## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Simpson

Mathematical Logic

Stephen G. Simpson

October 7, 2008

Department of Mathematics

The Pennsylvania State University

University Park, State College PA 16802

http://www.math.psu.edu/simpson/

This is a set of lecture notes for introductory courses in mathematical logic

oﬀered at the Pennsylvania State University.

Contents

Contents 1

1 Propositional Calculus 3

1.1 Formulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.2 Assignments and Satisﬁability . . . . . . . . . . . . . . . . . . . . 6

1.3 Logical Equivalence . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.4 The Tableau Method . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.5 The Completeness Theorem . . . . . . . . . . . . . . . . . . . . . 18

1.6 Trees and K¨onig’s Lemma . . . . . . . . . . . . . . . . . . . . . . 20

1.7 The Compactness Theorem . . . . . . . . . . . . . . . . . . . . . 21

1.8 Combinatorial Applications . . . . . . . . . . . . . . . . . . . . . 22

2 Predicate Calculus 24

2.1 Formulas and Sentences . . . . . . . . . . . . . . . . . . . . . . . 24

2.2 Structures and Satisﬁability . . . . . . . . . . . . . . . . . . . . . 26

2.3 The Tableau Method . . . . . . . . . . . . . . . . . . . . . . . . . 31

2.4 Logical Equivalence . . . . . . . . . . . . . . . . . . . . . . . . . . 37

2.5 The Completeness Theorem . . . . . . . . . . . . . . . . . . . . . 40

2.6 The Compactness Theorem . . . . . . . . . . . . . . . . . . . . . 46

2.7 Satisﬁability in a Domain . . . . . . . . . . . . . . . . . . . . . . 47

3 Proof Systems for Predicate Calculus 50

3.1 Introduction to Proof Systems . . . . . . . . . . . . . . . . . . . . 50

3.2 The Companion Theorem . . . . . . . . . . . . . . . . . . . . . . 51

3.3 Hilbert-Style Proof Systems . . . . . . . . . . . . . . . . . . . . . 56

3.4 Gentzen-Style Proof Systems . . . . . . . . . . . . . . . . . . . . 61

3.5 The Interpolation Theorem . . . . . . . . . . . . . . . . . . . . . 66

4 Extensions of Predicate Calculus 71

4.1 Predicate Calculus with Identity . . . . . . . . . . . . . . . . . . 71

4.2 The Spectrum Problem . . . . . . . . . . . . . . . . . . . . . . . 75

4.3 Predicate Calculus With Operations . . . . . . . . . . . . . . . . 78

4.4 Predicate Calculus with Identity and Operations . . . . . . . . . 82

4.5 Many-Sorted Predicate Calculus . . . . . . . . . . . . . . . . . . 84

1

5 Theories, Models, Deﬁnability 87

5.1 Theories and Models . . . . . . . . . . . . . . . . . . . . . . . . . 87

5.2 Mathematical Theories . . . . . . . . . . . . . . . . . . . . . . . . 89

5.3 Deﬁnability over a Model . . . . . . . . . . . . . . . . . . . . . . 97

5.4 Deﬁnitional Extensions of Theories . . . . . . . . . . . . . . . . . 100

5.5 Foundational Theories . . . . . . . . . . . . . . . . . . . . . . . . 103

5.6 Axiomatic Set Theory . . . . . . . . . . . . . . . . . . . . . . . . 106

5.7 Interpretability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

5.8 Beth’s Deﬁnability Theorem . . . . . . . . . . . . . . . . . . . . . 112

6 Arithmetization of Predicate Calculus 114

6.1 Primitive Recursive Arithmetic . . . . . . . . . . . . . . . . . . . 114

6.2 Interpretability of PRA in Z

1

. . . . . . . . . . . . . . . . . . . . 114

6.3 G¨odel Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

6.4 Undeﬁnability of Truth . . . . . . . . . . . . . . . . . . . . . . . . 117

6.5 The Provability Predicate . . . . . . . . . . . . . . . . . . . . . . 118

6.6 The Incompleteness Theorems . . . . . . . . . . . . . . . . . . . . 119

6.7 Proof of Lemma 6.5.3 . . . . . . . . . . . . . . . . . . . . . . . . 121

Bibliography 122

Index 123

2

Chapter 1

Propositional Calculus

1.1 Formulas

Deﬁnition 1.1.1. The propositional connectives are negation (), conjunction

( ∧), disjunction ( ∨), implication ( ⇒), biimplication ( ⇔). They are read as

“not”, “and”, “or”, “if-then”, “if and only if” respectively. The connectives ∧,

∨, ⇒, ⇔ are designated as binary, while is designated as unary.

Deﬁnition 1.1.2. A propositional language L is a set of propositional atoms

p, q, r, . . .. An atomic L-formula is an atom of L.

Deﬁnition 1.1.3. The set of L-formulas is generated inductively according to

the following rules:

1. If p is an atomic L-formula, then p is an L-formula.

2. If A is an L-formula, then (A) is an L-formula.

3. If A and B are L-formulas, then (A∧B), (A∨B), (A⇒B), and (A⇔B)

are L-formulas.

Note that rule 3 can be written as follows:

3

′

. If A and B are L-formulas and b is a binary connective, then (Ab B) is an

L-formula.

Example 1.1.4. Assume that L contains propositional atoms p, q, r, s. Then

(((p ⇒q) ∧(q ∨r)) ⇒(p ∨r)) ⇒(q ∨s)

is an L-formula.

Deﬁnition 1.1.5. If A is a formula, the degree of A is the number of occurrences

of propositional connectives in A. This is the same as the number of times rules

2 and 3 had to be applied in order to generate A.

3

Example 1.1.6. The degree of the formula of Example 1.1.4 is 8.

Remark 1.1.7 (omitting parentheses). As in the above example, we omit

parentheses when this can be done without ambiguity. In particular, outermost

parentheses can always be omitted, so instead of ((A) ⇒B) we may write

(A) ⇒B. But we may not write A⇒B, because this would not distinguish

the intended formula from (A⇒B).

Deﬁnition 1.1.8. Let L be a propositional language. A formation sequence is

ﬁnite sequence A

1

, A

2

, . . . , A

n

such that each term of the sequence is obtained

from previous terms by application of one of the rules in Deﬁnition 1.1.3. A

formation sequence for A is a formation sequence whose last term is A. Note

that A is an L-formula if and only if there exists a formation sequence for A.

Example 1.1.9. A formation sequence for the L-formula of Example 1.1.4 is

p, q, p ⇒q, r, q ∨r, (p ⇒q) ∧(q ∨r), p ∨r, ((p ⇒q) ∧(q ∨r)) ⇒(p ∨r),

s, q ∨s, (q ∨s), (((p ⇒q) ∧(q ∨r)) ⇒(p ∨r)) ⇒(q ∨s) .

Remark 1.1.10. In contexts where the language L does not need to be speci-

ﬁed, an L-formula may be called a formula.

Deﬁnition 1.1.11. A formation tree is a ﬁnite rooted dyadic tree where each

node carries a formula and each non-atomic formula branches to its immediate

subformulas (see the example below). If A is a formula, the formation tree for

A is the unique formation tree which carries A at its root.

Example 1.1.12. The formation tree for the formula of Example 1.1.4 is

(((p ⇒q) ∧(q ∨r)) ⇒(p ∨r)) ⇒(q ∨s)

/ ¸

((p ⇒q) ∧(q ∨r)) ⇒(p ∨r)

/ ¸

(p ⇒q) ∧(q ∨r)

/ ¸

p ⇒q q ∨r

/ ¸ / ¸

p q q r

p ∨r

/ ¸

p r

(q ∨s)

[

q ∨s

/ ¸

q s

or, in an abbreviated style,

⇒

_ _

⇒

/ ¸

∧ ∨

/ ¸ / ¸

⇒ ∨ p r

/ ¸ / ¸

p q q r

[

∨

/ ¸

q s

4

Remark 1.1.13. Note that, if we identify formulas with formation trees in the

abbreviated style, then there is no need for parentheses.

Remark 1.1.14. Another way to avoid parentheses is to use Polish notation.

In this case the set of L-formulas is generated as follows:

1. If p is an atomic L-formula, then p is an L-formula.

2. If A is an L-formula, then A is an L-formula.

3. If A and B are L-formulas and b is a binary connective, then b AB is an

L-formula.

For example, (p) ⇒q becomes ⇒p q, and (p ⇒q) becomes ⇒p q. The

formula of Example 1.1.4 becomes

⇒ ⇒ ∧ ⇒ p q ∨ q r ∨ p r ∨ q s

and a formation sequence for this is

p, q, ⇒p q, r, ∨q r, ∧ ⇒p q ∨q r, ∨p r, ⇒ ∧ ⇒p q ∨q r ∨p r,

s, ∨q s, ∨q s, ⇒ ⇒ ∧ ⇒p q ∨q r ∨p r ∨q s .

Obviously Polish notation is diﬃcult to read, but it has the advantages of being

linear and of not using parentheses.

Remark 1.1.15. In our study of formulas, we shall be indiﬀerent to the ques-

tion of which system of notation is actually used. The only point of interest for

us is that each non-atomic formula is uniquely of the form A or Ab B, where

A and B are formulas and b is a binary connective.

Exercises 1.1.16. Let C be the formula (p ∧q) ⇒(p ∨r).

1. Restore all the omitted parentheses to C. (See Remark 1.1.7.)

2. Exhibit a formation sequence for C.

3. List the immediate subformulas of C, their immediate subformulas, etc.,

i.e., all subformulas of C.

4. Calculate the degrees of C and its subformulas.

5. Display the formation tree of C.

6. Write C according to various notation systems:

(a) The rules 1–3 of Deﬁnition 1.1.3:

1. Each atom is a formula.

2. If A is a formula then (A) is a formula.

3. If A and B are formulas and b is a binary connective, then (AbB)

is a formula.

5

(b) The following alternative set of rules:

1. Each atom is a formula.

2. If A is a formula then (A) is a formula.

3. If A and B are formulas and b is a binary connective, then

(A)b(B) is a formula.

(c) Polish notation.

(d) Reverse Polish notation.

1.2 Assignments and Satisﬁability

Deﬁnition 1.2.1. There are two truth values, T and F, denoting truth and

falsity.

Deﬁnition 1.2.2. Let L be a propositional language. An L-assignment is a

mapping

M : ¦p [ p is an atomic L-formula¦ → ¦T, F¦ .

Note that if L has exactly n atoms then there are exactly 2

n

diﬀerent L-

assignments.

Lemma 1.2.3. Given an L-assignment M, there is a unique L-valuation

v

M

: ¦A [ A is an L-formula¦ → ¦T, F¦

given by the following clauses:

1. v

M

(A) =

_

T if v

M

(A) = F,

F if v

M

(A) = T.

2. v

M

(A∧B) =

_

T if v

M

(A) = v

M

(B) = T,

F if at least one of v

M

(A), v

M

(B) = F.

3. v

M

(A∨B) =

_

T if at least one of v

M

(A), v

M

(B) = T,

F if v

M

(A) = v

M

(B) = F.

4. v

M

(A⇒B) = v

M

((A∧B)) .

5. v

M

(A⇔B) =

_

T if v

M

(A) = v

M

(B) ,

F if v

M

(A) ,= v

M

(B) .

Proof. The truth value v

M

(A) is deﬁned by recursion on L-formulas, i.e., by

induction on the degree of A where A is an arbitrary L-formula.

Remark 1.2.4. Note that each clause of Lemma 1.2.3 corresponds to the fa-

miliar truth table for the corresponding propositional connective. Thus clause

3 corresponds to the truth table

6

A B A∨B

T T T

T F T

F T T

F F F

for ∨, and clause 4 corresponds to the truth table

A B A⇒B

T T T

T F F

F T T

F F T

for ⇒.

Remark 1.2.5. Lemma 1.2.3 may be visualized in terms of formation trees.

To deﬁne v

M

(A) for a formula A, one begins with an assignment of truth values

to the atoms, i.e., the end nodes of the formation tree for A, and then proceeds

upward to the root, assigning truth values to the nodes, each step being given

by the appropriate clause.

Example 1.2.6. Consider the formula (p ⇒q) ⇒(q ⇒r) under an assignment

M with M(p) = T, M(q) = F, M(r) = T. In terms of the formation tree, this

looks like

(p ⇒q) ⇒(q ⇒r)

/ ¸

p ⇒q q ⇒r

/ ¸ / ¸

p q q r

T F F T

and by applying clause 4 three times we get

(p ⇒q) ⇒(q ⇒r)

T

_ _

p ⇒q q ⇒r

F T

/ ¸ / ¸

p q q r

T F F T

and from this we see that v

M

((p ⇒q) ⇒(q ⇒r)) = T.

Remark 1.2.7. The above formation tree with truth values can be compressed

and written linearly as

(p ⇒q) ⇒(q ⇒r)

T F F T F T T .

7

This illustrates a convenient method for calculating v

M

(A), where M is an

arbitrary L-assignment.

Remark 1.2.8. Lemma 1.2.3 implies that there is an obvious one-to-one cor-

respondence between L-assignments and L-valuations. If the language L is

understood from context, we may speak simply of assignments and valuations.

We now present some key deﬁnitions. Fix a propositional language L.

Deﬁnition 1.2.9. Let M be an assignment. A formula A is said to be true

under M if v

M

(A) = T, and false under M if v

M

(A) = F.

Deﬁnition 1.2.10. A set of formulas S is said to be satisﬁable if there exists

an assignment M which satisﬁes S, i.e., v

M

(A) = T for all A ∈ S.

Deﬁnition 1.2.11. Let S be a set of formulas. A formula B is said to be a

logical consequence of S if it is true under all assignments which satisfy S.

Deﬁnition 1.2.12. A formula B is said to be logically valid (or a tautology) if

B is true under all assignments. Equivalently, B is a logical consequence of the

empty set.

Remark 1.2.13. B is a logical consequence of A

1

, . . . , A

n

if and only if

(A

1

∧ ∧A

n

) ⇒B

is logically valid. B is logically valid if and only if B is not satisﬁable.

Exercises 1.2.14.

1. Use truth tables to show that ((A⇒B) ⇒A) ⇒A is logically valid.

2. Use truth tables to show that (A∧B) ⇒C is logically equivalent to

A⇒(B ⇒C).

Exercises 1.2.15. Prove the following. (See Remarks 1.2.13 and 1.3.2.)

1. B is logically valid if and only if B is not satisﬁable.

2. B is satisﬁable if and only if B is not logically valid.

3. B is a logical consequence of A

1

, . . . , A

n

if and only if

(A

1

∧ ∧A

n

) ⇒B is logically valid.

4. A is logically equivalent to B if and only if A⇔B is logically valid.

Exercise 1.2.16. Brown, Jones, and Smith are suspected of a crime. They

testify as follows:

Brown: Jones is guilty and Smith is innocent.

Jones: If Brown is guilty then so is Smith.

8

Smith: I’m innocent, but at least one of the others is guilty.

Let b, j, and s be the statements “Brown is innocent,” “Jones is innocent,”

“Smith is innocent”.

1. Express the testimony of each suspect as a propositional formula. Write

a truth table for the three testimonies.

2. Use the above truth table to answer the following questions:

(a) Are the three testimonies consistent?

(b) The testimony of one of the suspects follows from that of another.

Which from which?

(c) Assuming everybody is innocent, who committed perjury?

(d) Assuming all testimony is true, who is innocent and who is guilty?

(e) Assuming that the innocent told the truth and the guilty told lies,

who is innocent and who is guilty?

Solution.

1. The testimonies are:

B : (j) ∧s

J : (b) ⇒(s)

S : s ∧((b) ∨(j))

The truth table is:

b j s B J S

1 T T T F T F

2 T T F F T F

3 T F T T T T

4 T F F F T F

5 F T T F F T

6 F T F F T F

7 F F T T F T

8 F F F F T F

2. (a) Yes, by line 3 of the table.

(b) The table shows that S is a logical consequence of B. In other words,

Smith’s testimony follows from Brown’s.

(c) If everybody is innocent, we are in line 1 of the table. Hence B and

S are false, i.e., Brown and Smith lied.

(d) If all the testimony is true, we are in line 3 of the table. Thus Brown

and Smith are innocent, while Jones is guilty.

(e) Our assumption is v

M

(b) = v

M

(B), v

M

(j) = v

M

(J), v

M

(s) = v

M

(S).

Hence we are in line 6 of the table. Thus Jones is innocent, and Brown

and Smith are guilty.

9

1.3 Logical Equivalence

Deﬁnition 1.3.1. Two formulas A and B are said to be logically equivalent,

written A ≡ B, if each is a logical consequence of the other.

Remark 1.3.2. A ≡ B holds if and only if A⇔B is logically valid.

Exercise 1.3.3. Assume A

1

≡ A

2

. Show that

1. A

1

≡ A

2

;

2. A

1

∧B ≡ A

2

∧B;

3. B∧A

1

≡ B∧A

2

;

4. A

1

∨B ≡ A

2

∨B;

5. B∨A

1

≡ B∨A

2

;

6. A

1

⇒B ≡ A

2

⇒B;

7. B⇒A

1

≡ B⇒A

2

;

8. A

1

⇔B ≡ A

2

⇔B;

9. B⇔A

1

≡ B⇔A

2

.

Exercise 1.3.4. Assume A

1

≡ A

2

. Show that for any formula C containing

A

1

as a part, if we replace one or more occurrences of the part A

1

by A

2

, then

the resulting formula is logically equivalent to C. (Hint: Use the results of the

previous exercise, plus induction on the degree of C.)

Remark 1.3.5. Some useful logical equivalences are:

1. commutative laws:

(a) A∧B ≡ B∧A

(b) A∨B ≡ B∨A

(c) A⇔B ≡ B⇔A

Note however that A⇒B ,≡ B⇒A.

2. associative laws:

(a) A∧(B∧C) ≡ (A∧B) ∧C

(b) A∨(B∨C) ≡ (A∨B) ∨C

(c) A⇔(B⇔C) ≡ (A⇔B) ⇔C

Note however that A⇒(B ⇒C) ,≡ (A⇒B) ⇒C.

3. distributive laws:

10

(a) A∧(B∨C) ≡ (A∧B) ∨(A∧C)

(b) A∨(B∧C) ≡ (A∨B) ∧(A∨C)

(c) A⇒(B∧C) ≡ (A⇒B) ∧(A⇒C)

(d) (A∨B) ⇒C ≡ (A⇒C) ∧(B ⇒C)

Note however that A⇒(B∨C) ,≡ (A⇒B) ∨(A⇒C), and (A∧B) ⇒C ,≡

(A⇒C) ∧(B⇒C).

4. negation laws:

(a) (A∧B) ≡ (A) ∨(B)

(b) (A∨B) ≡ (A) ∧(B)

(c) A ≡ A

(d) (A⇒B) ≡ A∧B

(e) (A⇔B) ≡ (A) ⇔B

(f) (A⇔B) ≡ A⇔(B)

5. implication laws:

(a) A⇒B ≡ (A∧B)

(b) A⇒B ≡ (A) ∨B

(c) A⇒B ≡ (B) ⇒(A)

(d) A⇔B ≡ (A⇒B) ∧(B ⇒A)

(e) A⇔B ≡ (A) ⇔(B)

Deﬁnition 1.3.6. A formula is said to be in disjunctive normal form if it is

of the form A

1

∨ ∨A

m

, where each clause A

i

, i = 1, . . . , m, is of the form

B

1

∧ ∧B

n

, and each B

j

, j = 1, . . . , n is either an atom or the negation of

an atom.

Example 1.3.7. Writing p as an abbreviation for p, the formula

(p

1

∧p

2

∧p

3

) ∨(p

1

∧p

2

∧p

3

) ∨(p

1

∧p

2

∧p

3

)

is in disjunctive normal form.

Exercise 1.3.8. Show that every propositional formula C is logically equivalent

to a formula in disjunctive normal form.

Remark 1.3.9. There are two ways to do Exercise 1.3.8.

1. One way is to apply the equivalences of Remark 1.3.5 to subformulas of C

via Exercise 1.3.4, much as one applies the commutative and distributive

laws in algebra to reduce every algebraic expression to a polynomial.

11

2. The other way is to use a truth table for C. The disjunctive normal form

of C has a clause for each assignment making C true. The clause speciﬁes

the assignment.

Example 1.3.10. Consider the formula (p ⇒q) ⇒r. We wish to put this in

disjunctive normal form.

Method 1. Applying the equivalences of Remark 1.3.5, we obtain

(p ⇒q) ⇒r ≡ r ∨(p ⇒q)

≡ r ∨(p ∧q)

≡ r ∨(p ∧q)

and this is in disjunctive normal form.

Method 2. Consider the truth table

p q r p ⇒q (p ⇒q) ⇒r

1 T T T T T

2 T T F T F

3 T F T F T

4 T F F F T

5 F T T T T

6 F T F T F

7 F F T T T

8 F F F T F

Each line of this table corresponds to a diﬀerent assignment. From lines 1, 3,

4, 5, 7 we read oﬀ the following formula equivalent to (p ⇒q) ⇒r in disjunctive

normal form:

(p ∧q ∧r) ∨(p ∧q ∧r) ∨(p ∧q ∧r) ∨(p ∧q ∧r) ∨(p ∧q ∧r) .

1.4 The Tableau Method

Remark 1.4.1. A more descriptive name for tableaux is satisﬁability trees. We

follow the approach of Smullyan [4].

Deﬁnition 1.4.2. A signed formula is an expression of the form TA or FA,

where A is a formula. An unsigned formula is simply a formula.

Deﬁnition 1.4.3. A signed tableau is a rooted dyadic tree where each node

carries a signed formula. An unsigned tableau is a rooted dyadic tree where

each node carries an unsigned formula. The signed tableau rules are presented

in Table 1.1. The unsigned tableau rules are presented in Table 1.2. If τ is a

(signed or unsigned) tableau, an immediate extension of τ is a larger tableau τ

′

obtained by applying a tableau rule to a ﬁnite path of τ.

12

.

.

.

TA∧B

.

.

.

[

TA

TB

.

.

.

FA∧B

.

.

.

/ ¸

FA FB

.

.

.

TA∨B

.

.

.

/ ¸

TA TB

.

.

.

FA∨B

.

.

.

[

FA

FB

.

.

.

TA⇒B

.

.

.

/ ¸

FA TB

.

.

.

FA⇒B

.

.

.

[

TA

FB

.

.

.

TA⇔B

.

.

.

/ ¸

TA FA

TB FB

.

.

.

FA⇔B

.

.

.

/ ¸

TA FA

FB TB

.

.

.

TA

.

.

.

[

FA

.

.

.

FA

.

.

.

[

TA

Table 1.1: Signed tableau rules for propositional connectives.

13

.

.

.

A∧B

.

.

.

[

A

B

.

.

.

(A∧B)

.

.

.

/ ¸

A B

.

.

.

A∨B

.

.

.

/ ¸

A B

.

.

.

(A∨B)

.

.

.

[

A

B

.

.

.

A⇒B

.

.

.

/ ¸

A B

.

.

.

(A⇒B)

.

.

.

[

A

B

.

.

.

A⇔B

.

.

.

/ ¸

A A

B B

.

.

.

(A⇔B)

.

.

.

/ ¸

A A

B B

.

.

.

A

.

.

.

[

A

Table 1.2: Unsigned tableau rules for propositional connectives.

14

Deﬁnition 1.4.4. Let X

1

, . . . , X

k

be a ﬁnite set of signed or unsigned formulas.

A tableau starting with X

1

, . . . , X

k

is a tableau obtained from

X

1

.

.

.

X

k

by repeatedly applying tableau rules.

Deﬁnition 1.4.5. A path of a tableau is said to be closed if it contains a

conjugate pair of signed or unsigned formulas, i.e., a pair such as TA, FA in

the signed case, or A, A in the unsigned case. A path of a tableau is said to

be open if it is not closed. A tableau is said to be closed if each of its paths is

closed.

The tableau method:

1. To test a formula A for validity, form a signed tableau starting with FA, or

equivalently an unsigned tableau starting with A. If the tableau closes

oﬀ, then A is logically valid.

2. To test whether B is a logical consequence of A

1

, . . . , A

k

, form a signed

tableau starting with TA

1

, . . . , TA

k

, FB, or equivalently an unsigned

tableau starting with A

1

, . . . , A

k

, B. If the tableau closes oﬀ, then B is

indeed a logical consequence of A

1

, . . . , A

k

.

3. To test A

1

, . . . , A

k

for satisﬁability, form a signed tableau starting with

TA

1

, . . . , TA

k

, or equivalently an unsigned tableau starting with A

1

, . . . , A

k

.

If the tableau closes oﬀ, then A

1

, . . . , A

k

is not satisﬁable. If the tableau

does not close oﬀ, then A

1

, . . . , A

k

is satisﬁable, and from any open path

we can read oﬀ an assignment satisfying A

1

, . . . , A

k

.

The correctness of these tests will be proved in Section 1.5. See Corollaries

1.5.9, 1.5.10, 1.5.11 below.

Example 1.4.6. Using the signed tableau method to test (p ∧q) ⇒(q ∧p) for

logical validity, we have

F(p ∧q) ⇒(q ∧p)

Tp ∧q

Fq ∧p

Tp

Tq

/ ¸

Fq Fp

Since (every path of) the tableau is closed, (p ∧q) ⇒(q ∧p) is logically valid.

Exercises 1.4.7.

15

1. Use a signed tableau to show that (A⇒B) ⇒(A⇒C) is a logical conse-

quence of A⇒(B ⇒C).

Solution.

TA⇒(B⇒C)

F(A⇒B) ⇒(A⇒C)

TA⇒B

FA⇒C

TA

FC

/ ¸

FA TB⇒C

/ ¸

FB TC

/ ¸

FA TB

2. Use a signed tableau to show that A⇒B is logically equivalent to

(B) ⇒(A).

Solution.

F(A⇒B) ⇔((B) ⇒(A))

/

TA⇒B

F(B) ⇒(A)

TB

FA

FB

TA

/ ¸

FA TB

¸

FA⇒B

T(B) ⇒(A)

TA

FB

/ ¸

FB TA

TB FA

3. Use an unsigned tableau to show that A⇒(B ⇒C) is logically equivalent

to (A∧B) ⇒C.

Solution.

((A⇒(B⇒C)) ⇔((A∧B) ⇒C))

/ ¸

A⇒(B ⇒C)

((A∧B) ⇒C)

A∧B

C

A

B

/ ¸

A B⇒C

/ ¸

B C

(A⇒(B ⇒C))

(A∧B) ⇒C

A

(B⇒C)

B

C

/ ¸

(A∧B) C

/ ¸

A B

16

4. Use an unsigned tableau to test (p ∨q) ⇒(p ∧q) for logical validity. If this

formula is not logically valid, use the tableau to ﬁnd all assignments which

falsify it.

Solution.

((p ∨q) ⇒(p ∧q))

p ∨q

(p ∧q)

/ ¸

p q

/ ¸ / ¸

p q p q

1 2 3 4

The open paths 2 and 3 provide the assignments M

2

and M

3

which falsify

our formula. M

2

(p) = T, M

2

(q) = F, M

3

(p) = F, M

3

(q) = T.

5. Redo the previous problem using a signed tableau.

Solution.

F(p ∨q) ⇒(p ∧q)

Tp ∨q

Fp ∧q

/ ¸

Tp Tq

/ ¸ / ¸

Fp Fq Fp Fq

1 2 3 4

The open paths 2 and 3 provide the assignments M

2

and M

3

which falsify

our formula. M

2

(p) = T, M

2

(q) = F, M

3

(p) = F, M

3

(q) = T.

Exercise 1.4.8.

1. Formulate the following argument as a propositional formula.

If it has snowed, it will be poor driving. If it is poor driving, I

will be late unless I start early. Indeed, it has snowed. Therefore,

I must start early to avoid being late.

Solution. Use the following atoms.

s: it has snowed

p: it is poor driving

l: I will be late

e: I start early

17

The argument can be translated as follows: s ⇒p, p ⇒(l ∨e), s, therefore

(l) ⇒e. Written as a single propositional formula, this becomes:

((s ⇒p) ∧(p ⇒(l ∨e)) ∧s) ⇒((l) ⇒e).

2. Use the tableau method to demonstrate that this formula is logically valid.

Solution.

F((s ⇒p) ∧(p ⇒(l ∨e)) ∧s) ⇒((l) ⇒e)

T(s ⇒p) ∧(p ⇒(l ∨e)) ∧s

F(l) ⇒e

Ts ⇒p

T(p ⇒(l ∨e)) ∧s

Tp ⇒(l ∨e)

Ts

Tl

Fe

Fl

/ ¸

Fs Tp

/ ¸

Fp Tl ∨e

/ ¸

Tl Te

1.5 The Completeness Theorem

Let X

1

, . . . , X

k

be a ﬁnite set of signed formulas, or a ﬁnite set of unsigned

formulas.

Lemma 1.5.1 (the Soundness Theorem). If τ is a ﬁnite closed tableau starting

with X

1

, . . . , X

k

, then X

1

, . . . , X

k

is not satisﬁable.

Proof. Straightforward.

Deﬁnition 1.5.2. A path of a tableau is said to be replete if, whenever it

contains the top formula of a tableau rule, it also contains at least one of the

branches. A replete tableau is a tableau in which every path is replete.

Lemma 1.5.3. Any ﬁnite tableau can be extended to a ﬁnite replete tableau.

Proof. Apply tableau rules until they cannot be applied any more.

Deﬁnition 1.5.4. A tableau is said to be open if it is not closed, i.e., it has at

least one open path.

Lemma 1.5.5. Let τ be a replete tableau starting with X

1

, . . . , X

k

. If τ is

open, then X

1

, . . . , X

k

is satisﬁable.

18

In order to prove Lemma 1.5.5, we introduce the following deﬁnition.

Deﬁnition 1.5.6. Let S be a set of signed or unsigned formulas. We say that

S is a Hintikka set if

1. S “obeys the tableau rules”, in the sense that if it contains the top formula

of a rule then it contains at least one of the branches;

2. S contains no pair of conjugate atomic formulas, i.e., Tp, Fp in the signed

case, or p, p in the unsigned case.

Lemma 1.5.7 (Hintikka’s Lemma). If S is a Hintikka set, then S is satisﬁable.

Proof. Deﬁne an assignment M by

M(p) =

_

T if Tp belongs to S

F otherwise

in the signed case, or

M(p) =

_

T if p belongs to S

F otherwise

in the unsigned case. It is not diﬃcult to see that v

M

(X) = T for all X ∈ S.

To prove Lemma 1.5.5, it suﬃces to note that a replete open path is a Hin-

tikka set. Thus, if a replete tableau starting with X

1

, . . . , X

k

is open, Hintikka’s

Lemma implies that X

1

, . . . , X

k

is satisﬁable.

Combining Lemmas 1.5.1 and 1.5.3 and 1.5.5, we obtain:

Theorem 1.5.8 (the Completeness Theorem). X

1

, . . . , X

k

is satisﬁable if and

only if there is no ﬁnite closed tableau starting with X

1

, . . . , X

k

.

Corollary 1.5.9. A

1

, . . . , A

k

is not satisﬁable if and only if there exists a ﬁnite

closed signed tableau starting with TA

1

, . . . , TA

k

, or equivalently a ﬁnite closed

unsigned tableau starting with A

1

, . . . , A

k

.

Corollary 1.5.10. A is logically valid if and only if there exists a ﬁnite closed

signed tableau starting with FA, or equivalently a ﬁnite closed unsigned tableau

starting with A.

Corollary 1.5.11. B is a logical consequence of A

1

, . . . , A

k

if and only if there

exists a ﬁnite closed signed tableau starting with TA

1

, . . . , TA

k

, FB, or equiv-

alently a ﬁnite closed unsigned tableau starting with A

1

, . . . , A

k

, B.

Exercise 1.5.12. Consider the following argument.

The attack will succeed only if the enemy is taken by surprise or the

position is weakly defended. The enemy will not be taken by surprise

unless he is overconﬁdent. The enemy will not be overconﬁdent if the

position is weakly defended. Therefore, the attack will not succeed.

1. Translate the argument into propositional calculus.

2. Use an unsigned tableau to determine whether the argument is logically

valid.

19

1.6 Trees and K¨onig’s Lemma

Up to this point, our discussion of trees has been informal. We now pause to

make our tree terminology precise.

Deﬁnition 1.6.1. A tree consists of

1. a set T

2. a function ℓ : T →N

+

,

3. a binary relation P on T.

The elements of T are called the nodes of the tree. For x ∈ T, ℓ(x) is the level

of x. The relation xPy is read as x immediately precedes y, or y immediately

succeeds x. We require that there is exactly one node x ∈ T such that ℓ(x) = 1,

called the root of the tree. We require that each node other than the root has

exactly one immediate predecessor. We require that ℓ(y) = ℓ(x) + 1 for all

x, y ∈ T such that xPy.

Deﬁnition 1.6.2. A subtree of T is a nonempty set T

′

⊆ T such that for all

y ∈ T

′

and xPy, x ∈ T

′

. Note that T

′

is itself a tree, under the restriction of ℓ

and P to T

′

. Moreover, the root of T

′

is the same as the root of T.

Deﬁnition 1.6.3. An end node of T is a node with no (immediate) successors.

A path in T is a set S ⊆ T such that (1) the root of T belongs to S, (2) for each

x ∈ S, either x is an end node of T or there is exactly one y ∈ S such that xPy.

Deﬁnition 1.6.4. Let P

∗

be the transitive closure of P, i.e., the smallest re-

ﬂexive and transitive relation on T containing P. For x, y ∈ T, we have xP

∗

y

if and only if x precedes y, i.e., y succeeds x, i.e., there exists a ﬁnite sequence

x = x

0

Px

1

Px

2

x

n−1

Px

n

= y. Note that the relation P

∗

is reﬂexive (xP

∗

x

for all x ∈ T), antisymmetric (xP

∗

y and yP

∗

x imply x = y), and transitive

(xP

∗

y and yP

∗

z imply xP

∗

z). Thus P

∗

is a partial ordering of T.

Deﬁnition 1.6.5. T is ﬁnitely branching if each node of T has only ﬁnitely

many immediate successors in T. T is dyadic if each node of T has at most two

immediate successors in T. Note that a dyadic tree is ﬁnitely branching.

Theorem 1.6.6 (K¨ onig’s Lemma). Let T be an inﬁnite, ﬁnitely branching tree.

Then T has an inﬁnite path.

Proof. Let

´

T be the set of all x ∈ T such that x has inﬁnitely many successors

in T. Note that

´

T is a subtree of T. Since T is ﬁnitely branching, it follows by

the pigeonhole principle that each x ∈

´

T has at least one immediate successor

y ∈

´

T. Now deﬁne an inﬁnite path S = ¦x

1

, x

2

, . . . , x

n

, . . .¦ in

´

T inductively by

putting x

1

= the root of T, and x

n+1

= one of the immediate successors of x

n

in

´

T. Clearly S is an inﬁnite path of T.

20

1.7 The Compactness Theorem

Theorem 1.7.1 (the Compactness Theorem, countable case). Let S be a count-

able set of propositional formulas. If each ﬁnite subset of S is satisﬁable, then

S is satisﬁable.

Proof. In brief outline: Form an inﬁnite tableau. Apply K¨onig’s Lemma to get

an inﬁnite path. Apply Hintikka’s Lemma.

Details: Let S = ¦A

1

, A

2

, . . . , A

i

, . . .¦. Start with A

1

and generate a ﬁnite

replete tableau, τ

1

. Since A

1

is satisﬁable, τ

1

has at least one open path. Append

A

2

to each of the open paths of τ

1

, and generate a ﬁnite replete tableau, τ

2

.

Since ¦A

1

, A

2

¦ is satisﬁable, τ

2

has at least one open path. Append A

3

to each

of the open paths of τ

2

, and generate a ﬁnite replete tableau, τ

3

. . . . . Put

τ =

∞

i=1

τ

i

. Thus τ is a replete tableau. Note also that τ is an inﬁnite, ﬁnitely

branching tree. By K¨onig’s Lemma (Theorem 1.6.6), let S

′

be an inﬁnite path

in τ. Then S

′

is a Hintikka set containing S. By Hintikka’s Lemma, S

′

is

satisﬁable. Hence S is satisﬁable.

Theorem 1.7.2 (the Compactness Theorem, uncountable case). Let S be an

uncountable set of propositional formulas. If each ﬁnite subset of S is satisﬁable,

then S is satisﬁable.

Proof. We present three proofs. The ﬁrst uses Zorn’s Lemma. The second uses

transﬁnite induction. The third uses Tychonoﬀ’s Theorem.

Let L be the (necessarily uncountable) propositional language consisting of

all atoms occurring in formulas of S. If S is a set of L-formulas, we say that

S is ﬁnitely satisﬁable if each ﬁnite subset of S is satisﬁable. We are trying to

prove that, if S is ﬁnitely satisﬁable, then S is satisﬁable.

First proof. Consider the partial ordering F of all ﬁnitely satisﬁable sets of

L-formulas which include S, ordered by inclusion. It is easy to see that any

chain in F has a least upper bound in F. Hence, by Zorn’s Lemma, F has a

maximal element, S

∗

. Thus S

∗

is a set of L-formulas, S

∗

⊇ S, S

∗

is ﬁnitely

satisﬁable, and for each L-formula A / ∈ S

∗

, S

∗

∪ ¦A¦ is not ﬁnitely satisﬁable.

From this it is straightforward to verify that S

∗

is a Hintikka set. Hence, by

Hintikka’s Lemma, S

∗

is satisﬁable. Hence S is satisﬁable.

Second proof. Let A

ξ

, ξ < α, be a transﬁnite enumeration of all L-formulas.

By transﬁnite recursion, put S

0

= S, S

ξ+1

= S

ξ

∪ ¦A

ξ

¦ if S

ξ

∪ ¦A

ξ

¦ is ﬁnitely

satisﬁable, S

ξ+1

= S

ξ

otherwise, and S

η

=

ξ<η

S

ξ

for limit ordinals η ≤ α.

Using transﬁnite induction, it is easy to verify that S

ξ

is ﬁnitely satisﬁable for

each ξ ≤ α. In particular, S

α

is ﬁnitely satisﬁable. It is straightforward to

verify that S

α

is a Hintikka set. Hence, by Hintikka’s Lemma, S

α

is satisﬁable.

Hence S is satisﬁable.

Third proof. Let M = ¦T, F¦

L

be the space of all L-assignments M : L →

¦T, F¦. Make M a topological space with the product topology where ¦T, F¦

has the discrete topology. Since ¦T, F¦ is compact, it follows by Tychonoﬀ’s

Theorem that M is compact. For each L-formula A, put M

A

= ¦M ∈ M [

v

M

(A) = T¦. It is easy to check that each M

A

is a topologically closed set

21

in M. If S is ﬁnitely satisﬁable, then the family of sets M

A

, A ∈ S has the

ﬁnite intersection property, i.e.,

A∈S0

M

A

,= ∅ for each ﬁnite S

0

⊆ S. By

compactness of M it follows that

A∈S

M

A

,= ∅. Thus S is satisﬁable.

1.8 Combinatorial Applications

In this section we present some combinatorial applications of the Compactness

Theorem for propositional calculus.

Deﬁnition 1.8.1.

1. A graph consists of a set of vertices together with a speciﬁcation of certain

pairs of vertices as being adjacent. We require that a vertex may not be

adjacent to itself, and that u is adjacent to v if and only if v is adjacent

to u.

2. Let G be a graph and let k be a positive integer. A k-coloring of G is a

function f : ¦vertices of G¦ → ¦c

1

, . . . , c

k

¦ such that f(u) ,= f(v) for all

adjacent pairs of vertices u, v.

3. G is said to be k-colorable if there exists a k-coloring of G. This notion is

much studied in graph theory.

Exercise 1.8.2. Let G be a graph and let k be a positive integer. For each

vertex v and each i = 1, . . . , k, let p

vi

be a propositional atom expressing that

vertex v receives color c

i

. Deﬁne C

k

(G) to be the following set of propositional

formulas: p

v1

∨ ∨p

vk

for each vertex v; (p

vi

∧p

vj

) for each vertex v and

1 ≤ i < j ≤ k; (p

ui

∧p

vi

) for each adjacent pair of vertices u, v and 1 ≤ i ≤ k.

1. Show that there is a one-to-one correspondence between k-colorings of G

and assignments satisfying C

k

(G).

2. Show that G is k-colorable if and only if C

k

(G) is satisﬁable.

3. Show that G is k-colorable if and only if each ﬁnite subgraph of G is

k-colorable.

Deﬁnition 1.8.3. A partial ordering consists of a set P together with a binary

relation ≤

P

such that

1. a ≤

P

a for all a ∈ P (reﬂexivity);

2. a ≤

P

b, b ≤

P

c imply a ≤

P

c (transitivity);

3. a ≤

P

b, b ≤

P

a imply a = b (antisymmetry).

Example 1.8.4. Let P = N

+

= ¦1, 2, 3, . . . , n, . . .¦ = the set of positive inte-

gers.

1. Let ≤

P

be the usual order relation on P, i.e., m ≤

P

n if and only if m ≤ n.

22

2. Let ≤

P

be the divisibility ordering of P, i.e., m ≤

P

n if and only if m is

a divisor of n.

Deﬁnition 1.8.5. Let P, ≤

P

be a partial ordering.

1. Two elements a, b ∈ P are comparable if either a ≤

P

b or b ≤

P

a. Other-

wise they are incomparable.

2. A chain is a set X ⊆ P such that any two elements of X are comparable.

3. An antichain is a set X ⊆ P such that any two distinct elements of X are

incomparable.

Exercise 1.8.6. Let P, ≤

P

be a partial ordering, and let k be a positive integer.

1. Use the Compactness Theorem to show that P is the union of k chains if

and only if each ﬁnite subset of P is the union of k chains.

2. Dilworth’s Theorem says that P is the union of k chains if and only if

every antichain is of size ≤ k. Show that Dilworth’s Theorem for arbi-

trary partial orderings follows from Dilworth’s Theorem for ﬁnite partial

orderings.

23

Chapter 2

Predicate Calculus

2.1 Formulas and Sentences

Deﬁnition 2.1.1 (languages). A language L is a set of predicates, each predi-

cate P of L being designated as n-ary for some nonnegative

1

integer n.

Deﬁnition 2.1.2 (variables and quantiﬁers). We assume the existence of a

ﬁxed, countably inﬁnite set of symbols x, y, z, . . . known as variables. We intro-

duce two new symbols: the universal quantiﬁer (∀) and the existential quantiﬁer

(∃). They are read as “for all” and “there exists”, respectively.

Deﬁnition 2.1.3 (formulas). Let L be a language, and let U be a set. It is

understood that U is disjoint from the set of variables. The set of L-U-formulas

is generated as follows.

1. An atomic L-U-formula is an expression of the form Pe

1

e

n

where P

is an n-ary predicate of L and each of e

1

, . . . , e

n

is either a variable or an

element of U.

