencies

Chapter 1

Chapter 9

Chapter 2 Chapter 3

Chapter 10

Chapter 6 Chapter 4 Chapter 11

Chapter 5 Chapter 13 Chapter 12

Chapter 7 Chapter 16 Chapter 14 Chapter 15

Chapter 8
Contents

Preface i

I Truth and Proof 1

1 Languages, Structures and Theories 3

2 Embeddings and Substructures 27

3 Formal Proofs 37

4 Gödel’s Completeness Theorem 45

II Elements of Model Theory 57

5 Compactness and Complete Theories 59

6 Ultraproducts 75

7 Quantifier Elimination 87

8 Model Theory of the Real Field 109

III Computability 125

9 Models of Computation 127

10 Universal Machines and Undecidability 147

11 Computably Enumerable and Arithmetic Sets 157

12 Turing Reducibility 169

ix
x CONTENTS

IV Arithmetic and Incompleteness 185

13 Gödel’s Incompleteness Theorems 187

14 Hilbert’s 10th Problem 213

15 Peano Arithmetic and ✏0 229

16 Models of Arithmetic and Independence Results 263

Appendicies 287
A Set Theory 287

B Unique Readability 301

C Real Algebra 305