2. Each atomic L-U-formula is an L-U-formula.

3. If A is an L-U-formula, then A is an L-U-formula.

4. If A and B are L-U-formulas, then A∧B, A∨B, A⇒B, A⇔B are L-

U-formulas.

5. If x is a variable and A is an L-U-formula, then ∀xA and ∃xA are L-U-

formulas.

Deﬁnition 2.1.4 (degree). The degree of a formula is the number of occurrences

of propositional connectives , ∧, ∨, ⇒, ⇔ and quantiﬁers ∀, ∃ in it.

1

It will be seen that 0-ary predicates behave as propositional atoms. Thus the predicate

calculus is an extension of the propositional calculus.

24

Deﬁnition 2.1.5. An L-formula is an L-∅-formula, i.e., an L-U-formula where

U = ∅, the empty set.

Remark 2.1.6. If U is a subset of U

′

, then every L-U-formula is automati-

cally an L-U

′

-formula. In particular, every L-formula is automatically an L-U-

formula, for any set U.

Deﬁnition 2.1.7. In situations where the language L is understood from con-

text, an L-U-formula may be called a U-formula, and an L-formula a formula.

Deﬁnition 2.1.8 (substitution). If A is an L-U-formula and x is a variable and

a ∈ U, we deﬁne an L-U-formula A[x/a] as follows.

1. If A is atomic, then A[x/a] = the result of replacing each occurrence of x

in A by a.

2. (A)[x/a] = A[x/a].

3. (A∧B)[x/a] = A[x/a] ∧B[x/a].

4. (A∨B)[x/a] = A[x/a] ∨B[x/a].

5. (A⇒B)[x/a] = A[x/a] ⇒B[x/a].

6. (A⇔B)[x/a] = A[x/a] ⇔B[x/a].

7. (∀xA)[x/a] = ∀xA.

8. (∃xA)[x/a] = ∃xA.

9. If y is a variable other than x, then (∀y A)[x/a] = ∀y A[x/a].

10. If y is a variable other than x, then (∃y A)[x/a] = ∃y A[x/a].

Deﬁnition 2.1.9 (free variables). An occurrence of a variable x in an L-U-

formula A is said to be bound in A if it is within the scope of a quantiﬁer ∀x or

∃x in A. An occurrence of a variable x in an L-U-formula A is said to be free

in A if it is not bound in A. A variable x is said to occur freely in A if there is

at least one occurrence of x in A which is free in A.

Exercise 2.1.10.

1. Show that A[x/a] is the result of substituting a for all free occurrences of

x in A.

2. Show that x occurs freely in A if and only if A[x/a] ,= A.

Deﬁnition 2.1.11 (sentences). An L-U-sentence is an L-U-formula in which no

variables occur freely. An L-sentence is an L-∅-sentence, i.e., an L-U-sentence

where U = ∅, the empty set.

25

Remark 2.1.12. If U is a subset of U

′

, then every L-U-sentence is automat-

ically an L-U

′

-sentence. In particular, every L-sentence is automatically an

L-U-sentence, for any set U.

Deﬁnition 2.1.13. In situations where the language L is understood from

context, an L-U-sentence may be called a U-sentence, and an L-sentence a

sentence.

2.2 Structures and Satisﬁability

Deﬁnition 2.2.1. Let U be a nonempty set, and let n be a nonnegative

2

integer.

U

n

is the set of all n-tuples of elements of U, i.e.,

U

n

= ¦¸a

1

, . . . , a

n

) [ a

1

, . . . , a

n

∈ U¦ .

An n-ary relation on U is a subset of U

n

.

Deﬁnition 2.2.2. Let L be a language. An L-structure M consists of a non-

empty set U

M

, called the domain or universe of M, together with an n-ary

relation P

M

on U

M

for each n-ary predicate P of L. An L-structure may be

called a structure, in situations where the language L is understood from context.

Deﬁnition 2.2.3. Two L-structures M and M

′

are said to be isomorphic if

there exists an isomorphism of M onto M

′

, i.e., a one-to-one correspondence

φ : U

M

∼

= U

M

′ such that for all n-ary predicates P of L and all n-tuples

¸a

1

, . . . , a

n

) ∈ (U

M

)

n

, ¸a

1

, . . . , a

n

) ∈ P

M

if and only if ¸φ(a

1

), . . . , φ(a

n

)) ∈ P

M

′ .

As usual in abstract mathematics, we are mainly interested in properties of

structures that are invariant under isomorphism.

Lemma 2.2.4. Given an L-structure M, there is a unique valuation or assign-

ment of truth values

v

M

: ¦A [ A is an L-U

M

-sentence¦ → ¦T, F¦

deﬁned as follows:

1. v

M

(Pa

1

a

n

) =

_

T if ¸a

1

, . . . , a

n

) ∈ P

M

,

F if ¸a

1

, . . . , a

n

) / ∈ P

M

.

2. v

M

(A) =

_

T if v

M

(A) = F,

F if v

M

(A) = T.

3. v

M

(A∧B) =

_

T if v

M

(A) = v

M

(B) = T,

F if at least one of v

M

(A), v

M

(B) = F.

2

In the special case n = 0 we obtain the notion of a 0-ary relation, i.e., a subset of {}.

There are only two 0-ary relations, {} and {}, corresponding to T and F respectively. Thus

a 0-ary predicate behaves as a propositional atom.

26

4. v

M

(A∨B) =

_

T if at least one of v

M

(A), v

M

(B) = T,

F if v

M

(A) = v

M

(B) = F.

5. v

M

(A⇒B) = v

M

((A∧B)) .

6. v

M

(A⇔B) =

_

T if v

M

(A) = v

M

(B) ,

F if v

M

(A) ,= v

M

(B) .

7. v

M

(∀xA) =

_

T if v

M

(A[x/a]) = T for all a ∈ U

M

,

F if v

M

(A[x/a]) = F for at least one a ∈ U

M

.

8. v

M

(∃xA) =

_

T if v

M

(A[x/a]) = T for at least one a ∈ U

M

,

F if v

M

(A[x/a]) = F for all a ∈ U

M

.

Proof. The truth value v

M

(A) is deﬁned by recursion on L-U

M

-sentences, i.e.,

by induction on the degree of A where A is an arbitrary L-U

M

-sentence.

Deﬁnition 2.2.5 (truth and satisfaction). Let M be an L-structure.

1. Let A be an L-U

M

-sentence. We say that A is true in M if v

M

(A) = T.

We say that A is false in M if v

M

(A) = F.

2. Let S be a set of L-U

M

-sentences. We say that M satisﬁes S, abbreviated

M [= S, if all of the sentences of S are true in M.

Theorem 2.2.6.

1. If M and M

′

are isomorphic L-structures and φ : M

∼

= M

′

is an isomor-

phism of M onto M

′

, then for all L-U

M

-sentences A we have v

M

(A) =

v

M

′ (A

′

) where A

′

= A[a

1

/φ(a

1

), . . . , a

k

/φ(a

k

)].

3

Here a

1

, . . . , a

k

are the

elements of U

M

which occur in A.

2. If M and M

′

are isomorphic L-structures, then they are elementarily equiv-

alent, i.e., they satisfy the same L-sentences. We shall see later that the

converse does not hold in general.

Proof. We omit the proof of part 1. A more general result will be proved later

as Theorem 2.7.3. Part 2 follows immediately from part 1.

Deﬁnition 2.2.7 (satisﬁability). Let S be a set of L-sentences. S is said to be

satisﬁable

4

if there exists an L-structure M which satisﬁes S.

3

We have extended the substitution notation 2.1.8 in an obvious way.

4

Similarly, the notions of logical validity and logical consequence are deﬁned for L-

sentences, in the obvious way, using L-structures. An L-sentence is said to be logically valid

if it is satisﬁed by all L-structures. An L-sentence is said to be a logical consequence of S if

it is satisﬁed by all L-structures satisfying S.

27

Remark 2.2.8. Satisﬁability is one of the most important concepts of mathe-

matical logic. A key result known as the Compactness Theorem

5

states that a

set S of L-sentences is satisﬁable if and only every ﬁnite subset of S is satisﬁable.

The following related notion is of technical importance.

Deﬁnition 2.2.9 (satisﬁability in a domain). Let U be a nonempty set. A set

S of L-U-sentences is said to be satisﬁable in the domain U if there exists an

L-structure M such that M [= S and U

M

= U.

Remark 2.2.10. Let S be a set of L-sentences. Then S is satisﬁable (according

to Deﬁnition 2.2.7) if and only if S is satisﬁable in some domain U.

Theorem 2.2.11. Let S be a set of L-sentences. If S is satisﬁable in a domain

U, then S is satisﬁable in any domain of the same cardinality as U.

Proof. Suppose S is satisﬁable in a domain U. Let M be an L-structure M

satisfying S with U

M

= U. Let U

′

be any set of the same cardinality as U.

Then there exists a one-to-one correspondence φ : U → U

′

. Let M

′

be the

L-structure with U

M

′ = U

′

, P

M

′ = ¦¸φ(a

1

), . . . , φ(a

n

)) [ ¸a

1

, . . . , a

n

) ∈ P

M

¦ for

all n-ary predicates P of L. Then M is isomorphic to M

′

. Hence, by Theorem

2.2.6, M

′

[= S. Thus S is satisﬁable in the domain U

′

.

Example 2.2.12. We exhibit a sentence A

∞

which is satisﬁable in an inﬁnite

domain but not in any ﬁnite domain. Our sentence A

∞

is (1) ∧(2) ∧(3) with

(1) ∀x∀y ∀z ((Rxy ∧Ryz) ⇒Rxz)

(2) ∀x∀y (Rxy ⇒Ryx)

(3) ∀x∃y Rxy

See also Example 2.5.9.

Exercise 2.2.13. Let L be the language consisting of one binary predicate, R.

Consider the following sentences:

(a) ∀xRxx

(b) ∀xRxx

(c) ∀x∀y (Rxy ⇒Ryx)

(d) ∀x∀y (Rxy ⇒Ryx)

(e) ∀x∀y ∀z ((Rxy ∧Ryz) ⇒Rxz)

(f) ∀x∃y Rxy

Which of subsets of this set of sentences are satisﬁable? Verify your claims by

exhibiting appropriate structures. Use the simplest possible structures.

5

See Theorems 2.6.1 and 2.6.2 below.

28

Solution.

(a,c,e,f) is satisﬁable: U = ¦1¦, R = ¦¸1, 1)¦.

(b,c,d,e) is satisﬁable: U = ¦1¦, R = ¦ ¦.

(b,c,f) is satisﬁable: U = ¦1, 2¦, R = ¦¸1, 2), ¸2, 1)¦.

(b,d,e,f) is satisﬁable: U = ¦1, 2, 3, . . .¦, R = <.

These are the only maximal satisﬁable sets, because:

(a,b) is not satisﬁable.

(a,d) is not satisﬁable.

(b,c,e,f) is not satisﬁable.

(c,d,f) is not satisﬁable.

Note: (d,e,f) is not satisﬁable in any ﬁnite domain.

Exercise 2.2.14.

1. Assume the following predicates:

Hx: x is a human

Cx: x is a car

Tx: x is a truck

Dxy: x drives y

Write formulas representing the obvious assumptions: no human is a car,

no car is a truck, humans exist, cars exist, only humans drive, only cars

and trucks are driven, etc.

2. Write formulas representing the following statements:

(a) Everybody drives a car or a truck.

(b) Some people drive both.

(c) Some people don’t drive either.

(d) Nobody drives both.

3. Assume in addition the following predicate:

Ixy: x is identical to y

Write formulas representing the following statements:

(a) Every car has at most one driver.

(b) Every truck has exactly two drivers.

29

(c) Everybody drives exactly one vehicle (car or truck).

Solution.

1. No human is a car. ∃x(Hx∧Cx).

No car is a truck. ∃x(Cx∧Tx).

Humans exist. ∃xHx.

Cars exist. ∃xCx.

Only humans drive. ∀x((∃y Dxy) ⇒Hx).

Only cars and trucks are driven. ∀x((∃y Dyx) ⇒(Cx∨Tx)).

Some humans drive. ∃x(Hx∧∃y Dxy).

Some humans do not drive. ∃x(Hx∧∃y Dxy).

Some cars are driven. ∃x(Cx∧∃y Dyx).

Some cars are not driven (e.g., old wrecks). ∃x(Cx∧∃y Dyx).

etc, etc.

2. (a) ∀x(Hx⇒∃y (Dxy ∧(Cy ∨Ty))).

(b) ∃x(Hx∧∃y ∃z (Dxy ∧Cy ∧Dxz ∧Tz)).

(c) ∃x(Hx∧∃y (Dxy ∧(Cy ∨Ty))).

(d) ∃x(Hx∧∃y ∃z (Dxy ∧Dxz ∧Cy ∧Tz)).

3. (a) ∀x(Cx⇒∀y ∀z ((Dyx∧Dzx) ⇒Iyz)).

(b) ∀x(Tx⇒∃y ∃z ((Iyz) ∧Dyx∧Dzx∧∀w(Dwx⇒(Iwy ∨Iwz)))).

(c) ∀x(Hx⇒∃y (Dxy ∧(Cy ∨Ty) ∧∀z ((Dxz ∧(Cz ∨Tz)) ⇒Iyz))).

Exercise 2.2.15. Assume the following predicates:

Ixy: x = y

Pxyz: x y = z

Write formulas representing the axioms for a group: axioms for equality, ex-

istence and uniqueness of products, associative law, existence of an identity

element, existence of inverses.

Solution.

1. equality axioms:

(a) ∀xIxx (reﬂexivity)

(b) ∀x∀y (Ixy ⇔Iyx) (symmetry)

(c) ∀x∀y ∀z ((Ixy ∧Iyz) ⇒Ixz) (transitivity)

(d) ∀x∀x

′

∀y ∀y

′

∀z ∀z

′

((Ixx

′

∧Iyy

′

∧Izz

′

) ⇒(Pxyz ⇔Px

′

y

′

z

′

))

(congruence with respect to P).

30

2. existence and uniqueness of products:

(a) ∀x∀y ∃z Pxyz (existence)

(b) ∀x∀y ∀z ∀w((Pxyz ∧Pxyw) ⇒Izw) (uniqueness).

3. associative law:

∀x∀y ∀z ∃u ∃v ∃w(Pxyu ∧Pyzv ∧Puzw∧Pxvw).

4. existence of identity element:

∃u ∀x(Puxx∧Pxux).

5. existence of inverses:

∃u ∀x∃y (Puxx∧Pxux∧Pxyu ∧Pyxu).

Exercise 2.2.16. Let G be a group. The order of an element a ∈ G is the

smallest positive integer n such that a

n

= e. Here e denotes the identity element

of G, and

a

n

= a a

. ¸¸ .

n times

.

Using only the predicates Pxyz (“x y = z”) and Ixy (“x = y”), write a

predicate calculus sentence S such that, for any group G, G satisﬁes S if and

only if G has no elements of order 2 or 3.

Solution. ∃x∃y (Pxxx∧(Pyyy) ∧(Pyyx∨∃z (Pyyz ∧Pzyx))).

2.3 The Tableau Method

Deﬁnition 2.3.1. Fix a countably inﬁnite set V = ¦a

1

, a

2

, . . . , a

n

, . . .¦ =

¦a, b, c, . . .¦. The elements of V will be called parameters. If L is a language,

L-V -sentences will be called sentences with parameters.

Deﬁnition 2.3.2. A (signed or unsigned) tableau is a rooted dyadic tree where

each node carries a (signed or unsigned) L-V -sentence. The tableau rules for

the predicate calculus are the same as those for the propositional calculus, with

the following additional rules.

Signed:

.

.

.

T∀xA

.

.

.

[

TA[x/a]

.

.

.

F∃xA

.

.

.

[

FA[x/a]

where a is an arbitrary parameter

31

.

.

.

T∃xA

.

.

.

[

TA[x/a]

.

.

.

F∀xA

.

.

.

[

FA[x/a]

where a is a new parameter

Unsigned:

.

.

.

∀xA

.

.

.

[

A[x/a]

.

.

.

∃xA

.

.

.

[

A[x/a]

where a is an arbitrary parameter

.

.

.

∃xA

.

.

.

[

A[x/a]

.

.

.

∀xA

.

.

.

[

A[x/a]

where a is a new parameter

Remark 2.3.3. In the above tableau rules, “a is new” means that a does not

occur in the path that is being extended. Or, we can insist that a not occur in

the tableau that is being extended.

Remark 2.3.4. We are going to prove that the tableau method for predicate

calculus is sound (Theorem 2.3.13) and complete (Theorem 2.5.5). In particular,

a sentence A of the predicate calculus is logically valid if and only if there exists

a ﬁnite closed signed tableau starting with FA, or equivalently a ﬁnite closed

unsigned tableau starting with A.

32

Example 2.3.5. The signed tableau

F(∃x∀y Rxy) ⇒(∀y ∃xRxy)

T∃x∀y Rxy

F∀y ∃xRxy

T∀y Ray

F∃xRxb

TRab

FRab

is closed. Therefore, by the Soundness Theorem, (∃x∀y Rxy) ⇒(∀y ∃xRxy) is

logically valid.

Example 2.3.6. The unsigned tableau

((∃x(Px∨Qx)) ⇔((∃xPx) ∨(∃xQx)))

/ ¸

∃x(Px∨Qx)

((∃xPx) ∨(∃xQx))

∃xPx

∃xQx

Pa ∨Qa

/ ¸

Pa Qa

Pa Qa

∃x(Px∨Qx)

(∃xPx) ∨(∃xQx)

/ ¸

∃xPx

Pb

(Pb ∨Qb)

Pb

Qb

∃xQx

Qc

(Pc ∨Qc)

Pc

Qc

is closed. Therefore, by the Soundness Theorem,

(∃x(Px∨Qx)) ⇔((∃xPx) ∨(∃xQx))

is logically valid.

Exercises 2.3.7. Use signed tableaux to show that the following are logically

valid.

1. (∀x(A⇒B)) ⇒((∀xA) ⇒(∀xB))

Solution.

F(∀x(A⇒B)) ⇒((∀xA) ⇒(∀xB))

T ∀x(A⇒B)

F(∀xA) ⇒(∀xB)

T ∀xA

F ∀xB

FB[x/a]

TA[x/a]

T(A⇒B)[x/a]

/ ¸

FA[x/a] TB[x/a]

33

2. (∃x(A∨B)) ⇔((∃xA) ∨(∃xB))

Solution.

F(∃x(A∨B)) ⇔((∃xA) ∨(∃xB))

/ ¸

T ∃x(A∨B)

F(∃xA) ∨(∃xB)

T(A∨B)[x/a]

F ∃xA

F ∃xB

FA[x/a]

FB[x/a]

/ ¸

TA[x/a] TB[x/a]

F ∃x(A∨B)

T(∃xA) ∨(∃xB)

/ ¸

T ∃xA T∃xB

TA[x/a] TB[x/a]

F(A∨B)[x/a] F(A∨B)[x/a]

FA[x/a] FA[x/a]

FB[x/a] FB[x/a]

3. (∃xA) ⇔(∀xA)

Solution.

F(∃xA) ⇔(∀xA)

/ ¸

T ∃xA F ∃xA

F∀xA T∀xA

TA[x/a] F ∀xA

T ∀xA F(A)[x/a]

T(A)[x/a] TA[x/a]

FA[x/a] FA[x/a]

4. (∀x(A∨C)) ⇔((∀xA) ∨C), provided x is not free in C

Solution.

F(∀x(A∨C)) ⇔((∀xA) ∨C)

/ ¸

T ∀x(A∨C)

F(∀xA) ∨C

F ∀xA

FC

FA[x/a]

T(A∨C)[x/a]

/ ¸

TA[x/a] TC

F ∀x(A∨C)

T(∀xA) ∨C

F(A∨C)[x/a]

FA[x/a]

FC

/ ¸

T ∀xA TC

TA[x/a]

Exercise 2.3.8. Using the predicates Bx (“x is a barber in Podunk”), Cx

(“x is a citizen of Podunk”), and Sxy (“x shaves y”), translate the following

argument into a sentence of the predicate calculus.

34

There is a barber in Podunk who shaves exactly those citizens of

Podunk who do not shave themselves. Therefore, there is a barber

in Podunk who is not a citizen of Podunk.

Use an unsigned tableau to test this argument for logical validity.

Solution. (∃x(Bx∧∀y (Cy ⇒(Sxy ⇔Syy)))) ⇒∃x(Bx∧Cx). A tableau

starting with the negation of this sentence (left to the reader) closes oﬀ to show

that the sentence is logically valid.

Exercise 2.3.9. Translate the following argument into the predicate calculus,

and use appropriate methods to establish its validity or invalidity.

Anyone who can solve all logic problems is a good student. No

student can solve every logic problem. Therefore, there are logic

problems that no student can solve.

Exercise 2.3.10. Use an unsigned tableau to show that ∃x(Px⇔∀y Py) is

logically valid.

The rest of this section is devoted to proving the Soundness Theorem 2.3.13.

Deﬁnition 2.3.11.

1. An L-V -structure consists of an L-structure M together with a mapping

φ : V → U

M

. If A is an L-V -sentence, we write

A

φ

= A[a

1

/φ(a

1

), . . . , a

k

/φ(a

k

)]

where a

1

, . . . , a

k

are the parameters occurring in A. Note that A

φ

is an

L-U

M

-sentence. Note also that, if A is an L-sentence, then A

φ

= A.

2. Let S be a ﬁnite or countable set of (signed or unsigned) L-V -sentences.

An L-V -structure M, φ is said to satisfy S if v

M

(A

φ

) = T for all A ∈ S.

S is said to be satisﬁable

6

if there exists an L-V -structure satisfying S.

Note that this deﬁnition is compatible with Deﬁnition 2.2.7.

3. Let τ be an L-tableau. We say that τ is satisﬁable if at least one path of

τ is satisﬁable.

Lemma 2.3.12. Let τ and τ

′

be tableaux such that τ

′

is an immediate extension

of τ, i.e., τ

′

is obtained from τ by applying a tableau rule to a path of τ. If τ

is satisﬁable, then τ

′

is satisﬁable.

Proof. The proof consists of one case for each tableau rule. We consider some

representative cases.

6

Similarly, the notions of logical validity and logical consequence are extended to L-V -

sentences, in the obvious way, using L-V -structures. An L-V -sentence is said to be logically

valid if it satisﬁed by all L-V -structures. An L-V -sentence is said to be a logical consequence

of S if it is satisﬁed by all L-V -structures satisfying S.

35

Case 1. Suppose that τ

′

is obtained from τ by applying the rule

.

.

.

A∨B

.

.

.

/ ¸

A B

to the path θ in τ. Since τ is satisﬁable, it has at least one satisﬁable path, S. If

S ,= θ, then S is a path of τ

′

, so τ

′

is satisﬁable. If S = θ, then θ is satisﬁable, so

let M and φ : V → U

M

satisfy θ. In particular v

M

((A∨B)

φ

) = T, so we have

at least one of v

M

(A

φ

) = T and v

M

(B

φ

) = T. Thus M and φ satisfy at least

one of θ, A and θ, B. Since these are paths of τ

′

, it follows that τ

′

is satisﬁable.

Case 2. Suppose that τ

′

is obtained from τ by applying the rule

.

.

.

∀xA

.

.

.

[

A[x/a]

to the path θ in τ, where a is a parameter. Since τ is satisﬁable, it has at least

one satisﬁable path, S. If S ,= θ, then S is a path of τ

′

, so τ

′

is satisﬁable. If

S = θ, then θ is satisﬁable, so let M and φ : V → U

M

satisfy θ. In particular

v

M

(∀x(A

φ

)) = v

M

((∀xA)

φ

) = T, so v

M

(A

φ

[x/c]) = T for all c ∈ U

M

. In

particular

v

M

(A[x/a]

φ

) = v

M

(A

φ

[x/φ(a)]) = T.

Thus M and φ satisfy θ, A[x/a]. Since this is a path of τ

′

, it follows that τ

′

is

satisﬁable.

Case 3. Suppose that τ

′

is obtained from τ by applying the rule

.

.

.

∃xA

.

.

.

[

A[x/a]

to the path θ in τ, where a is a new parameter. Since τ is satisﬁable, it has at

least one satisﬁable path, S. If S ,= θ, then S is a path of τ

′

, so τ

′

is satisﬁable.

If S = θ, then θ is satisﬁable, so let M and φ : V → U

M

satisfy θ. In particular

v

M

(∃x(A

φ

)) = v

M

((∃xA)

φ

) = T, so v

M

(A

φ

[x/c]) = T for at least one c ∈ U

M

.

Fix such a c and deﬁne φ

′

: V → U

M

by putting φ

′

(a) = c, and φ

′

(b) = φ(b) for

all b ,= a, b ∈ V . Since a is new, we have B

φ

′

= B

φ

for all B ∈ θ, and A

φ

′

= A

φ

,

36

hence A[x/a]

φ

′

= A

φ

′

[x/φ

′

(a)] = A

φ

[x/c]. Thus v

M

(B

φ

′

) = v

M

(B

φ

) = T for all

B ∈ θ, and v

M

(A[x/a]

φ

′

) = v

M

(A

φ

[x/c]) = T. Thus M and φ

′

satisfy θ, A[x/a].

Since this is a path of τ

′

, it follows that τ

′

is satisﬁable.

Theorem 2.3.13 (the Soundness Theorem). Let X

1

, . . . , X

k

be a ﬁnite set of

(signed or unsigned) sentences with parameters. If there exists a ﬁnite closed

tableau starting with X

1

, . . . , X

k

, then X

1

, . . . , X

k

is not satisﬁable.

Proof. Let τ be a closed tableau starting with X

1

, . . . , X

k

. Thus there is a ﬁnite

sequence of tableaux τ

0

, τ

1

, . . . , τ

n

= τ such that

τ

0

=

X

1

.

.

.

X

k

and each τ

i+1

is an immediate extension of τ

i

. Suppose X

1

, . . . , X

k

is satisﬁable.

Then τ

0

is satisﬁable, and by induction on i using Lemma 2.3.12, it follows that

all of the τ

i

are satisﬁable. In particular τ

n

= τ is satisﬁable, but this is

impossible since τ is closed.

2.4 Logical Equivalence

Deﬁnition 2.4.1. Given a formula A, let A

′

= A[x

1

/a

1

, . . . , x

k

/a

k

], where

x

1

, . . . , x

k

are the variables which occur freely in A, and a

1

, . . . , a

k

are parame-

ters not occurring in A. Note that A

′

has no free variables, i.e., it is a sentence.

We deﬁne A to be satisﬁable if and only if A

′

is satisﬁable, in the sense of Deﬁ-

nition 2.3.11. We deﬁne A to be logically valid if and only if A

′

is logically valid,

in the sense of Deﬁnition 2.3.11.

Exercises 2.4.2. Let A be a formula.

1. Show that A is logically valid if and only if A is not satisﬁable. Show

that A is satisﬁable if and only if A is not logically valid.

2. Let x be a variable. Show that A is logically valid if and only if ∀xA is

logically valid. Show that A is satisﬁable if and only if ∃xA is satisﬁable.

3. Let x be a variable, and let a be a parameter not occurring in A. Show

that A is logically valid if and only if A[x/a] is logically valid. Show that

A is satisﬁable if and only if A[x/a] is satisﬁable.

Deﬁnition 2.4.3. Let A and B be formulas. A and B are said to be logically

equivalent, written A ≡ B, if A⇔B is logically valid.

Exercise 2.4.4. Assume A ≡ B. Show that for any variable x, ∀xA ≡ ∀xB

and ∃xA ≡ ∃xB. Show that for any variable x and parameter a, A[x/a] ≡

B[x/a].

37

Exercise 2.4.5. For a formula A, it is not in general true that A ≡ A

′

, where

A

′

is as in Deﬁnition 2.4.1. Also, it is not in general true that A ≡ ∀xA, or that

A ≡ ∃xA, or that A ≡ A[x/a]. Give examples illustrating these remarks.

Exercise 2.4.6. If A and B are formulas, put A

′

= A[x

1

/a

1

, . . . , x

k

/a

k

] and

B

′

= B[x

1

/a

1

, . . . , x

k

/a

k

], where x

1

, . . . , x

k

are the variables occurring freely in

A and B, and a

1

, . . . , a

k

are parameters not occurring in A or in B. Show that

A ≡ B if only if A

′

≡ B

′

.

Remark 2.4.7. The results of Exercises 1.3.3 and 1.3.4 and Remark 1.3.5 for

formulas of the propositional calculus, also hold for formulas of the predicate

calculus. In particular, if A

1

≡ A

2

, then for any formula C containing A

1

as

a part, if we replace one or more occurrences of the part A

1

by A

2

, then the

resulting formula is logically equivalent to C.

Remark 2.4.8. Some useful logical equivalences are:

1. (a) ∀xA ≡ A, provided x does not occur freely in A

(b) ∃xA ≡ A, provided x does not occur freely in A

(c) ∀xA ≡ ∀y A[x/y], provided y does not occur in A

(d) ∃xA ≡ ∃y A[x/y], provided y does not occur in A

Note that the last two equivalences provide for “change of bound vari-

ables”. In this way, we can convert any formula into a logically equivalent

formula where no variable occurs both free and bound, and each bound

variable is bound by at most one quantiﬁer.

2. (a) ∀x(A∧B) ≡ (∀xA) ∧(∀xB)

(b) ∃x(A∨B) ≡ (∃xA) ∨(∃xB)

(c) ∃x(A⇒B) ≡ (∀xA) ⇒(∃xB)

Note however that, in general, ∃x(A∧B) ,≡ (∃xA) ∧(∃xB), and

∀x(A∨B) ,≡ (∀xA) ∨(∀xB), and ∀x(A⇒B) ,≡ (∃xA) ⇒(∀xB).

On the other hand, we have:

3. (a) ∃x(A∧B) ≡ A∧(∃xB), provided x does not occur freely in A

(b) ∃x(A∧B) ≡ (∃xA) ∧B, provided x does not occur freely in B

(c) ∀x(A∨B) ≡ A∨(∀xB), provided x does not occur freely in A

(d) ∀x(A∨B) ≡ (∀xA) ∨B, provided x does not occur freely in B

(e) ∃x(A⇒B) ≡ A⇒(∃xB), provided x does not occur freely in A

(f) ∀x(A⇒B) ≡ A⇒(∀xB), provided x does not occur freely in A

(g) ∃x(A⇒B) ≡ (∀xA) ⇒B, provided x does not occur freely in B

(h) ∀x(A⇒B) ≡ (∃xA) ⇒B, provided x does not occur freely in B

4. (a) ∃xA ≡ ∀xA

38

(b) ∀xA ≡ ∃xA

(c) ∀xA ≡ ∃xA

(d) ∃xA ≡ ∀xA

Deﬁnition 2.4.9 (variants). Let A be a formula. A variant of A is any formula

´

A obtained from A by replacing all the bound variables of A by distinct new

variables. Note that

´

A has the same free variables as A and is logically equivalent

to A, in view of Remark 2.4.8, parts 1(c) and 1(d).

Example 2.4.10. Let A be ∀x∃y Rxyz, and let

´

A be ∀u ∃v Ruvz. Then A and

´

A are variants of each other, hence logically equivalent to each other.

Deﬁnition 2.4.11 (prenex form). A formula is said to be quantiﬁer-free if it

contains no quantiﬁers. A formula is said to be in prenex form if it is of the form

Q

1

x

1

Q

n

x

n

B, where each Q

i

is a quantiﬁer (∀ or ∃), each x

i

is a variable,

and B is quantiﬁer-free.

Example 2.4.12. The sentence

∀x∀y ∃z ∀w(Rxy ⇒(Rxz ∧Rzy ∧(Rzw∧Rwy)))

is in prenex form.

Exercise 2.4.13. Show that every formula is logically equivalent to a formula

in prenex form. (Hint: Use the equivalences of Remark 2.4.8. First replace the

given formula by a variant in which each bound variable is quantiﬁed only once

and does not occur freely. Then use parts 3 and 4 to move all quantiﬁers to the

front.)

Example 2.4.14. Consider the sentence (∃xPx) ∧(∃xQx). We wish to put

this into prenex form. Applying the equivalences of Remark 2.4.8, we have

(∃xPx) ∧(∃xQx) ≡ (∃xPx) ∧(∃y Qy)

≡ ∃x(Px∧(∃y Qy))

≡ ∃x∃y (Px∧Qy)

and this is in prenex form.

Exercise 2.4.15. Find a sentence in prenex normal form which is logically

equivalent to (∀x∃y Rxy) ⇒∃xPx.

Solution. ∃x∀y ∀z (Rxy ⇒Pz).

Exercises 2.4.16. Let A and B be quantiﬁer-free formulas. Put the following

into prenex form.

1. (∃xA) ∧(∃xB)

2. (∀xA) ⇔(∀xB)

39

3. (∀xA) ⇔(∃xB)

Deﬁnition 2.4.17 (universal closure). Let A be a formula. The universal clo-

sure of A is the sentence A

∗

= ∀x

1

∀x

k

A, where x

1

, . . . , x

k

are the variables

which occur freely in A. Note that A

∗∗

≡ A

∗

.

Exercises 2.4.18. Let A be a formula.

1. Show that A is logically valid if and only if A

∗

, the universal closure of A,

is logically valid.

2. It is not true in general that A ≡ A

∗

. Give an example illustrating this.

3. It is not true in general that A is satisﬁable if and only if A

∗

is satisﬁable.

Give an example illustrating this.

4. For formulas A and B, it is not true in general that A ≡ B if and only if

A

∗

≡ B

∗

. Give an example illustrating this.

For completeness we state the following deﬁnition.

Deﬁnition 2.4.19. Let A

1

, . . . , A

k

, B be formulas. We say that B is a logi-

cal consequence of A

1

, . . . , A

k

if (A

1

∧ ∧A

k

) ⇒B is logically valid. This is

equivalent to saying that the universal closure of (A

1

∧ ∧A

k

) ⇒B is logically

valid.

Remark 2.4.20. A and B are logically equivalent if and only if each is a

logical consequence of the other. A is logically valid if and only if A is a logical

consequence of the empty set. ∃xA is a logical consequence of A[x/a], but the

converse does not hold in general. A[x/a] is a logical consequence of ∀xA, but

the converse does not hold in general.

2.5 The Completeness Theorem

Let U be a nonempty set, and let S be a set of (signed or unsigned) L-U-

sentences.

Deﬁnition 2.5.1. S is closed if S contains a conjugate pair of L-U-sentences.

In other words, for some L-U-sentence A, S contains TA, FA in the signed

case, A, A in the unsigned case. S is open if it is not closed.

Deﬁnition 2.5.2. S is U-replete if S “obeys the tableau rules” with respect to

U. We list some representative clauses of the deﬁnition.

1. If S contains TA, then S contains FA. If S contains FA, then S

contains TA. If S contains A, then S contains A.

2. If S contains TA∧B, then S contains both TA and TB. If S contains

FA∧B, then S contains at least one of FA and FB. If S contains A∧B,

then S contains both A and B. If S contains (A∧B), then S contains

at least one of A and B.

40

3. If S contains T∃xA, then S contains TA[x/a] for at least one a ∈ U. If

S contains F∃xA, then S contains FA[x/a] for all a ∈ U. If S contains

∃xA, then S contains A[x/a] for at least one a ∈ U. If S contains ∃xA,

then S contains A[x/a] for all a ∈ U.

4. If S contains T∀xA, then S contains TA[x/a] for all a ∈ U. If S contains

F∀xA, then S contains FA[x/a] for at least one a ∈ U. If S contains

∀xA, then S contains A[x/a] for all a ∈ U. If S contains ∀xA, then S

contains A[x/a] for at least one a ∈ U.

Lemma 2.5.3 (Hintikka’s Lemma). If S is U-replete and open

7

, then S is

satisﬁable. In fact, S is satisﬁable in the domain U.

Proof. Assume S is U-replete and open. We deﬁne an L-structure M by putting

U

M

= U and, for each n-ary predicate P of L,

P

M

= ¦¸a

1

, . . . , a

n

) ∈ U

n

[ TPa

1

a

n

∈ S¦

in the signed case, and

P

M

= ¦¸a

1

, . . . , a

n

) ∈ U

n

[ Pa

1

a

n

∈ S¦

in the unsigned case.

We claim that for all L-U-sentences A,

(a) if S contains TA, then v

M

(A) = T

(b) if S contains FA, then v

M

(A) = F

in the signed case, and

(c) if S contains A, then v

M

(A) = T

(d) if S contains A, then v

M

(A) = F

in the unsigned case.

In both cases, the claim is easily proved by induction on the degree of A.

We give the proof for some representative cases.

1. deg(A) = 0. In this case A is atomic, say A = Pa

1

a

n

.

(a) If S contains TPa

1

a

n

, then by deﬁnition of M we have (a

1

, . . . , a

n

) ∈

P

M

, so v

M

(Pa

1

a

n

) = T.

(b) If S contains FPa

1

a

n

, then S does not contain TPa

1

a

n

since

S is open. Thus by deﬁnition of M we have (a

1

, . . . , a

n

) / ∈ P

M

, so

v

M

(Pa

1

a

n

) = F.

(c) If S contains Pa

1

a

n

, then by deﬁnition of M we have (a

1

, . . . , a

n

) ∈

P

M

, so v

M

(Pa

1

a

n

) = T.

7

See also Exercise 2.5.7.

41

(d) If S contains Pa

1

a

n

, then S does not contain Pa

1

a

n

since

S is open. Thus by deﬁnition of M we have (a

1

, . . . , a

n

) / ∈ P

M

, so

v

M

(Pa

1

a

n

) = F.

2. deg(A) > 0 and A = B. Note that deg(B) < deg(A) so the inductive

hypothesis applies to B.

3. deg(A) > 0 and A = B∧C. Note that deg(B) and deg(C) are < deg(A)

so the inductive hypothesis applies to B and C.

(a) If S contains TB∧C, then by repleteness of S we see that S contains

both TB and TC. Hence by inductive hypothesis we have v

M

(B) =

v

M

(C) = T. Hence v

M

(B∧C) = T.

(b) If S contains FB∧C, then by repleteness of S we see that S contains

at least one of FB and FC. Hence by inductive hypothesis we have

at least one of v

M

(B) = F and v

M

(C) = F. Hence v

M

(B ∧C) = F.

(c) If S contains B∧C, then by repleteness of S we see that S contains

both B and C. Hence by inductive hypothesis we have v

M

(B) =

v

M

(C) = T. Hence v

M

(B∧C) = T.

(d) If S contains (B∧C), then by repleteness of S we see that S con-

tains at least one of B and C. Hence by inductive hypothe-

sis we have at least one of v

M

(B) = F and v

M

(C) = F. Hence

v

M

(B∧C) = F.

4. deg(A) > 0 and A = ∃xB. Note that for all a ∈ U we have deg(B[x/a]) <

deg(A), so the inductive hypothesis applies to B[x/a].

5. deg(A) > 0 and A = ∀xB. Note that for all a ∈ U we have deg(B[x/a]) <

deg(A), so the inductive hypothesis applies to B[x/a].

**We shall now use Hintikka’s Lemma to prove the completeness of the tableau
**

method. As in Section 2.3, Let V = ¦a

1

, . . . , a

n

, . . .¦ be the set of parameters.

Recall that a tableau is a tree whose nodes carry L-V -sentences.

Lemma 2.5.4. Let τ

0

be a ﬁnite tableau. By applying tableau rules, we can

extend τ

0

to a (possibly inﬁnite) tableau τ with the following properties: every

closed path of τ is ﬁnite, and every open path of τ is V -replete.

Proof. The idea is to start with τ

0

and use tableau rules to construct a sequence

of ﬁnite extensions τ

0

, τ

1

, . . . , τ

i

, . . .. If some τ

i

is closed, then the construction

halts, i.e., τ

j

= τ

i

for all j ≥ i, and we set τ = τ

i

. In any case, we set

τ = τ

∞

=

∞

i=0

τ

i

. In the course of the construction, we apply tableau rules

systematically to ensure that τ

∞

will have the desired properties, using the fact

that V = ¦a

1

, a

2

, . . . , a

n

, . . .¦ is countably inﬁnite.

Here are the details of the construction. Call a node X of τ

i

quasiuniversal

if it is of the form T∀xA or F∃xA or ∀xA or ∃xA. Our construction begins

42

with τ

0

. Suppose we have constructed τ

2i

. For each quasiuniversal node X of

τ

2i

and each n ≤ 2i, apply the appropriate tableau rule to extend each open

path of τ

2i

containing X by TA[x/a

n

] or FA[x/a

n

] or A[x/a

n

] or A[x/a

n

] as

the case may be. Let τ

2i+1

be the ﬁnite tableau so obtained. Next, for each

non-quasiuniversal node X of τ

2i+1

, extend each open path containing X by

applying the appropriate tableau rule. Again, let τ

2i+2

be the ﬁnite tableau so

obtained.

In this construction, a closed path is never extended, so all closed paths of

τ

∞

are ﬁnite. In addition, the construction ensures that each open path of τ

∞

is V -replete. Thus τ

∞

has the desired properties. This proves our lemma.

Theorem 2.5.5 (the Completeness Theorem). Let X

1

, . . . , X

k

be a ﬁnite set of

(signed or unsigned) sentences with parameters. If X

1

, . . . , X

k

is not satisﬁable,

then there exists a ﬁnite closed tableau starting with X

1

, . . . , X

k

. If X

1

, . . . , X

k

is satisﬁable, then X

1

, . . . , X

k

is satisﬁable in the domain V .

Proof. By Lemma 2.5.4 there exists a (possibly inﬁnite) tableau τ starting with

X

1

, . . . , X

k

such that every closed path of τ is ﬁnite, and every open path of τ

is V -replete. If τ is closed, then by K¨onig’s Lemma (Theorem 1.6.6), τ is ﬁnite.

If τ is open, let S be an open path of τ. Then S is V -replete. By Hintikka’s

Lemma 2.5.3, S is satisﬁable in V . Hence X

1

, . . . , X

k

is satisﬁable in V .

Deﬁnition 2.5.6. Let L, U, and S be as in Deﬁnition 2.5.1. S is said to

be atomically closed if S contains a conjugate pair of atomic L-U-sentences.

In other words, for some n-ary L-predicate P and a

1

, . . . , a

n

∈ U, S contains

TPa

1

a

n

, FPa

1

a

n

in the signed case, and Pa

1

a

n

, Pa

1

a

n

in the

unsigned case. S is atomically open if it is not atomically closed.

Exercise 2.5.7. Show that Lemmas 2.5.3 and 2.5.4 and Theorem 2.5.5 continue

to hold with “closed” (“open”) replaced by “atomically closed” (“atomically

open”).

Remark 2.5.8. Corollaries 1.5.9, 1.5.10, 1.5.11 carry over from the proposi-

tional calculus to the predicate calculus. In particular, the tableau method

provides a test for logical validity of sentences of the predicate calculus.

Note however that the test is only partially eﬀective. If a sentence A is

logically valid, we will certainly ﬁnd a ﬁnite closed tableau starting with A.

But if A is not logically valid, we will not necessarily ﬁnd a ﬁnite tableau which

demonstrates this. See the following example.

Example 2.5.9. In 2.2.12 we have seen an example of a sentence A

∞

which

is satisﬁable in a countably inﬁnite domain but not in any ﬁnite domain. It is

43

instructive to generate a tableau starting with A

∞

.

A

∞

.

.

.

∀x∀y ∀z ((Rxy ∧Ryz) ⇒Rxz)

∀x∀y (Rxy ⇒Ryx)

∀x∃y Rxy

∃y Ra

1

y

Ra

1

a

2

∀y (Ra

1

y ⇒Rya

1

)

Ra

1

a

2

⇒Ra

2

a

1

/ ¸

Ra

1

a

2

Ra

2

a

1

∃y Ra

2

y

Ra

2

a

3

.

.

.

Ra

3

a

2

∀y ∀z ((Ra

1

y ∧Ryz) ⇒Ra

1

z)

∀z ((Ra

1

a

2

∧Ra

2

z) ⇒Ra

1

z)

(Ra

1

a

2

∧Ra

2

a

3

) ⇒Ra

1

a

3

)

/ ¸

(Ra

1

a

2

∧Ra

2

a

3

) Ra

1

a

3

/ ¸

.

.

.

Ra

1

a

2

Ra

2

a

3

Ra

3

a

1

∃y Ra

3

y

Ra

3

a

4

.

.

.

An inﬁnite open path gives rise (via the proof of Hintikka’s Lemma) to an inﬁnite

L-structure M with U

M

= ¦a

1

, a

2

, . . . , a

n

, . . .¦, R

M

= ¦¸a

m

, a

n

) [ 1 ≤ m < n¦.

Clearly M [= A

∞

.

Remark 2.5.10. In the course of applying a tableau test, we will sometimes ﬁnd

a ﬁnite open path which is U-replete for some ﬁnite set of parameters U ⊆ V .

In this case, the proof of Hintikka’s Lemma provides a ﬁnite L-structure with

domain U.

Example 2.5.11. Let A be the sentence (∀x(Px∨Qx)) ⇒((∀xPx) ∨(∀xQx)).

44

Testing A for logical validity, we have:

A

∀x(Px∨Qx)

((∀xPx) ∨(∀xQx))

∀xPx

∀xQx

Pa

Qb

Pa ∨Qa

Pb ∨Qb

/ ¸

Pa Qa

/ ¸

Pb Qb

This tableau has a unique open path, which gives rise (via the proof of Hintikka’s

Lemma) to a ﬁnite L-structure M with U

M

= ¦a, b¦, P

M

= ¦b¦, Q

M

= ¦a¦.

Clearly M falsiﬁes A.

Exercise 2.5.12. Using the predicate Rxy (“x is an ancestor of y”), translate

the following argument into a sentence of the predicate calculus.

Every ancestor of an ancestor of an individual is an ancestor of the

same individual. No individual is his own ancestor. Therefore, there

is an individual who has no ancestor.

Is this argument valid? Justify your answer by means of an appropriate structure

or tableau.

Solution. ((∀x∀y ((∃z (Rxz ∧Rzy)) ⇒Rxy)) ∧∃xRxx) ⇒∃x∃y Ryx.

A tableau starting with the negation of this sentence (left to the reader) fails

to close oﬀ. The structure (N, >

N

) falsiﬁes the sentence, thus showing that it is

not logically valid.

Exercise 2.5.13. Using the predicates Sx (“x is a set”) and Exy (“x is a

member of y”), translate the following into a sentence of the predicate calculus.

There exists a set whose members are exactly those sets which are

not members of themselves.

Use an unsigned tableau to test your sentence for consistency, i.e., satisﬁability.

Exercise 2.5.14. Using the predicates Sx (“x is Socrates”), Hx (“x is a man”),

Mx (“x is mortal”), translate the following argument into a sentence of the

predicate calculus.

Socrates is a man. All men are mortal. Therefore, Socrates is mortal.

Use an unsigned tableau to test whether the argument is valid.

45

Exercises 2.5.15.

1. Using the predicates Sx (“x can solve this problem”), Mx (“x is a math-

ematician”), Jx (“x is Joe”), translate the following argument into a sen-

tence of the predicate calculus.

If anyone can solve this problem, some mathematician can solve

it. Joe is a mathematician and cannot solve it. Therefore, no-

body can solve it.

Use an unsigned tableau to test whether the argument is valid.

2. Using the same predicates as above, translate the following argument into

a sentence of the predicate calculus.

Any mathematician can solve this problem if anyone can. Joe

is a mathematician and cannot solve it. Therefore, nobody can

solve it.

Use an unsigned tableau to test whether the argument is valid.

2.6 The Compactness Theorem

Theorem 2.6.1 (the Compactness Theorem, countable case). Let S be a count-

ably inﬁnite set of sentences of the predicate calculus. S is satisﬁable if and only

if each ﬁnite subset of S is satisﬁable.

Proof. We combine the ideas of the proofs of the Countable Compactness Theo-

rem for propositional calculus (Theorem 1.7.1) and the Completeness Theorem

for predicate calculus (Theorem 2.5.5).

Details: Let S = ¦A

0

, A

1

, . . . , A

i

, . . .¦. Start by letting τ

0

be the empty

tableau. Suppose we have constructed τ

2i

. Extend τ

2i

to τ

′

2i

by appending A

i

to each open path of τ

2i

. Since ¦A

0

, A

1

, . . . , A

i

¦ is satisﬁable, τ

′

2i

has at least

one open path. Now extend τ

′

2i

to τ

2i+1

and then to τ

2i+2

as in the proof of

Lemma 2.5.4. Finally put τ = τ

∞

=

∞

i=1

τ

i

. As in Lemma 2.5.4 we have that

every closed path of τ is ﬁnite, and every open path of τ is V -replete. Note

also that τ is an inﬁnite, ﬁnitely branching tree. By K¨onig’s Lemma (Theorem

1.6.6), let S

′

be an inﬁnite path in τ. Then S

′

is a V -replete open set which

contains S. By Hintikka’s Lemma for the predicate calculus (Lemma 2.5.3), S

′

is satisﬁable. Hence S is satisﬁable.

Theorem 2.6.2 (the Compactness Theorem, uncountable case). Let S be an

uncountable set of sentences of the predicate calculus. S is satisﬁable if and

only if each ﬁnite subset of S is satisﬁable.

Proof. Assume that S is ﬁnitely satisﬁable. For each sentence A ∈ S of the form

∃xB or ∀xB, introduce a new parameter c

A

. Let U

S

be the set of parameters

so introduced. Let S

′

be S together with the sentences B[x/c

A

] or B[x/c

A

]

46

as the case may be, for all c

A

∈ U

S

. Then S

′

is a set of L-U

S

-sentences, and

it is easy to verify that S

′

is ﬁnitely satisﬁable. By Zorn’s Lemma, let S

′′

be a

maximal ﬁnitely satisﬁable set of L-U

S

-sentences extending S

′

.

Now inductively deﬁne S

0

= S, S

n+1

= S

′′

n

, S

∞

=

∞

n=0

S

n

, U =

∞

n=0

U

Sn

.

It is straightforward to verify that S

∞

is U-replete and open. Hence, by Hin-

tikka’s Lemma, S

∞

is satisﬁable in the domain U. Since S ⊆ S

∞

, it follows

that S is satisﬁable in U.

Exercise 2.6.3. Let L be a language consisting of a binary predicate R and

some additional predicates. Let M = (U

M

, R

M

, . . .) be an L-structure such

that (U

M

, R

M

) is isomorphic to (N, <

N

). Note that M contains no inﬁnite

R-descending sequence. Show that there exists an L-structure M

′

such that:

1. M and M

′

satisfy the same L-sentences.

2. M

′

contains an inﬁnite R-descending sequence. In other words, there

exist elements a

′

1

, a

′

2

, . . . , a

′

n

, . . . ∈ U

M

′ such that ¸a

′

n+1

, a

′

n

) ∈ R

M

′ for all

n = 1, 2, . . ..

Hint: Use the Compactness Theorem.

Exercise 2.6.4. Generalize Exercise 2.6.3 replacing (N, <

N

) by an arbitrary

inﬁnite linear ordering with no inﬁnite descending sequence. Show that M

′

can

be obtained such that (U

M

′ , R

M

′ ) is a linear ordering which contains an inﬁnite

descending sequence.

Exercise 2.6.5. Let L = ¦R, . . .¦ be a language which includes a binary pred-

icate R. Let S be a set of L-sentences. Assume that for each n ≥ 1 there exists

an L-structure (U

n

, R

n

, . . .) satisfying S and containing elements a

n1

, . . . , a

nn

such that ¸a

ni

, a

nj

) ∈ R

n

for all i and j with 1 ≤ i < j ≤ n. Prove that there

there exists an L-structure (U

∞

, R

∞

, . . .) satisfying S and containing elements

a

∞i

, i = 1, 2, . . . such that ¸a

∞i

, a

∞j

) ∈ R

∞

for all i and j with 1 ≤ i < j.

Solution. Let L

∗

= L∪¦P

1

, P

2

, . . .¦ where P

1

, P

2

, . . . are new unary predicates.

Let S

∗

be S plus ∃xP

i

x plus ∀x∀y ((P

i

x∧P

j

y) ⇒Rxy), 1 ≤ i < j. Consider

the L

∗

-structures (U

n

, R

n

, . . . , P

n1

, . . . , P

nn

, . . .), n ≥ 1, where P

ni

= ¦a

ni

¦ for

1 ≤ i ≤ n, and P

ni

= ¦¦ for i > n. Clearly each ﬁnite subset of S

∗

is satisﬁed

by all but ﬁnitely many of these structures. It follows by the Compactness The-

orem that S

∗

is satisﬁable. Let (U

∞

, R

∞

, . . . , P

∞1

, P

∞2

, . . .) be an L

∗

-structure

satisfying S

∗

. Clearly the L-structure (U

∞

, R

∞

, . . .) has the desired properties.

2.7 Satisﬁability in a Domain

The notion of satisﬁability in a domain was introduced in Deﬁnition 2.2.9.

Theorem 2.7.1. Let S be a set of L-sentences.

1. Assume that S is ﬁnite or countably inﬁnite. If S is satisﬁable, then S is

satisﬁable in a countably inﬁnite domain.

47

2. Assume that S is of cardinality κ ≥ ℵ

0

. If S is satisﬁable, then S is

satisﬁable in a domain of cardinality κ.

Proof. Parts 1 and 2 follow easily from the proofs of Compactness Theorems

2.6.1 and 2.6.2, respectively. In the countable case we have that S is satisﬁable

in V , which is countably inﬁnite. In the uncountable case we have that S is

satisﬁable in U, where U is as in the proof of 2.6.2. By the arithmetic of inﬁnite

cardinal numbers, the cardinality of U is κ ℵ

0

= κ.

In Example 2.2.12 we have seen a sentence A

∞

which is satisﬁable in a

countably inﬁnite domain but not in any ﬁnite domain. Regarding satisﬁability

in ﬁnite domains, we have:

Example 2.7.2. Given a positive integer n, we exhibit a sentence A

n

which

is satisﬁable in a domain of cardinality n but not in any domain of smaller

cardinality. Our sentence A

n

is (1) ∧(2) ∧(3) with

(1) ∀x∀y ∀z ((Rxy ∧Ryz) ⇒Ryz)

(2) ∀x∀y (Rxy ⇒Ryx)

(3) ∃x

1

∃x

n

(Rx

1

x

2

∧Rx

2

x

3

∧ ∧Rx

n−1

x

n

)

On the other hand, we have:

Theorem 2.7.3. Let M and M

′

be L-structures. Assume that there ex-

ists an onto mapping φ : U

M

→ U

M

′ such that for all n-ary predicates P

of L and all n-tuples ¸a

1

, . . . , a

n

) ∈ (U

M

)

n

, ¸a

1

, . . . , a

n

) ∈ P

M

if and only if

¸φ(a

1

), . . . , φ(a

n

)) ∈ P

M

′ . Then as in Theorem 2.2.6 we have v

M

(A) = v

M

′ (A

′

)

for all L-U

M

-sentences A, where A

′

= A[a

1

/φ(a

1

), . . . , a

k

/φ(a

k

)]. In particular,

M and M

′

satisfy the same L-sentences.

Proof. The proof is by induction on the degree of A. Suppose for example that

A = ∀xB. Then by deﬁnition of v

M

we have that v

M

(A) = T if and only if

v

M

(B[x/a]) = T for all a ∈ U

M

. By inductive hypothesis, this holds if and only

if v

M

′ (B[x/a]

′

) = T for all a ∈ U

M

. But for all a ∈ U

M

we have B[x/a]

′

=

B

′

[x/φ(a)]. Thus our condition is equivalent to v

M

′ (B

′

[x/φ(a)]) = T for all

a ∈ U

M

. Since φ : U

M

→ U

M

′ is onto, this is equivalent to v

M

′ (B

′

[x/b]) = T

for all b ∈ U

M

′ . By deﬁnition of v

M

′ this is equivalent to v

M

′ (∀xB

′

) = T. But

∀xB

′

= A

′

, so our condition is equivalent to v

M

′ (A

′

) = T.

Corollary 2.7.4. Let S be a set of L-sentences. If S is satisﬁable in a domain

U, then S is satisﬁable in any domain of the same or larger cardinality.

Proof. Suppose S is satisﬁable in domain U. Let U

′

be a set of cardinality

greater than or equal to that of U. Let φ : U

′

→ U be onto. If M is any

L-structure with U

M

= U, we can deﬁne an L-structure M

′

with U

M

′ = U

′

by putting P

M

′ = ¦¸a

1

, . . . , a

n

) [ ¸φ(a

1

), . . . , φ(a

n

)) ∈ P

M

¦ for all n-ary pred-

icates P of L. By Theorem 2.7.3, M and M

′

satisfy the same L-sentences. In

particular, if M [= S, then M

′

[= S.

48

Remark 2.7.5. We shall see later

8

that Theorem 2.7.3 and Corollary 2.7.4 fail

for normal satisﬁability.

8

See Section 4.1.

49

Chapter 3

Proof Systems for Predicate

Calculus

3.1 Introduction to Proof Systems

Deﬁnition 3.1.1. An abstract proof system consists of a set X together with a

relation R ⊆

∞

k=0

X

k+1

. Elements of X are called objects. Elements of R are

called rules of inference. An object X ∈ X is said to be derivable, or provable,

if there exists a ﬁnite sequence of objects X

1

, . . . , X

n

such that X

n

= X and,

for each i ≤ n, there exist j

1

, . . . , j

k

< i such that ¸X

j1

, . . . , X

j

k

, X

i

) ∈ R. The

sequence X

1

, . . . , X

n

is called a derivation of X, or a proof of X.

Notation 3.1.2. For k ≥ 1 it is customary to write

X

1

X

k

Y

indicating that ¸X

1

, . . . , X

k

, Y ) ∈ R. This is to be understood as “from the

premises X

1

, . . . , X

k

we may immediately infer the conclusion Y ”. For k = 0

we may write

Y

or simply Y , indicating that ¸Y ) ∈ R. This is to be understood as “we may

immediately infer Y from no premises”, or “we may assume Y ”.

Deﬁnition 3.1.3. Let L be a language. Recall that V is the set of parameters.

A Hilbert-style proof system for L is a proof system with the following properties:

1. The objects are sentences with parameters. In other words,

X = ¦A [ A is an L-V -sentence¦ .

2. For each rule of inference

A

1

A

k

B

50

(i.e., ¸A

1

, . . . , A

k

, B) ∈ R), we have that B is a logical consequence of

A

1

, . . . , A

k

. This property is known as soundness. It implies that every

L-V -sentence which is derivable is logically valid.

3. For all L-V -sentences A, B, we have a rule of inference ¸A, A⇒B, B) ∈ R,

i.e.,

A A⇒B

B

.

In other words, fromA and A⇒B we immediately infer B. This collection

of inference rules is known as modus ponens.

4. An L-V -sentence A is logically valid if and only if A is derivable. This

property is known as completeness.

Remark 3.1.4. In Section 3.3 we shall exhibit a particular Hilbert-style proof

system, LH. The soundness of LH will be obvious. In order to verify the

completeness of LH, we shall ﬁrst prove a result known as the Companion

Theorem, which is also of interest in its own right.

3.2 The Companion Theorem

In this section we shall comment on the notion of logical validity for sentences of

the predicate calculus. We shall analyze logical validity into two components: a

propositional component (quasitautologies), and a quantiﬁcational component

(companions).

Deﬁnition 3.2.1 (quasitautologies).

1. A tautology is a propositional formula which is logically valid.

2. A quasitautology is an L-V -sentence of the form F[p

1

/A

1

, . . . , p

k

/A

k

],

where F is a tautology, p

1

, . . . , p

k

are the atoms occurring in F, and

A

1

, . . . , A

k

are L-V -sentences.

For example, p ⇒(q ⇒p) is a tautology. This implies that, for all L-V -sentences

A and B, A⇒(B⇒A) is a quasitautology.

Remarks 3.2.2.

1. Obviously, every quasitautology is logically valid.

2. There is a decision procedure

1

for quasitautologies. One such decision

procedure is based on truth tables. Another is based on propositional

tableaux.

1

In other words, there is a Turing algorithm which, given an L-V -sentence A as input, will

eventually halt with output 1 if A is a quasitautology, 0 if A is not a quasitautology.

51

3. It can be shown that there is no decision procedure for logical validity.

(This result is known as Church’s Theorem.) Therefore, in relation to the

problem of characterizing logical validity, we regard the quasitautologies

as trivial.

Let A be an L-V -sentence.

Deﬁnition 3.2.3 (companions). A companion of A is any L-V -sentence of one

of the forms

(1) (∀xB) ⇒B[x/a]

(2) B[x/a] ⇒(∀xB)

(3) (∃xB) ⇒B[x/a]

(4) B[x/a] ⇒(∃xB)

where, in (2) and (3), the parameter a may not occur in A or in B.

Lemma 3.2.4. Let C be a companion of A.

1. A is satisﬁable if and only if C ∧A is satisﬁable.

2. A is logically valid if and only if C ⇒A is logically valid.

Proof. Let C be a companion of A.

For part 1, assume that A is satisﬁable. In accordance with Deﬁnition 2.3.11,

let M, φ be an L-V -structure satisfying A. If C is of the form 3.2.3(1) or

3.2.3(4), then C is logically valid, hence M, φ satisﬁes C ∧A. Next, consider

the case when C is of the form 3.2.3(2). If M, φ satisﬁes ∀xB, then M, φ

satisﬁes C. Otherwise we have v

M

(∀xB

φ

) = F, so let c ∈ U

M

be such that

v

M

(B

φ

[x/c]) = F. Deﬁne φ

′

: V → U

M

by putting φ

′

(a) = c, φ

′

(b) = φ(b)

for b ,= a. Since a does not occur in A, we have that M, φ

′

satisﬁes A. Also,

since a does not occur in B, we have B[x/a]

φ

′

= B

φ

′

[x/c] = B

φ

[x/c], hence

v

M

(B[x/a]

φ

′

) = v

M

(B

φ

[x/c]) = F, i.e., M, φ

′

satisﬁes B[x/a]. Thus M, φ

′

satisﬁes C ∧A. The case when C is of the form 3.2.3(3) is handled similarly.

For part 2 note that, since C is a companion of A, C is a companion of A.

Thus we have that A is logically valid if and only if A is not satisﬁable, if

and only if C ∧A is not satisﬁable (by part 1), if and only if (C ∧A) is

logically valid, i.e., C ⇒A is logically valid.

Deﬁnition 3.2.5 (companion sequences). A companion sequence of A is a ﬁnite

sequence C

1

, . . . , C

n

such that, for each i < n, C

i+1

is a companion of

(C

1

∧ ∧C

i

) ⇒A.

Lemma 3.2.6. If C

1

, . . . , C

n

is a companion sequence of A, then A is logically

valid if and only if (C

1

∧ ∧C

n

) ⇒A is logically valid.

Proof. Note that (C

1

∧ ∧C

n

) ⇒A is quasitautologically equivalent to

C

n

⇒(C

n−1

⇒ ⇒(C

1

⇒A)).

52

Our lemma follows by n applications of part 2 of Lemma 3.2.4.

Theorem 3.2.7 (the Companion Theorem). A is logically valid if and only if

there exists a companion sequence C

1

, . . . , C

n

of A such that

(C

1

∧ ∧C

n

) ⇒A

is a quasitautology.

Proof. The “if” part is immediate from Lemma 3.2.6. For the “only if” part,

assume that A is logically valid. By Theorem 2.5.5 let τ be a ﬁnite closed

unsigned tableau starting with A. Thus we have a ﬁnite sequence of tableaux

τ

0

, τ

1

, . . . , τ

n

where τ

0

= A, τ

n

= τ, and each τ

i+1

is obtained by applying

a tableau rule R

i

to τ

i

. If R

i

is a quantiﬁer rule, let C

i

be an appropriate

companion. Thus C

1

, . . . , C

n

is a companion sequence for A, and we can easily

transform τ into a closed tableau τ

′

starting with

A

C

1

.

.

.

C

n

in which only propositional tableau rules are applied. Thus (C

1

∧ ∧C

n

) ⇒A

is a quasitautology. This proves our theorem.

For instance, if R

i

is the tableau rule

(∗)

.

.

.

∀xB

.

.

.

[

B[x/a] ,

where a is an arbitrary parameter, let C

i

be the companion (∀xB) ⇒B[x/a],

and replace the application of (∗) by

.

.

.

(∀xB) ⇒B[x/a]

.

.

.

∀xB

.

.

.

/ ¸

∀xB B[x/a]

noting that the left-hand path is closed.

53

Similarly, if R

i

is the tableau rule

(∗∗)

.

.

.

∀xB

.

.

.

[

B[x/a]

where a is a new parameter, let C

i

be the companion B[x/a] ⇒(∀xB), and

replace the application of (∗∗) by

.

.

.

B[x/a] ⇒(∀xB)

.

.

.

∀xB

.

.

.

/ ¸

B[x/a] ∀xB

noting that the right-hand path is closed.

Example 3.2.8. As an example illustrating Theorem 3.2.7 and its proof, let

A be the sentence (∃x(Px∨Qx)) ⇒((∃xPx) ∨(∃xQx)). Let τ be the closed

tableau

A

∃x(Px∨Qx)

((∃xPx) ∨(∃xQx))

∃xPx

∃xQx

Pa ∨Qa

Pa

Qa

/ ¸

Pa Qa

which shows that A is logically valid. Examining the applications of quantiﬁer

rules in τ, we obtain the companion sequence C

1

, C

2

, C

3

for A, where C

1

is

(∃x(Px∨Qx)) ⇒(Pa ∨Qa), C

2

is Pa ⇒∃xPx, C

3

is Qa ⇒∃xQx. Clearly

(C

1

∧C

2

∧C

3

) ⇒A is a quasitautology.

Exercise 3.2.9. Let A be the logically valid sentence ∃x(Px⇒∀y Py). Find

a companion sequence C

1

, . . . , C

n

for A such that (C

1

∧ ∧C

n

) ⇒A is a qu-

asitautology.

54

Solution. The unsigned tableau

∃x(Px⇒∀y Py)

(Pa ⇒∀y Py)

Pa

∀y Py

Pb

(Pb ⇒∀y Py)

Pb

is closed and shows that A is logically valid. From this tableau we read oﬀ the

companion sequence C

1

, C

2

, C

3

, where

C

1

is (Pa ⇒∀y Py) ⇒∃x(Px⇒∀y Py),

C

2

is Pb ⇒∀y Py,

C

3

is (Pb ⇒∀y Py) ⇒∃x(Px⇒∀y Py).

A simpler companion sequence for A is C

′

1

, C

′

2

where

C

′

1

is Pa ⇒∀y Py,

C

′

2

is (Pa ⇒∀y Py) ⇒∃x(Px⇒∀y Py).

Exercise 3.2.10. Let A be the logically valid sentence ∃x(Px⇔∀y Py). Find

a companion sequence C

1

, . . . , C

n

for A such that (C

1

∧ ∧C

n

) ⇒A is a qu-

asitautology.

Solution. To ﬁnd a companion sequence for A, we ﬁrst construct a closed un-

signed tableau starting with A.

∃x(Px⇔∃y Py)

(Pa ⇔∃y Py)

/ ¸

Pa

∃y Py

Pa

Pa

∃y Py

Pb

(Pb ⇔∃y Py)

/ ¸

Pb

∃y Py

Pb

∃y Py

From this tableau, we read oﬀ the companion sequence C

1

, C

2

, C

3

, C

4

, where

C

1

is (Pa ⇔∃y Py) ⇒∃x(Px⇔∃y Py),

C

2

is Pa ⇒∃y Py,

C

3

is (∃y Py) ⇒Pb,

C

4

is (Pb ⇔∃y Py) ⇒∃x(Px⇔∃y Py).

55

A simpler companion sequence for A is C

′

1

, C

′

2

, C

′

3

where

C

′

1

is (∃y Py) ⇒Pa,

C

′

2

is Pa ⇒∃y Py,

C

′

3

is (Pa ⇔∃y Py) ⇒∃x(Px⇔∃y Py).

3.3 Hilbert-Style Proof Systems

Let L be a language. Recall that V is the set of parameters.

Deﬁnition 3.3.1 (the system LH). Our Hilbert-style proof system LH for the

predicate calculus is as follows:

1. The objects are L-V -sentences.

2. For each quasitautology A, ¸A) is a rule of inference.

3. ¸(∀xB) ⇒B[x/a]) and ¸B[x/a] ⇒(∃xB)) are rules of inference.

4. ¸A, A⇒B, B) is a rule of inference.

5. ¸A⇒B[x/a], A⇒(∀xB)) and ¸B[x/a] ⇒A, (∃xB) ⇒A) are rules of in-

ference, provided the parameter a does not occur in A or in B.

Schematically, LH consists of:

1. A, where A is any quasitautology

2. (a) (∀xB) ⇒B[x/a] (universal instantiation)

(b) B[x/a] ⇒(∃xB) (existential instantiation)

3.

A A⇒B

B

(modus ponens)

4. (a)

A⇒B[x/a]

A⇒(∀xB)

(universal generalization)

(b)

B[x/a] ⇒A

(∃xB) ⇒A

(existential generalization),

where a does not occur in A, B.

Lemma 3.3.2 (soundness of LH). LH is sound. In other words, for all L-V -

sentences A, if A is derivable, then A is logically valid.

Proof. The proof is straightforward by induction on the length of a derivation.

The induction step is similar to the proof of Lemma 3.2.4.

Example 3.3.3. In LH we have the following derivation:

1. (∀xA) ⇒A[x/a] (by universal instantiation)

56

2. A[x/a] ⇒(∃xA) (by existential instantiation)

3. ((∀xA) ⇒A[x/a]) ⇒((A[x/a] ⇒(∃xA)) ⇒((∀xA) ⇒(∃xA)))

(This is a quasitautology, obtained from the tautology

(p ⇒q) ⇒((q ⇒r) ⇒(p ⇒r)).)

4. (A[x/a] ⇒(∃xA)) ⇒((∀xA) ⇒(∃xA)) (from 1, 3, and modus ponens)

5. (∀xA) ⇒(∃xA) (from 2, 4, and modus ponens)

Thus, by Lemma 3.3.2, (∀xA) ⇒(∃xA) is logically valid.

Example 3.3.4. In LH we have the following derivation:

1. B[x/a] ⇒(∃xB) (by existential instantiation)

2. (B[x/a] ⇒(∃xB)) ⇒((A∧B)[x/a] ⇒(∃xB)) (a quasitautology)

3. (A∧B)[x/a] ⇒(∃xB) (from 1, 2, and modus ponens)

4. (∃x(A∧B)) ⇒(∃xB) (from 3 and existential generalization)

Thus, by Lemma 3.3.2, (∃x(A∧B)) ⇒(∃xB) is logically valid.

We now turn to the proof that LH is complete.

Lemma 3.3.5. LH is closed under quasitautological consequence. In other

words, if A

1

, . . . , A

k

are derivable, and if B is a quasitautological consequence

of A

1

, . . . , A

k

, then B is derivable.

Proof. We are assuming that B is a quasitautological consequence of A

1

, . . . , A

k

.

Thus A

1

⇒(A

2

⇒ ⇒(A

k

⇒B)) is a quasitautology, hence derivable. We are

also assuming that A

1

, . . . , A

k

are derivable. Thus we obtain B by k applications

of modus ponens.

Lemma 3.3.6. If C is a companion of A, and if C ⇒A is derivable in LH, then

A is derivable in LH.

Proof. First, suppose C is of the form 3.2.3(1), namely (∀xB) ⇒B[x/a]. By

universal instantiation, C is derivable. In addition, we are assuming that C ⇒A

is derivable. Hence, by modus ponens, A is derivable.

Next, suppose C is of the form 3.2.3(2), namely B[x/a] ⇒(∀xB), where

a does not occur in A, B. We are assuming that C ⇒A is derivable, i.e.,

(B[x/a] ⇒(∀xB)) ⇒A is derivable. It follows by Lemma 3.3.5 that both (i)

(A) ⇒B[x/a] and (ii) (A) ⇒(∀xB) are derivable. Applying universal gen-

eralization to (i), we see that (A) ⇒(∀xB) is derivable. From this plus (ii),

it follows by Lemma 3.3.5 that A is derivable.

The other cases, where C is of the form 3.2.3(3) or 3.2.3(4), are handled

similarly.

57

Theorem 3.3.7 (completeness of LH). LH is sound and complete. In other

words, for all L-V -sentences A, A is derivable if and only if A is logically valid.

Proof. The “only if” part is Lemma 3.3.2. For the “if” part, assume that

A is logically valid. By Theorem 3.2.7, there exists a companion sequence

C

1

, . . . , C

n

for A such that (C

1

∧ ∧C

n

) ⇒A is a quasitautology. Hence

C

n

⇒(C

n−1

⇒ ⇒(C

1

⇒A)) is a quasitautology, hence derivable. From this

and n applications of Lemma 3.3.6, we obtain derivability of A.

Remark 3.3.8. For convenience in writing proofs, we supplement the rules of

LH with

A

1

A

k

B

whenever B is a quasitautological consequence of A

1

, . . . , A

k

. This is justiﬁed by

Lemma 3.3.5. We indicate applications of this rule by QT, for quasitautology.

Similarly we use UI, EI, UG, EG to indicate universal instantiation, existential

instantiation, universal generalization, existential generalization, respectively.

Exercise 3.3.9. Construct a Hilbert-style proof of the sentence

(∃x∀y Rxy) ⇒(∀y ∃xRxy).

Solution. A proof in LH is

1. (∀y Ray) ⇒Rab UI

2. Rab ⇒(∃xRxb) EI

3. (∀y Ray) ⇒(∃xRxb) 1,2,QT

4. (∀y Ray) ⇒(∀y ∃xRxy) 3,UG

5. (∃x∀y Rxy) ⇒(∀y ∃xRxy) 4,EG

Exercise 3.3.10. Construct a Hilbert-style proof of the sentence

(∀x(Px∧Qx)) ⇔((∀xPx) ∧(∀xQx))

Solution. A proof in LH is

1. (∀x(Px∧Qx)) ⇒(Pa ∧Qa) UI

2. (∀x(Px∧Qx)) ⇒Pa 1,QT

3. (∀x(Px∧Qx)) ⇒∀xPx 2,UG

4. (∀x(Px∧Qx)) ⇒Qa 1,QT

5. (∀x(Px∧Qx)) ⇒∀xQx 4,UG

6. (∀xPx) ⇒Pa UI

7. (∀xQx) ⇒Qa UI

8. ((∀xPx) ∧(∀xQx)) ⇒(Pa ∧Qa) 6,7,QT

9. ((∀xPx) ∧(∀xQx)) ⇒∀x(Px∧Qx) 8,UG

10. (∀x(Px∧Qx)) ⇔((∀xPx) ∧(∀xQx)) 3,5,9,QT

Exercise 3.3.11. Construct a Hilbert-style proof of ∃x∀y (Eyx⇔Eyy).

58

Solution. A proof in LH is

1. ∀y (Eya ⇔Eyy) ⇒(Eaa ⇔Eaa) UI

2. ∀y (Eya ⇔Eyy) 1,QT

3. (∀y (Eya ⇔Eyy)) ⇒∃x∀y (Eyx⇔Eyy) 2,QT

4. (∃x∀y (Eyx⇔Eyy)) ⇒∃x∀y (Eyx⇔Eyy) 3,EG

5. ∃x∀y (Eyx⇔Eyy) 4,QT

Exercise 3.3.12. Construct a Hilbert-style proof of the sentence

∃x(Sx∧∀y (Eyx⇔(Sy ∧Eyy))).

Solution. A proof in LH is

1. (∀y (Eya ⇔(Sy ∧Eyy))) ⇒(Eaa ⇔(Sa ∧Eaa)) UI

2. (Sa ∧(∀y (Eya ⇔(Sy ∧Eyy)))) ⇒(Sb ∧Sb) 1,QT

3. (∃x(Sx∧(∀y (Eyx⇔(Sy ∧Eyy))))) ⇒(Sb ∧Sb) 2,EG

4. ∃x(Sx∧(∀y (Eyx⇔(Sy ∧Eyy)))) 3,QT

Exercise 3.3.13. Construct a Hilbert-style proof of ∃x(Px⇒∀y Py).

Solution. A proof in LH is

1. (Pa ⇒∀y Py) ⇒∃x(Px⇒∀y Py) EI

2. (∃x(Px⇒∀y Py)) ⇒Pa 1,QT

3. (∃x(Px⇒∀y Py)) ⇒∀y Py 2,UG

4. ∃x(Px⇒∀y Py) 1,3,QT

Exercise 3.3.14. Construct a Hilbert-style proof of ∃x(Px⇔∀y Py).

Exercise 3.3.15. Consider the following proof system LH

′

, which is a “stripped

down” version of LH. The objects of LH

′

are L-V -sentences containing only ∀,

⇒, (i.e., not containing ∃, ⇔, ∧, ∨). The rules of LH

′

are:

(a) quasitautologies

(b) (∀xB) ⇒B[x/a]

(c) (∀x(A⇒B)) ⇒(A⇒∀xB)

(d)

A A⇒B

B

(modus ponens)

(e)

B[x/a]

∀xB

(generalization), where a does not occur in B.

Show that LH

′

is sound and complete.

59

Solution. Soundness is proved just as for LH.

Just as for the full tableau method, we can prove soundness and complete-

ness of the restricted tableau method with ∀, ⇒, , and from this we obtain

the restricted Companion Theorem. There are now only two kinds of compan-

ions, the ones involving ∀. It remains to prove the following lemma: If C is a

companion of A, and if C ⇒A is derivable in LH

′

, then A is derivable in LH

′

.

Consider a companion of the form B[x/a] ⇒(∀xB). Assume that

(B[x/a] ⇒(∀xB)) ⇒A

is derivable in LH

′

, where a does not occur in A, B. It follows quasitautologically

that both (1) (A) ⇒B[x/a] and (2) (A) ⇒∀xB are derivable in LH

′

. From

(1) and the generalization rule (e) of LH

′

, we see that ∀x((A) ⇒B) is derivable

in LH

′

. Also, by rule (c) of LH

′

,

(∀x((A) ⇒B)) ⇒((A) ⇒∀xB)

is derivable in LH

′

. Hence, by modus ponens, (A) ⇒∀xB is derivable in LH

′

.

It follows quasitautologically from this and (2) that A is derivable in LH

′

. This

completes the proof.

Exercise 3.3.16.

1. Let S be a set of L-sentences. Consider a proof system LH(S) consisting

of LH with additional rules of inference ¸A), A ∈ S. Show that an L-V -

sentence B is derivable in LH(S) if and only if B is a logical consequence

of S.

2. Indicate the modiﬁcations needed when S is a set of L-V -sentences.

Solution.

1. By induction on the length of derivations, it is straightforward to prove

that each sentence derivable in LH(S) is a logical consequence of S. The

assumption that S is a set of L-sentences (not L-V -sentences) is used in

the inductive steps corresponding to rules 4(a) and 4(b), universal and

existential generalization, because we need to know that the parameter a

does not occur in S.

Conversely, assume B is a logical consequence of S. By the Compactness

Theorem, it follows that B is a logical consequence of a ﬁnite subset of

S, say A

1

, . . . , A

n

. Hence (A

1

∧ ∧A

n

) ⇒B is logically valid. Hence,

by completeness of LH, (A

1

∧ ∧A

n

) ⇒B is derivable in LH. Since

LH(S) includes LH, we have that

(A

1

∧ ∧A

n

) ⇒B

is derivable in LH(S). But A

1

, . . . , A

n

are derivable in LH(S). It follows

quasitautologically that B is derivable in LH(S). This completes the

proof.

60

2. If S is a set of L-V -sentences, we modify our system as follows. Let V

′

be a countably inﬁnite set of new parameters, disjoint from V . Deﬁne

LH(S) as before, but allowing parameters from V ∪ V

′

. The objects are

L-V ∪V

′

-sentences. In rules 4(a) and 4(b), one must impose the restriction

that a does not occur in A, B, S. With this modiﬁcation, everything goes

through as before.

Notation 3.3.17. We write S ⊢ B to indicate that B is derivable in LH(S).

Exercise 3.3.18. Let S be an inﬁnite set of L-sentences, and let B be an L-

sentence. Prove that S [= B (i.e., B is true in all L-structures satisfying S)

if and only if there exists a ﬁnite set of L-sentences A

1

, . . . , A

k

∈ S such that

A

1

, . . . , A

k

⊢ B (i.e., B is provable from A

1

, . . . , A

k

).

Solution. The “if” part follows from the soundness of our proof system. For the

“only if” part, assume that S [= B, i.e., S ∪ ¦B¦ is not satisﬁable. It follows

by the Compactness Theorem that there exists a ﬁnite set ¦A

1

, . . . , A

k

¦ ⊂ S

such that A

1

, . . . , A

k

, B is not satisﬁable. Thus B is a logical consequence of

A

1

, . . . , A

k

. It follows by completeness of our proof system that A

1

, . . . , A

k

⊢ B.

3.4 Gentzen-Style Proof Systems

Throughout this section, let L be a language. As usual, V is the set of param-

eters.

Before presenting our Gentzen-style proof system for L, we ﬁrst discuss the

block tableau method, a trivial variant of the signed tableau method.

Deﬁnition 3.4.1. A block is a ﬁnite set of signed L-V -sentences. A block is

said to be closed if it contains TA and FA for some L-V -sentence A.

Notation 3.4.2. If S is a block and X is a signed L-V -sentence, we write S, X

instead of S ∪ ¦X¦, etc.

Deﬁnition 3.4.3. A block tableau is a rooted dyadic tree where each node car-

ries a block. A block tableau is said to be closed if each of its end nodes is closed.

Given a block S, a block tableau starting with S is a block tableau generated

from S by means of block tableau rules. The block tableau rules are obtained

from the signed tableau rules (pages 13 and 31) as follows. Corresponding to

signed tableau rules of the form

.

.

.

X

.

.

.

[

Y

.

.

.

X

.

.

.

[

Y

1

Y

2

.

.

.

X

.

.

.

/ ¸

Y Z

.

.

.

X

.

.

.

/ ¸

Y

1

Z

1

Y

2

Z

2

61

we have block tableau rules

S, X

[

S, X, Y

S, X

[

S, X, Y

1

, Y

2

S, X

/ ¸

S, X, Y S, X, Z

S, X

/ ¸

S, X, Y

1

, Y

2

S, X, Z

1

, Z

2

respectively.

For example, we have the following block tableau rules:

S, TA∧B

[

S, TA∧B, TA, TB

S, FA∧B

/ ¸

S, FA∧B, FA S, FA∧B, FB

S, TA⇒B

/ ¸

S, TA⇒B, FA S, TA⇒B, TB

S, FA⇒B

[

S, FA⇒B, TA, FB

S, T∀xA

[

S, T∀xA, TA[x/a]

S, F∀xA

[

S, F∀xA, FA[x/a]

where a is new.

Example 3.4.4. We exhibit a closed block tableau demonstrating that ∃xA is

a logical consequence of ∀xA.

T∀xA, F∃xA

[

T∀xA, F∃xA, TA[x/a]

[

T∀xA, F∃xA, TA[x/a], FA[x/a]

This block tableau is of course similar to the signed tableau

T∀xA

F∃xA

TA[x/a]

FA[x/a]

which demonstrates the same thing.

We now deﬁne our Gentzen-style system, LG.

Deﬁnition 3.4.5. A sequent is an expression of the form Γ → ∆ where Γ and

∆ are ﬁnite sets of L-V -sentences. If S = TA

1

, . . . , TA

m

, FB

1

, . . . , FB

n

is a

block, let [S[ be the sequent A

1

, . . . , A

m

→ B

1

, . . . , B

n

. This gives a one-to-one

correspondence between blocks and sequents.

62

Deﬁnition 3.4.6 (the system LG). Our Gentzen-style proof system LG for the

predicate calculus is as follows.

1. The objects of LG are sequents.

2

2. For each closed block S, we have a rule of inference ¸[S[). In other words,

for all ﬁnite sets of L-V -sentences Γ and ∆ and all L-V -sentences A, we

assume the sequent Γ, A → A, ∆.

3. For each non-branching block tableau rule

S

[

S

′

we have a rule of inference ¸[S

′

[, [S[), i.e.,

[S

′

[

[S[

.

4. For each branching block tableau rule

S

/ ¸

S

′

S

′′

we have a rule of inference ¸[S

′

[, [S

′′

[, [S[), i.e.,

[S

′

[ [S

′′

[

[S[

.

Thus LG includes the following rules of inference:

Γ, A → A, ∆

Γ, A → A, ∆

Γ, A → ∆

Γ, A → A, ∆

Γ → A, ∆

Γ, A∧B, A, B → ∆

Γ, A∧B → ∆

Γ → A∧B, A, ∆ Γ → A∧B, B, ∆

Γ → A∧B, ∆

Γ, A⇒B → A, ∆ Γ, A⇒B, B → ∆

Γ, A⇒B → ∆

Γ, A → A⇒B, B, ∆

Γ → A⇒B, ∆

Γ, ∀xA, A[x/a] → ∆

Γ, ∀xA → ∆

Γ → ∀xA, A[x/a], ∆

Γ → ∀xA, ∆

where a does not occur in the conclusion.

2

For this reason, LG is sometimes called a sequent calculus.

63

Exercise 3.4.7. Explicitly display the remaining inference rules of LG.

Deﬁnition 3.4.8. A sequent A

1

, . . . , A

m

→ B

1

, . . . , B

n

is said to be logically

valid if and only if the L-V -sentence

(A

1

∧ ∧A

m

) ⇒(B

1

∨ ∨B

n

)

is logically valid.

3

Theorem 3.4.9 (soundness and completeness of LG). LG is sound and com-

plete. In other words, a sequent Γ → ∆ is logically valid if and only if it is

derivable in LG. In particular, an L-V -sentence A is logically valid if and only

if the sequent

→ A

is derivable in LG.

Proof. Note that the sequent A

1

, . . . , A

m

→ B

1

, . . . , B

n

is logically valid if

and only if the block TA

1

, . . . , TA

m

, FB

1

, . . . , FB

n

is not satisﬁable. Thus,

soundness and completeness of LG is equivalent to soundness and completeness

of the block tableau method. The latter is in turn easily seen to be equivalent

to soundness and completeness of the signed tableau method, as presented in

Theorems 2.3.13 and 2.5.5.

Exercise 3.4.10. Construct a Gentzen-style proof of the sequent

∃x∀y Rxy → ∀y ∃xRxy.

Solution. A proof in LG is

1. ∃x∀y Rxy, ∀y Ray, Rab → Rab, ∃xRxb, ∀y ∃xRxy

2. ∃x∀y Rxy, ∀y Ray, Rab → ∃xRxb, ∀y ∃xRxy

3. ∃x∀y Rxy, ∀y Ray → ∃xRxb, ∀y ∃xRxy

4. ∃x∀y Rxy → ∃xRxb, ∀y ∃xRxy

5. ∃x∀y Rxy → ∀y ∃xRxy

Deﬁnition 3.4.11. In order to simplify the writing of Gentzen-style proofs, let

LG

+

be LG augmented with the so-called weakening rules or padding rules:

Γ → ∆

Γ, A → ∆

Γ → ∆

Γ → A, ∆

where A is an L-V -sentence. Clearly LG

+

is sound and complete.

Remark 3.4.12. Any proof in LG is a proof in LG

+

, and any proof in LG

+

may be straightforwardly “padded out” to a proof in LG. Thus LG

+

diﬀers only

slightly from LG. However, proofs in LG

+

are easier. For example, patterned

on the above proof in LG, we have the following proof in LG

+

:

3

In particular, the sequents A

1

, . . . , Am → and → B

1

, . . . , Bn are said to be logically

valid if and only if the L-V -sentences ¬(A

1

∧ · · · ∧Am) and B

1

∨ · · · ∨Bn are logically valid,

respectively. The empty sequent → is deemed not logically valid.

64

1. Rab → Rab

1.5. ∀y Ray, Rab → Rab

2. ∀y Ray → Rab

2.5. ∀y Ray → Rab, ∃xRxb

3. ∀y Ray → ∃xRxb

3.5. ∃x∀y Rxy, ∀y Ray → ∃xRxb

4. ∃x∀y Rxy → ∃xRxb

4.5. ∃x∀y Rxy → ∃xRxb, ∀y ∃xRxy

5. ∃x∀y Rxy → ∀y ∃xRxy

or, omitting the applications of the padding rules,

1. Rab → Rab

2. ∀y Ray → Rab

3. ∀y Ray → ∃xRxb

4. ∃x∀y Rxy → ∃xRxb

5. ∃x∀y Rxy → ∀y ∃xRxy

Exercise 3.4.13. Construct a Gentzen-style proof of the sequent

→ (∃x∀y Rxy) ⇒(∀y ∃xRxy).

Solution. A proof in LG

+

consists of the previous proof followed by

5.5. ∃x∀y Rxy → ∀y ∃xRxy, (∃x∀y Rxy) ⇒(∀y ∃xRxy)

6. → (∃x∀y Rxy) ⇒(∀y ∃xRxy)

Exercise 3.4.14. Construct a Gentzen-style proof of the sequent

→ ∃x(Sx∧∀y (Eyx⇔(Sy ∧Eyy))).

Solution. A proof in LG

+

with padding rules omitted is

1. Eaa → Eaa axiom

2. Eaa, Eaa → from 1

3. Eaa, Sa ∧Eaa → from 2

4. Sa → Sa axiom

5. → Eaa, Eaa from 1

6. Sa → Eaa, Sa ∧Eaa from 4 and 5

7. Sa, Eaa ⇔(Sa ∧Eaa) → from 4 and 6

8. Sa, ∀y (Eya ⇔(Sy ∧Eyy)) → from 7

9. Sa ∧∀y (Eya ⇔(Sy ∧Eyy)) → from 8

10. ∃x(Sx∧∀y (Eyx⇔(Sy ∧Eyy)) → from 9

11. → ∃x(Sx∧∀y (Eyx⇔(Sy ∧Eyy)) from 10

Exercise 3.4.15. Construct a Gentzen-style proof of ∃x(Px⇒∀y Py).

Solution. A proof in LG

+

with padding rules omitted is

65

1. Pa → Pa axiom

2. → Pa, Pa ⇒∀y Py from 1

3. → Pa, ∃x(Px⇒∀y Py) from 2

4. → ∀y Py, ∃x(Px⇒∀y Py) from 3

5. → Pa ⇒∀y Py, ∃x(Px⇒∀y Py) from 4

6. → ∃x(Px⇒∀y Py) from 5

Exercise 3.4.16. Construct a Gentzen-style proof of ∃x(Px⇔∀y Py).

Exercise 3.4.17. Let LG(atomic) be a variant of LG in which Γ, A → A, ∆ is

assumed only for atomic L-V -sentences A. Show that LG(atomic) is sound and

complete. (Hint: Use the result of Exercise 2.5.7.)

Exercise 3.4.18.

1. The modiﬁed block tableau rules are a variant of the block tableau rules of

Deﬁnition 3.4.3, replacing each non-branching rule of the form

S, X

[

S, X, Y

1

, Y

2

by a pair of rules

S, X

[

S, X, Y

1

S, X

[

S, X, Y

2

Show that the modiﬁed block tableau rules are sound and complete.

2. Let LG

′

be the variant of LG corresponding to the modiﬁed block tableau

rules. Write out all the rules of LG

′

explicitly. Show that LG

′

is sound

and complete.

3.5 The Interpolation Theorem

As usual, let L be a language and let V be the set of parameters.

Theorem 3.5.1 (the Interpolation Theorem). Let A and B be L-V -sentences.

If A⇒B is logically valid, we can ﬁnd an L-V -sentence I such that:

1. A⇒I and I ⇒B are logically valid.

2. Each predicate and parameter occurring in I occurs in both A and B.

Such an I is called an interpolant for A⇒B. We indicate this by writing

A

I

⇒ B.

66

Remark 3.5.2. If A and B have no predicates in common, then obviously the

theorem is incorrect as stated, because all L-V -sentences necessarily contain at

least one predicate. In this case, we modify the conclusion of the theorem to

say that at least one of A and B is logically valid.

4

The conclusion is obvious

in this case.

In order to prove the Interpolation Theorem, we introduce a “symmetric”

variant of LG, wherein sentences do not move from one side of → to the other.

Deﬁnition 3.5.3. A signed sequent is an expression of the form M → N where

M and N are ﬁnite sets of signed L-V -sentences. A variant of M → N is a

signed sequent obtained from M → N by transferring sentences from one side

of → to the other, changing signs. In particular, M, X → N and M → X, N

are variants of each other, where we use an overline to denote conjugation, i.e.,

TA = FA, FA = TA.

Deﬁnition 3.5.4. Let

C

1

, . . . , C

m

→ D

1

, . . . , D

n

be an unsigned sequent

5

. A signed variant of C

1

, . . . , C

m

→ D

1

, . . . , D

n

is any

variant of the signed sequent

TC

1

, . . . , TC

m

→ TD

1

, . . . , TD

n

.

Note that each signed sequent is a signed variant of one and only one unsigned

sequent. We deﬁne a signed sequent to be logically valid if and only if the

corresponding unsigned sequent is logically valid.

Deﬁnition 3.5.5. LG(symmetric) is the following proof system.

1. The objects are signed sequents.

2. We have

M, X → X, N

and

M, X, X → N

and

M → X, X, N

for all X.

4

This amounts to saying that at least one of the truth values T and F is an interpolant for

A⇒B.

5

An unsigned sequent is just what we have previously called a sequent.

67

3. For each signed tableau rule of the form

.

.

.

X

.

.

.

[

Y

.

.

.

X

.

.

.

[

Y

1

Y

2

.

.

.

X

.

.

.

/ ¸

Y Z

.

.

.

X

.

.

.

/ ¸

Y

1

Z

1

Y

2

Z

2

we have a corresponding pair of signed sequent rules

M, X, Y → N

M, X → N

M → X, Y , N

M → X, N

M, X, Y

1

, Y

2

→ N

M, X → N

M → X, Y

1

, Y

2

, N

M → X, N

M, X, Y → N M, X, Z → N

M, X → N

M → X, Y , N M → X, Z, N

M → X, N

M, X, Y

1

, Y

2

→ N M, X, Z

1

, Z

2

→ N

M, X → N

M → X, Y

1

, Y

2

, N M → X, Z

1

, Z

2

, N

M → X, N

respectively.

Lemma 3.5.6. An unsigned sequent is derivable in LG if and only if all of its

signed variants are derivable in LG(symmetric).

Proof. The proof is by induction on the length of derivations in LG. The

base step consists of noting that all signed variants of Γ, A → A, ∆ are of

the form M, X → X, N or M, X, X → N or M → X, X, N, hence deriv-

able in LG(symmetric). The inductive step consists of checking that, for each

rule of inference of LG, if all signed variants of the premises are derivable

in LG(symmetric), then so are all signed variants of the conclusion. This is

straightforward.

Theorem 3.5.7. LG(symmetric) is sound and complete. In other words, a

signed sequent is logically valid if and only if it is derivable in LG(symmetric).

In particular, an L-V -sentence A⇒B is logically valid if and only if the signed

sequent TA → TB is derivable in LG(symmetric).

Proof. Soundness and completeness of LG(symmetric) follows from Theorem

3.4.9, soundness and completeness of LG, using Lemma 3.5.6.

We now prove the Interpolation Theorem.

68

Deﬁnition 3.5.8. Let M → N be a signed sequent. An interpolant for M → N

is an L-V -sentence I such that the signed sequents M → TI and TI → N are

logically valid, and all predicates and parameters occurring in I occur in both

M and N.

6

We indicate this by writing M

I

→ N.

In order to prove the Interpolation Theorem, it suﬃces by Theorem 3.5.7 to

prove that every signed sequent derivable in LG(symmetric) has an interpolant.

We prove this by induction on the length of derivations.

For the base step, we note that X is an interpolant for M, X → X, N, and

that M, X, X →and → X, X, N are logically valid. Thus we have M, X

X

→ X, N

and M, X, X

F

→ N and M

T

→ X, X, N.

For the induction step we show that, for each rule of LG(symmetric), given

interpolants for the premises of the rule, we can ﬁnd an interpolant for the

conclusion. We present some representative special cases.

M, TA∧B, TA, TB

I

→ N

M, TA∧B

I

→ N

M

I

→ FA∧B, FA, FB, N

M

I

→ FA∧B, N

M, FA∧B, FA

I

→ N M, FA∧B, FB

J

→ N

M, FA∧B

I ∨J

−→ N

M

I

→ TA∧B, TA, N M

J

→ TA∧B, TB, N

M

I ∧J

−→ TA∧B, N

M, TA, FA

I

→ N

M, TA

I

→ N

M

I

→ FA, TA, N

M

I

→ FA, N

M, FA, TA

I

→ N

M, FA

I

→ N

M

I

→ TA, FA, N

M

I

→ TA, N

M, F∀xA, FA[x/a]

I

→ N

M, F∀xA

I

→ N

where a does not occur in the conclusion.

M

I

→ T∀xA, TA[x/a], N

M

I

→ T∀xA, N

6

In the special case when M and N have no predicates in common, we require instead

that at least one of the signed sequents M → and → N be logically valid. This amounts to

requiring that at least one of T, F be an interpolant for M → N.

69

where a does not occur in the conclusion.

M, T∀xA, TA[x/a]

I

→ N

M, T∀xA

K

→ N

where K = I if a occurs in M, T∀xA, otherwise K = ∀z I[a/z] where z is a

new variable.

M

I

→ F∀xA, FA[x/a], N

M

K

→ F∀xA, N

where K = I if a occurs in F∀xA, N, otherwise K = ∃z I[a/z] where z is a new

variable.

This completes the proof.

Example 3.5.9. We give an example illustrating the Interpolation Theorem.

Let n be a large positive integer, say n = 1000. Let A say that the universe

consists of the vertices of a simple, undirected graph with a clique of size n. Let

B say that the graph is not (n−1)-colorable. Both A and B contain a predicate

R denoting adjacency in the graph. A contains a unary predicate Q denoting a

clique. B contains a binary predicate E saying that two vertices get the same

color.

A is:

R and G are irreﬂexive relations on the universe, R is symmet-

ric, G is transitive, ∀x∀y ((Qx∧Qy ∧Gxy) ⇒Rxy), and there exist

x

1

, . . . , x

n

such that Qx

1

∧ ∧Qx

n

∧Gx

1

x

2

∧ ∧Gx

n−1

x

n

.

B is the negation of:

E is an equivalence relation on the universe, ∀x∀y (Rxy ⇒Exy),

and there exist x

1

, . . . , x

n−1

such that ∀y (Ex

1

y ∨ ∨Ex

n−1

y).

Clearly A⇒B is logically valid. Note that the lengths of A and B are O(n), i.e.,

proportional to n. The obvious interpolant I says there exists a clique of size

n, i.e., there exist x

1

, . . . , x

n

such that Rx

1

x

2

∧Rx

1

x

3

∧ ∧Rx

n−1

x

n

. Note

that the length of I is O(n

2

), i.e., proportional to n

2

. It appears that there is

no interpolant of length O(n).

70

Chapter 4

Extensions of Predicate

Calculus

In this chapter we consider various extensions of the predicate calculus. These

extensions may be regarded as inessential features or “bells and whistles” which

are introduced solely in order to make the predicate calculus more user-friendly.

4.1 Predicate Calculus with Identity

Deﬁnition 4.1.1. A language with identity consists of a language L with a

particular binary predicate, I, designated as the identity predicate.

Deﬁnition 4.1.2. Let L be a language with identity. The identity axioms for

L are the following sentences:

1. ∀xIxx (reﬂexivity)

2. ∀x∀y (Ixy ⇔Iyx) (symmetry)

3. ∀x∀y ∀z ((Ixy ∧Iyz) ⇒Ixz) (transitivity)

4. For each n-ary predicate P of L, we have an axiom

∀x

1

∀x

n

∀y

1

∀y

n

((Ix

1

y

1

∧ ∧Ix

n

y

n

) ⇒(Px

1

x

n

⇔Py

1

y

n

))

(congruence).

Exercise 4.1.3. Show that the identity predicate is unique in the following

sense. If L contains two identity predicates I

1

and I

2

, then ∀x∀y (I

1

xy ⇔I

2

xy)

is a logical consequence of the identity axioms for I

1

and I

2

.

Let L be a language with identity.

Deﬁnition 4.1.4. An L-structure M is said to be normal if the identity pred-

icate denotes the identity relation, i.e., I

M

= ¦¸a, a) [ a ∈ U

M

¦.

71

Note that any normal L-structure automatically satisﬁes the identity axioms

for L. Conversely, we have:

Theorem 4.1.5. Let M be an L-structure satisfying the identity axioms for

L. For each a ∈ U

M

put a = ¦b ∈ U

M

[ v

M

(Iab) = T¦. Then we have a normal

L-structure M and an onto mapping φ : U

M

→ U

M

as in Theorem 2.7.3, deﬁned

by putting U

M

= ¦a [ a ∈ U

M

¦, and P

M

= ¦¸a

1

, . . . , a

n

) [ ¸a

1

, . . . , a

n

) ∈ P

M

¦

for all n-ary predicates P.

Proof. This is straightforward, using the fact that I

M

is a congruence with

respect to each of the relations P

M

, P ∈ L.

Theorem 4.1.6. If M is an L-structure satisfying the identity axioms for L,

then we have a normal L-structure M satisfying the same sentences as M.

Proof. This is immediate from Theorems 4.1.5 and 2.7.3.

Let S be a set of L-sentences.

Deﬁnition 4.1.7. S is normally satisﬁable if there exists a normal L-structure

which satisﬁes S.

Corollary 4.1.8. S is normally satisﬁable if and only if

S ∪ ¦identity axioms for L¦

is satisﬁable.

We also have the Compactness Theorem for normal satisﬁability:

Corollary 4.1.9. S is normally satisﬁable if and only if each ﬁnite subset of S

is normally satisﬁable.

Proof. This is immediate from Corollary 4.1.8 plus the Compactness Theorem

for predicate calculus without identity (Theorems 2.6.1 and 2.6.2), applied to

the set S ∪ ¦identity axioms for L¦.

Regarding normal satisﬁability in particular domains, we have:

Example 4.1.10. Given a positive integer n, we exhibit a sentence E

n

which

is normally satisﬁable in domains of cardinality n but not in domains of any

other cardinality. The sentence

∃x

1

∃x

n

(∀y (Ix

1

y ∨ ∨Ix

n

y) ∧(Ix

1

x

2

∨Ix

1

x

3

∨ ∨Ix

n−1

x

n

))

has this property. Intuitively, E

n

says that there exist exactly n things.

Exercise 4.1.11. Let M = (U

M

, f

M

, g

M

, I

M

) where U

M

= ¦0, 1, 2, 3, 4¦, I

M

is

the identity relation on U

M

, and f

M

, g

M

are the binary operations of addition

and multiplication modulo 5. Thus M is essentially just the ring of integers

modulo 5. Let L be the language consisting of f, g, I. Note that M is a normal

L-structure.

Write an L-sentence A such that for all normal L-structures M

′

, M

′

satisﬁes

A if and only if M

′

is isomorphic to M.

72

Solution. A brute force solution is to let A be ∃x

0

∃x

1

∃x

2

∃x

3

∃x

4

B, where B

is the conjunction of ¦∀y (Ix

0

y ∨Ix

1

y ∨Ix

2

y ∨Ix

3

y ∨Ix

4

y)¦ ∪ ¦Ix

i

x

j

: i ,=

j¦ ∪ ¦Ifx

i

x

j

x

k

[ i + j = k mod 5¦ ∪ ¦Igx

i

x

j

x

k

[ ij = k mod 5¦ with i, j, k

ranging over 0, 1, 2, 3, 4.

Another solution is to let A be a sentence describing a ﬁeld consisting of 5

elements. Namely, let A be the conjunction of the ﬁeld axioms plus “there exist

exactly 5 things”. We are using the algebraic fact that, up to isomorphism,

there is exactly one ﬁeld of 5 elements.

Exercise 4.1.12. Let L be a ﬁnite language with identity, and let M be a

ﬁnite normal L-structure. Construct an L-sentence A such that, for all normal

L-structures M

′

, M

′

[= A if and only if M

′

is isomorphic to M.

Solution. Let a

1

, . . . , a

k

be the elements of U

M

, and let P, . . . , Q be the predi-

cates of L. As A we may take

∃x

1

∃x

k

(D

P

∧ ∧D

Q

∧∀y (Ix

1

y ∨ ∨Ix

k

y))

where for each n-ary predicate P of L, D

P

is the conjunction of Px

i1

x

in

for

each n-tuple ¸a

i1

, . . . , a

in

) ∈ P

M

, and Px

i1

x

in

for each n-tuple

¸a

i1

, . . . , a

in

) / ∈ P

M

.

On the other hand, we have:

Theorem 4.1.13 (L¨owenheim/Skolem Theorem).

1. If S is normally satisﬁable in arbitrarily large ﬁnite domains, then S is

normally satisﬁable in some inﬁnite domain.

2. If S is normally satisﬁable in some inﬁnite domain, then S is normally

satisﬁable in all inﬁnite domains of cardinality ≥ the cardinality of S.

Proof. For the ﬁrst part, let S

∗

= S ∪ ¦H

n

[ n = 1, 2, . . .¦ where H

n

is the

sentence

∃x

1

∃x

n

(Ix

1

x

2

∨Ix

1

x

3

∨ ∨Ix

n−1

x

n

)

saying that there exist at least n things. Since S is normally satisﬁable in

arbitrarily large ﬁnite domains, each ﬁnite subset of S

∗

is normally satisﬁ-

able. Hence, by Corollary 4.1.9, S

∗

is normally satisﬁable. But any normal

L-structure satisfying S

∗

satisﬁes S and has an inﬁnite domain.

For the second part, let κ be a cardinal number ≥ the cardinality of S. Let

L

∗

= L∪¦Q

i

[ i ∈ X¦, where each Q

i

is a new 1-ary predicate, and X is a set of

cardinality κ. Let S

∗

= S ∪¦∃xQ

i

x [ i ∈ X¦ ∪ ¦∃x(Q

i

x∧Q

j

x) [ i, j ∈ X, i ,=

j¦ ∪¦identity axioms for L

∗

¦. Thus S

∗

is a set of L

∗

-sentences of cardinality κ.

Furthermore, any domain in which S

∗

is satisﬁable will contain pairwise distinct

elements a

i

, i ∈ X, and will therefore have cardinality ≥ κ. By assumption, S

is normally satisﬁable in some inﬁnite domain. It follows that each ﬁnite subset

of S

∗

is satisﬁable. Hence, by the Compactness Theorems 2.6.1 and 2.6.2, S

∗

is

satisﬁable. Hence, by part 2 of Theorem 2.7.1, S

∗

is satisﬁable in a domain of

73

cardinality κ. Therefore, by Theorems 4.1.5 and 4.1.6, S

∗

is normally satisﬁable

in a domain of cardinality ≤ κ, hence = κ. Let M

∗

be a normal L

∗

-structure

with U

M

∗ of cardinality κ. Let M be the reduct of M

∗

to L, i.e., M is the

L-structure with U

M

= U

M

∗ and P

M

= P

M

∗ for each predicate P in L. Then

M normally satisﬁes S and U

M

is of cardinality κ.

Exercise 4.1.14. Let L be the following language:

Ox: x = 1

Pxyz: x +y = z

Qxyz: x y = z

Rxy: x < y

Sxy: x + 1 = y

Ixy: x = y (identity predicate)

For each positive integer n, let M

n

be the normal L-structure

M

n

= (U

n

, O

n

, P

n

, Q

n

, R

n

, S

n

, I

n

)

where

U

n

= ¦1, . . . , n¦

O

n

= ¦1¦

P

n

= ¦¸i, j, k) ∈ (U

n

)

3

[ i +j = k¦

Q

n

= ¦¸i, j, k) ∈ (U

n

)

3

[ i j = k¦

R

n

= ¦¸i, j) ∈ (U

n

)

2

[ i < j¦

S

n

= ¦¸i, j) ∈ (U

n

)

2

[ i + 1 = j¦

I

n

= ¦¸i, j) ∈ (U

n

)

2

[ i = j¦

Exhibit an L-sentence Z such that, for all ﬁnite normal L-structures M

′

, M

′

[=

Z if and only if M

′

is isomorphic to M

n

for some n.

Solution. As Z we may take the conjunction of the following clauses.

(a) ∀x∀y (Rxy ∨Ryx∨Ixy)

(b) ∀x∀y (Rxy ⇒Ryx)

(c) ∀x∀y ∀z ((Rxy ∧Ryz) ⇒Rxz)

(d) ∀x∀z (Sxz ⇔(Rxz ∧∃y (Rxy ∧Ryz)))

(e) ∀u (Ou ⇔∃xRxu)

74

(f) ∀u (Ou ⇒∀x∀z (Puxz ⇔Sxz))

(g) ∀v ∀w(Svw⇒∀x∀z (Pwxz ⇔∃y (Syz ∧Pvxy)))

(h) ∀u (Ou ⇒∀x∀z (Quxz ⇔Ixz))

(i) ∀v ∀w(Svw⇒∀x∀z (Qwxz ⇔∃y (Qvxy ∧Pxyz)))

Clauses (a), (b) and (c) say that R is an irreﬂexive linear ordering of the universe.

Clause (d) says that S is the immediate successor relation, with respect to R.

Clause (e) says that 1 is the ﬁrst element of the universe, with respect to R.

Clauses (f) and (g) deﬁne the addition predicate P, by induction along R,

in terms of S. Clauses (h) and (i) deﬁne the multiplication predicate Q, by

induction along R, in terms of S and P.

Exercise 4.1.15. Let L and M

n

be as in Exercise 4.1.14. Show that there

exists an inﬁnite normal L-structure M = M

∞

with the following property: for

all L-sentences A, if M

p

[= A for all suﬃciently large primes p, then M

∞

[= A.

(Hint: Use the Compactness Theorem.)

Solution. Let S be the set of L-sentences A with the following property: there

exists n = n

A

such that for all primes p > n

A

, M

p

satisﬁes A. We claim that

every ﬁnite subset of S is normally satisﬁable. To see this, let S

0

= ¦A

1

, . . . , A

k

¦

be a ﬁnite subset of S. Put n = max(n

A1

, . . . , n

A

k

). Let p be any prime > n.

Then M

p

satisﬁes A

1

, . . . , A

k

. This proves our claim. By the Compactness

Theorem for normal satisﬁability (Corollary 4.1.9), it follows that S is normally

satisﬁable. Let M

∞

be a normal L-structure satisfying S. Among the sentences

of S are those asserting that the universe has at least k elements, for each

positive integer k. Since M

∞

satisﬁes these sentences, M

∞

is inﬁnite.

4.2 The Spectrum Problem

Deﬁnition 4.2.1. Let A be a sentence of the predicate calculus with identity.

The spectrum of A is the set of positive integers n such that A is normally

satisﬁable in a domain of cardinality n. A spectrum is a set X of positive

integers, such that X = spectrum(A) for some A.

Remark 4.2.2. The spectrum problem is the problem of characterizing the

spectra, among all sets of positive integers. This is a famous and apparently

diﬃcult open problem.

1

In particular, it is unknown whether the complement

of a spectrum is necessarily a spectrum.

Example 4.2.3. We show that the set ¦n ≥ 1 [ n is even¦ is a spectrum.

1

Jones/Selman [1] show that X is a spectrum if and only if there exists a nondeterministic

Turing machine which accepts X in time 2

ck

, where k is the length of the input. Since the

input is a positive integer n, we have k = [log

2

n], as usual in computational number theory.

75

Let U be a nonempty set. A binary relation R ⊆ U

2

is said to be an

equivalence relation on U if it is reﬂexive, symmetric, and transitive, i.e., if the

structure (U, R) satisﬁes (1) ∧(2) ∧(3):

(1) ∀xRxx

(2) ∀x∀y (Rxy ⇔Ryx)

(3) ∀x∀y ∀z ((Rxy ∧Ryz) ⇒Rxz)

In this situation, the equivalence classes [a]

R

= ¦b ∈ U [ ¸a, b) ∈ R¦, a ∈ U,

form a partition of U, i.e., a decomposition of the set U into pairwise disjoint,

nonempty subsets.

Let A be the following sentence of the predicate calculus with identity:

(1) ∧ (2) ∧ (3) ∧ ∀x∃y ((Ixy) ∧∀z (Rxz ⇔(Ixz ∨Iyz)))

Intuitively, A says that R is an equivalence relation with the property that each

equivalence class consists of exactly two elements. Obviously, a ﬁnite set U

admits an equivalence relation with this property if and only if the cardinality

of U is even. Thus the spectrum of A is the set of even numbers.

Exercises 4.2.4. Prove the following.

1. If X is a ﬁnite or coﬁnite

2

set of positive integers, then X is a spectrum.

2. The set of even numbers is a spectrum.

3. The set of odd numbers is a spectrum.

4. If r and m are positive integers, ¦n ≥ 1 [ n ≡ r mod m¦ is a spectrum.

5. If X and Y are spectra, X ∪ Y and X ∩ Y are spectra.

Solution.

1. Let E

n

be sentence in the language with only the identity predicate I,

saying that the universe consists of exactly n elements (Exercise 4.1.10).

If X = ¦n

1

, . . . , n

k

¦, then X is the spectrum of E

n1

∨ ∨E

n

k

, and the

complement of X is spectrum of (E

n1

∨ ∨E

n

k

).

2. The even numbers are the spectrum of a sentence which says: R is an

equivalence relation on the universe, such that each equivalence class con-

sists of exactly two elements. For more details, see Example 4.2.3.

3. The odd numbers are the spectrum of a sentence which says: R is an equiv-

alence relation on the universe, such that each equivalence class consists

of exactly two elements, except for one equivalence class, which consists

of exactly one element.

4. We may assume that 0 ≤ r < m. If r = 0, the set

2

A set of positive integers is said to be coﬁnite if its complement is ﬁnite.

76

¦n ≥ 1 : n ≡ 0 mod m¦ = ¦n ≥ 1 : m divides n with no remainder¦

is the spectrum of a sentence which says: R is an equivalence relation

on the universe, such that each equivalence class consists of exactly m

elements. If r > 0, the set

¦n ≥ 1 : n ≡ r mod m¦ = ¦n ≥ 1 : m divides n with remainder r¦

is the spectrum of a sentence which says: R is an equivalence relation

on the universe, such that each equivalence class consists of exactly m

elements, except for one equivalence class, which consists of exactly r

elements.

5. Assume that X is the spectrum of A and Y is the spectrum of B. Then X∪

Y is the spectrum of A∨B. Also, X∩Y is the spectrum of A∧B, provided

A and B have no predicates in common except the identity predicate.

To arrange for this, replace B by an analogous sentence in a diﬀerent

language.

Exercise 4.2.5. Prove that, for any sentence A of the predicate calculus with

identity, at least one of spectrum(A) and spectrum(A) is coﬁnite. (Hint: Use

part 1 of Theorem 4.1.13.)

Example 4.2.6. We show that the set of composite numbers

3

is a spectrum.

Let L be a language consisting of two binary predicates, R and S, as well

as the identity predicate, I. Let A be an L-sentence saying that R and S are

equivalence relations, each with more than one equivalence class, and

∀x∀y (∃ exactly one z)(Rxz ∧Syz).

Thus, for any normal L-structure M = (U

M

, R

M

, S

M

, I

M

) satisfying A, we have

that R

M

and S

M

partition U

M

into “rows” and “columns”, respectively, in such

a way that the intersection of any “row” with any “column” consists of exactly

one element of U

M

. Thus, if U

M

is ﬁnite, the elements of U

M

are arranged

in an m n “matrix”, where m, n ≥ 2. Therefore, the number of elements

in U

M

is mn, a composite number. Conversely, for any m, n ≥ 2, there is an

L-structure M as above, which satisﬁes A. Thus spectrum(A) is the set of

composite numbers.

Exercise 4.2.7. Use the result of Exercise 4.1.14 to prove the following:

1. The set of prime numbers and its complement are spectra.

2. The set of squares ¦1, 4, 9, . . .¦ and its complement are spectra.

3. The set of powers of 2, ¦2

n

[ n = 1, 2, 3, . . .¦, and its complement, are

spectra.

3

A composite number is an integer greater than 1 which is not prime.

77

4. The set of prime powers ¦p

n

[ p prime, n = 1, 2, . . .¦ and its complement

are spectra.

Solution. Let Z be as in Exercise 4.1.14 above. For each of the given sets X, we

exhibit a sentence A with the following properties: X is the spectrum of Z ∧A,

and the complement of X is the spectrum of Z ∧A.

1. ∃z ((∃wRzw) ∧(∃x∃y (Rxz ∧Ryz ∧Qxyz)) ∧(Oz)).

2. ∃z ((∃wRzw) ∧∃xQxxz).

3. ∃z ∃v ((∃wRzw) ∧(∃u (Ou ∧Suv))

∧∀x((Ox∧ ∃y Qxyz) ⇒∃wQvwx)).

4. ∃z ∃v ((∃wRzw) ∧(∃x∃y (Rxv ∧Ryv ∧Qxyv)) ∧(Ov)

∧∀x((Ox∧ ∃y Qxyz) ⇒∃wQvwx)).

Exercise 4.2.8.

1. The Fibonacci numbers are deﬁned recursively by F

1

= 1, F

2

= 2, F

n

=

F

n−1

+F

n−2

for n ≥ 3. Show that the set of Fibonacci numbers

¦F

n

[ n = 1, 2, . . .¦ = ¦1, 2, 3, 5, 8, 13, 21, 34, 55, . . .¦

and its complement are spectra.

2. Show that ¦x

y

[ x, y ≥ 2¦ and its complement are spectra.

Exercise 4.2.9. Let X be a subset of ¦1, 2, 3, . . .¦. Prove that if X is a spectrum

then ¦n

2

[ n ∈ X¦ is a spectrum.

4.3 Predicate Calculus With Operations

In this section we extend the syntax and semantics of the predicate calculus,

to encompass operations. As examples of operations, we may cite the familiar

mathematical operations of addition (+) and multiplication (). Such opera-

tions are considered binary, because they take two arguments. More generally,

we consider n-ary operations.

Deﬁnition 4.3.1 (languages). A language is a set of predicates P, Q, R, . . . and

operations f, g, h, . . .. Each predicate and each operation is designated as n-ary

for some nonnegative

4

integer n.

Deﬁnition 4.3.2 (terms, formulas, sentences). Let L be a language, and let U

be a set. The set of L-U-terms is generated as follows.

1. Each variable is an L-U-term.

4

A 0-ary operation is known as a constant. Syntactically, constants behave as parameters.

78

2. Each element of U is an L-U-term.

3. If f is an n-ary operation of L, and if t

1

, . . . , t

n

are L-U-terms, then

ft

1

t

n

is an L-U-term.

An L-U-term is said to be variable-free if no variables occur in it. An atomic

L-U-formula is an expression of the form

Pt

1

t

n

where P is an n-ary predicate of L, and t

1

, . . . , t

n

are L-U-terms. The set

of L-U-formulas is generated as in clauses 2, 3, 4 and 5 of Deﬁnition 2.1.3.

The notions of substitution, free variables, and L-U-sentences are deﬁned as in

Section 2.1. Note that Pt

1

t

n

is a sentence if and only if it is variable-free.

Deﬁnition 4.3.3 (structures). An L-structure M consists of a nonempty set

U

M

, an n-ary relation P

M

⊆ (U

M

)

n

for each n-ary predicate P of L, and an

n-ary function f

M

: (U

M

)

n

→ U

M

for each n-ary operation f of L.

Deﬁnition 4.3.4 (isomorphism). Two L-structures M and M

′

are said to be

isomorphic if there exists an isomorphism of M onto M

′

, i.e., a one-to-one

correspondence φ : U

M

∼

= U

M

′ such that:

1. for all n-ary predicates P of L and all n-tuples ¸a

1

, . . . , a

n

) ∈ (U

M

)

n

,

¸a

1

, . . . , a

n

) ∈ P

M

if and only if ¸φ(a

1

), . . . , φ(a

n

)) ∈ P

M

′ .

2. for all n-ary operations f of L and all n-tuples ¸a

1

, . . . , a

n

) ∈ (U

M

)

n

,

φ(f

M

(a

1

, . . . , a

n

) = f

M

′ (φ(a

1

), . . . , φ(a

n

)).

Lemma 4.3.5 (valuations). Let M be an L-structure.

1. There is a unique valuation

v

M

: ¦t [ t is a variable-free L-U

M

-term¦ → U

M

deﬁned as follows:

(a) v

M

(a) = a for all a ∈ U

M

.

(b) v

M

(ft

1

t

n

) = f

M

(v

M

(t

1

), . . . , v

M

(t

n

)) for all n-ary operations f

of L and all variable-free L-U

M

-terms t

1

, . . . , t

n

.

2. There is a unique valuation

v

M

: ¦A [ A is an L-U

M

-sentence¦ → ¦T, F¦

deﬁned as follows. For atomic L-U-sentences, we have

v

M

(Pt

1

t

n

) =

_

T if ¸v

M

(t

1

), . . . , v

M

(t

n

)) ∈ P

M

,

F if ¸v

M

(t

1

), . . . , v

M

(t

n

)) / ∈ P

M

.

For non-atomic L-U

M

-sentences, v

M

(A) is deﬁned as in clauses 2 through

8 of Lemma 2.2.4.

79

Proof. The proof is as for Lemma 2.2.4.

Deﬁnition 4.3.6 (tableau method). The signed and unsigned tableau methods

carry over to predicate calculus with operations. We modify the tableau rules

as follows.

Signed:

.

.

.

T∀xA

.

.

.

[

TA[x/t]

.

.

.

F∃xA

.

.

.

[

FA[x/t]

where t is a variable-free term

.

.

.

T∃xA

.

.

.

[

TA[x/a]

.

.

.

F∀xA

.

.

.

[

FA[x/a]

where a is a new parameter

Unsigned:

.

.

.

∀xA

.

.

.

[

A[x/t]

.

.

.

∃xA

.

.

.

[

A[x/t]

where t is a variable-free term

.

.

.

∃xA

.

.

.

[

A[x/a]

.

.

.

∀xA

.

.

.

[

A[x/a]

where a is a new parameter

80

Remark 4.3.7 (soundness and completeness). With the tableau rules as above,

the Soundness Theorem 2.3.13 carries over unchanged to the context of predicate

calculus with operations. The results of Section 2.4 on logical equivalence also

carry over. The notion of U-repleteness (Deﬁnition 2.5.2) is modiﬁed to say

that, for example, if S contains ∀xA then S contains A[x/t] for all variable-free

L-U-terms t. The conclusion of Hintikka’s Lemma 2.5.3 is modiﬁed to say that

S is satisﬁable in the domain of variable-free L-U-terms. The conclusion of the

Completeness Theorem 2.5.5 is modiﬁed to say that X

1

, . . . , X

k

is satisﬁable in

the domain of variable-free L-V -terms. The Compactness Theorems 2.6.1 and

2.6.2 carry over unchanged.

Remark 4.3.8 (satisﬁability in a domain). The notion of satisﬁability in a

domain carries over unchanged to the context of predicate calculus with op-

erations. Theorems 2.2.6 and 2.2.11 on isomorphism, and Theorem 2.7.1 on

satisﬁability in inﬁnite domains, also carry over. Theorem 2.7.3 carries over in

an appropriately modiﬁed form. See Theorem 4.3.9 and Exercise 4.3.10 below.

Theorem 4.3.9. Let M and M

′

be L-structures. Assume that φ : U

M

→ U

M

′

is an onto mapping such that conditions 1 and 2 of Deﬁnition 4.3.4 hold. Then

as in Theorem 2.2.6 we have v

M

(A) = v

M

′ (A

′

) for all L-U

M

-sentences A, where

A

′

= A[a

1

/φ(a

1

), . . . , a

k

/φ(a

k

)]. In particular, M and M

′

satisfy the same

L-sentences.

Proof. The proof is similar to that of Theorem 2.7.3.

Exercise 4.3.10. Use Theorem 4.3.9 to show that Corollary 2.7.4 carries over

to the context of predicate calculus with operations.

Remark 4.3.11 (companions and proof systems). In our notion of companion

(Deﬁnition 3.2.3), clauses (1) and (4) are modiﬁed as follows:

(1) (∀xB) ⇒B[x/t]

(4) B[x/t] ⇒(∃xB)

where t is any variable-free term. In our Hilbert-style proof system LH, the

instantiation rules are modiﬁed as follows:

(a) (∀xB) ⇒B[x/t] (universal instantiation)

(b) B[x/t] ⇒(∃xB) (existential instantiation)

where t is any variable-free term. Also, our Gentzen-style proof system LG is

modiﬁed in accordance with the modiﬁed tableau rules. With these changes,

the soundness and completeness of LG and LH carry over.

Exercise 4.3.12 (the Interpolation Theorem). Strengthen the Interpolation

Theorem 3.5.1 to say that each operation, predicate and parameter occurring in

I occurs in both A and B. (Hint: The version with operations can be deduced

from the version without operations.)

Exercise 4.3.13. Skolemization.

81

4.4 Predicate Calculus with Identity and Oper-

ations

Remark 4.4.1 (predicate calculus with identity and operations). We augment

the identity axioms (Deﬁnition 4.1.2) as follows:

5. For each n-ary operation f of L, we have an axiom

∀x

1

∀x

n

∀y

1

∀y

n

((Ix

1

y

1

∧ ∧Ix

n

y

n

) ⇒Ifx

1

x

n

fy

1

y

n

).

The notions of normal structure and normal satisﬁability are deﬁned as before.

The results of Section 4.1 on the predicate calculus with identity carry over

unchanged to the predicate calculus with identity and operations. See also

Exercise 4.4.2 below.

Exercise 4.4.2 (elimination of operations). Let L be a language with identity

and operations. Let L

o

be the language with identity and without operations,

obtained by replacing each n-ary operation f belonging to L by a new (n + 1)-

ary predicate P

f

belonging to L

o

. Each normal L-structure M gives rise to a

normal L

o

-structure M

o

where

(P

f

)

M

′ = ¦¸a

1

, . . . , a

n

, b) ∈ (U

M

)

n+1

[ f

M

(a

1

, . . . , a

n

) = b¦.

For each n-ary operation f of L, let G

f

be the L

o

-sentence

∀x

1

∀x

n

∃y ∀z (Iyz ⇔P

f

x

1

x

n

z).

1. Show that to each L-sentence A we may associate an L

o

-sentence A

o

such

that, for all L-structures M, M [= A if and only if M

o

[= A

o

.

2. Show that a normal L

o

-structure satisﬁes the sentences G

f

, f in L, if and

only if it is of the form M

o

for some L-structure M.

Exercise 4.4.3. Show that the spectrum problem for predicate calculus with

identity and operations is equivalent to the spectrum problem for predicate

calculus with identity and without operations, as previously discussed in Section

4.2. In other words, given a sentence A involving some operations, construct a

sentence A

oo

involving no operations, such that spectrum(A) = spectrum(A

oo

).

(Hint: Use the result of Exercise 4.4.2. Note that A

oo

will not be the same as

the A

o

of Exercise 4.4.2.)

Remark 4.4.4 (predicate calculus with equality). The predicate calculus with

identity and operations is well suited for the study of algebraic structures such

as number systems, groups, rings, etc. In such a context, one often writes t

1

= t

2

instead of It

1

t

2

, and one refers to predicate calculus with equality rather than

predicate calculus with identity. In this notation, the equality axioms (i.e., the

identity axioms) read as follows:

∀x(x = x),

82

∀x∀y (x = y ⇔y = x),

∀x∀y ∀z ((x = y ∧y = z) ⇒x = z),

∀x

1

∀y

1

∀x

n

∀y

n

((x

1

= y

1

∧ ∧x

n

= y

n

) ⇒(Px

1

x

n

⇔Py

1

y

n

)),

for each n-ary predicate P,

∀x

1

∀y

1

∀x

n

∀y

n

((x

1

= y

1

∧ ∧x

n

= y

n

) ⇒fx

1

x

n

= fy

1

y

n

),

for each n-ary operation f .

One also uses customary algebraic notation, e.g., t

1

+t

2

instead of +t

1

t

2

, t

1

t

2

or t

1

t

2

instead of t

1

t

2

, etc. To avoid ambiguity, parentheses are used.

Examples 4.4.5 (groups and rings). Using predicate calculus with identity

and operations, a group may be viewed as a normal L-structure

G = (U

G

, f

G

, i

G

, e

G

, I

G

).

Here U

G

is the underlying set of the group, and L is the language ¦f, i, e, I¦,

where f is the group composition law (a binary operation), i is group inversion

(a unary operation), e is the group identity element (a 0-ary operation, i.e., a

constant), and I is the identity predicate (a binary predicate). We could refer

to L as the language of groups. It is customary to write G instead of U

G

, t

1

t

2

or t

1

t

2

instead of ft

1

t

2

, t

−1

instead of it, 1 instead of e, and t

1

= t

2

instead of

It

1

t

2

. Thus

G = (G,

G

,

−1

G

, 1

G

, =

G

)

and G is required to satisfy the group axioms, consisting of the identity axioms

for L, plus ∀x∀y ∀z ((xy)z = x(yz)), ∀x(x

−1

x = xx

−1

= 1), ∀x(1x = x1 = x).

Similarly, a ring may be viewed as a normal structure

R = (R, +

R

,

R

, −

R

, 0

R

, 1

R

, =

R

)

where + and are binary operations, − is a unary operation, 0 and 1 are

constants, and = is the equality predicate. We could refer to the language

¦+, , −, 0, 1, =¦ as the language of rings. R is required to satisfy the ring axioms,

consisting of the identity axioms plus ∀x∀y ∀y (x + (y + z) = (x + y) + z),

∀x∀y (x +y = y +x), ∀x(x + 0 = x), ∀x(x + (−x) = 0), ∀x∀y ∀z (x (y z) =

(x y) z), ∀x(x 1 = 1 x = x), ∀x∀y ∀z (x (y + z) = (x y) + (x z)),

∀x∀y ∀z ((x + y) z = (x z) + (y z)), 0 ,= 1.

Exercise 4.4.6. Let G be a group. For a ∈ G write a

n

= a a (n times).

Thus a

1

= a and a

n+1

= a

n

a. We say that G is a torsion group if for all a ∈ G

there exists a positive integer n such that a

n

= 1. We say that G is torsion-free

if for all a ∈ G, if a ,= 1 then a

n

,= 1 for all positive integers n.

1. Show that the class of torsion-free groups can be characterized by a set of

sentences. I.e., there is a set of sentences S such that, for all groups G, G

is torsion-free if and only if G [= S.

Solution. Let S = ¦A

n

: n ≥ 2¦, where A

n

is the sentence

83

∀x(x ,= 1 ⇒x

n

,= 1).

Clearly the groups satisfying S are exactly the torsion-free groups.

2. Show that the class of torsion-free groups cannot be characterized by a

ﬁnite set of sentences.

Solution. Suppose S

′

were a ﬁnite of sentences such that the groups sat-

isfying S

′

are exactly the torsion-free groups. In particular, each sentence

in S

′

is a logical consequence of the group axioms plus S = ¦A

n

: n ≥ 2¦

as above. By the Compactness Theorem, each sentence in S

′

is a logical

consequence of the group axioms plus ¦A

2

, . . . , A

n

¦ for suﬃciently large n.

Since S

′

is ﬁnite, there is a ﬁxed n such that all of the sentences in S

′

are

logical consequences of the group axioms plus ¦A

2

, . . . , A

n

¦. Now let G

be a torsion group satisfying ¦A

2

, . . . , A

n

¦. (For example, we may take G

to be the additive group of integers modulo p, where p is a prime number

greater than n.) Then G satisﬁes S

′

yet is not torsion-free, contradicting

our assumption on S

′

.

3. Show that the class of torsion groups cannot be characterized by a set of

sentences. I.e., there is no set of sentences S with the property that, for

all groups G, G is a torsion group if and only if G [= S.

Exercise 4.4.7. Let L be the language of groups. Let S be the set of L-

sentences which are true in all ﬁnite groups. Deﬁne a pseudo-ﬁnite group to be

a group which satisﬁes S. Note that every ﬁnite group is pseudo-ﬁnite.

Does there exist an inﬁnite, pseudo-ﬁnite group? Prove your answer.

4.5 Many-Sorted Predicate Calculus

Deﬁnition 4.5.1 (many-sorted languages). A many-sorted language consists

of

1. a set of sorts σ, τ, . . .,

2. a set of predicates P, Q, . . ., each designated as n-ary of type

σ

1

σ

n

for some nonnegative integer n and sorts σ

1

, . . . , σ

n

,

3. a set of operations f, g, . . ., each designated as n-ary of type

σ

1

σ

n

→ σ

n+1

for some nonnegative integer n and sorts σ

1

, . . . , σ

n

, σ

n+1

.

84

Deﬁnition 4.5.2 (terms, formulas, sentences). Let L be a many-sorted lan-

guage. For each sort σ, we assume a ﬁxed, countably inﬁnite set of variables

of sort σ, denoted x

σ

, y

σ

, z

σ

, . . .. Let U = (U

σ

, U

τ

, . . .) consist of a set U

σ

for

each sort σ of L. The L-U-terms are generated as follows.

1. Each variable of sort σ is a term of sort σ.

2. Each element of U

σ

is a term of sort σ.

3. If f is an n-ary operaton of type σ

1

σ

n

→ σ

n+1

, and if t

1

, . . . , t

n

are terms of sort σ

1

, . . . , σ

n

respectively, then ft

1

. . . t

n

is a term of sort

σ

n+1

.

An atomic L-U-formula is an expression of the form Pt

1

. . . t

n

, where P is an

n-ary predicate of type σ

1

σ

n

, and t

1

, . . . , t

n

are terms of sort σ

1

, . . . , σ

n

respectively. The L-U-formulas are generated as in Deﬁnition 2.1.3, with clause

5 modiﬁed as follows:

5

′

. If x

σ

is a variable of sort σ, and if A is an L-U-formula, then ∀x

σ

A and

∃x

σ

A are L-U-formulas.

Our notions of substitution, free and bound variables, sentences, etc., are ex-

tended in the obvious way to the many-sorted context. Naturally, the substitu-

tion A[x

σ

/t] makes sense only when t is a term of sort σ. An L-formula is an

L-U-formula where U

σ

= ∅ for each sort σ.

Deﬁnition 4.5.3 (many-sorted structures). An L-structure M consists of

1. a nonempty set U

σ

M

for each sort σ of L,

2. an n-ary relation P

M

⊆ U

σ1

M

U

σn

M

for each n-ary predicate P of

type σ

1

σ

n

belonging to L,

3. an n-ary function f

M

: U

σ1

M

U

σn

M

→ U

σn+1

M

for each n-ary operation

of type σ

1

σ

n

→ σ

n+1

belonging to L.

Notions such as isomorphism, valuation, truth, satisﬁability, and results such

as Theorem 2.2.6 on isomorphism, and Theorem 4.3.9 on onto mappings, carry

over to the many-sorted context in the obvious way.

Deﬁnition 4.5.4 (many-sorted domains). We deﬁne a domain or universe for

L to be an indexed family of nonempty sets U = (U

σ

, U

τ

, . . .), where σ, τ, . . . are

the sorts of L. In this way, the notion of satisﬁability in a domain generalizes

to the many-sorted context.

Remark 4.5.5 (tableau method, proof systems). For each sort σ of L, ﬁx a

countably inﬁnite set V

σ

= ¦a

σ

, b

σ

, . . .¦, the set of parameters of sort σ. Then

the tableau method carries over in the obvious way, generalizing Remark 4.3.7.

In the Completeness Theorem for the tableau method, we obtain satisﬁability

in the domain U = (U

σ

, U

τ

, . . .), where U

σ

is the set of variable-free L-V -terms

of sort σ, with V = (V

σ

, V

τ

, . . .). The soundness and completeness of our proof

systems LH and LG and the Interpolation Theorem also carry over, just as in

Section 4.3.

85

Remark 4.5.6 (identity predicates). For each sort σ of L, L may or may not

contain a binary predicate I

σ

of type σ σ designated as the identity predicate

for σ. As identity axioms we may take the universal closures of all L-formulas

of the form

∀x

σ

∀y

σ

(I

σ

xy ⇒(A⇔A[x/y]))

where A is atomic. An L-structure M is said to be normal if I

σ

M

= ¦¸a, a) [ a ∈

U

σ

M

¦ for all σ such that I

σ

belongs to L. The results of Section 4.1 concerning

normal satisﬁability carry over to the many-sorted context.

Deﬁnition 4.5.7 (languages with identity). A many-sorted language with iden-

tity is a many-sorted language which contains an identity predicate for each sort.

Remark 4.5.8 (many-sorted spectrum problem). Let L be a many-sorted lan-

guage with identity. If A is an L-sentence and σ

1

, . . . , σ

k

are the sorts occurring

in A, the spectrum of A is the set of k-tuples of positive integers (n

1

, . . . , n

k

)

such that there exists a normal L-structure M with U

σi

M

of cardinality n

i

, for

i = 1, . . . , k. In this way, the spectrum problem carries over to many-sorted

predicate calculus. So far as I know, the problem of characterizing many-sorted

spectra has not been investigated thoroughly.

Remark 4.5.9 (many-sorted L¨ owenheim/Skolem theorems). It is natural to try

to generalize the L¨ owenheim/Skolem Theorem 4.1.13 to many-sorted predicate

calculus. This is straightforward provided we consider only normal structures

M where all of the domains U

σ

M

, U

τ

M

, . . . are of the same inﬁnite cardinality.

However, if we require U

σ

M

, U

τ

M

, . . . to be of speciﬁed distinct cardinalities, then

this leads to diﬃcult issues. Even for two sorts, the topic of so-called two-

cardinal theorems turns out to be rather delicate and complicated. See for

example the model theory textbook of Marker [2].

Remark 4.5.10. Our reasons for including many-sorted predicate calculus in

this course are as follows:

1. it is more useful . . . .

FIXME

Remark 4.5.11 (one-sorted languages). A language or structure is said to

be one-sorted if it has only one sort. This term is used for contrast with the

many-sorted generalization which we are considering in this section. Generally

speaking, one-sorted logic tends to be a little simpler than many-sorted logic.

86

Chapter 5

Theories, Models,

Deﬁnability

5.1 Theories and Models

Deﬁnition 5.1.1. A theory T consists of a language L, called the language of

T, together with a set of L-sentences called the axioms of T. Thus T = (L, S),

where L is the language of T, and S is the set of axioms of T.

Deﬁnition 5.1.2. Let T = (L, S) be a theory.

1. A model of T is an L-structure M such that M satisﬁes S. If L contains

identity predicates, then M is required to be normal with respect to these

predicates.

2. A theorem of T is an L-sentence A such that A is true in all models of

T, i.e., A is a logical consequence of the axioms of T. Equivalently, A is

derivable in LH(S ∪ ¦identity axioms for L¦).

If A is a theorem of T, we denote this by T ⊢ A.

3. T is ﬁnitely axiomatized if S is ﬁnite.

4. Two theories are equivalent if they have the same language and the same

theorems. I.e., they have the same language and the same models.

5. T is ﬁnitely axiomatizable if it is equivalent to a ﬁnitely axiomatized theory.

Deﬁnition 5.1.3 (consistency, categoricity, completeness). Let T = (L, S) be

a theory.

1. T is consistent if there exists at least one model of T. Equivalently, T is

consistent if and only if there is no L-sentence A such that both T ⊢ A

and T ⊢ A. Equivalently, T is consistent if and only if there exists an

L-sentence A such that T ,⊢ A.

87

2. T is categorical if T is consistent and all models of T are isomorphic.

3. T is complete if T is consistent and all models of T are elementarily equiv-

alent. Equivalently, T is complete if and only if for all L-sentences A either

T ⊢ A or T ⊢ A but not both.

Remark 5.1.4. Our formal notion of theory, as deﬁned above, is intended as a

precise explication of the informal notion of “deductive scientiﬁc theory”. The

language of T is the vocabulary of our theory. The theorems of T are the

assertions of our theory. The axioms of T are the basic assertions, from which

all others are deduced. Consistency of T means that our theory is free of internal

contradictions. Categoricity of T means that our theory is fully successful in that

it fully captures the structure of the underlying reality described by the theory.

Completeness of T means that our theory is suﬃciently successful to decide the

truth values of all statements expressible in the language of the theory.

Remark 5.1.5 (mathematical theories, foundational theories). Later in this

chapter we shall present several interesting examples of theories. Loosely speak-

ing, the examples are of two kinds.

The ﬁrst kind consists of mathematical theories. By a mathematical theory

we mean a theory which is introduced in order to describe a certain class of

mathematical structures. See Section 5.2. Since the class is diverse, the the-

ory is typically not intended to be complete. Nevertheless we shall see that,

remarkably, several of these mathematical theories turn out to be complete.

The second kind consists of foundational theories, i.e., theories which are

introduced in order to serve as a general axiomatic foundation for all of mathe-

matics, or at least a large part of it. See Sections 5.5 and 5.6. Each such theory

is intended to fully describe a speciﬁc, foundationally important model, known

as the intended model of the theory. Although such theories are intended to

be complete, we shall see in Chapter 6 that, regrettably, most of these theories

turn out to be incomplete.

One way to show that a theory is complete is to show that it is categorical.

Theorem 5.1.6. If T is categorical, then (1) T is complete, and (2) the language

of T is a language with identity.

Proof. Assume that T is categorical. Then any two models of T are isomorphic.

Hence by Theorem 2.2.6 (see also Deﬁnition 4.5.3), any two models of T are

elementarily equivalent. Hence T is complete. Also, by Theorem 2.7.3 (see also

Theorem 4.3.9 and Deﬁnition 4.5.3), T contains an identity predicate for each

sort in the language of T.

On the other hand, we have:

Exercise 5.1.7. Let T be a complete theory in a language with identity. Let

M be a model of T.

1. In the one-sorted case, show that T is categorical if and only if the universe

U

M

is ﬁnite.

88

2. In the many-sorted case with sorts σ, τ, . . ., show that T is categorical if

and only if each of the universes U

σ

M

, U

τ

M

, . . . is ﬁnite.

Solution. If one of the universes U

σ

M

is inﬁnite, we can use the L¨ owenheim/Skolem

Theorem 4.1.13 to blow it up to an arbitrarily large uncountable cardinality.

Remark 5.1.8. Theorem 5.1.6 and Exercise 5.1.7 show that we cannot use

categoricity as a test for completeness, except in very special circumstances. A

similar but more useful test is provided by the following theorem.

Deﬁnition 5.1.9. Let κ be an inﬁnite cardinal number. A one-sorted theory

T is said to be κ-categorical if all models of T of cardinality κ are isomorphic.

Theorem 5.1.10 (Vaught’s Test). Let T be a one-sorted theory. Assume that

(a) T is consistent, (b) all models of T are inﬁnite, and (c) there exists an inﬁnite

cardinal κ ≥ the cardinality of the language of T such that T is κ-categorical.

Then T is complete.

Proof. Suppose T is not complete. Since T is consistent, there exist M

1

and

M

2

which are models of T and not elementarily equivalent. By assumption (b),

M

1

and M

2

are inﬁnite. Let κ be an inﬁnite cardinal ≥ the cardinality of the

language of T. By the L¨ owenheim/Skolem Theorem 4.1.13, there exist models

M

′

1

, M

′

2

of cardinality κ elementarily equivalent to M

1

, M

2

respectively. Clearly

M

′

1

, M

′

2

are not elementarily equivalent. Hence, by Theorem 2.2.6, M

′

1

and M

′

2

are not isomorphic. This contradicts assumption (c).

Exercise 5.1.11. Generalize Vaught’s Test to many-sorted theories.

5.2 Mathematical Theories

In this section we give several examples of theories suggested by abstract algebra

and other speciﬁc mathematical topics. We point out that several of these

mathematical theories are complete.

Example 5.2.1 (groups). The language of groups consists of a binary operation

(multiplication), a unary operation

−1

(inverse), a constant 1 (the identity

element), and a binary predicate = (equality). The theory of groups consists of

the equality axioms plus

∀x∀y ∀z (x (y z) = (x y) z) (associativity),

∀x(x x

−1

= x

−1

x = 1) (inverses),

∀x(x 1 = 1 x = x) (identity).

A group is a model of the theory of groups. A group is said to be Abelian if it

satisﬁes the additional axiom

∀x∀y (x y = y x) (commutativity).

89

A torsion group is a group G such that for all a ∈ G there exists a positive

integer n such that a

n

= 1. A group G is torsion-free if for all a ∈ G, if a ,= 1

then a

n

,= 1 for all positive integers n. Note that G is torsion-free if and only if

it satisﬁes the axioms ∀x(x

n

= 1 ⇒x = 1) for n = 2, 3, . . .. A group is said to

be divisible if it satisﬁes the axioms ∀x∃y (y

n

= x), for n = 2, 3, . . ..

Exercise 5.2.2. Show that the theory of torsion-free Abelian groups is not

ﬁnitely axiomatizable. Deduce that the theory of torsion-free groups is not

ﬁnitely axiomatizable.

Solution. Suppose that the theory of torsion-free Abelian groups were ﬁnitely

axiomatizable. By Compactness, the axioms would be logical consequences

of the Abelian group axioms plus ﬁnitely many axioms of the form ∀x(x

n

=

1 ⇒x = 1), say n = 1, . . . , k. Let p be a prime number greater than k. Then the

additive group of integers modulo p satisﬁes these axioms but is not torsion-free.

This is a contradiction.

If the theory of torsion-free groups were ﬁnitely axiomatizable, then the

theory of torsion-free Abelian groups would also be ﬁnitely axiomatizable, by

adjoining the single axiom ∀x∀y (x y = y x).

Exercise 5.2.3. Show that there exist Abelian groups G

1

and G

2

such that

G

1

is a torsion group, G

1

is elementarily equivalent to G

2

, yet G

2

is not a

torsion group. Deduce that there is no theory in the language of groups whose

models are precisely the Abelian torsion groups. Hence, there is no theory in

the language of groups whose models are precisely the torsion groups.

Solution. Let G

1

be a torsion group with elements of arbitrarily large ﬁnite

order. (For example, we could take G

1

to be the additive group of rational

numbers modulo 1.) Let L be the language of groups, and let S be the set

of all L-sentences true in G

1

. Let L

∗

= L ∪ ¦c¦ where c is a new constant,

and let S

∗

= S ∪ ¦c

n

,= 1 [ n = 1, 2, . . .¦. By choosing c ∈ G

1

appropriately,

we see that any ﬁnite subset of S

∗

is normally satisﬁable. By the Compactness

Theorem for normal satisﬁability (Corollary 4.1.9), it follows that S

∗

is normally

satisﬁable, so let (G

2

, c) be a model of S

∗

. Then G

2

is an Abelian group which is

elementarily equivalent to G

1

yet contains an element c of inﬁnite order, hence

is not a torsion group.

If T were an L-theory whose models are just the Abelian torsion groups,

then G

1

would be a model of T but G

2

would not, contradicting the fact that

G

1

and G

2

are elementarily equivalent.

If T were an L-theory whose models are just the torsion groups, then T ∪

¦∀x∀y (x y = y x)¦ would be an L-theory whose models are just the Abelian

torsion groups.

Remark 5.2.4. Let DAG

0

be the theory of inﬁnite torsion-free divisible Abelian

groups. It can be shown that DAG

0

is κ-categorical for all uncountable cardinals

κ. (This is because such groups may be viewed as vector spaces over the rational

ﬁeld, Q.) It follows by Vaught’s Test that DAG

0

is complete.

90

Example 5.2.5 (linear orderings). The language of linear orderings consists

of a binary predicate < plus the equality predicate =. The axioms for lin-

ear orderings are ∀x∀y ∀z ((x < y ∧y < z) ⇒x < z), and ∀x(x < x), and

∀x∀y (x < y ∨x = y ∨x > y). A linear ordering is a model of these axioms.

A linear ordering is said to be nontrivial if it satisﬁes ∃x∃y (x < y). It is

said to be dense if it is nontrivial and satisﬁes ∀x∀y (x < y ⇒∃z (x < z < y)).

It is said to be without end points if it satisﬁes ∀x∃y (y < x) and ∀x∃y (y > x).

It is said to with end points if it satisﬁes ∃x∃y (y < x) and ∃x∃y (y > x).

An example of a dense linear ordering without end points is (Q, <), where Q is

the set of rational numbers, and < is the usual ordering of Q.

Remark 5.2.6. It can be shown that, up to isomorphism, (Q, <) is the unique

countable dense linear ordering without end points. (This is proved by a back-

and-forth argument.) Hence, if we let DLO be the theory of dense linear ordering

without end points, DLO is ℵ

0

-categorical. It follows by Vaught’s Test that DLO

is complete.

Example 5.2.7 (graphs). The language of graphs consists of a binary predicate,

R, plus the equality predicate, =. The theory of graphs consists of the equality

axioms plus ∀x∀y (Rxy ⇔Ryx) and ∀xRxx. A graph is a model of the theory

of graphs.

Thus a graph is essentially an ordered pair G = (V

G

, R

G

), where V

G

is a

nonempty set and R

G

is a symmetric, irreﬂexive relation on V

G

. The elements

of V

G

are called vertices. Two vertices a, b ∈ V

G

are said to be adjacent if

¸a, b) ∈ R

G

. A path from a to b is a ﬁnite sequence of pairwise distinct vertices

a = v

0

, v

1

, . . . , v

n

= b such that a = v

0

is adjacent to v

1

, v

1

is adjacent to v

2

,

. . . , v

n−1

is adjacent to v

n

= b. G is said to be connected if for all a, b ∈ V

G

there exists a path from a to b. Equivalently, G is connected if and only if, for

all partitions of V

G

into two disjoint nonempty sets X and Y , there exist a ∈ X

and b ∈ Y such that a and b are adjacent.

Exercise 5.2.8. Show that there exist graphs G

1

and G

2

such that G

1

is

connected, G

1

is elementarily equivalent to G

2

, yet G

2

is not connected. Deduce

that there is no theory T in the language of graphs such that the models of T

are exactly the connected graphs.

Solution. If a and b are two vertices in a graph, we deﬁne d(a, b), the distance

from a to b, to be the smallest length of a path from a to b, or ∞ if there is no

such path. Let G

1

be a graph which is connected yet contains pairs of vertices

which are at distance n for arbitrarily large n. (For example, we may take

G

1

= (N, R

1

) where R

1

= ¦¸n, n + 1), ¸n + 1, n) [ n ∈ N¦.) Let L = ¦R, =¦ be

the language of graphs, and let S be the set of L-sentences satisﬁed by G

1

. Let

L

∗

= L∪¦a, b¦ where a, b are new constants, and let S

∗

= S∪¦A

n

[ n = 1, 2, . . .¦

where A

n

is an L

∗

-sentence saying that there is no path of length n from a to

b. By choosing a, b ∈ G

1

appropriately, we see that all ﬁnite subsets of S

∗

are normally satisﬁable. Hence by the Compactness Theorem S

∗

is normally

satisﬁable, so let (G

2

, a, b) be a model of S

∗

. Then G

2

is a graph which is

91

elementarily equivalent to G

1

, yet a, b ∈ G

2

are such that d(a, b) = ∞, hence

G

2

is not connected.

If there were an L-theory T whose models are exactly the connected graphs,

then G

1

would be a model of T but G

2

would not, contradicting the fact that

G

1

and G

2

are elementarily equivalent.

Exercise 5.2.9. A graph G is said to be random if for all ﬁnite sets of distinct

vertices a

1

, . . . , a

m

, b

1

, . . . , b

n

there exists a vertex c such that c is adjacent

to a

1

, . . . , a

m

and not adjacent to b

1

, . . . , b

n

. Show that the theory of random

graphs is ℵ

0

-categorical. It follows by Vaught’s Test that this theory is complete.

Solution. Let G and G

′

be two random graphs of cardinality ℵ

0

, say G = ¦a

k

[

k ∈ N¦ and G

′

= ¦a

′

k

[ k ∈ N¦. We use a back-and-forth argument to construct

an isomorphism of G onto G

′

. At stage n of the construction, we have a ﬁnite

partial isomorphism, f

n

, which maps a ﬁnite subset of G isomorphically onto a

ﬁnite subset of G

′

. Start with f

0

= ∅. Suppose we have already constructed f

n

.

To construct f

n+1

, consider two cases. If n is even, let k

n

be the least k such

that a

k

/ ∈ dom(f

n

), and put c = a

kn

. By randomness of G

′

, ﬁnd c

′

∈ G

′

such

that for all a ∈ dom(f

n

), c

′

is adjacent to f

n

(a) if and only if c is adjacent to

a. If n is odd, let k

n

be the least k such that a

′

k

/ ∈ ran(f

n

), and put c

′

= a

′

kn

.

By randomness of G, ﬁnd c ∈ G such that for all a ∈ dom(f

n

), c is adjacent to

a if and only if c

′

is adjacent to f

n

(a). In either case, let f

n+1

= f

n

∪ ¦(c, c

′

)¦.

Finally, by construction, f =

∞

n=0

f

n

is an isomorphism of G onto G

′

.

Example 5.2.10 (rings). The language of rings consists of binary operations

+ and (addition and multiplication), a unary operation − (subtraction), con-

stants 0 and 1 (the additive and multiplicative identity elements), and a binary

predicate = (equality). The theory of rings consists of the equality axioms plus

∀x∀y ∀z (x + (y +z) = (x +y) +z),

∀x∀y (x +y = y +x),

∀x(x + (−x) = 0),

∀x(x + 0 = x),

∀x∀y ∀z (x (y z) = (x y) z),

∀x∀y ∀z (x (y +z) = (x y) + (x z)) (left distributivity),

∀x∀y ∀z ((x +y) z = (x z) + (y z)) (right distributivity),

∀x(x 1 = 1 x = x),

∀x(x 0 = 0 x = 0),

0 ,= 1.

A ring is a model of the theory of rings. A ring is said to be commutative if it

satisﬁes the additional axiom

92

∀x∀y (x y = y x).

An example of a commutative ring is the ring of integers,

Z = ¦. . . , −2, −1, 0, 1, 2, . . .¦.

An example of a non-commutative ring is the ring of 2 2 matrices.

Example 5.2.11 (ﬁelds). A ﬁeld is a commutative ring satisfying the additional

axiom

∀x(x ,= 0 ⇒∃y (x y = 1)).

A ﬁeld is said to be of characteristic 0 if it satisﬁes

1 + + 1

. ¸¸ .

n

,= 0

for all positive integers n. Familiar ﬁelds such as the ﬁeld of rational numbers,

Q, the ﬁeld of real numbers, R, and the ﬁeld of complex numbers, C, are of

characteristic 0. It can be shown that if a ﬁeld satisﬁes 1 + + 1

. ¸¸ .

n

= 0 for some

positive integer n, then the least such n is a prime number, p. In this case, our

ﬁeld is said to be of characteristic p. An example of a ﬁeld of characteristic p

is the ring of integers modulo p.

A ﬁeld F is said to be algebraically closed if for all nontrivial polynomials

f(z) = a

n

z

n

+ + a

1

z + a

0

, a

n

, . . . , a

1

, a

0

∈ F, a

n

,= 0, n ≥ 1, there exists

z ∈ F such that f(z) = 0. It is known that the complex ﬁeld C is algebraically

closed. (This theorem is known as the Fundamental Theorem of Algebra.) Note

that a ﬁeld is algebraically closed if and only if it satisﬁes the axioms

∀x

0

∀x

1

∀x

n

(x

n

,= 0 ⇒∃z (x

n

z

n

+ +x

1

z +x

0

= 0))

for n = 1, 2, 3, . . ..

Exercise 5.2.12. Show that the theory of ﬁelds of characteristic 0 is not ﬁnitely

axiomatizable. Show that the theory ACF of algebraically closed ﬁelds is not

ﬁnitely axiomatizable. Show that the theory ACF

0

of algebraically closed ﬁelds

of characteristic 0 is not ﬁnitely axiomatizable.

Solution. Suppose that the theory of ﬁelds of characteristic 0 were ﬁnitely

axiomatizable. By Compactness, the axioms would be logical consequences of

the ﬁeld axioms plus ﬁnitely many axioms of the form 1 + + 1

. ¸¸ .

n

,= 0, say

n = 1, . . . , k. Let p be a prime number greater than k. The ﬁeld of integers

modulo p satisﬁes these axioms but is not of characteristic 0, a contradiction.

Suppose ACF or ACF

0

were ﬁnitely axiomatizable. By Compactness, the

axioms would be logical consequences of the ﬁeld axioms plus ﬁnitely many of

the ACF

0

axioms as presented in Example 5.2.11 above. But, for any ﬁnite

subset of the ACF

0

axioms, we can construct a ﬁeld which satisﬁes these axioms

yet is not algebraically closed. (The construction of such a ﬁeld is perhaps

somewhat delicate.) This gives a contradiction.

93

Exercise 5.2.13. Let L be the language of rings. Let S be the set of L-sentences

which are true in the ring of integers modulo p for all but ﬁnitely many prime

numbers p. Does there exist a ﬁeld of characteristic 0 satisfying S? Prove your

answer.

Solution. The answer is yes. Note ﬁrst that, for each prime p, the ring of

integers modulo p is actually a ﬁeld. Hence S includes the ﬁeld axioms. Let

S

0

= S ∪ ¦1 + + 1

. ¸¸ .

n

,= 0 [ n = 1, 2, . . .¦. Any ﬁnite subset of S

0

is normally

satisﬁable, e.g., in the integers modulo p for all suﬃciently large primes p. By

the Compactness Theorem for normal satisﬁability (Corollary 4.1.9), it follows

that S

0

is normally satisﬁable. The normal structures which satisfy S

0

are ﬁelds

of characteristic 0 satisfying S.

Remark 5.2.14. Let ACF

0

be the theory of algebraically closed ﬁelds of char-

acteristic 0. For each prime number p, let ACF

p

be the theory of algebraically

closed ﬁelds of characteristic p. It can be shown that the theories ACF

0

and

ACF

p

are κ-categorical for all uncountable cardinals κ. It follows by Vaught’s

Test that these theories are complete.

Example 5.2.15 (vector spaces). The language of vector spaces is a 2-sorted

language with sorts σ and τ, denoting scalars and vectors respectively. For the

scalars we have binary operations + and of type σ σ → σ, a unary operation

− of type σ → σ, constants 0 and 1 of type σ, and an equality predicate = of

type σ σ. For the vectors we have a binary operation + of type τ τ → τ,

a unary operation − of type τ → τ, a constant 0 of type τ, and an equality

predicate = of type τ τ. In addition we have a binary operation of “mixed”

type σ τ → τ, denoting scalar multiplication.

The theory of vector spaces consists of the ﬁeld axioms for scalars, the

Abelian group axioms for vectors, and the axioms

∀x

σ

∀v

τ

∀w

τ

(x (v +w) = (x v) + (x w)),

∀x

σ

∀v

τ

(x (−v) = −(x v)),

∀x

σ

(x 0

τ

= 0),

∀x

σ

∀y

σ

∀v

τ

((x +y) v = (x v) + (y v)),

∀x

σ

∀v

τ

((−x) v = −(x v)),

∀x

σ

∀y

σ

∀v

τ

((x y) v = x (y v)),

∀v

τ

(1 v = v),

∀v

τ

(0 v = 0)

for scalar multiplication. A vector space is a model of these axioms.

If V is a vector space, a set of vectors S in V is said to span V if every

vector v in V can be written as a linear combination of vectors in S, i.e., v =

94

a

1

v

1

+ +a

n

v

n

for some v

1

, . . . , v

n

∈ S and scalars a

1

, . . . , a

n

. An important

invariant of a vector space is its dimension, i.e., the minimum cardinality of a

spanning set. It can be shown that, up to isomorphism over a ﬁeld F, the

unique vector space over F of dimension κ is the familiar space

i<κ

F. Here

the vectors are sequences ¸a

i

[ i < κ), with a

i

∈ F for all i, and a

i

= 0 for all

but ﬁnitely many i. Vector addition is given by

¸a

i

[ i < κ) +¸b

i

[ i < κ) = ¸a

i

+b

i

[ i < κ),

and scalar multiplication is given by

c ¸a

i

[ i < κ) = ¸c a

i

[ i < κ).

Example 5.2.16 (ordered algebraic structures). The language of ordered rings

consists of the language of rings +, , −, 0, 1, =, together with <. The ordered

ﬁeld axioms consist of the ﬁeld axioms, plus the linear ordering axioms, plus

∀x∀y ∀z (x < y ⇔x +z < y +z),

∀x∀y ((x > 0 ∧ y > 0) ⇒x y > 0).

An ordered ﬁeld is a model of these axioms. An example of an ordered ﬁeld is

the ﬁeld of rational numbers (Q, <) with its usual ordering.

An ordered ﬁeld (F, <) is said to be real-closed if for all polynomials f(x) =

a

n

x

n

+ +a

1

x +a

0

, a

n

, . . . , a

1

, a

0

∈ F, and for all b, c ∈ F, if f(b) < 0 < f(c)

then there exists x ∈ F between b and c such that f(x) = 0. Clearly the

ordered ﬁeld of real numbers (R, <) is real-closed. Note that an ordered ﬁeld is

real-closed if and only if it satisﬁes the axioms

∀u ∀v ∀w

0

∀w

1

∀w

n

((u < v ∧ w

n

u

n

+ +w

1

u +w

0

< 0 < w

n

v

n

+ +w

1

v +w

0

)

⇒∃x(u < x < v ∧ w

n

x

n

+ +w

1

x +w

0

= 0))

for n = 1, 2, 3, . . ..

Remark 5.2.17 (elimination of quantiﬁers). Let RCOF be the theory of real-

closed ordered ﬁelds. A famous and important theorem of Tarski says that

RCOF is complete. This holds despite the fact that RCOF is not κ-categorical

for any κ.

Tarski’s method of proof is as follows. A theory T = (L, S) is said to admit

elimination of quantiﬁers if for all L-formulas A there exists a quantiﬁer-free L-

formula A

∗

such that T ⊢ the universal closure of A⇔A

∗

. Tarski uses algebraic

methods to show that the theory RCOF admits elimination of quantiﬁers. For

example, the formula ∃x(ax

2

+ bx + c = 0) is equivalent over RCOF to the

quantiﬁer-free formula (a = b = c = 0) ∨(a = 0 ,= b) ∨(a ,= 0 ≤ b

2

−4ac).

As a special case of quantiﬁer elimination, we have that each sentence in

the language of RCOF is equivalent over RCOF to a quantiﬁer-free sentence. On

the other hand, it is evident that the quantiﬁer-free sentences of the language

95

of RCOF are of a very simple nature, e.g., 1 + 0 = 1 ∧(1 1) + 1 < 1 + 1 + 1.

Since the truth values of such sentences are decided by the axioms of RCOF, it

follows that RCOF is complete.

Exercise 5.2.18. Which of the following theories are complete? Justify your

answers.

1. The theory of dense linear orderings with end points.

2. The theory of ﬁelds of characteristic 0.

3. The theory of inﬁnite, torsion-free, Abelian groups.

4. The theory of ﬁnite-dimensional vector spaces over a ﬁeld of 5 elements.

5. The theory of inﬁnite-dimensional vector spaces over a ﬁeld of 5 elements.

Solution.

1. As noted in Remark 5.2.6, the theory of dense linear orderings without end

points is ℵ

0

-categorical. It follows immediately that the theory of dense

linear orderings with end points is ℵ

0

-categorical. Hence, by Vaught’s

Test, each of these theories is complete.

2. The ﬁelds Q, R, and C are of characteristic 0, yet they are not elementarily

equivalent, as can be seen by considering the sentences ∃x(x x = 1 + 1)

and ∃z (z z = −1). Therefore, the theory of ﬁelds of characteristic 0 is

incomplete.

3. The additive groups of Z and Q are inﬁnite, Abelian, and torsion-free,

yet they are not elementary equivalent, as can be seen by considering

the sentence ∃x(x + x = 1). Therefore, the theory of inﬁnite, Abelian,

torsion-free groups is incomplete.

4. The vector spaces of dimension n = 0, 1, 2, . . . over a particular ﬁnite

ﬁeld F are not elementarily equivalent, because they have diﬀerent ﬁnite

cardinalities 1, q, q

2

, . . . where q is the cardinality of F. Therefore, the

theory of ﬁnite-dimensional vector spaces over F is incomplete.

5. As noted in Example 5.2.15, any two vector spaces over the same ﬁeld of

the same dimension are isomorphic. On the other hand, for any inﬁnite

cardinal number κ, any vector space of cardinality κ over a ﬁnite ﬁeld

is of dimension κ. Combining these facts, we see that for any particular

ﬁnite ﬁeld F, the theory T

F

of inﬁnite-dimensional vector spaces over F

is κ-categorical. Hence, by Vaught’s Test, T

F

is complete.

96

5.3 Deﬁnability over a Model

Deﬁnition 5.3.1 (explicit deﬁnability). Let L be a language, let M be an L-

structure, and let R be an n-ary relation on U

M

. We say that R is explicitly

deﬁnable over M, or just deﬁnable over M, if there exists an L-formula D with

n free variables x

1

, . . . , x

n

such that

R = ¦¸a

1

, . . . , a

n

) [ M [= D[x

1

/a

1

, . . . , x

n

/a

n

]¦.

Example 5.3.2. Consider the binary relation < = ¦¸a, b) ∈ R

2

[ a < b¦ on

the set R of real numbers. Viewing R as a commutative ring, we see that for

all a, b ∈ R, a < b if and only if R [= ∃x(x ,= 0 ∧ a + x

2

= b). Thus < is

explicitly deﬁnable over the commutative ring R = (R, +, −, , 0, 1, =) by the

formula ∃z (z ,= 0 ∧ x +z

2

= y) with free variables x, y.

On the other hand, in view of Tarski’s theorem on elimination of quantiﬁers

for RCOF (see Remark 5.2.17), any subset of R which is deﬁnable over R consists

of a union of ﬁnitely many points and intervals. From this it follows that, for

example, the set of integers is not deﬁnable over R.

Example 5.3.3. Let N = (N, +, , 0, 1, <, =) be the natural number system. It

can be shown that the class of relations which are explicitly deﬁnable over N

includes all relations which are computable in the sense of Turing. In particular,

the 3-ary relation ¦¸m, n, k) ∈ N

3

[ m = n

k

¦ is deﬁnable over N, as we now show.

Theorem 5.3.4. The exponential function (n, k) → n

k

is explicitly deﬁnable

over the natural number system.

Proof. The proof uses some number-theoretic lemmas, as follows.

Lemma 5.3.5 (the Chinese Remainder Theorem). Let m

1

, . . . , m

k

be pairwise

relatively prime. Given r

1

, . . . , r

k

, we can ﬁnd r such that r ≡ r

i

mod m

i

for

all i = 1, . . . , k.

Proof. We omit the proof of this well-known result.

Lemma 5.3.6. For each k ∈ N we can ﬁnd inﬁnitely many a ∈ N such that the

integers a + 1, 2a + 1, . . . , ka + 1 are pairwise relatively prime.

Proof. Let a be any multiple of k!. Suppose p is a prime number which divides

both ai + 1 and aj + 1 where 1 ≤ i < j ≤ k. Since p divides ai + 1 and a is

a multiple of k!, we clearly have p > k. On the other hand, p divides a(j − i),

and j −i < k, hence p divides a, a contradiction since p divides ai + 1.

Deﬁnition 5.3.7 (G¨ odel’s beta-function). We deﬁne β(a, r, i) = the remainder

of r upon division by a(i+1)+1. Note that the β-function is explicitly deﬁnable

over the natural number system.

Lemma 5.3.8. Given a ﬁnite sequence of natural numbers n

0

, n

1

, . . . , n

k

, we

can ﬁnd natural numbers a, r such that β(a, r, i) = n

i

for all i = 0, 1, . . . , k.

97

Proof. Let a ≥ max¦n

i

[ i ≤ k¦ be such that a (i + 1) + 1, i ≤ k, are pairwise

relatively prime. By the Chinese Remainder Theorem, we can ﬁnd r such that

r ≡ n

i

mod a (i + 1) + 1 for all i ≤ k. Since 0 ≤ n

i

≤ a < a (i + 1) + 1, it

follows that β(a, r, i) = n

i

.

Now, to prove Theorem 5.3.4, note that m = n

k

holds if and only if ∃a ∃r

(β(a, r, 0) = 1 ∧β(a, r, k) = m∧∀i (i < k ⇒β(a, r, i) n = β(a, r, i + 1))).

Exercise 5.3.9. Let p

n

be the nth prime number. Thus p

0

= 2, p

1

= 3, p

2

= 5,

p

3

= 7, p

4

= 11, etc. Show that the function k → p

k

is explicitly deﬁnable over

the natural number system N. (This means, show that the binary relation

¦¸k, p

k

) [ k ∈ N¦ is explicitly deﬁnable over N.)

Solution. Let A be a formula expressing that x is prime, for example

1 < x∧∃u ∃v (u < x∧v < x∧u v = x).

We then have p

k

= n if and only if ∃a ∃r (B ∧C), where B is

β(a, r, k) = n∧∀i (i < k ⇒β(a, r, i) < β(a, r, i + 1))

and C is

∀x(x ≤ n⇒(A⇔∃i (i ≤ k ∧β(a, r, i) = x))).

We now turn to implicit deﬁnability.

Deﬁnition 5.3.10 (implicit deﬁnability). Let L be a language, let M be an

L-structure, and let R be an n-ary relation on U

M

. We say that R is implicitly

deﬁnable over M if there exists a sentence D

∗

in the language L

∗

= L ∪ ¦R¦

with an additional n-ary predicate R, such that for all n-ary relations R

′

on

U

M

, (M, R

′

) [= D

∗

if and only if R

′

= R.

Example 5.3.11. Let R = (R, +, −, , 0, 1, <, =) be the ordered ﬁeld of real

numbers. Consider the set of integers, Z ⊂ R. As noted above, Z is not

explicitly deﬁnable over R. However, Z is implicitly deﬁnable over R, by the

following sentence with an additional unary predicate Z:

Z0 ∧Z1 ∧(∃x(Zx∧0 < x < 1)) ∧(∀x(Zx⇔Z(x + 1))).

Example 5.3.12. It can be shown that there exists a subset of N which is

implicitly deﬁnable over the natural number system N = (N, +, , 0, 1, <, =) but

is not explicitly deﬁnable over N. See Remark 6.4.6 and Exercise 6.4.7 below.

Deﬁnition 5.3.13 (automorphisms). Let M be a structure. An automorphism

of M is an isomorphism of M onto itself. An n-ary relation R on U

M

is said to

be invariant if

R = ¦¸f(a

1

), . . . , f(a

n

)) [ ¸a

1

, . . . , a

n

) ∈ R¦

for all automorphisms f of M.

98

Exercise 5.3.14. Let M be a structure. Show that any relation which is ex-

plicitly deﬁnable over M is implicitly deﬁnable over M. Show that any relation

which is implicitly deﬁnable over M is invariant under all automorphisms of

M. Give counterexamples showing that the converses of these assertions fail in

general.

Solution. Let R be an n-ary relation over U

M

. If R is explicitly deﬁned over

M by a formula D with free variables x

1

, . . . , x

n

, then R is implicitly deﬁned

over M by the L∪¦R¦-sentence ∀x

1

∀x

n

(Rx

1

x

n

⇔D). Now suppose R is

implicitly deﬁned over M by an L∪¦R¦-sentence D

∗

. Let f be an automorphism

of M, and put R

′

= ¦¸f(a

1

), . . . , f(a

n

)) [ ¸a

1

, . . . , a

n

) ∈ R¦. Then f is an

isomorphism of (M, R) onto (M, R

′

). Since (M, R) [= D

∗

, it follows by Theorem

2.2.6 that (M, R

′

) [= D

∗

. Hence R

′

= R, i.e., R is invariant under f.

Consider the ordered ﬁeld R of real numbers. Example 5.3.11 shows that the

subset Z of R is implicitly deﬁnable over R but not explicitly deﬁnable over R.

Since there are only countably many sentences, only countably many subsets of

R are implicitly deﬁnable over R. However, all subsets of R are invariant under

automorphisms of R, inasmuch as R has no automorphisms except the identity.

Exercise 5.3.15. Let R = (R, +, −, , 0, 1, <, =) be the ordered ﬁeld of real

numbers. Show that the relations y = e

x

and y = sin x are implicitly deﬁnable

over R. It can be shown that these relations are not explicitly deﬁnable over R.

(Hint: The relations y = e

x

and y = sinx are implicitly deﬁned by diﬀerential

equations which can be expressed as formulas of the predicate calculus, using

the ǫ-δ-method.)

Exercise 5.3.16. Show that if U

M

is ﬁnite, then any relation which is implicitly

deﬁnable over M is explicitly deﬁnable over M.

Solution. Let us say that ¸a

1

, . . . , a

n

), ¸b

1

, . . . , b

n

) ∈ (U

M

)

n

are of the same n-

type if (M, a

1

, . . . , a

n

) is elementarily equivalent to (M, b

1

, . . . , b

n

). Thus (U

M

)

n

is partitioned into equivalence classes, the n-types. Since U

M

is ﬁnite, there

are only ﬁnitely many n-types, and each n-type is explicitly deﬁnable over M.

Moreover, by Theorem 2.2.6 and Exercise 4.1.12, the n-types are just the orbits

of (U

M

)

n

under the automorphism group of M. If R ⊆ (U

M

)

n

is implicitly

deﬁnable over M, then R is invariant under automorphisms of M, hence R is

the union of some of the n-types, hence R is explicitly deﬁnable over M.

Exercise 5.3.17. (In this exercise we assume familiarity with saturated mod-

els.) Show that if a structure M is saturated, then any relation which is implic-

itly deﬁnable over M is explicitly deﬁnable over M.

Exercise 5.3.18. Let G be a group which has inﬁnitely many distinct sub-

groups. Prove that there exists a countable group G

′

such that

1. G

′

is elementarily equivalent to G, and

2. G

′

has a subgroup which is not explicitly deﬁnable over G

′

.

99

Solution. Let S be the set of sentences which are true in G. Introduce a

new unary predicate P, and let S

′

consist of S plus the sentences P1 and

∀x(Px⇒Px

−1

) and ∀x∀y ((Px∧Py) ⇒P(x y)) and ∀x(Px⇔D) where D

is any formula with x as its only free variable. Since G has inﬁnitely many

subgroups, each ﬁnite subset of S

′

is normally satisﬁable in G by letting P

be an appropriately chosen subgroup of G. It follows by Compactness plus

L¨ owenheim/Skolem that we can ﬁnd a countable normal structure (G

′

, P

′

) sat-

isfying S

′

. Then G

′

is a countable group which is elementarily equivalent to G,

and P

′

is a subgroup of G

′

which is not explicitly deﬁnable over G

′

.

Remark 5.3.19. In this section we have considered explicit and implicit de-

ﬁnability over a model, M. We have given examples showing that, in general,

implicit deﬁnability over M does not imply explicit deﬁnability over M.

In Sections 5.4 and 5.8 below, we shall consider the related but more re-

strictive notions of explicit and implicit deﬁnability over a theory, T. It will be

obvious that explicit deﬁnability over T implies explicit deﬁnability over any

model of T, and implicit deﬁnability over T implies implicit deﬁnability over

any model of T. A pleasant surprise is that explicit deﬁnability over T is equiv-

alent to implicit deﬁnability over T. This is the content of Beth’s Deﬁnability

Theorem, Theorem 5.8.2 below.

5.4 Deﬁnitional Extensions of Theories

We now show how theories can be usefully extended by adding new predicates

and operations which are explicitly deﬁnable in terms of old predicates and op-

erations. This method of extending a theory is known as deﬁnitional extension.

In this section we are considering only explicit deﬁnitional extensions. Later,

in Section 5.8, we shall consider implicit deﬁnitional extensions. It will turn

out that, in principle, an implicit deﬁnitional extension of a theory is always

equivalent to an explicit deﬁnitional extension of the same theory.

Deﬁnition 5.4.1 (deﬁning a new predicate). Let T = (L, S) be a theory,

and let D be an L-formula with free variables x

1

, . . . , x

n

. Introduce a new

n-ary predicate P, and let T

′

= (L

′

, S

′

) where L

′

= L ∪ ¦P¦ and S

′

= S ∪

¦∀x

1

∀x

n

(Px

1

x

n

⇔D)¦.

Deﬁnition 5.4.2 (deﬁning a new operation). Let T = (L, S) be a theory with

an identity predicate I. Let D be an L-formula with free variables x

1

, . . . , x

n

, y

such that

T ⊢ ∀x

1

∀x

n

(∃ exactly one y) D,

i.e.,

T ⊢ ∀x

1

∀x

n

∃z ∀y (Iyz ⇔D)

where z is a new variable. Introduce a new n-ary operation f, and let T

′

=

(L

′

, S

′

) where L

′

= L ∪ ¦f¦ and S

′

= S ∪ ¦∀x

1

∀x

n

∀y (Ifx

1

. . . x

n

y ⇔D)¦.

100

Remark 5.4.3. In Deﬁnition 5.4.2 above, we have assumed for simplicity that

T is one-sorted. In case T is many-sorted, we make the obvious modiﬁcations.

Namely, letting σ

1

, . . . , σ

n

, σ

n+1

be the sorts of the variables x

1

, . . . , x

n

, y re-

spectively, we require that z is a new variable of sort σ

n+1

, I is an identity

predicate of sort σ

n+1

, and f is an operation of type σ

1

σ

n

→ σ

n+1

.

Remark 5.4.4. We are going to prove that these extensions of T are “trivial” or

“harmless” or “inessential”, in the sense that each formula in the extended lan-

guage L∪¦P¦ or L∪¦f¦ can be straightforwardly translated into an equivalent

formula of the original language L. See Theorem 5.4.10 below.

Lemma 5.4.5. Let T = (L, S) and T

′

= (L

′

, S

′

) be as in Deﬁnition 5.4.1 or

5.4.2. Then for all L-sentences A we have T

′

⊢ A if and only if T ⊢ A.

Proof. It suﬃces to show that, for each model M of T, there exists a model M

′

of T

′

such that M = M

′

↾L, i.e., M is the reduct of M

′

to L. In the case of

Deﬁnition 5.4.1, let M

′

= (M, P

M

′ ) where

P

M

′ = ¦¸a

1

, . . . , a

n

) [ M [= D[x

1

/a

1

, . . . , x

n

/a

n

]¦.

In the case of Deﬁnition 5.4.2, let M

′

= (M, f

M

′ ) where f

M

′ (a

1

, . . . , a

n

) = the

unique b such that M [= D[x

1

/a

1

, . . . , x

n

/a

n

, y/b]. Clearly M

′

is as desired.

Deﬁnition 5.4.6 (translation). Let T = (L, S) and T

′

= (L

′

, S

′

) be as in

Deﬁnition 5.4.1 or 5.4.2. To each L

′

-formula A we associate an L-formula A

′

,

the translation of A into L. In order to deﬁne A

′

, we ﬁrst modify D, replacing

the free and bound variables of D by new variables which do not occur in A.

(Compare Deﬁnition 2.4.9.) In the case of Deﬁnition 5.4.1, where we are adding

a new predicate P, we obtain A

′

from A by replacing each atomic formula of

the form Pt

1

. . . t

n

by D[x

1

/t

1

, . . . , x

n

/t

n

].

In the case of Deﬁnition 5.4.2, where we are adding a new operation f, the

translation is more complicated. For non-atomic A we obtain A

′

by induction

on the degree of A, putting (A)

′

= A

′

, (A∧B)

′

= A

′

∧B

′

, (∀xA)

′

= ∀xA

′

,

etc. For atomic A we obtain A

′

by induction on the number of occurrences of f

in A. If there are no occurrences of f in A, let A

′

be just A. Otherwise, write

A in the form B[w/ft

1

t

n

] where w is a new variable and f does not occur

in t

1

, . . . , t

n

, and let A

′

be

∃w(D[x

1

/t

1

, . . . , x

n

/t

n

, y/w] ∧B

′

)

or equivalently

∀w(D[x

1

/t

1

, . . . , x

n

/t

n

, y/w] ⇒B

′

).

Lemma 5.4.7. Let T and T

′

be as in Deﬁnition 5.4.1 or 5.4.2, and let A → A

′

be our translation of L

′

-formulas to L-formulas as in Deﬁnition 5.4.6. Then:

1. A

′

has the same free variables as A and is equivalent to it over T

′

, i.e.,

T

′

⊢ A⇔A

′

.

101

2. Propositional connectives and quantiﬁers are respected, i.e., we have

(A)

′

≡ A

′

, (A∧B)

′

≡ A

′

∧B

′

, (∀xA)

′

≡ ∀xA

′

, etc.

3. If A happens to be an L-formula, then A

′

is just A.

Consequently, for all L

′

-sentences A we have T

′

⊢ A if and only if T ⊢ A

′

.

Proof. The ﬁrst part is straightforward. For the last part, since T

′

⊢ A⇔A

′

,

we obviously have T

′

⊢ A if and only if T

′

⊢ A

′

. But then, since A

′

is an

L-sentence, it follows by Lemma 5.4.5 that T

′

⊢ A if and only if T ⊢ A

′

.

Deﬁnition 5.4.8 (deﬁnitional extensions). Let T be a theory. A deﬁnitional ex-

tension of T is a theory T

∗

which is a union of sequences of theories T

0

, T

1

, . . . , T

k

beginning with T

0

= T such that each T

i+1

is obtained by extending T

i

as in

Deﬁnition 5.4.1 or 5.4.2.

Deﬁnition 5.4.9 (conservative extensions). Let T = (L, S) and T

∗

= (L

∗

, S

∗

)

be theories such that L

∗

⊇ L. We say that T

∗

is a conservative extension of T

if, for all L-sentences A, T

∗

⊢ A if and only if T ⊢ A.

Theorem 5.4.10. Let T

∗

be a deﬁnitional extension of T. Then T

∗

is a conser-

vative extension of T. Moreover, there is a straightforward translation A → A

∗

of L

∗

-formulas to L-formulas, with the following properties.

1. A

∗

has the same free variables as A and is equivalent to it over T

∗

, i.e.,

T

∗

⊢ A⇔A

∗

.

2. Propositional connectives and quantiﬁers are respected, i.e., we have

(A)

∗

≡ A

∗

, (A∧B)

∗

≡ A

∗

∧B

∗

, (∀xA)

∗

≡ ∀xA

∗

, etc.

3. If A happens to be an L-formula, then A

∗

is just A.

Consequently, for all L

∗

-sentences A, we have

T

∗

⊢ A if and only if T ⊢ A

∗

.

Proof. This is clear from Lemmas 5.4.5 and 5.4.7.

Remark 5.4.11. In practice, when it comes to working with speciﬁc theories

T, the technique of deﬁnitional extensions is very useful. This is because for-

mulas written in the extended language L

∗

tend to be much shorter than their

translations in L. This is particularly important for the foundational theories

Z

1

, Z

2

, . . . , Z

n

, . . . , Z

∞

, ZC, ZFC which are discussed in Sections 5.5 and 5.6

below.

Exercise 5.4.12. Show that for any theory T there is a deﬁnitional extension

of T which admits elimination of quantiﬁers.

Solution. Let L be the language of T. For each L-formula A with free variables

x

1

, . . . , x

n

introduce a new n-ary predicate P

A

and a new axiom

∀x

1

∀x

n

(P

A

x

1

x

n

⇔A).

The resulting theory admits elimination of quantiﬁers.

102

Exercise 5.4.13. Let T = (L, S) be a theory. Let A be an L-formula with free

variables x

1

, . . . , x

n

, y. Let T

′

= (L

′

, S

′

) where L

′

consists of L plus a new n-

ary operation f, and S

′

consists of S plus ∀x

1

∀x

n

(A[y/fx

1

x

n

] ⇔∃y A).

Show that T

′

is a conservative extension of T.

Solution. Let B be an L-sentence such that T

′

⊢ B. We must show that T ⊢ B.

If T ,⊢ B, let M be a model of T ∪ ¦B¦. Let M

′

= (M, f

M

′ ) where f

M

′ is

an n-ary function on U

M

with the following property: for all a

1

, . . . , a

n

∈ U

M

such that M [= ∃y A[x

1

/a

1

, . . . , x

n

/a

n

], we have M [= A[x

1

/a

1

, . . . , x

n

/a

n

, y/b]

where b = f

M

′ (a

1

, . . . , a

n

). (Note that f

M

′ is not necessarily unique and is not

necessarily deﬁnable over M. The existence of an f

M

′ with the desired property

follows from the axiom of choice.) Clearly M

′

is a model of T

′

∪ ¦B¦, hence

T

′

,⊢ B, a contradiction.

5.5 Foundational Theories

In this section and the next, we present several speciﬁc theories which are of

signiﬁcance in the foundations of mathematics. We point out that these theories

are, in certain senses, “almost complete” or “practically complete.” On the other

hand, we shall see in Chapter 6 that these theories are not complete.

Deﬁnition 5.5.1 (ﬁrst-order arithmetic). One of the most famous and impor-

tant foundational theories is ﬁrst-order arithmetic, Z

1

, also known as Peano

Arithmetic, PA. The language of ﬁrst-order arithmetic is L

1

= ¦+, , 0, S, =¦

where + and are binary operations, S is a unary operation, 0 is a constant,

and = is the equality predicate. Let Q be the ﬁnitely axiomatizable L

1

-theory

with axioms

∀x(Sx ,= 0),

∀x∀y (Sx = Sy ⇒x = y),

∀x(x + 0 = x),

∀x∀y (x +Sy = S(x +y)),

∀x(x 0 = 0),

∀x∀y (x Sy = (x y) +x).

The axioms of Z

1

consist of the axioms of Q plus the induction scheme, i.e., the

universal closure of

(A[x/0] ∧∀x(A⇒A[x/Sx])) ⇒∀xA

where A is any L

1

-formula.

Note that the induction scheme consists of an inﬁnite set of axioms. It can

be shown that Z

1

is not ﬁnitely axiomatizable.

103

Remark 5.5.2 (practical completeness of ﬁrst-order arithmetic). The intended

model of Z

1

is the natural number system

N = (N, +, , 0, S, =)

where N = ¦0, 1, 2, . . .¦ and +, , 0, = are as expected, and S is the successor

function, S(n) = n + 1 for all n ∈ N. Obviously the axioms of Z

1

are true in

this model. The idea behind the axioms of Z

1

is that we are trying to write

down a set S

1

of L

1

-sentences with the property that, for all L

1

-sentences A,

S

1

⊢ A if and only if N [= A. In particular, the axioms of Z

1

are intended

to be complete. Unfortunately this intention cannot be fulﬁlled, as shown by

G¨odel’s First Incompleteness Theorem (see Chapter 6). However, it is known

that Z

1

“almost” fulﬁlls the intention. This is because it has been found that, in

practice, all or most of the theorems of number theory which can be written in

(deﬁnitional extensions of) Z

1

are either provable or refutable in Z

1

. Exceptions

are known, but the exceptions are obscure and marginal. In this sense, we can

say that Z

1

is “practically complete.”

Remark 5.5.3 (deﬁnitional extensions of ﬁrst-order arithmetic). We have seen

in Theorem 5.3.4 and Exercise 5.3.9 that the exponential function (n, k) → n

k

and the “kth prime” function k → p

k

are deﬁnable over N, the intended model

of Z

1

. It can be shown that the same deﬁnitions work abstractly over the

theory Z

1

. Thus we have a deﬁnitional extension of Z

1

in which basic properties

such as ∀m∀n∀k (m

k

n

k

= (mn)

k

) and ∀n∀i ∀j (n

i

n

j

= n

i+j

) can be stated and

proved. Similarly, many other number-theoretical operations can be introduced,

and their properties proved, in deﬁnitional extensions of Z

1

. Our remarks above

concerning practical completeness of Z

1

apply all the more to these deﬁnitional

extensions of Z

1

.

Exercise 5.5.4. Show that the commutative laws ∀x∀y (x + y = y + x) and

∀x∀y (x y = y x) are theorems of Z

1

. Similarly it can be shown that the

associative laws ∀x∀y ∀z (x + (y + z) = (x + y) + z) and ∀x∀y ∀z (x (y z) =

(x y) z) and the distributive law ∀x∀y ∀z (x (y + z) = (x y) + (x z)) are

theorems of Z

1

.

Deﬁnition 5.5.5 (second-order arithmetic). Another important foundational

theory is second-order arithmetic, Z

2

. See also my book [3].

The language of Z

2

is a 2-sorted language, L

2

, with sorts σ and τ designating

numbers and sets respectively. The number variables x

σ

, y

σ

, . . . are written as

i, j, k, l, m, n, . . ., while the set variables x

τ

, y

τ

, . . . are written as X, Y, Z, . . ..

The predicates and operations of L

2

are +, , 0, S, =, ∈. Here + and are

binary number operations, S is a unary number operation, and 0 is a numerical

constant. Thus + and are of type σ σ → σ, S is of type σ → σ, and 0 is of

type σ. In addition, the numerical equality predicate = is a binary predicate of

type σ σ, and the membership predicate ∈ is a binary predicate of “mixed”

type σ τ.

We identify the variables x, y, . . . of L

1

with the number variables m, n, . . .

of L

2

. Also, we identify the operations and predicates +, , 0, S, = of L

1

with

the +, , 0, S, = of L

2

. Thus L

1

is a sublanguage of L

2

.

104

The axioms of Z

2

consist of Q (Deﬁnition 5.5.1), i.e.,

∀m(Sm ,= 0),

∀m∀n(Sm = Sn⇒m = n),

∀m(m + 0 = m),

∀m∀n(m +Sn = S(m+n)),

∀m(m 0 = 0),

∀m∀n(m Sn = (m n) +m),

plus the induction axiom

∀X ((0 ∈ X ∧∀n(n ∈ X⇒Sn ∈ X)) ⇒∀n(n ∈ X)),

plus the comprehension scheme, i.e., the universal closure of

∃X ∀n(n ∈ X⇔A)

where A is any L

2

-formula in which X does not occur.

Note that the comprehension scheme consists of an inﬁnite set of axioms. It

can be shown that Z

2

is not ﬁnitely axiomatizable.

Remark 5.5.6 (practical completeness of second-order arithmetic). The in-

tended model of Z

2

is the 2-sorted structure

P(N) = (N, P(N), +, , 0, S, =, ∈)

where (N, +, , 0, S, =) is the natural number system (see Remark 5.5.2), P(N)

is the power set of N, i.e., the set of all subsets of N,

P(N) = ¦X [ X ⊆ N¦,

and ∈ is the membership relation between natural numbers and sets of natural

numbers, i.e.,

∈ = ¦¸n, X) ∈ N P(N) [ n ∈ X¦.

As in the case of Z

1

(compare Remark 5.5.2), it is obvious that the axioms of

Z

2

are true in the 2-sorted structure P(N), and again, the idea behind Z

2

is

that we are trying to write down axioms for the complete theory of P(N). This

intention cannot be fulﬁlled because of the G¨odel Incompleteness Theorem, but

again, Z

2

is “practically complete.”

Remark 5.5.7 (deﬁnitional extensions of second-order arithmetic). The foun-

dational signiﬁcance of Z

2

is that, within deﬁnitional extensions of Z

2

, it is

possible and convenient to develop the bulk of ordinary countable and sep-

arable mathematics. This includes diﬀerential equations, analysis, functional

analysis, algebra, geometry, topology, combinatorics, descriptive set theory, etc.

For details of how this can be done, see my book [3].

105

Exercise 5.5.8. As noted in Deﬁntion 5.5.5, L

1

is a sublanguage of L

2

. Show

that all of the axioms of Z

1

are theorems of Z

2

. In this sense Z

1

is a subtheory

of Z

2

.

Solution. Recall that Q is a subtheory of Z

2

. It remains to show that, for any

L

1

-formula A, the universal closure of (A[n/0] ∧∀n(A⇒A[n/Sn])) ⇒∀nA is a

theorem of Z

2

. More generally, we shall prove this when A is any L

2

-formula. Let

the free variables of A be among n, n

1

, . . . , n

k

, X

1

, . . . , X

l

. Within Z

2

we reason

as follows. Given n

1

, . . . , n

k

, X

1

, . . . , X

l

, we use the comprehension scheme of

Z

2

to prove the existence of a set X such that ∀n(n ∈ X⇔A). From the

induction axiom of Z

2

we have (0 ∈ X∧∀n(n ∈ X⇒Sn ∈ X)) ⇒∀n(n ∈ X).

From this it follows that (A[n/0] ∧∀n(A⇒A[n/Sn])) ⇒∀nA. Since this holds

for arbitrary n

1

, . . . , n

k

, X

1

, . . . , X

l

, we obtain the universal closure.

Remark 5.5.9 (nth order arithmetic). Similarly we can deﬁne third-order arith-

metic, Z

3

, a 3-sorted theory with variables intended to range over numbers, sets

of numbers, and sets of sets of numbers. The intended model of Z

3

is the 3-sorted

structure

P

2

(N) = (N, P(N), P(P(N)), +, , 0, S, =, ∈

1

, ∈

2

)

where

∈

1

= ¦¸n, X) ∈ N P(N) [ n ∈ X¦

and

∈

2

= ¦¸X, /) ∈ P(N) P(P(N)) [ X ∈ /¦.

More generally, for each positive integer n ≥ 1 we can deﬁne nth-order arith-

metic, Z

n

, an n-sorted theory whose intended model is the n-sorted structure

P

n−1

(N) = P(P (P

. ¸¸ .

n−1

(N)) )

where P is the power set operation. Thus we have a sequence of theories

Z

1

⊆ Z

2

⊆ ⊆ Z

n

⊆ Z

n+1

⊆ .

The union Z

∞

=

∞

n=1

Z

n

is known as simple type theory. This theory is his-

torically important, because it or something like it was oﬀered by Russell as a

solution of the Russell Paradox.

5.6 Axiomatic Set Theory

We now turn to another kind of foundational theory, known as axiomatic set

theory.

Deﬁnition 5.6.1 (the language of set theory). The language of set theory, L

set

,

consists of two binary predicates, ∈ and =, the membership predicate and the

identity predicate. The variables u, v, w, x, y, z, . . . are intended to range over

106

sets. Note that u ∈ x means that u is an element of the set x, i.e., a member of

x. We also use notations such as x = ¦u, v, . . .¦ meaning that x is a set whose

elements are u, v, . . ., and x = ¦u [ . . .¦ meaning that x is a set whose elements

are all u such that . . . .

Remark 5.6.2 (pure well-founded sets). In order to motivate and clarify our

presentation of the axioms of set theory, we ﬁrst note that the axioms are

intended to apply only to sets which are pure and well-founded. A set x is said

to be pure if all elements of x are sets, all elements of elements of x are sets,

all elements of elements of elements of x are sets, etc. A set x is said to be

well-founded if there is no inﬁnite descending ∈-chain

∈ u

n+1

∈ u

n

∈ ∈ u

2

∈ u

1

∈ u

0

= x.

Remark 5.6.3. In order to state the axioms and theorems of set theory eﬃ-

ciently, we shall tacitly employ the technique of deﬁnitional extensions, which

has been discussed in Section 5.4.

Deﬁnition 5.6.4 (Zermelo set theory). Zermelo set theory with the axiom of

choice, ZC, is a theory in the language L

set

consisting of the following axioms.

1. The axiom of extensionality: ∀x∀y (x = y ⇔∀u (u ∈ x⇔u ∈ y)).

We deﬁne a binary predicate ⊆ by x ⊆ y ⇔∀u (u ∈ x⇒u ∈ y), i.e., x is a

subset of y. Extensionality

1

says that x = y is equivalent to x ⊆ y ∧y ⊆ x.

2. The pairing axiom: ∀x∀y ∃z ∀u (u ∈ z ⇔(u = x∨u = y)).

By extensionality this z is unique, so we deﬁne a binary operation ¦x, y¦ =

this z. Note that ¦x, y¦ is the unordered pair consisting of x and y. In

addition, we deﬁne a unary operation ¦x¦ = ¦x, x¦. Note that ¦x¦ is the

singleton set consisting of x.

We also deﬁne a binary operation (x, y) = ¦¦x¦, ¦x, y¦¦, the ordered pair

consisting of x and y. Using extensionality, we can prove the basic property

∀x∀y ∀u ∀v ((x, y) = (u, v) ⇔(x = u ∧y = v)).

3. The union axiom: ∀x∃z ∀u (u ∈ z ⇔∃v (u ∈ v ∧v ∈ x)).

By extensionality this z is unique, so we deﬁne a unary operation

x =

this z. Note that

**x is the union of x, i.e., the union of all of the sets
**

which are elements of x. We also deﬁne a binary operation x∪y =

¦x, y¦,

the union of x and y.

4. The power set axiom: ∀x∃z ∀y (y ∈ z ⇔y ⊆ x).

1

Recall that Lset is a one-sorted language with only set variables. This fact together with

the axiom of extensionality embodies our restriction to pure sets. Later we shall introduce

another axiom, the axiom of foundation, which embodies our restriction to well-founded sets.

107

By extensionality this z is unique, so we deﬁne a unary operation P(x) =

this z. Note that P(x) is the power set of x,

P(x) = ¦y [ y ⊆ x¦,

the set of all subsets of x.

5. The comprehension scheme: the universal closure of

∀x∃z ∀u (u ∈ z ⇔(u ∈ x∧A))

where A is any L

set

-formula in which z does not occur.

The idea here is that x is a given set, and A expresses a property of

elements u ∈ x. The comprehension scheme asserts the existence of a set

z = ¦u ∈ x [ A¦,

i.e., z is a subset of x consisting of all u ∈ x such that A holds. By

extensionality, this z is unique.

Note that the comprehension scheme consists of an inﬁnite set of axioms.

It can be shown that ZC is not ﬁnitely axiomatizable.

Using comprehension, we deﬁne binary operations

(a) x ∩ y = ¦u ∈ x [ u ∈ y¦, the intersection of x and y,

(b) x ¸ y = ¦u ∈ x [ u / ∈ y¦, the set-theoretic diﬀerence of x and y, and

(c) x y = ¦w ∈ P(P(x ∪ y)) [ ∃u ∃v (u ∈ x∧v ∈ y ∧(u, v) = w)¦, the

Cartesian product of x and y.

We also deﬁne a constant ∅ = ¦¦ = x ¸ x, the empty set.

We deﬁne a unary predicate Fcn(f) saying that f is a function, i.e., a set

f such that

∀w(w ∈ f ⇒∃x∃y (w = (x, y)))

and

∀x∀y ∀z (((x, y) ∈ f ∧(x, z) ∈ f) ⇒y = z).

Using comprehension, the domain and range of f are deﬁned as unary

operations

dom(f) = ¦x ∈

__

f [ ∃y ((x, y) ∈ f)¦

and

ran(f) = ¦y ∈

__

f [ ∃x((x, y) ∈ f)¦.

We also deﬁne f(x), the value of f at x ∈ dom(f), to be the unique y

such that (x, y) ∈ f.

108

6. The axiom of inﬁnity: ∃z (∅ ∈ z ∧∀x∀y ((x ∈ z ∧y ∈ z) ⇒x ∪ ¦y¦ ∈ z)).

Using comprehension, we can prove the existence of a unique smallest set

z as above, namely the intersection of all such sets. We deﬁne a constant

HF = this z. Note that HF is an inﬁnite set. The elements of HF are the

hereditarily ﬁnite sets, i.e., those pure well-founded sets x such that C(x)

is ﬁnite. Here

C(x) = x ∪

_

x ∪

__

x ∪ ,

i.e., C(x) is the set consisting of all elements of x, elements of elements of

x, elements of elements of elements of x, . . . .

Similarly, we can prove that there exists a unique smallest set w such

that ∅ ∈ w∧∀x(x ∈ w⇒x ∪ ¦x¦ ∈ w). We deﬁne a constant ω = this

w. Note that ω ⊆ HF. The elements of ω are just the natural numbers,

inductively identiﬁed with hereditarily ﬁnite sets via n = ¦0, 1, . . . , n−1¦.

Thus ω = N, the set of natural numbers, and we have 0 = ∅ and, for all

n, n + 1 = n ∪ ¦n¦. We deﬁne a ﬁnite sequence to be a function f such

that dom(f) ∈ ω. We deﬁne an inﬁnite sequence to be a function f such

that dom(f) = ω.

A set x is said to be ﬁnite if

∃n∃f (n ∈ ω ∧Fcn(f) ∧dom(f) = n∧ran(f) = x).

We deﬁne [x[, the cardinality of x, to be the least such n. We can prove

that, for all n ∈ ω, [n[ = n. For m, n ∈ ω we deﬁne

m+n = [(m¦0¦) ∪ (n ¦1¦)[

and m n = [mn[. On this basis, we can prove the essential properties

of + and on ω. Moreover, from the deﬁnition of ω it follows that

∀x((x ⊆ ω ∧0 ∈ x∧∀n(n ∈ x⇒n ∪ ¦n¦ ∈ x)) ⇒x = ω).

Thus we have a copy of the natural number system.

7. The axiom of choice:

∀f ((Fcn(f) ∧∀x(x ∈ dom(f) ⇒f(x) ,= ∅)) ⇒

∃g (Fcn(g) ∧dom(g) = dom(f) ∧∀x(x ∈ dom(f) ⇒g(x) ∈ f(x)))).

The axiom of choice says that, given an indexed family of nonempty sets,

there exists a function which chooses one element from each of the sets.

There is a history of controversy surrounding this axiom.

8. The axiom of foundation: ∀x(x ,= ∅ ⇒∃u (u ∈ x∧u ∩ x = ∅)).

The axiom of foundation amounts to saying that all sets are well-founded.

To see this, note that if there were an inﬁnite descending ∈-chain

∈ u

n+1

∈ u

n

∈ ∈ u

2

∈ u

1

∈ u

0

109

then we would have a counterexample to the axiom of foundation, namely

x = ¦u

0

, u

1

, . . . , u

n

, u

n+1

, . . .¦. Conversely, if x were a counterexample to

the axiom of foundation, i.e., x ,= ∅ and ∀u (u ∈ x⇒u ∩ x ,= ∅), then

we could use the axiom of choice to obtain an inﬁnite descending ∈-chain

u

1

∈ x, u

2

∈ u

1

∩ x, u

3

∈ u

2

∩ x, . . . .

Remark 5.6.5 (foundational signiﬁcance of set theory). The signiﬁcance of

ZC and similar theories is that they can serve as an axiomatic, set-theoretical

foundation for virtually all of mathematics. As already noted, within ZC we

have the natural number system, N. On this basis, it is possible to follow the

usual Dedekind construction of the integers Z, the rational numbers Q, and

the real numbers R. We can also develop the theory of higher mathematical

objects such as manifolds, topological spaces, operators on Banach spaces, etc.,

all within (deﬁnitional extensions of) ZC.

Deﬁnition 5.6.6 (Zermelo/Fraenkel set theory). ZFC, Zermelo/Fraenkel set

theory with the axiom of choice, consists of ZC, Zermelo set theory with the

axiom of choice, plus

9. The replacement scheme: universal closure of

(∀u (∃ exactly one v) A) ⇒∀x∃y ∀v (v ∈ y ⇔∃u (u ∈ x∧A))

where A is any L

set

-formula in which y does not occur.

The idea here is that to each u is associated exactly one v such that A

holds. Under this assumption, we assert that for all sets x there exists a

set y consisting of all v associated to some u ∈ x. By extensionality, this

y is unique.

Note that the replacement scheme consists of an inﬁnite set of axioms. It can

be shown that ZFC is not ﬁnitely axiomatizable.

Remark 5.6.7 (practical completeness of set theory). The most obvious model

of Zermelo set theory is the set

P

ω

(HF) =

_

n∈ω

P

n

(HF) =

_

n∈N

P(P (P

. ¸¸ .

n

(HF)) )

consisting of HF plus all subsets of HF plus all subsets of subsets of HF plus

. . . . Thus P

ω

(HF) is the set of all sets of ﬁnite order over HF. The axioms of

ZC express evident properties of P

ω

(HF).

The existence of the set P

ω

(HF) cannot be proved in Zermelo set theory.

However, the existence of P

ω

(HF) can be proved in Zermelo/Fraenkel set theory,

as follows. Let A be a formula which associates to each n ∈ N the set P

n

(HF).

Since N is a set, the replacement scheme for A gives us the set ¦P

n

(HF) [ n ∈ N¦,

and then the union axiom gives us

¦P

n

(HF) [ n ∈ N¦ = P

ω

(HF).

The intended model of Zermelo/Fraenkel set theory is the collection V of all

pure, well-founded sets. V is also known as the universe of set theory. It can be

110

shown that V =

α

P

α

(HF), where α ranges over transﬁnite ordinal numbers.

Thus V is the collection of all sets of all transﬁnite orders over HF.

The axioms of ZFC express evident properties of V . Moreover, it has been

found that ZFC is “practically complete” in the sense that all L

set

-sentences

expressing evident properties of V are provable in ZFC. At the same time, there

are many interesting L

set

-sentences, e.g., the Continuum Hypothesis, which are

neither evidently true nor evidently false in V according to our current under-

standing, and which are known to be neither provable nor refutable in ZFC.

Thus it appears that ZFC accurately reﬂects our current understanding of V .

5.7 Interpretability

Deﬁnition 5.7.1. Let T

1

and T

2

be theories. We say that T

1

is a subtheory of

T

2

if the language of T

1

is included in the language of T

2

and the theorems of

T

1

are included in the theorems of T

2

.

Deﬁnition 5.7.2 (interpretability). Let T

1

and T

2

be theories. We say that T

1

is interpretable in T

2

if T

1

is a subtheory of some deﬁnitional extension of T

2

.

Intuitively, this means that T

2

is “at least as strong as” T

1

, in some abstract

sense. We sometimes write T

1

≤ T

2

to mean that T

1

is interpretable in T

2

.

Remark 5.7.3. It is straightforward to show that the interpretability relation

is transitive. In other words, T

1

≤ T

2

and T

2

≤ T

3

imply T

1

≤ T

3

. Thus we have

equivalence classes of theories under mutual interpretability, partially ordered

by the interpretability relation.

Examples 5.7.4. First-order arithmetic is interpretable in second-order arith-

metic, and second-order arithmetic is interpretable in set theory. More generally,

for all n ≥ 1, nth-order arithmetic is interpretable in (n +1)st-order arithmetic

and in set theory. It can be shown that (n+1)st-order arithmetic and set theory

are not interpretable in nth order arithmetic. In particular, second-order arith-

metic is not interpretable in ﬁrst-order arithmetic. Results of this kind follow

from G¨odel’s Second Incompleteness Theorem. We have

Z

1

< Z

2

< < Z

n

< Z

n+1

< < Z

∞

< ZC < ZFC

where T

1

< T

2

means that T

1

is “weaker than” T

2

, i.e., T

2

is “stronger than”

T

1

, i.e., T

1

is interpretable in T

2

and not vice versa.

Remark 5.7.5 (the G¨odel hierarchy). The partial ordering of foundational

theories under interpretability is sometimes known as the G¨ odel hierarchy. This

hierarchy is of obvious foundational interest.

Remark 5.7.6. The foundational signiﬁcance of interpretability is highlighted

by the following observations. If T

1

is interpretable in T

2

, then:

1. Consistency of T

2

implies consistency of T

1

.

111

2. Essential incompleteness of T

1

implies essential incompleteness of T

2

.

3. Eﬀective essential incompleteness of T

1

implies eﬀective essential incom-

pleteness of T

2

.

5.8 Beth’s Deﬁnability Theorem

In this section we consider implicit deﬁnitional extensions of theories. We state

and prove Beth’s Deﬁnability Theorem, which says that an implicit deﬁnitional

extension of a theory T is always equivalent to an explicit deﬁnitional extension

of the same theory, T.

We consider only the case of predicates, but operations can be handled

similarly. Let T = (L, S) be a theory, and let P be an n-ary predicate of L.

Deﬁnition 5.8.1.

1. We say that T explicitly deﬁnes P if there exists an L-formula D not

involving P with free variables x

1

, . . . , x

n

such that

T ⊢ ∀x

1

∀x

n

(Px

1

x

n

⇔D).

2. We say that T implicitly deﬁnes P if, letting P

′

be a new n-ary predicate

and letting T

′

= T[P/P

′

], we have

T ∪ T

′

⊢ ∀x

1

∀x

n

(Px

1

x

n

⇔P

′

x

1

x

n

).

Theorem 5.8.2 (Beth’s Deﬁnability Theorem). T explicitly deﬁnes P if and

only if T implicitly deﬁnes P.

Proof. Assume ﬁrst that T explicitly deﬁnes P, say

T ⊢ ∀x

1

∀x

n

(Px

1

x

n

⇔D).

It follows that

T

′

⊢ ∀x

1

∀x

n

(P

′

x

1

x

n

⇔D).

Hence

T ∪ T

′

⊢ ∀x

1

∀x

n

(Px

1

x

n

⇔P

′

x

1

x

n

),

i.e., T implicitly deﬁnes P.

Conversely, assume that T implicitly deﬁnes P, i.e.,

T ∪ T

′

⊢ ∀x

1

∀x

n

(Px

1

x

n

⇔P

′

x

1

x

n

).

By the Compactness Theorem, there are ﬁnitely many axioms A

1

, . . . , A

k

∈ S

such that

(A∧A

′

) ⇒∀x

1

∀x

n

(Px

1

x

n

⇔P

′

x

1

x

n

)

112

is logically valid, where A = A

1

∧ ∧A

k

and A

′

= A[P/P

′

]. Introducing

parameters a

1

, . . . , a

n

, we see that

(A∧A

′

) ⇒(Pa

1

a

n

⇔P

′

a

1

a

n

) (5.1)

is logically valid. It follows quasitautologically that

(A∧Pa

1

a

n

) ⇒(A

′

⇒P

′

a

1

a

n

) (5.2)

is logically valid. By the Interpolation Theorem 3.5.1, we can ﬁnd an L-formula

D with free variables x

1

, . . . , x

n

such that D[x

1

/a

1

, . . . , x

n

/a

n

] is an interpolant

for (5.2). Thus

(A∧Pa

1

a

n

) ⇒D[x

1

/a

1

, . . . , x

n

/a

n

] (5.3)

and

D[x

1

/a

1

, . . . , x

n

/a

n

] ⇒(A

′

⇒P

′

a

1

a

n

) (5.4)

are logically valid, and P and P

′

do not occur in D. Since (5.4) is logically

valid, it follows that

D[x

1

/a

1

, . . . , x

n

/a

n

] ⇒(A⇒Pa

1

a

n

) (5.5)

is logically valid. From the logical validity of (5.3) and (5.5), it follows quasitau-

tologically that

A⇒(Pa

1

a

n

⇔D[x

1

/a

1

, . . . , x

n

/a

n

]) (5.6)

is logically valid. Hence

A⇒∀x

1

∀x

n

(Px

1

x

n

⇔D)

is logically valid, so

T ⊢ ∀x

1

∀x

n

(Px

1

x

n

⇔D).

Thus we see that T explicitly deﬁnes P.

113

Chapter 6

Arithmetization of

Predicate Calculus

6.1 Primitive Recursive Arithmetic

Deﬁnition 6.1.1. To each natural number n we associate a variable-free PRA-

term n as follows: 0 = 0, n + 1 = S n. Thus

n = S S

. ¸¸ .

n

0.

These terms are known as numerals.

Theorem 6.1.2. Let f be a k-ary primitive recursive function. Then for all

k-tuples of natural numbers m

1

, . . . , m

k

we have

PRA ⊢ f m

1

m

k

= f(m

1

, . . . , m

k

) .

Proof.

6.2 Interpretability of PRA in Z

1

6.3 G¨odel Numbers

Let L be a countable language. Assume that to all the sorts σ, predicates P,

and operations f of L have been assigned distinct positive integers #(σ), #(P),

#(f) respectively. As usual, let V be the set of parameters.

Deﬁnition 6.3.1 (G¨ odel numbers). To each L-V -term t and L-V -formula A

we assign a positive integer, the G¨ odel number of t or of A, denoted #(t) or

114

#(A), respectively.

#(v

σ

i

) = 2 3

#(σ)

5

i

#(a

σ

i

) = 2

2

3

#(σ)

5

i

#(ft

1

t

n

) = 2

3

3

#(f)

p

#(t1)

2

p

#(tn)

n+1

if f is an n-ary operation

#(Pt

1

t

n

) = 2

4

3

#(P)

p

#(t1)

2

p

#(tn)

n+1

if P is an n-ary predicate

#(A) = 2

5

3

#(A)

#(A∧B) = 2

6

3

#(A)

5

#(B)

#(A∨B) = 2

7

3

#(A)

5

#(B)

#(A⇒B) = 2

8

3

#(A)

5

#(B)

#(A⇔B) = 2

9

3

#(A)

5

#(B)

#(∀v A) = 2

10

3

#(v)

5

#(A)

#(∃v A) = 2

11

3

#(v)

5

#(A)

Deﬁnition 6.3.2. The language L is said to be primitive recursive if the fol-

lowing items are primitive recursive.

Sort(x) ≡ x = #(σ) for some sort σ

Pred(x) ≡ x = #(P) for some predicate P

Op(x) ≡ x = #(f) for some operation f

arity(#(P)) = n if P is an n-ary predicate

arity(#(f)) = n if f is an n-ary operation

sort(#(P), i) = #(σ

i

) if 1 ≤ i ≤ n and P is an n-ary predicate of

type σ

1

σ

n

sort(#(f), i) = #(σ

i

) if 1 ≤ i ≤ n + 1 and f is an n-ary operation

of type σ

1

σ

n

→ σ

n+1

Lemma 6.3.3. If L is primitive recursive, then the following are primitive

recursive.

Var(x) ≡ x = #(v) for some variable v

Param(x) ≡ x = #(a) for some parameter a

Term(x) ≡ x = #(t) for some term t

ClTerm(x) ≡ x = #(t) for some closed term t

AtFml(x) ≡ x = #(A) for some atomic formula A

Fml(x) ≡ x = #(A) for some formula A

sort(#(t)) = #(σ) if t is a term of sort σ

Proof. We have

Var(x) ≡ (x)

0

= 1 ∧ x = 2

(x)0

3

(x)1

5

(x)2

∧ Sort((x)

1

)

115

and

Param(x) ≡ (x)

0

= 2 ∧ x = 2

(x)0

3

(x)1

5

(x)2

∧ Sort((x)

1

) .

To show that the predicate Term(x) is primitive recursive, we ﬁrst show that the

function sort(x) is primitive recursive, where sort(#(t)) = #(σ) if t is a term of

sort σ, sort(x) = 0 otherwise. Put lh(x) = least w < x such that (x)

w

= 0. We

then have

sort(x) =

_

¸

_

¸

_

(x)

1

if Var(x) ∨ Param(x) ,

sort((x)

1

, lh(x)

−1) if (x)

0

= 3 ∧ Op((x)

1

) ∧ (+) ,

0 otherwise ,

where

(+) arity((x)

1

) = lh(x)

−2 ∧ x =

lh(x)

−1

i=0

p

(x)i

i

∧ (∀i < lh(x)

−2) (sort((x)

i+2

) = sort((x)

1

, i + 1)) .

Then

Term(x) ≡ sort(x) > 0 .

For closed terms, deﬁne clsort(x) like sort(x) replacing Var(x) ∨Param(x) by

Param(x). We then have

ClTerm(x) ≡ clsort(x) > 0 .

For formulas we have

AtFml(x) ≡ ((x)

0

= 4 ∧ Pred((x)

1

) ∧ (+))

and

Fml(x) ≡ AtFml(x) ∨ ((x)

0

= 5 ∧ x = 2

(x)0

3

(x)1

∧ Fml((x)

1

))

∨ (6 ≤ (x)

0

≤ 9 ∧ x = 2

(x)0

3

(x)1

5

(x)2

∧ Fml((x)

1

) ∧ Fml((x)

2

))

∨ (10 ≤ (x)

0

≤ 11 ∧ x = 2

(x)0

3

(x)1

5

(x)2

∧ Var((x)

1

) ∧ Fml((x)

2

))

and this completes the proof.

Lemma 6.3.4 (substitution). There is a primitive recursive function sub(x, y, z)

such that for any formula A and any variable v and any closed term t,

sub(#(A), #(v), #(t)) = #(A[v/t]) .

Proof.

sub(x, y, z) =

_

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

_

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

_

z if x = y,

2

(x)0

3

(x)1

lh(x)

−1

i=2

p

sub((x)i,y,z)

i

if 3 ≤ (x)

0

≤ 4,

2

(x)0

3

sub((x)1,y,z)

if (x)

0

= 5,

2

(x)0

3

sub((x)1,y,z)

5

sub((x)2,y,z)

if 6 ≤ (x)

0

≤ 9,

2

(x)0

3

(x)1

5

sub((x)2,y,z)

if 10 ≤ (x)

0

≤ 11 ∧(x)

1

,= y,

x otherwise.

116

**Lemma 6.3.5. If L is primitive recursive, then the predicate
**

Snt(x) ≡ x = #(A) for some sentence A

is primitive recursive.

Proof. Recall that, by Exercise 2.1.10, a formula A is a sentence if and only if

A[v/a] = A for all variables v occurring in A. Note also that if y = #(v

σ

i

) then

2y = #(a

σ

i

). Thus we have

Snt(x) ≡ Fml(x) ∧ (∀y < x) (Var(y) ⇒ x = sub(x, y, 2y)) .

**Lemma 6.3.6. There is a primitive recursive function num(x) such that
**

num(n) = #(n)

for any nonnegative integer n.

Proof. The recursion equations for num(x) are

num(0) = #(0) ,

num(x + 1) = 2

3

3

#(S)

5

num(x)

.

6.4 Undeﬁnability of Truth

In this section, let T be a theory which includes PRA. For example, we could

take T to be PRA itself. Or, by Section 6.2, we could take T to be an appropriate

deﬁnitional extension of Z

1

or Z

2

or ZFC.

Lemma 6.4.1 (Self-Reference Lemma). Let L be the language of T. Let A be

an L-formula with a free number variable x. Then we can ﬁnd an L-formula B

such that

T ⊢ B⇔A[x/#(B) ] .

The free variables of B are those of A except for x. In particular, if x is the

only free variable of A, then B is an L-sentence.

Proof. Put d(z) = sub(z, #(x), num(z)). Thus d is a 1-ary primitive recursive

function such that, if A is any L-formula containing the number variable x

as a free variable, then d(#(A)) = #(A[x/#(A) ]. Now given A as in the

hypothesis of the lemma, let D be the formula A[x/d x], and let B be the

formula A[x/d #(D) ], i.e., D[x/#(D) ]. Note that d(#(D)) = #(B). It follows

by Theorem 6.1.2 that PRA ⊢ d #(D) = #(B). Since T includes PRA, it

follows that T ⊢ d #(D) = #(B). Hence T ⊢ A[x/d #(D)] ⇔A[x/#(B) ]. In

other words, T ⊢ B⇔A[x/#(B) ]. This completes the proof.

117

Deﬁnition 6.4.2. If M is any model of T, let True

M

be the set of G¨odel

numbers of sentences that are true in M, i.e.,

True

M

= ¦#(B) [ B is a sentence and M [= B¦.

Deﬁnition 6.4.3. An ω-model of T is a model M of T such that the number

domain of M is ω = ¦0, 1, 2, . . .¦ and 0

M

= 0 and S

M

(n) = n + 1 for all n ∈ ω.

More generally, if M is any model of T, we may assume that ω is identiﬁed

with a subset of the number domain of M in such a way that 0

M

= 0 and

S

M

(n) = n + 1 for each n ∈ ω. Thus each n ∈ ω is identiﬁed with the element

of M that is denoted by n, i.e., n = v

M

(n).

Theorem 6.4.4 (undeﬁnability of truth). If M is an ω-model of T, then True

M

is not explicitly deﬁnable over M. More generally, if M is any model of T, then

the characteristic function of True

M

is is not included in the characteristic

function of any subset of M that is explicitly deﬁnable over M.

Proof. Let X be a subset of the number domain of M which is explicitly deﬁnable

over M. Let A be an L-formula with a free number variable x and no other free

variables, such that A explicitly deﬁnes X over M. Applying Lemma 6.4.1 to

the negation of A, we obtain an L-sentence B such that T ⊢ B⇔A[x/#(B) ].

Since M is a model of T, it follows that #(B) ∈ True

M

if and only if #(B) / ∈ X.

Hence the characteristic function of True

M

is not included in the characteristic

function of X, q.e.d.

Corollary 6.4.5. Let M = (ω, +, , 0, 1, =) be the standard model of ﬁrst-order

arithmetic, Z

1

. Then True

M

is not explicitly deﬁnable over M. This

1

may be

paraphrased by saying that arithmetical truth is not arithmetically deﬁnable.

Remark 6.4.6. With M = (ω, +, , 0, 1, =) as above, it can be shown that

True

M

is implicitly deﬁnable over M. (See also Exercise 6.4.7.) Thus the

Beth’s Deﬁnability Theorem does not hold for deﬁnability over this particular

model.

Exercise 6.4.7. Let M be an ω-model of Z

1

or Z

2

or ZFC. Let Sat

M

be the

satisfaction relation on M. Show that Sat

M

is implicitly deﬁnable over M.

6.5 The Provability Predicate

In this section, let L be a primitive recursive language, and let T be an L-

theory which is primitive recursively axiomatizable. For example, T could be

PRA itself, or T could be any of the mathematical or foundational theories

discussed in Sections 5.2, 5.5, 5.6.

Deﬁnition 6.5.1. Choose a primitive recursive predicate Ax

T

for the set of

G¨odel numbers of axioms of T. In terms of Ax

T

show that various predicates

1

This result is due to Tarski [5].

118

associated with T are primitive recursive. Introduce the provability predicate

Pvbl

T

by deﬁnition:

Pvbl

T

(x) ⇔ ∃y Prf

T

(x, y) .

Note that, for all L-sentences B, Pvbl

T

(#(B)) is true if and only if T ⊢ B.

Lemma 6.5.2 (derivability condition 1). For any L-sentence A, if T ⊢ A then

PRA ⊢ Pvbl

T

(#(A)) .

Proof. Suppose T ⊢ A. Let p be a proof of A in T. Then Prf

T

(#(A), #(p))

holds. Since Prf

T

(x, y) is a primitive recursive predicate, it follows by Theorem

6.1.2 that PRA ⊢ Prf

T

(#(A), #(p)). Hence PRA ⊢ Pvbl

T

(#(A)), q.e.d.

Lemma 6.5.3 (derivability condition 2). For any L-sentence A, we have

PRA ⊢ Pvbl

T

(#(A)) ⇒Pvbl

PRA

(#(Pvbl

T

(#(A)))) .

Proof. This is just Lemma 6.5.2 formalized in PRA. The details of the formal-

ization are in Section 6.7.

Lemma 6.5.4 (derivability condition 3). For any L-sentences A and B, we

have

PRA ⊢ Pvbl

T

(#(A⇒B)) ⇒ (Pvbl

T

(#(A)) ⇒Pvbl

T

(#(B))) .

Proof. This is a straightforward consequence of the fact that our rules of infer-

ence include modus ponens.

6.6 The Incompleteness Theorems

In this section, let T be a theory which is primitive recursively axiomatizable

and includes PRA. For example, T could be PRA itself, or it could be an

appropriate deﬁnitional extension of Z

1

or Z

2

or ZFC. As in Section 6.5, let

Pvbl

T

be a provability predicate for T.

Lemma 6.6.1. Let A(x) be a PRA-formula with one free variable x. Then we

can ﬁnd a PRA-sentence B such that PRA ⊢ B⇔A(#(B)).

Proof. This is the Self-Reference Lemma 6.4.1 specialized to PRA.

Lemma 6.6.2. We can ﬁnd a PRA-sentence S such that

PRA ⊢ S ⇔Pvbl

T

(#(S)) . (6.1)

Note that S is self-referential and says “I am not provable in T.”

Proof. This is an instance of Lemma 6.6.1 with A(x) ≡ Pvbl

T

(x).

119

Lemma 6.6.3. Let S be as in Lemma 6.6.2. If T is consistent, then T ,⊢ S.

Proof. Suppose for a contradiction that T ⊢ S. By Lemma 6.5.2 we have

PRA ⊢ Pvbl

T

(#(S)). Hence by (6.1) it follows that PRA ⊢ S. Since T

includes PRA, we have T ⊢ S. Thus T is inconsistent.

Theorem 6.6.4 (the First Incompleteness Theorem). If T is consistent, then

we can ﬁnd a sentence S

′

in the language of ﬁrst-order arithmetic such that S

′

is true yet S

′

is not a theorem of T.

Proof. Let S be a PRA-sentence as in Lemma 6.6.2. By Lemma 6.6.3, T ,⊢ S. It

follows by (6.1) that S is true. As in Section 6.2, let S

′

be the translation of S

into the language of ﬁrst-order arithmetic. Thus S

′

is also true. By the results

of Section 6.2, PRA ⊢ S ⇔S

′

. Hence T ⊢ S ⇔S

′

. Hence T ,⊢ S

′

.

Assume now that the primitive recursive predicate Ax

T

has been chosen

in such a way that PRA ⊢ ∀x(Ax

PRA

(x) ⇒Ax

T

(x)). It follows that PRA ⊢

∀x(Pvbl

PRA

(x) ⇒Pvbl

T

(x)). In particular we have:

Lemma 6.6.5. For all PRA-sentences A, we have

PRA ⊢ Pvbl

PRA

(#(A)) ⇒Pvbl

T

(#(A)) .

Deﬁnition 6.6.6. Con

T

is deﬁned to be the sentence Pvbl

T

(#(0 ,= 0)). Note

that Con

T

is a PRA-sentence which asserts the consistency of T.

Theorem 6.6.7 (the Second Incompleteness Theorem). If T is consistent, then

T ,⊢ Con

T

.

Proof. Let S be as in Lemma 6.6.2. By Theorem 6.6.4 we have T ,⊢ S. Therefore,

to show T ,⊢ Con

T

, it suﬃces to show T ⊢ Con

T

⇒S. Since T includes PRA, it

suﬃces to show PRA ⊢ Con

T

⇒S. By (6.1) it suﬃces to show

PRA ⊢ Con

T

⇒ Pvbl

T

(#(S)) . (6.2)

But this is just Lemma 6.6.3 formalized in PRA.

Details: We need to prove (6.2). Reasoning in PRA, suppose Pvbl

T

(#(S)).

By Lemma 6.5.3 we have Pvbl

PRA

(#(Pvbl

T

(#(S)))). Moreover, from (6.1)

and Lemma 6.5.2 we have Pvbl

PRA

(#(S ⇔Pvbl

T

(#(S)))). Hence by Lem-

mas 6.5.2 and 6.5.4 we have Pvbl

PRA

(#(S)). By Lemma 6.6.5 it follows that

Pvbl

T

(#(S)). Hence by Lemmas 6.5.2 and 6.5.4 we have Pvbl

T

(#(0 ,= 0)),

i.e., Con

T

. This completes the proof.

Exercise 6.6.8. Show that PRA ⊢ S ⇔Con

T

.

Exercise 6.6.9 (Rosser’s Theorem). Show that if T is as in Theorems 6.6.4

and 6.6.7, then T is incomplete.

Hint: Use the Self-Reference Lemma 6.4.1 to obtain a sentence B such that

PRA ⊢ B⇔A[x/#(B)], where A[x/#(B)] says that for any proof p of B in T

there exists a proof q of B in T such that #(q) < #(p). Using the assumption

that T is consistent, show that T ,⊢ B and T ,⊢ B.

120

Exercise 6.6.10. Give an example of a T as in Theorems 6.6.4 and 6.6.7 such

that T ⊢ Con

T

.

Solution. We may take T = PRA +Con

PRA

, or T = Z

1

+Con

Z1

, etc.

6.7 Proof of Lemma 6.5.3

FIXME Write this section!

121

Bibliography

[1] Neil D. Jones and Alan L. Selman. Turing machines and the spectra of

ﬁrst-order formulas. Journal of Symbolic Logic, 39:139–150, 1974.

[2] David Marker. Model Theory: An Introduction. Springer, 2002. VIII + 342

pages.

[3] Stephen G. Simpson. Subsystems of Second Order Arithmetic. Perspectives

in Mathematical Logic. Springer-Verlag, 1999. XIV + 445 pages.

[4] Raymond M. Smullyan. First-Order Logic. Dover Publications, New York,

1995. XII + 158 pages.

[5] Alfred Tarski. Introduction to Logic and to the Methodology of Deductive

Sciences. Oxford University Press, 4th edition, 1994. XXII + 229 pages.

122

Index

Abelian group, 89

ACF, 93

ACF

0

, 94

ACF

p

, 94

Algebra, Fundamental Theorem of, 93

algebraically closed ﬁeld, 93

antisymmetry, 22

arithmetic

ﬁrst-order, 103

second-order, 104

nth order, 106

arity, 24, 78, 84

assignment, 6

associativity, 10, 30, 104

atomically closed, 43

atomic formula, 3, 24, 79, 85

automorphism, 98

axiom of choice, 109

axiom of foundation, 109

Beth’s Deﬁnability Theorem, 100, 112,

118

binary, 3

block tableau, 61

modiﬁed, 66

bound variable, 25

C (complex numbers), 93

Cartesian product, 108

categoricity, 88, 89

change of bound variables, 38

characteristic, 93

choice, 109

Church’s Theorem, 52

clause, 11

closed, 15, 40

atomically, 43

coﬁnite, 76

commutativity, 10, 89, 92

Compactness Theorem, 72

companion, 51, 52

completeness, 19, 43, 51, 58, 64, 66, 81,

85

practical, 103–105, 111

complete theories, 88–92, 94–96, 103

complex numbers, 93

composite number, 77

comprehension, 105, 108

congruence, 71, 72

consequence

logical, 8, 10, 27, 35, 40, 60, 87

quasitautological, 57, 58

conservative extension, 102

consistency, 87

constant, 78

DAG

0

, 90

Dedekind, 110

deﬁnability, 97

deﬁnitional extension, 100, 102

degree, 3, 24

dense linear ordering, 91

diﬀerence, 108

diﬀerential equations, 99

disjunctive normal form, 11

distance, 91

distributivity, 10, 92, 104

divisible group, 90

DLO, 91

dom, 108

domain, 26, 85, 108

dyadic, 20

element, 107

123

elementary equivalence, 27, 88

elimination of quantiﬁers, 95

empty set, 108

end node, 20

end points, 91

equality, 82

equivalence relation, 76

e

x

, 99

explicit deﬁnability, 97, 112

exponential function, 97

extensionality, 107

falsity, 27

Fcn, 108

Fibonacci numbers, 78

ﬁeld, 93

ﬁnite, 109

ﬁnitely branching, 20

ﬁrst-order arithmetic, 103

formation sequence, 4

formation tree, 4

formula, 3, 24, 79, 85

foundation, axiom of, 109

foundational theories, 88, 102–111

free variable, 25

function, 79, 85, 108

Fundamental Theorem of Algebra, 93

Gentzen-style proof system, 63

G¨odel number, 114

G¨odel hierarchy, 111

graph, 22, 70, 91

group, 83

Hilbert-style proof system, 50

Hintikka’s Lemma, 19, 41

identity, 71, 86

identity axioms, 71, 82, 86

identity predicate, 71, 86

immediate extension, 12, 35

immediate predecessor, 20

immediate successor, 20

implicit deﬁnability, 98, 112

induction, 103

inﬁnity, 109

integers, 93, 97

intended model, 88, 104–106, 110

interpolation, 66, 69

interpretability, 111

intersection, 108

invariance, 98

irreﬂexivity, 75

isomorphism, 26, 79

K¨onig’s Lemma, 20

L

1

, 103

L

2

, 104

language, 3, 24, 78

primitive recursive, 115

LG, 63, 81, 85

LG

+

, 64

LG

′

, 66

LG(atomic), 66

LG(symmetric), 67

LH, 56, 81, 85

LH(S), 60

LH

′

, 59

linear ordering, 91

logical consequence, 8, 10, 27, 35, 40,

60, 87

logical equivalence, 10, 37

logical validity, 8, 27, 35, 37, 64, 67

many-sorted, 84

mathematical theories, 88–96

membership, 45, 104–106

mixed type, 94, 104

modus ponens, 51

N (natural numbers), 97, 104

n, 114

n-ary function, 79, 85

n-ary operation, 78, 84

n-ary predicate, 24, 84

n-ary relation, 26, 85

natural numbers, 97, 104

normal satisﬁability, 72, 82

normal structure, 71, 82, 86

nth-order arithmetic, 106

numeral, 114

124

∅ (empty set), 108

one-sorted, 86

open, 15, 18, 40

operation, 78, 84

ordered ﬁeld, 95

ordered pair, 107

ordering

linear, 91

partial, 22

P (power set), 108

PA, 103

padding, 64

pair

ordered, 107

unordered, 107

pairing, 107

parameter, 31, 85

partial ordering, 22

partition, 76

path, 20

Peano arithmetic, 103

power set, 105–108

practical completeness, 103–105, 111

predecessor, 20

predicate, 24, 84

prenex form, 39

prime numbers, 77

primitive recursive language, 115

product, Cartesian, 108

proof system, 50

Gentzen-style, 63

Hilbert-style, 50

pure set, 107

Q, 103

Q (rational numbers), 93

quantiﬁer, 24

quantiﬁer-free, 39

quantiﬁer elimination, 95

quasitautological consequence, 57, 58

quasitautology, 51, 53, 56

quasiuniversal, 42

R (real numbers), 93, 95, 97–99

ran, 108

random graph, 92

range, 108

rational numbers, 93

RCOF, 95

real-closed ﬁeld, 95

real numbers, 93, 95, 98, 99

reduct, 74, 101

reﬂexivity, 71

relation, 26, 85

replacement, 110

replete, 18, 40

ring, 83, 92

root, 20

S (successor function), 104

satisfaction, 8, 27, 35

satisﬁability, 8, 27, 28, 35, 37

scalar, 94

second-order arithmetic, 104

sentence, 25, 26

sequent, 62

sequent calculus, 63

set theory, 106, 107, 110

signed formula, 12

signed sequent, 67

signed variant, 67

simple type theory, 106

singleton, 107

sin x, 99

sort, 84

soundness, 51

spectrum, 75, 82, 86

strength, 111

structure, 26, 35

sub, 116

subtheory, 111

subtree, 20

successor, 20

successor function, 104

symmetric, 67

symmetry, 71

tableau, 12, 15, 31, 80

Tarski, 95, 118

tautology, 8, 51

term, 78, 85

125

theories, 87

complete, 88–92, 94–96, 103

foundational, 88, 102–111

mathematical, 88–96

practically complete, 103–105, 111

torsion, 83, 90

transitivity, 71

tree, 20

truth, 27

truth values, 6

type theory, 106

(union), 107

unary, 3

union, 107

universal closure, 40

universe, 26, 85, 110

unordered pair, 107

unsigned formula, 12

unsigned sequent, 67

valuation, 6, 26, 79

value, 108

variable, 24, 85

bound, 25

free, 25

variable-free, 79

variant, 39, 67

Vaught’s Test, 89

vector space, 94

weakening, 64

well-founded set, 107

Z (integers), 93

Z

1

, 103

Z

2

, 104

Z

n

, 106

Z

∞

, 106

ZC, 107

ZFC, 110

Zermelo, 107

Zermelo/Fraenkel, 110

126

Contents

Contents 1 Propositional Calculus 1.1 Formulas . . . . . . . . . . . 1.2 Assignments and Satisﬁability 1.3 Logical Equivalence . . . . . . 1.4 The Tableau Method . . . . . 1.5 The Completeness Theorem . 1.6 Trees and K¨nig’s Lemma . . o 1.7 The Compactness Theorem . 1.8 Combinatorial Applications . 2 Predicate Calculus 2.1 Formulas and Sentences . . 2.2 Structures and Satisﬁability 2.3 The Tableau Method . . . . 2.4 Logical Equivalence . . . . . 2.5 The Completeness Theorem 2.6 The Compactness Theorem 2.7 Satisﬁability in a Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 3 3 6 10 12 18 20 21 22 24 24 26 31 37 40 46 47 50 50 51 56 61 66 71 71 75 78 82 84

3 Proof Systems for Predicate Calculus 3.1 Introduction to Proof Systems . . . . . 3.2 The Companion Theorem . . . . . . . 3.3 Hilbert-Style Proof Systems . . . . . . 3.4 Gentzen-Style Proof Systems . . . . . 3.5 The Interpolation Theorem . . . . . . 4 Extensions of Predicate Calculus 4.1 Predicate Calculus with Identity . . 4.2 The Spectrum Problem . . . . . . . 4.3 Predicate Calculus With Operations 4.4 Predicate Calculus with Identity and 4.5 Many-Sorted Predicate Calculus . .

. . . . . . . . . . . . . . . . . . . . . Operations . . . . . . .

1

5 Theories, Models, Deﬁnability 5.1 Theories and Models . . . . . . . . 5.2 Mathematical Theories . . . . . . . 5.3 Deﬁnability over a Model . . . . . 5.4 Deﬁnitional Extensions of Theories 5.5 Foundational Theories . . . . . . . 5.6 Axiomatic Set Theory . . . . . . . 5.7 Interpretability . . . . . . . . . . . 5.8 Beth’s Deﬁnability Theorem . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

87 87 89 97 100 103 106 111 112 114 114 114 114 117 118 119 121 122 123

6 Arithmetization of Predicate Calculus 6.1 Primitive Recursive Arithmetic . . . . 6.2 Interpretability of PRA in Z1 . . . . . 6.3 G¨del Numbers . . . . . . . . . . . . . o 6.4 Undeﬁnability of Truth . . . . . . . . . 6.5 The Provability Predicate . . . . . . . 6.6 The Incompleteness Theorems . . . . . 6.7 Proof of Lemma 6.5.3 . . . . . . . . . Bibliography Index

2

r. biimplication ( ⇔ ).1. and (A ⇔ B) are L-formulas. disjunction ( ∨ ).1 Formulas Deﬁnition 1. If p is an atomic L-formula. If A and B are L-formulas and b is a binary connective. ∨ . conjunction ( ∧ ).1. ⇔ are designated as binary. 3 . q.5. s. The set of L-formulas is generated inductively according to the following rules: 1. “and”. “or”. ⇒ . 2. They are read as “not”. “if-then”. Deﬁnition 1. while ¬ is designated as unary. Deﬁnition 1. then (A ∧ B).1. “if and only if” respectively.2.1. then p is an L-formula. r. An atomic L-formula is an atom of L. then (¬ A) is an L-formula. Assume that L contains propositional atoms p. A propositional language L is a set of propositional atoms p. If A is an L-formula. (A ⇒ B). . Deﬁnition 1. . then (A b B) is an L-formula.4. (A ∨ B).3. . If A and B are L-formulas. The connectives ∧ . This is the same as the number of times rules 2 and 3 had to be applied in order to generate A. implication ( ⇒ ). The propositional connectives are negation (¬ ). q.Chapter 1 Propositional Calculus 1..1. the degree of A is the number of occurrences of propositional connectives in A.1. If A is a formula. Example 1. Then (((p ⇒ q) ∧ (q ∨ r)) ⇒ (p ∨ r)) ⇒ ¬ (q ∨ s) is an L-formula. Note that rule 3 can be written as follows: 3′ . 3.

Remark 1.1. A formation sequence for the L-formula of Example 1. . an L-formula may be called a formula.10.1.4 is p. ((p ⇒ q) ∧ (q ∨ r)) ⇒ (p ∨ r). A formation tree is a ﬁnite rooted dyadic tree where each node carries a formula and each non-atomic formula branches to its immediate subformulas (see the example below). s.1. A2 . ⇒ ⇒ / \ ∧ ∨ / \ /\ ⇒ ∨ p r /\ /\ p q q r 4 ¬ | ∨ / \ q s .1.11. In contexts where the language L does not need to be speciﬁed. Let L be a propositional language.4 is 8.Example 1. A formation sequence is ﬁnite sequence A1 . . we omit parentheses when this can be done without ambiguity.1. q ∨ s. so instead of ((¬ A) ⇒ B) we may write (¬ A) ⇒ B. But we may not write ¬ A ⇒ B. The formation tree for the formula of Example 1.8.3.1. in an abbreviated style.6. An such that each term of the sequence is obtained from previous terms by application of one of the rules in Deﬁnition 1.7 (omitting parentheses). ¬ (q ∨ s). because this would not distinguish the intended formula from ¬ (A ⇒ B). The degree of the formula of Example 1. (((p ⇒ q) ∧ (q ∨ r)) ⇒ (p ∨ r)) ⇒ ¬ (q ∨ s) . Example 1. (p ⇒ q) ∧ (q ∨ r).1. .4 is (((p ⇒ q) ∧ (q ∨ r)) ⇒ (p ∨ r)) ⇒ ¬ (q ∨ s) / \ ((p ⇒ q) ∧ (q ∨ r)) ⇒ (p ∨ r) ¬ (q ∨ s) / \ | (p ⇒ q) ∧ (q ∨ r) p∨r q∨s / \ / \ / \ p⇒q q∨r p r q s / \ / \ p q q r or.1. If A is a formula. p ⇒ q. r. q ∨ r.12. p ∨ r. q. Remark 1.1. Example 1. outermost parentheses can always be omitted. Note that A is an L-formula if and only if there exists a formation sequence for A. In particular.1.9. the formation tree for A is the unique formation tree which carries A at its root. Deﬁnition 1. Deﬁnition 1. As in the above example.1. . A formation sequence for A is a formation sequence whose last term is A.

1.1. Remark 1. List the immediate subformulas of C.1.Remark 1. If p is an atomic L-formula.14. 3. If A is an L-formula. then b A B is an L-formula. and ¬ (p ⇒ q) becomes ¬ ⇒ p q. Write C according to various notation systems: (a) The rules 1–3 of Deﬁnition 1. If A is a formula then (¬ A) is a formula..1. Remark 1.15. ∧ ⇒ p q ∨ q r.4 becomes ⇒ ⇒ ∧ ⇒ p q ∨ q r ∨ pr ¬ ∨ q s and a formation sequence for this is p. then ¬ A is an L-formula. (¬ p) ⇒ q becomes ⇒ ¬ p q. ⇒ p q. If A and B are L-formulas and b is a binary connective. 3. In this case the set of L-formulas is generated as follows: 1.) 2. ∨ q r. i. then (AbB) is a formula. 5 . Each atom is a formula.7. 5. Another way to avoid parentheses is to use Polish notation. ⇒ ⇒ ∧ ⇒ p q ∨ q r ∨ p r ¬ ∨ q s . ¬ ∨ q s. s. but it has the advantages of being linear and of not using parentheses. etc. then there is no need for parentheses.1. 3.3: 1. Exercises 1. r. if we identify formulas with formation trees in the abbreviated style. (See Remark 1. In our study of formulas. The only point of interest for us is that each non-atomic formula is uniquely of the form ¬ A or A b B.. where A and B are formulas and b is a binary connective. 2. all subformulas of C. Calculate the degrees of C and its subformulas. ⇒ ∧ ⇒ p q ∨ q r ∨ p r. 4. then p is an L-formula.16. Let C be the formula (p ∧ ¬ q) ⇒ ¬ (p ∨ r).1. Obviously Polish notation is diﬃcult to read. Display the formation tree of C. we shall be indiﬀerent to the question of which system of notation is actually used. If A and B are formulas and b is a binary connective. Restore all the omitted parentheses to C.e. Exhibit a formation sequence for C. 6.13. For example. ∨ p r.1. their immediate subformulas. ∨ q s. The formula of Example 1. Note that. 2.1. q.

vM (A ∨ B) = 4. if vM (A) = vM (B) = T . F} given by the following clauses: 1. i. Note that each clause of Lemma 1. 2. vM (B) = F .2. Lemma 1.3. 3. 5. if vM (A) = T . if vM (A) = vM (B) = F . vM (A ⇔ B) = T F if vM (A) = vM (B) . Proof. Note that if L has exactly n atoms then there are exactly 2n diﬀerent Lassignments.3 corresponds to the familiar truth table for the corresponding propositional connective. if at least one of vM (A). Given an L-assignment M .. vM (B) = T . denoting truth and falsity. There are two truth values. if at least one of vM (A). Deﬁnition 1. F} . Let L be a propositional language.2. The truth value vM (A) is deﬁned by recursion on L-formulas. by induction on the degree of A where A is an arbitrary L-formula.2. T and F. if vM (A) = vM (B) . An L-assignment is a mapping M : {p | p is an atomic L-formula} → {T.2. 1. there is a unique L-valuation vM : {A | A is an L-formula} → {T. Remark 1. If A is a formula then ¬ (A) is a formula. If A and B are formulas and b is a binary connective. (c) Polish notation.4.e. (d) Reverse Polish notation.2.2 Assignments and Satisﬁability Deﬁnition 1. vM (¬ A) = T F T F T F if vM (A) = F .2. then (A)b(B) is a formula. 2.(b) The following alternative set of rules: 1.1. Thus clause 3 corresponds to the truth table 6 . vM (A ∧ B) = 3. vM (A ⇒ B) = vM (¬ (A ∧ ¬ B)) . Each atom is a formula.

this looks like (p ⇒ q) ⇒ (q ⇒ r) / \ p⇒q q⇒r / \ / \ p q q r T F F T and by applying clause 4 three times we get (p ⇒ q) ⇒ (q ⇒ r) T B T F T F p⇒q F / \ p q T F q⇒r T / \ q r F T and from this we see that vM ((p ⇒ q) ⇒ (q ⇒ r)) = T. Example 1. Remark 1. the end nodes of the formation tree for A. one begins with an assignment of truth values to the atoms. Remark 1.2. assigning truth values to the nodes.7.5.2..3 may be visualized in terms of formation trees. In terms of the formation tree.6. The above formation tree with truth values can be compressed and written linearly as (p ⇒ q) ⇒ (q ⇒ r) TFF T FTT.2.A T T F F B T F T F A∨B T T T F A⇒B T F T T for ∨ . To deﬁne vM (A) for a formula A. Lemma 1. M (r) = T.2. 7 . and then proceeds upward to the root. and clause 4 corresponds to the truth table A T T F F for ⇒ . i.e. Consider the formula (p ⇒ q) ⇒ (q ⇒ r) under an assignment M with M (p) = T. each step being given by the appropriate clause. M (q) = F.

.15.3. Deﬁnition 1.2. .2.2. Lemma 1. 2.2. An if and only if (A1 ∧ · · · ∧ An ) ⇒ B is logically valid.14.. A set of formulas S is said to be satisﬁable if there exists an assignment M which satisﬁes S. Jones: If Brown is guilty then so is Smith.12. . A formula B is said to be logically valid (or a tautology) if B is true under all assignments. 3.3 implies that there is an obvious one-to-one correspondence between L-assignments and L-valuations. An if and only if (A1 ∧ · · · ∧ An ) ⇒ B is logically valid. Remark 1. B is a logical consequence of the empty set.This illustrates a convenient method for calculating vM (A). A formula A is said to be true under M if vM (A) = T.11. Fix a propositional language L. 4. Prove the following. Equivalently. (See Remarks 1. Remark 1. . B is logically valid if and only if ¬ B is not satisﬁable. B is satisﬁable if and only if ¬ B is not logically valid. . and Smith are suspected of a crime.10. vM (A) = T for all A ∈ S. If the language L is understood from context. and false under M if vM (A) = F.2.2.13 and 1. Use truth tables to show that ((A ⇒ B) ⇒ A) ⇒ A is logically valid.e. 2. 1.) 1. Exercise 1. B is a logical consequence of A1 . Use truth tables to show that (A ∧ B) ⇒ C is logically equivalent to A ⇒ (B ⇒ C). 8 .16. Deﬁnition 1.13. They testify as follows: Brown: Jones is guilty and Smith is innocent.2. where M is an arbitrary L-assignment. A formula B is said to be a logical consequence of S if it is true under all assignments which satisfy S. Let M be an assignment.8. Deﬁnition 1.2.9. We now present some key deﬁnitions. Jones. A is logically equivalent to B if and only if A ⇔ B is logically valid. B is a logical consequence of A1 .2.2. Let S be a set of formulas.2. Exercises 1. Exercises 1. . Deﬁnition 1.2. . i. . Brown. we may speak simply of assignments and valuations. B is logically valid if and only if ¬ B is not satisﬁable.

(a) Yes. who is innocent and who is guilty? (e) Assuming that the innocent told the truth and the guilty told lies. Which from which? (c) Assuming everybody is innocent.” “Jones is innocent.” “Smith is innocent”. we are in line 3 of the table. while Jones is guilty. by line 3 of the table. Let b. vM (s) = vM (S). (d) If all the testimony is true.Smith: I’m innocent.e. (c) If everybody is innocent. vM (j) = vM (J). Hence we are in line 6 of the table. Brown and Smith lied. who committed perjury? (d) Assuming all testimony is true.. j. In other words. Hence B and S are false. Express the testimony of each suspect as a propositional formula. Write a truth table for the three testimonies. but at least one of the others is guilty. (b) The table shows that S is a logical consequence of B. The testimonies are: B : (¬ j) ∧ s J : (¬ b) ⇒ (¬ s) S : s ∧ ((¬ b) ∨ (¬ j)) The truth table is: b 1 2 3 4 5 6 7 8 T T T T F F F F j s B F F T F F F T F J T T T T F T F T S F F T F T F T F T T T F F T F F T T T F F T F F 2. and Brown and Smith are guilty. 2. Use the above truth table to answer the following questions: (a) Are the three testimonies consistent? (b) The testimony of one of the suspects follows from that of another. who is innocent and who is guilty? Solution. 1. Smith’s testimony follows from Brown’s. 9 . 1. and s be the statements “Brown is innocent. i. Thus Jones is innocent. we are in line 1 of the table. Thus Brown and Smith are innocent. (e) Our assumption is vM (b) = vM (B).

7. Exercise 1.3.5. 3. commutative laws: (a) A ∧ B ≡ B ∧ A (b) A ∨ B ≡ B ∨ A (c) A ⇔ B ≡ B ⇔ A Note however that A ⇒ B ≡ B ⇒ A. 2. plus induction on the degree of C. ¬ A1 ≡ ¬ A2 . 6. Show that 1. if each is a logical consequence of the other. B ⇒ A1 ≡ B ⇒ A2 .3.1.3 Logical Equivalence Deﬁnition 1. B ∨ A1 ≡ B ∨ A2 .3. Assume A1 ≡ A2 . Remark 1. (Hint: Use the results of the previous exercise. Exercise 1. A1 ∨ B ≡ A2 ∨ B. 2.) Remark 1. 8. distributive laws: 10 . A1 ⇒ B ≡ A2 ⇒ B.3.3.1. Show that for any formula C containing A1 as a part. associative laws: (a) A ∧ (B ∧ C) ≡ (A ∧ B) ∧ C (b) A ∨ (B ∨ C) ≡ (A ∨ B) ∨ C (c) A ⇔ (B ⇔ C) ≡ (A ⇔ B) ⇔ C Note however that A ⇒ (B ⇒ C) ≡ (A ⇒ B) ⇒ C. Two formulas A and B are said to be logically equivalent. written A ≡ B.4. 5. A1 ∧ B ≡ A2 ∧ B. A ≡ B holds if and only if A ⇔ B is logically valid. if we replace one or more occurrences of the part A1 by A2 . B ∧ A1 ≡ B ∧ A2 . 9. 4. Some useful logical equivalences are: 1. A1 ⇔ B ≡ A2 ⇔ B. Assume A1 ≡ A2 . B ⇔ A1 ≡ B ⇔ A2 . then the resulting formula is logically equivalent to C.2.3. 3.

. n is either an atom or the negation of an atom. Example 1. There are two ways to do Exercise 1. . much as one applies the commutative and distributive laws in algebra to reduce every algebraic expression to a polynomial. i = 1. j = 1.3. .9. .3. .4. negation laws: (a) ¬ (A ∧ B) ≡ (¬ A) ∨ (¬ B) (b) ¬ (A ∨ B) ≡ (¬ A) ∧ (¬ B) (c) ¬ ¬ A ≡ A (d) ¬ (A ⇒ B) ≡ A ∧ ¬ B (e) ¬ (A ⇔ B) ≡ (¬ A) ⇔ B (f) ¬ (A ⇔ B) ≡ A ⇔ (¬ B) 5. 1.6. is of the form B1 ∧ · · · ∧ Bn . One way is to apply the equivalences of Remark 1.3.3.3. Writing p as an abbreviation for ¬ p. implication laws: (a) A ⇒ B ≡ ¬ (A ∧ ¬ B) (b) A ⇒ B ≡ (¬ A) ∨ B (c) A ⇒ B ≡ (¬ B) ⇒ (¬ A) (d) A ⇔ B ≡ (A ⇒ B) ∧ (B ⇒ A) (e) A ⇔ B ≡ (¬ A) ⇔ (¬ B) Deﬁnition 1. A formula is said to be in disjunctive normal form if it is of the form A1 ∨ · · · ∨ Am .3. Show that every propositional formula C is logically equivalent to a formula in disjunctive normal form. 4. .(a) A ∧ (B ∨ C) ≡ (A ∧ B) ∨ (A ∧ C) (b) A ∨ (B ∧ C) ≡ (A ∨ B) ∧ (A ∨ C) (c) A ⇒ (B ∧ C) ≡ (A ⇒ B) ∧ (A ⇒ C) (d) (A ∨ B) ⇒ C ≡ (A ⇒ C) ∧ (B ⇒ C) Note however that A ⇒ (B ∨ C) ≡ (A ⇒ B) ∨ (A ⇒ C). . m.7. . 11 .8.8. Exercise 1. and (A ∧ B) ⇒ C ≡ (A ⇒ C) ∧ (B ⇒ C). Remark 1. the formula (p1 ∧ p2 ∧ p3 ) ∨ (p1 ∧ p2 ∧ p3 ) ∨ (p1 ∧ p2 ∧ p3 ) is in disjunctive normal form.3. and each Bj . where each clause Ai .5 to subformulas of C via Exercise 1.

7 we read oﬀ the following formula equivalent to (p ⇒ q) ⇒ r in disjunctive normal form: (p ∧ q ∧ r) ∨ (p ∧ q ∧ r) ∨ (p ∧ q ∧ r) ∨ (p ∧ q ∧ r) ∨ (p ∧ q ∧ r) . We wish to put this in disjunctive normal form. The signed tableau rules are presented in Table 1. From lines 1.2. Consider the formula (p ⇒ q) ⇒ r.3. A signed tableau is a rooted dyadic tree where each node carries a signed formula.2.4 The Tableau Method Remark 1.3.5. Example 1. An unsigned tableau is a rooted dyadic tree where each node carries an unsigned formula. Deﬁnition 1. Method 1. where A is a formula. An unsigned formula is simply a formula. Method 2. Consider the truth table p 1 2 3 4 5 6 7 8 T T T T F F F F q r p⇒q T T F F T T T T (p ⇒ q) ⇒ r T F T T T F T F r ∨ ¬ (p ⇒ q) r ∨ ¬ ¬ (p ∧ ¬ q) r ∨ (p ∧ ¬ q) T T T F F T F F T T T F F T F F Each line of this table corresponds to a diﬀerent assignment.4. A signed formula is an expression of the form TA or FA. 1.4.2. an immediate extension of τ is a larger tableau τ ′ obtained by applying a tableau rule to a ﬁnite path of τ . We follow the approach of Smullyan [4]. The clause speciﬁes the assignment. The other way is to use a truth table for C. 5. 3.4.10. A more descriptive name for tableaux is satisﬁability trees. we obtain (p ⇒ q) ⇒ r ≡ ≡ ≡ and this is in disjunctive normal form. If τ is a (signed or unsigned) tableau. The unsigned tableau rules are presented in Table 1. Applying the equivalences of Remark 1. 12 .1.1. The disjunctive normal form of C has a clause for each assignment making C true.3. 4. Deﬁnition 1.

/ \ FA TB . TA∧B . . . .1: Signed tableau rules for propositional connectives. . . FA∧B . TA∨B . 13 . . . / FA \ FB . . . . . . TA⇔B . | TA FB . | FA FB . | FA . . . . . . TA⇒B . . . . . | TA Table 1. / \ TA FA TB FB .. . . . . . / TA FB \ FA TB . . . . F¬A . | TA TB . . . . . . / \ TA TB . . FA⇔B . T¬A . . FA⇒B . . . . FA∨B . .

. . ¬ (A ⇒ B) . . . . ¬ (A ∨ B) . . | A B . . . . | A ¬B . . . . ¬¬A . . . . . | A . | ¬A ¬B . . / \ ¬A B . . . . . . . 14 . . / A B \ ¬A ¬B Table 1. . . ¬ (A ∧ B) . . . A⇔B . . / \ A ¬A ¬B B . . A∧B .. / ¬A \ ¬B . . . . ¬ (A ⇔ B) .2: Unsigned tableau rules for propositional connectives. . . . / A \ B . A∨B . A⇒B .

7. . or equivalently an unsigned tableau starting with A1 . 1. Let X1 . Ak . . . . If the tableau closes oﬀ. .e. . . 2. . A tableau is said to be closed if each of its paths is closed. 1. .4. . . To test whether B is a logical consequence of A1 . . TAk . Exercises 1. . then A is logically valid. a pair such as TA.4. If the tableau closes oﬀ. .6. See Corollaries 1. we have F (p ∧ q) ⇒ (q ∧ p) Tp∧q Fq∧p Tp Tq / \ Fq Fp Since (every path of) the tableau is closed. . . . Example 1. A tableau starting with X1 . . . If the tableau does not close oﬀ.5. . Xk by repeatedly applying tableau rules. Xk is a tableau obtained from X1 . Ak for satisﬁability. . . . then A1 . TAk . ¬ B. i. Ak .. or A. . . To test A1 . A path of a tableau is said to be open if it is not closed. . Using the signed tableau method to test (p ∧ q) ⇒ (q ∧ p) for logical validity. . . Xk be a ﬁnite set of signed or unsigned formulas. A path of a tableau is said to be closed if it contains a conjugate pair of signed or unsigned formulas. . and from any open path we can read oﬀ an assignment satisfying A1 .5. Ak is not satisﬁable. form a signed tableau starting with FA.4. .9.4. . 3. . .11 below. Deﬁnition 1. The correctness of these tests will be proved in Section 1. Ak is satisﬁable. . . ¬ A in the unsigned case. . .10. then A1 . Ak . FA in the signed case. then B is indeed a logical consequence of A1 . or equivalently an unsigned tableau starting with ¬ A. Ak .4. . .5. . . Ak . form a signed tableau starting with TA1 . . . . . .5. . . To test a formula A for validity. or equivalently an unsigned tableau starting with A1 .Deﬁnition 1. .5. form a signed tableau starting with TA1 . . 15 . If the tableau closes oﬀ. (p ∧ q) ⇒ (q ∧ p) is logically valid. FB. The tableau method: 1.

Use an unsigned tableau to show that A ⇒ (B ⇒ C) is logically equivalent to (A ∧ B) ⇒ C.1. Solution. Use a signed tableau to show that (A ⇒ B) ⇒ (A ⇒ C) is a logical consequence of A ⇒ (B ⇒ C). T A ⇒ (B ⇒ C) F (A ⇒ B) ⇒ (A ⇒ C) TA⇒B FA⇒C TA FC / \ FA TB ⇒C / \ FB TC / \ FA TB 2. F (A ⇒ B) ⇔ ((¬ B) ⇒ (¬ A)) / \ TA⇒B FA⇒B F (¬ B) ⇒ (¬ A) T (¬ B) ⇒ (¬ A) T¬B TA F¬A FB FB / \ TA F¬B T¬A / \ TB FA FA TB 3. Use a signed tableau to show that A ⇒ B is logically equivalent to (¬ B) ⇒ (¬ A). Solution. ¬ ((A ⇒ (B ⇒ C)) ⇔ ((A ∧ B) ⇒ C)) / \ ¬ (A ⇒ (B ⇒ C)) A ⇒ (B ⇒ C) (A ∧ B) ⇒ C ¬ ((A ∧ B) ⇒ C) A A∧B ¬ (B ⇒ C) ¬C B A ¬C B / \ / \ ¬ (A ∧ B) C ¬A B ⇒C / \ / \ ¬A ¬B ¬B C 16 . Solution.

If it is poor driving. M3 (q) = T. Therefore. M3 (p) = F. it will be poor driving. F (p ∨ q) ⇒ (p ∧ q) Tp∨q Fp∧q / \ Tp Tq / \ / \ Fp Fq Fp Fq 1 2 3 4 The open paths 2 and 3 provide the assignments M2 and M3 which falsify our formula. Use an unsigned tableau to test (p ∨ q) ⇒ (p ∧ q) for logical validity. I must start early to avoid being late. Indeed. M2 (q) = F. Use the following atoms. Redo the previous problem using a signed tableau. I will be late unless I start early. 5. s: it has snowed p: it is poor driving l: I will be late e: I start early 17 . If this formula is not logically valid. ¬ ((p ∨ q) ⇒ (p ∧ q)) p∨q ¬ (p ∧ q) / \ p q / \ / \ ¬p ¬q ¬p ¬q 1 2 3 4 The open paths 2 and 3 provide the assignments M2 and M3 which falsify our formula. Solution. Exercise 1. M3 (q) = T. 1. M2 (p) = T. use the tableau to ﬁnd all assignments which falsify it.4. M3 (p) = F. M2 (q) = F. Solution. it has snowed.8. Formulate the following argument as a propositional formula. Solution. M2 (p) = T. If it has snowed.4.

5 The Completeness Theorem Let X1 . Solution.3. . Xk . .2. . whenever it contains the top formula of a tableau rule. . 2.e.5. Xk is satisﬁable. A path of a tableau is said to be replete if. . Lemma 1. F ((s ⇒ p) ∧ (p ⇒ (l ∨ e)) ∧ s) ⇒ ((¬ l) ⇒ e) T (s ⇒ p) ∧ (p ⇒ (l ∨ e)) ∧ s F (¬ l) ⇒ e Ts⇒p T (p ⇒ (l ∨ e)) ∧ s T p ⇒ (l ∨ e) Ts T¬l Fe Fl / \ Fs Tp / \ Fp Tl∨e / \ Tl Te 1. this becomes: ((s ⇒ p) ∧ (p ⇒ (l ∨ e)) ∧ s) ⇒ ((¬ l) ⇒ e). A tableau is said to be open if it is not closed. . . .. it also contains at least one of the branches. Proof. . 18 . . Deﬁnition 1.The argument can be translated as follows: s ⇒ p. Lemma 1. Written as a single propositional formula. s.5. Any ﬁnite tableau can be extended to a ﬁnite replete tableau. A replete tableau is a tableau in which every path is replete.5. Lemma 1. If τ is open.5. . If τ is a ﬁnite closed tableau starting with X1 . or a ﬁnite set of unsigned formulas. Let τ be a replete tableau starting with X1 . . Straightforward. Deﬁnition 1. . Xk be a ﬁnite set of signed formulas. . . then X1 . then X1 . Proof. Xk is not satisﬁable. . . i. it has at least one open path. . p ⇒ (l ∨ e). .5.4. . Use the tableau method to demonstrate that this formula is logically valid. Xk .1 (the Soundness Theorem).5. Apply tableau rules until they cannot be applied any more. therefore (¬ l) ⇒ e.

. If S is a Hintikka set. Translate the argument into propositional calculus. . . . . Let S be a set of signed or unsigned formulas. Use an unsigned tableau to determine whether the argument is logically valid. It is not diﬃcult to see that vM (X) = T for all X ∈ S.9. . Xk is open.6. . TAk . .5. . Corollary 1. the attack will not succeed. . The attack will succeed only if the enemy is taken by surprise or the position is weakly defended. .5. or equivalently a ﬁnite closed unsigned tableau starting with ¬ A. .1 and 1. we introduce the following deﬁnition.5.12.5.7 (Hintikka’s Lemma). . we obtain: Theorem 1. . Corollary 1.5. The enemy will not be taken by surprise unless he is overconﬁdent.5. Ak . Tp. or p. . We say that S is a Hintikka set if 1. . .. Thus.5. . . . Therefore. S “obeys the tableau rules”. . 1. Xk is satisﬁable if and only if there is no ﬁnite closed tableau starting with X1 . Deﬁnition 1. or equivalently a ﬁnite closed unsigned tableau starting with A1 . Deﬁne an assignment M by M (p) = in the signed case.5.8 (the Completeness Theorem). Fp in the signed case. then S is satisﬁable. Hintikka’s Lemma implies that X1 .5. S contains no pair of conjugate atomic formulas. A1 . Ak is not satisﬁable if and only if there exists a ﬁnite closed signed tableau starting with TA1 . . . .5. . B is a logical consequence of A1 . . in the sense that if it contains the top formula of a rule then it contains at least one of the branches.3 and 1. To prove Lemma 1. . Corollary 1. it suﬃces to note that a replete open path is a Hintikka set. FB. .10. ¬ p in the unsigned case. . X1 . A is logically valid if and only if there exists a ﬁnite closed signed tableau starting with FA. Proof.5.11. ¬ B.5. Xk is satisﬁable. or equivalently a ﬁnite closed unsigned tableau starting with A1 . . .e. or M (p) = T F if p belongs to S otherwise T F if Tp belongs to S otherwise in the unsigned case.5. . Combining Lemmas 1. .5. Lemma 1. 19 . TAk . if a replete tableau starting with X1 . Ak . . .5. . The enemy will not be overconﬁdent if the position is weakly defended. 2. . . Exercise 1. . . Ak if and only if there exists a ﬁnite closed signed tableau starting with TA1 . Consider the following argument.In order to prove Lemma 1. i. 2. Xk .

6. Let T be an inﬁnite. Let T be the set of all x ∈ T such that x has inﬁnitely many successors in T . . the root of T ′ is the same as the root of T . Now deﬁne an inﬁnite path S = {x1 . Since T is ﬁnitely branching. x ∈ T ′ .5. . Deﬁnition 1. A path in T is a set S ⊆ T such that (1) the root of T belongs to S. o Theorem 1. y ∈ T such that xP y. y ∈ T . Let P ∗ be the transitive closure of P . or y immediately succeeds x. we have xP ∗ y if and only if x precedes y. Deﬁnition 1. An end node of T is a node with no (immediate) successors. called the root of the tree. ℓ(x) is the level of x.6. We require that each node other than the root has exactly one immediate predecessor. T is dyadic if each node of T has at most two immediate successors in T . a binary relation P on T . Thus P ∗ is a partial ordering of T . it follows by the pigeonhole principle that each x ∈ T has at least one immediate successor y ∈ T . the smallest reﬂexive and transitive relation on T containing P . our discussion of trees has been informal. i. We require that there is exactly one node x ∈ T such that ℓ(x) = 1. . antisymmetric (xP ∗ y and yP ∗ x imply x = y). and transitive (xP ∗ y and yP ∗ z imply xP ∗ z). Note that T is a subtree of T . under the restriction of ℓ and P to T ′ .6.e. Note that the relation P ∗ is reﬂexive (xP ∗ x for all x ∈ T ).6. A subtree of T is a nonempty set T ′ ⊆ T such that for all y ∈ T ′ and xP y.4.2. 20 . For x ∈ T . y succeeds x.6. Clearly S is an inﬁnite path of T . . T is ﬁnitely branching if each node of T has only ﬁnitely many immediate successors in T . Moreover. Deﬁnition 1. A tree consists of 1. We require that ℓ(y) = ℓ(x) + 1 for all x.6 (K¨nig’s Lemma). Note that T ′ is itself a tree. a function ℓ : T → N+ . Deﬁnition 1. Then T has an inﬁnite path. ﬁnitely branching tree.6 Trees and K¨nig’s Lemma o Up to this point. there exists a ﬁnite sequence x = x0 P x1 P x2 · · · xn−1 P xn = y. . and xn+1 = one of the immediate successors of xn in T . Deﬁnition 1.} in T inductively by putting x1 = the root of T .e. xn ..1. i. either x is an end node of T or there is exactly one y ∈ S such that xP y. . Note that a dyadic tree is ﬁnitely branching.3. .6. x2 . The elements of T are called the nodes of the tree. a set T 2. 3.. For x.. We now pause to make our tree terminology precise. i. (2) for each x ∈ S. Proof.e.1. The relation xP y is read as x immediately precedes y.

7. S ∗ . By K¨nig’s Lemma (Theorem 1. Proof. S ∗ is ﬁnitely satisﬁable.2 (the Compactness Theorem. Since {A1 . Let M = {T. by Hintikka’s Lemma. We are trying to prove that. Third proof. For each L-formula A. and generate a ﬁnite replete tableau. S ∗ ∪ {A} is not ﬁnitely satisﬁable. . Sξ+1 = Sξ otherwise. If S is a set of L-formulas. by Hintikka’s Lemma. Sξ+1 = Sξ ∪ {Aξ } if Sξ ∪ {Aξ } is ﬁnitely satisﬁable. . F has a maximal element. .7 The Compactness Theorem Theorem 1. τ1 . and for each L-formula A ∈ S ∗ . we say that S is ﬁnitely satisﬁable if each ﬁnite subset of S is satisﬁable. Let S be an uncountable set of propositional formulas. be a transﬁnite enumeration of all L-formulas. It is easy to see that any chain in F has a least upper bound in F. Since A1 is satisﬁable. In brief outline: Form an inﬁnite tableau. Hence S is satisﬁable. Let S be a countable set of propositional formulas. Sα is ﬁnitely satisﬁable. In particular. The second uses transﬁnite induction. By Hintikka’s Lemma. Using transﬁnite induction. it follows by Tychonoﬀ’s Theorem that M is compact. and generate a ﬁnite replete tableau. Then S ′ is a Hintikka set containing S. τ3 . . ordered by inclusion. Note also that τ is an inﬁnite. .1. F} has the discrete topology. Second proof.7. Proof. Append A3 to each of the open paths of τ2 . then S is satisﬁable. Theorem 1. countable case). . Thus τ is a replete tableau. . Let L be the (necessarily uncountable) propositional language consisting of all atoms occurring in formulas of S. First proof. ξ < α. . Put ∞ τ = i=1 τi . Let Aξ . The ﬁrst uses Zorn’s Lemma.6. Hence. F}. let S ′ be an inﬁnite path o in τ . S ∗ is satisﬁable. put S0 = S. τ1 has at least one open path. Apply Hintikka’s Lemma. A2 } is satisﬁable. S ∗ ⊇ S. . It is straightforward to verify that Sα is a Hintikka set. and Sη = ξ<η Sξ for limit ordinals η ≤ α. put MA = {M ∈ M | vM (A) = T}. Details: Let S = {A1 . Make M a topological space with the product topology where {T. Since {T. Hence S is satisﬁable. The third uses Tychonoﬀ’s Theorem. if S is ﬁnitely satisﬁable. by Zorn’s Lemma. If each ﬁnite subset of S is satisﬁable. By transﬁnite recursion. / From this it is straightforward to verify that S ∗ is a Hintikka set. then S is satisﬁable.}. Sα is satisﬁable. F} is compact. Hence. It is easy to check that each MA is a topologically closed set 21 . S ′ is satisﬁable. Consider the partial ordering F of all ﬁnitely satisﬁable sets of L-formulas which include S. F}L be the space of all L-assignments M : L → {T. Apply K¨nig’s Lemma to get o an inﬁnite path. Append A2 to each of the open paths of τ1 . then S is satisﬁable.1 (the Compactness Theorem.6). Start with A1 and generate a ﬁnite replete tableau. it is easy to verify that Sξ is ﬁnitely satisﬁable for each ξ ≤ α. τ2 . . uncountable case). . τ2 has at least one open path. Hence. Thus S ∗ is a set of L-formulas. If each ﬁnite subset of S is satisﬁable. We present three proofs. Hence S is satisﬁable. ﬁnitely branching tree. A2 . Ai .

.e. 22 . . Show that G is k-colorable if and only if each ﬁnite subgraph of G is k-colorable. .8. . .4. Show that there is a one-to-one correspondence between k-colorings of G and assignments satisfying Ck (G). . 3. 1. G is said to be k-colorable if there exists a k-coloring of G. .in M. i. . 2. ¬ (pui ∧ pvi ) for each adjacent pair of vertices u. and that u is adjacent to v if and only if v is adjacent to u. then the family of sets MA . 1. 2. 1. . A graph consists of a set of vertices together with a speciﬁcation of certain pairs of vertices as being adjacent. Exercise 1. For each vertex v and each i = 1. Let G be a graph and let k be a positive integer.8. a ≤P a for all a ∈ P (reﬂexivity). . m ≤P n if and only if m ≤ n. A ∈ S has the ﬁnite intersection property. ck } such that f (u) = f (v) for all adjacent pairs of vertices u. A k-coloring of G is a function f : {vertices of G} → {c1 .8 Combinatorial Applications In this section we present some combinatorial applications of the Compactness Theorem for propositional calculus. a ≤P b. k. 2. Deﬁnition 1. Deﬁnition 1. Let ≤P be the usual order relation on P . Deﬁne Ck (G) to be the following set of propositional formulas: pv1 ∨ · · · ∨ pvk for each vertex v. Example 1. . v. n. . If S is ﬁnitely satisﬁable. i. .8.2. A partial ordering consists of a set P together with a binary relation ≤P such that 1. 3. Let G be a graph and let k be a positive integer. b ≤P c imply a ≤P c (transitivity). This notion is much studied in graph theory. 1. ¬ (pvi ∧ pvj ) for each vertex v and 1 ≤ i < j ≤ k. We require that a vertex may not be adjacent to itself. . Let P = N+ = {1. v and 1 ≤ i ≤ k.1. let pvi be a propositional atom expressing that vertex v receives color ci . By compactness of M it follows that A∈S MA = ∅. 3. Show that G is k-colorable if and only if Ck (G) is satisﬁable.e..8. 2. b ≤P a imply a = b (antisymmetry). Thus S is satisﬁable. .. A∈S0 MA = ∅ for each ﬁnite S0 ⊆ S. a ≤P b. 3.} = the set of positive integers.3.

2. Dilworth’s Theorem says that P is the union of k chains if and only if every antichain is of size ≤ k. Show that Dilworth’s Theorem for arbitrary partial orderings follows from Dilworth’s Theorem for ﬁnite partial orderings. 23 . Deﬁnition 1. 2. b ∈ P are comparable if either a ≤P b or b ≤P a.8. m ≤P n if and only if m is a divisor of n. An antichain is a set X ⊆ P such that any two distinct elements of X are incomparable. Exercise 1. Use the Compactness Theorem to show that P is the union of k chains if and only if each ﬁnite subset of P is the union of k chains. ≤P be a partial ordering. 1. ≤P be a partial ordering. Otherwise they are incomparable. i.e. 3. 1.8.. Let ≤P be the divisibility ordering of P . Let P. and let k be a positive integer.6. 2.5. Let P. Two elements a. A chain is a set X ⊆ P such that any two elements of X are comparable.

1 Formulas and Sentences Deﬁnition 2. They are read as “for all” and “there exists”.1. ∧ . A ⇔ B are LU -formulas. Each atomic L-U -formula is an L-U -formula.3 (formulas). respectively. . An atomic L-U -formula is an expression of the form P e1 · · · en where P is an n-ary predicate of L and each of e1 .1 (languages). 4. If A and B are L-U -formulas. . 3. We introduce two new symbols: the universal quantiﬁer (∀) and the existential quantiﬁer (∃).1. A language L is a set of predicates. . 1. ∃ in it. ⇒ . and let U be a set.Chapter 2 Predicate Calculus 2. If A is an L-U -formula. If x is a variable and A is an L-U -formula. Thus the predicate calculus is an extension of the propositional calculus. 5. Let L be a language. The set of L-U -formulas is generated as follows. A ⇒ B. countably inﬁnite set of symbols x. .1.4 (degree). known as variables. y. 1 It will be seen that 0-ary predicates behave as propositional atoms. . z. then ∀x A and ∃x A are L-U formulas. en is either a variable or an element of U . each predicate P of L being designated as n-ary for some nonnegative1 integer n. The degree of a formula is the number of occurrences of propositional connectives ¬ . We assume the existence of a ﬁxed. ⇔ and quantiﬁers ∀. Deﬁnition 2. 24 . . then ¬ A is an L-U -formula. Deﬁnition 2. then A ∧ B. Deﬁnition 2. .1. 2. It is understood that U is disjoint from the set of variables. A ∨ B. ∨ .2 (variables and quantiﬁers).

the empty set. we deﬁne an L-U -formula A[x/a] as follows. An occurrence of a variable x in an L-U -formula A is said to be free in A if it is not bound in A. and an L-formula a formula. In situations where the language L is understood from context. An L-formula is an L-∅-formula.e. for any set U . an L-U -sentence where U = ∅. A variable x is said to occur freely in A if there is at least one occurrence of x in A which is free in A. (∃x A)[x/a] = ∃x A. Exercise 2.7. Deﬁnition 2. every L-formula is automatically an L-U formula. (¬ A)[x/a] = ¬ A[x/a]. If y is a variable other than x. 2. (A ⇒ B)[x/a] = A[x/a] ⇒ B[x/a].Deﬁnition 2.e..1. then A[x/a] = the result of replacing each occurrence of x in A by a. an L-U -formula may be called a U -formula. If U is a subset of U ′ . Show that A[x/a] is the result of substituting a for all free occurrences of x in A. 8. If y is a variable other than x. An L-sentence is an L-∅-sentence.10. Deﬁnition 2. An L-U -sentence is an L-U -formula in which no variables occur freely.11 (sentences). 9.8 (substitution). (∀x A)[x/a] = ∀x A. Deﬁnition 2. If A is atomic.6. then (∃y A)[x/a] = ∃y A[x/a].5.1.1. 7. (A ∨ B)[x/a] = A[x/a] ∨ B[x/a].1. 1. Remark 2. i. 1. In particular. the empty set. (A ⇔ B)[x/a] = A[x/a] ⇔ B[x/a].1. then (∀y A)[x/a] = ∀y A[x/a]. i. 4.9 (free variables). 2. Show that x occurs freely in A if and only if A[x/a] = A.. An occurrence of a variable x in an L-U formula A is said to be bound in A if it is within the scope of a quantiﬁer ∀x or ∃x in A. 25 . (A ∧ B)[x/a] = A[x/a] ∧ B[x/a].1. 3. 10. 6.1. an L-U -formula where U = ∅. then every L-U -formula is automatically an L-U ′ -formula. Deﬁnition 2. 5. If A is an L-U -formula and x is a variable and a ∈ U .

Remark 2.2.e. an ∈ PM . .12. and an L-sentence a sentence. Thus a 0-ary predicate behaves as a propositional atom. . .. for any set U .2. Deﬁnition 2. 2. we are mainly interested in properties of structures that are invariant under isomorphism. and let n be a nonnegative2 integer.1.4.e. { } and {}. . Let L be a language. φ(an ) ∈ PM ′ . an ∈ PM if and only if φ(a1 ). Two L-structures M and M ′ are said to be isomorphic if there exists an isomorphism of M onto M ′ . Deﬁnition 2. .13.1. . a one-to-one correspondence φ : UM ∼ UM ′ such that for all n-ary predicates P of L and all n-tuples = a1 . i. Let U be a nonempty set. 3. . vM (¬ A) = if vM (A) = F .e. U n is the set of all n-tuples of elements of U . Given an L-structure M . a subset of { }. if a1 . .. every L-sentence is automatically an L-U -sentence. .2. 26 . F} deﬁned as follows: 1. vM (P a1 · · · an ) = T F T F T F if a1 . if at least one of vM (A). . . Lemma 2. an | a 1 . there is a unique valuation or assignment of truth values vM : {A | A is an L-UM -sentence} → {T. . in situations where the language L is understood from context.. an L-U -sentence may be called a U -sentence. . .2. . . a1 . . . an ∈ PM . An L-structure may be called a structure. In particular. U n = { a 1 . . if vM (A) = T . together with an n-ary relation PM on UM for each n-ary predicate P of L. . / 2. . vM (A ∧ B) = 2 In the special case n = 0 we obtain the notion of a 0-ary relation. . . If U is a subset of U ′ . i. corresponding to T and F respectively.2. There are only two 0-ary relations.3. . An n-ary relation on U is a subset of U n . . if vM (A) = vM (B) = T . .2 Structures and Satisﬁability Deﬁnition 2. then every L-U -sentence is automatically an L-U ′ -sentence. an ∈ (UM )n . As usual in abstract mathematics. called the domain or universe of M . vM (B) = F .1. . i. An L-structure M consists of a nonempty set UM . . Deﬁnition 2. an ∈ U } . In situations where the language L is understood from context.

have extended the substitution notation 2. We say that A is true in M if vM (A) = T. if all of the sentences of S are true in M . if vM (A[x/a]) = F for all a ∈ UM . We say that A is false in M if vM (A) = F. The truth value vM (A) is deﬁned by recursion on L-UM -sentences.1. using L-structures. .8 in an obvious way. Deﬁnition 2. vM (∀x A) = 8. An L-sentence is said to be logically valid if it is satisﬁed by all L-structures. Let A be an L-UM -sentence. ak /φ(ak )]. We omit the proof of part 1. 2. . 2. 4 Similarly. . abbreviated M |= S. Deﬁnition 2. .3. vM (B) = T . S is said to be satisﬁable 4 if there exists an L-structure M which satisﬁes S.3 Here a1 .4. if vM (A[x/a]) = T for at least one a ∈ UM . A more general result will be proved later as Theorem 2. then for all L-UM -sentences A we have vM (A) = vM ′ (A′ ) where A′ = A[a1 /φ(a1 ). We say that M satisﬁes S. they satisfy the same L-sentences. if vM (A) = vM (B) = F . vM (A ⇔ B) = T F T F T F if vM (A) = vM (B) . vM (∃x A) = Proof. if vM (A) = vM (B) . If M and M ′ are isomorphic L-structures. the notions of logical validity and logical consequence are deﬁned for Lsentences.. An L-sentence is said to be a logical consequence of S if it is satisﬁed by all L-structures satisfying S.2. 3 We 27 . then they are elementarily equivalent. vM (A ∨ B) = T F if at least one of vM (A). 5. Proof. in the obvious way. Part 2 follows immediately from part 1.7 (satisﬁability). .2. If M and M ′ are isomorphic L-structures and φ : M ∼ M ′ is an isomor= phism of M onto M ′ . i.2. Let S be a set of L-UM -sentences. 1. 1.7. ak are the elements of UM which occur in A.e. Theorem 2.. by induction on the degree of A where A is an arbitrary L-UM -sentence.e. . We shall see later that the converse does not hold in general. i. Let S be a set of L-sentences. Let M be an L-structure. vM (A ⇒ B) = vM (¬ (A ∧ ¬ B)) . if vM (A[x/a]) = T for all a ∈ UM . .6. 7. .5 (truth and satisfaction). if vM (A[x/a]) = F for at least one a ∈ UM . 6.

PM ′ = { φ(a1 ). R. A set S of L-U -sentences is said to be satisﬁable in the domain U if there exists an L-structure M such that M |= S and UM = U . Thus S is satisﬁable in the domain U ′ .12. then S is satisﬁable in any domain of the same cardinality as U .2.10. Remark 2. Satisﬁability is one of the most important concepts of mathematical logic.7) if and only if S is satisﬁable in some domain U . Consider the following sentences: (a) ∀x Rxx (b) ∀x ¬ Rxx (c) ∀x ∀y (Rxy ⇒ Ryx) (d) ∀x ∀y (Rxy ⇒ ¬ Ryx) (e) ∀x ∀y ∀z ((Rxy ∧ Ryz) ⇒ Rxz) (f) ∀x ∃y Rxy Which of subsets of this set of sentences are satisﬁable? Verify your claims by exhibiting appropriate structures. . . Proof. Let L be the language consisting of one binary predicate. If S is satisﬁable in a domain U . A key result known as the Compactness Theorem5 states that a set S of L-sentences is satisﬁable if and only every ﬁnite subset of S is satisﬁable. .13.2.6. Hence. The following related notion is of technical importance. an ∈ PM } for all n-ary predicates P of L. Our sentence A∞ is (1) ∧ (2) ∧ (3) with (1) ∀x ∀y ∀z ((Rxy ∧ Ryz) ⇒ Rxz) (2) ∀x ∀y (Rxy ⇒ ¬ Ryx) (3) ∀x ∃y Rxy See also Example 2.Remark 2. Let M be an L-structure M satisfying S with UM = U . Then S is satisﬁable (according to Deﬁnition 2.5.2. M ′ |= S. . .2 below. Suppose S is satisﬁable in a domain U . Theorem 2. 5 See Theorems 2.2. . 28 .8.9 (satisﬁability in a domain).6. by Theorem 2. Let S be a set of L-sentences. Let U be a nonempty set.1 and 2. Let U ′ be any set of the same cardinality as U . . Then M is isomorphic to M ′ . Use the simplest possible structures. Then there exists a one-to-one correspondence φ : U → U ′ . Deﬁnition 2. We exhibit a sentence A∞ which is satisﬁable in an inﬁnite domain but not in any ﬁnite domain.6. Let S be a set of L-sentences.9.2.2.11. Example 2. . Exercise 2. Let M ′ be the L-structure with UM ′ = U ′ .2.2. φ(an ) | a1 .

(c) Some people don’t drive either. (b. (b.c. (b) Every truck has exactly two drivers. .f) is not satisﬁable in any ﬁnite domain. 2}.c.e) is satisﬁable: U = {1}.}.Solution.f) is satisﬁable: U = {1. 1 }. 2 .c. 29 . 3. Exercise 2. because: (a.d. . only cars and trucks are driven. (a. R = <.f) is satisﬁable: U = {1. Note: (d. etc. R = { }.d. (b.2. (c. cars exist. 2. humans exist.f) is satisﬁable: U = {1}. (a. (b.e. (b) Some people drive both. no car is a truck. .e. 2. Write formulas representing the following statements: (a) Everybody drives a car or a truck. These are the only maximal satisﬁable sets.d. 1. only humans drive. R = { 1.d) is not satisﬁable.e. 2.f) is not satisﬁable.14. R = { 1.f) is not satisﬁable. 1 }.b) is not satisﬁable.c. Assume the following predicates: Hx: x is a human Cx: x is a car T x: x is a truck Dxy: x drives y Write formulas representing the obvious assumptions: no human is a car.e. (d) Nobody drives both. 3. Assume in addition the following predicate: Ixy: x is identical to y Write formulas representing the following statements: (a) Every car has at most one driver.

¬ ∃x (Hx ∧ Cx).(c) Everybody drives exactly one vehicle (car or truck). Exercise 2. ∀x ((∃y Dxy) ⇒ Hx). existence of inverses. 1. (a) ∀x (Cx ⇒ ∀y ∀z ((Dyx ∧ Dzx) ⇒ Iyz)). etc. ∃x (Hx ∧ ¬ ∃y Dxy). Assume the following predicates: Ixy: x = y P xyz: x · y = z Write formulas representing the axioms for a group: axioms for equality. existence of an identity element. (c) ∃x (Hx ∧ ¬ ∃y (Dxy ∧ (Cy ∨ T y))). 30 . (c) ∀x (Hx ⇒ ∃y (Dxy ∧ (Cy ∨ T y) ∧ ∀z ((Dxz ∧ (Cz ∨ T z)) ⇒ Iyz))).g.2. Some humans do not drive. ¬ ∃x (Cx ∧ T x). ∃x (Cx ∧ ∃y Dyx). Humans exist. Some humans drive. (b) ∀x (T x ⇒ ∃y ∃z ((¬ Iyz) ∧ Dyx ∧ Dzx ∧ ∀w (Dwx ⇒ (Iwy ∨ Iwz)))). Only humans drive. ∃x (Cx ∧ ¬ ∃y Dyx). Solution. (a) ∀x (Hx ⇒ ∃y (Dxy ∧ (Cy ∨ T y))). Only cars and trucks are driven.15. No human is a car. 1. associative law. ∀x ((∃y Dyx) ⇒ (Cx ∨ T x)). equality axioms: (a) ∀x Ixx (reﬂexivity) (b) ∀x ∀y (Ixy ⇔ Iyx) (symmetry) (c) ∀x ∀y ∀z ((Ixy ∧ Iyz) ⇒ Ixz) (transitivity) (d) ∀x ∀x′ ∀y ∀y ′ ∀z ∀z ′ ((Ixx′ ∧ Iyy ′ ∧ Izz ′ ) ⇒ (P xyz ⇔ P x′ y ′ z ′ )) (congruence with respect to P ). etc. existence and uniqueness of products. ∃x Cx. 3. 2. Solution. ∃x Hx. (d) ¬ ∃x (Hx ∧ ∃y ∃z (Dxy ∧ Dxz ∧ Cy ∧ T z)). No car is a truck.. old wrecks). (b) ∃x (Hx ∧ ∃y ∃z (Dxy ∧ Cy ∧ Dxz ∧ T z)). ∃x (Hx ∧ ∃y Dxy). Some cars are driven. Cars exist. Some cars are not driven (e.

Fix a countably inﬁnite set V = {a1 . . . | T A[x/a] . b.2. . The order of an element a ∈ G is the smallest positive integer n such that an = e.2. L-V -sentences will be called sentences with parameters.16. Let G be a group. F ∃x A . Signed: . existence of inverses: ∃u ∀x ∃y (P uxx ∧ P xux ∧ P xyu ∧ P yxu). with the following additional rules. . 2. . | F A[x/a] where a is an arbitrary parameter 31 . The elements of V will be called parameters. . 3. existence and uniqueness of products: (a) ∀x ∀y ∃z P xyz (existence) (b) ∀x ∀y ∀z ∀w ((P xyz ∧ P xyw) ⇒ Izw) (uniqueness).}. . Here e denotes the identity element of G.1. . existence of identity element: ∃u ∀x (P uxx ∧ P xux). n times Using only the predicates P xyz (“x · y = z”) and Ixy (“x = y”). T ∀x A . . . 5. . Exercise 2. .3. an . 4. Solution. write a predicate calculus sentence S such that. . and an = a · · · · · a . .2. Deﬁnition 2. . If L is a language. for any group G.} = {a. ¬ ∃x ∃y (P xxx ∧ (¬ P yyy) ∧ (P yyx ∨ ∃z (P yyz ∧ P zyx))).3. a2 . G satisﬁes S if and only if G has no elements of order 2 or 3. A (signed or unsigned ) tableau is a rooted dyadic tree where each node carries a (signed or unsigned) L-V -sentence. . The tableau rules for the predicate calculus are the same as those for the propositional calculus. c. . associative law: ∀x ∀y ∀z ∃u ∃v ∃w (P xyu ∧ P yzv ∧ P uzw ∧ P xvw).3 The Tableau Method Deﬁnition 2. .

| A[x/a] . We are going to prove that the tableau method for predicate calculus is sound (Theorem 2. . . or equivalently a ﬁnite closed unsigned tableau starting with ¬ A. . | F A[x/a] where a is a new parameter Unsigned: . .. . . F ∀x A . | A[x/a] . ∃x A . . Or. .5). ¬ ∀x A .3. . . “a is new” means that a does not occur in the path that is being extended. .3. | ¬ A[x/a] where a is a new parameter Remark 2. . .4. . In particular. ¬ ∃x A . In the above tableau rules. . .13) and complete (Theorem 2. | ¬ A[x/a] where a is an arbitrary parameter .3. T ∃x A . ∀x A . .3. . 32 . | T A[x/a] . . . . . Remark 2. a sentence A of the predicate calculus is logically valid if and only if there exists a ﬁnite closed signed tableau starting with F A.5. we can insist that a not occur in the tableau that is being extended. . .

Use signed tableaux to show that the following are logically valid. by the Soundness Theorem. F (∀x (A ⇒ B)) ⇒ ((∀x A) ⇒ (∀x B)) T ∀x(A ⇒ B) F (∀x A) ⇒ (∀x B) T ∀x A F ∀x B F B[x/a] T A[x/a] T (A ⇒ B)[x/a] / \ F A[x/a] T B[x/a] 33 . The signed tableau F (∃x ∀y Rxy) ⇒ (∀y ∃x Rxy) T ∃x ∀y Rxy F ∀y ∃x Rxy T ∀y Ray F ∃x Rxb T Rab F Rab is closed. Therefore. (∃x ∀y Rxy) ⇒ (∀y ∃x Rxy) is logically valid. Example 2. (∃x (P x ∨ Qx)) ⇔ ((∃x P x) ∨ (∃x Qx)) is logically valid. The unsigned tableau ¬ ((∃x (P x ∨ Qx)) ⇔ ((∃x P x) ∨ (∃x Qx))) / \ ∃x (P x ∨ Qx) ¬ ∃x (P x ∨ Qx) ¬ ((∃x P x) ∨ (∃x Qx)) (∃x P x) ∨ (∃x Qx) ¬ ∃x P x / \ ¬ ∃x Qx ∃x P x ∃x Qx P a ∨ Qa Pb Qc / \ ¬ (P b ∨ Qb) ¬ (P c ∨ Qc) Pa Qa ¬Pb ¬Pc ¬Pa ¬ Qa ¬ Qb ¬ Qc is closed. (∀x (A ⇒ B)) ⇒ ((∀x A) ⇒ (∀x B)) Solution.Example 2.3. Therefore.3. by the Soundness Theorem.6.7. Exercises 2.5.3. 1.

translate the following argument into a sentence of the predicate calculus. F (∃x A) ⇔ (¬ ∀x ¬ A) / \ T ∃x A F ∃x A F ¬ ∀x ¬ A T ¬ ∀x ¬ A T A[x/a] F ∀x ¬ A T ∀x ¬ A F (¬ A)[x/a] T (¬ A)[x/a] T A[x/a] F A[x/a] F A[x/a] 4. (∃x (A ∨ B)) ⇔ ((∃x A) ∨ (∃x B)) Solution. F (∃x (A ∨ B)) ⇔ ((∃x A) ∨ (∃x B)) / \ T ∃x (A ∨ B) F ∃x (A ∨ B) F (∃x A) ∨ (∃x B) T (∃x A) ∨ (∃x B) T (A ∨ B)[x/a] \ / F ∃x A T ∃x A T ∃x B F ∃x B T A[x/a] T B[x/a] F A[x/a] F (A ∨ B)[x/a] F (A ∨ B)[x/a] F B[x/a] F A[x/a] F A[x/a] / \ F B[x/a] F B[x/a] T A[x/a] T B[x/a] 3. F (∀x (A ∨ C)) ⇔ ((∀x A) ∨ C) / \ T ∀x (A ∨ C) F ∀x (A ∨ C) F (∀x A) ∨ C T (∀x A) ∨ C F ∀x A F (A ∨ C)[x/a] FC F A[x/a] F A[x/a] FC T (A ∨ C)[x/a] / \ / \ T ∀x A T C T A[x/a] T C T A[x/a] Exercise 2. provided x is not free in C Solution. and Sxy (“x shaves y”). 34 . (∃x A) ⇔ (¬ ∀x ¬ A) Solution. Cx (“x is a citizen of Podunk”).8.2. (∀x (A ∨ C)) ⇔ ((∀x A) ∨ C).3. Using the predicates Bx (“x is a barber in Podunk”).

Use an unsigned tableau to test this argument for logical validity.e. Therefore. No student can solve every logic problem. .11. (∃x (Bx ∧ ∀y (Cy ⇒ (Sxy ⇔ ¬ Syy)))) ⇒ ∃x (Bx ∧ ¬ Cx). If A is an L-V -sentence. An L-V -structure consists of an L-structure M together with a mapping φ : V → UM . S is said to be satisﬁable 6 if there exists an L-V -structure satisfying S. φ is said to satisfy S if vM (Aφ ) = T for all A ∈ S. then Aφ = A. Exercise 2. The rest of this section is devoted to proving the Soundness Theorem 2. in the obvious way. Proof. Note that Aφ is an L-UM -sentence.12.2.9. then τ ′ is satisﬁable. 6 Similarly. i. An L-V -sentence is said to be logically valid if it satisﬁed by all L-V -structures.3. ak are the parameters occurring in A. Let τ and τ ′ be tableaux such that τ ′ is an immediate extension of τ . Therefore. τ ′ is obtained from τ by applying a tableau rule to a path of τ . 35 . We consider some representative cases. 2. . Exercise 2. We say that τ is satisﬁable if at least one path of τ is satisﬁable. . . Lemma 2. ak /φ(ak )] where a1 .3. The proof consists of one case for each tableau rule. Use an unsigned tableau to show that ∃x (P x ⇔ ∀y P y) is logically valid.There is a barber in Podunk who shaves exactly those citizens of Podunk who do not shave themselves. 1.10. . Let τ be an L-tableau..3. . Note also that. Translate the following argument into the predicate calculus. An L-V -structure M. Solution. and use appropriate methods to establish its validity or invalidity. . Deﬁnition 2. we write Aφ = A[a1 /φ(a1 ).3. A tableau starting with the negation of this sentence (left to the reader) closes oﬀ to show that the sentence is logically valid. Anyone who can solve all logic problems is a good student. If τ is satisﬁable.13. 3. Note that this deﬁnition is compatible with Deﬁnition 2. the notions of logical validity and logical consequence are extended to L-V sentences. if A is an L-sentence. there is a barber in Podunk who is not a citizen of Podunk.3. An L-V -sentence is said to be a logical consequence of S if it is satisﬁed by all L-V -structures satisfying S. using L-V -structures. . Let S be a ﬁnite or countable set of (signed or unsigned) L-V -sentences.7. there are logic problems that no student can solve.

then θ is satisﬁable. Fix such a c and deﬁne φ′ : V → UM by putting φ′ (a) = c.Case 1. then θ is satisﬁable. ∀x A . then S is a path of τ ′ . so τ ′ is satisﬁable. 36 . Suppose that τ ′ is obtained from τ by applying the rule . b ∈ V . . | A[x/a] to the path θ in τ . where a is a parameter. Since this is a path of τ ′ . . it follows that τ ′ is satisﬁable. so let M and φ : V → UM satisfy θ. If S = θ. it follows that τ ′ is satisﬁable. so τ ′ is satisﬁable. . . so let M and φ : V → UM satisfy θ. and Aφ = Aφ . S. / \ A B to the path θ in τ . Since these are paths of τ ′ . Suppose that τ ′ is obtained from τ by applying the rule . Since τ is satisﬁable. so let M and φ : V → UM satisfy θ. B. S. then S is a path of τ ′ . If S = θ. then θ is satisﬁable. it has at least one satisﬁable path. . Thus M and φ satisfy θ. In particular vM (∃x (Aφ )) = vM ((∃x A)φ ) = T. If S = θ. . . Since τ is satisﬁable. . | A[x/a] to the path θ in τ . S. we have B φ = B φ for all B ∈ θ. . so τ ′ is satisﬁable. In particular vM ((A ∨ B)φ ) = T. so vM (Aφ [x/c]) = T for at least one c ∈ UM . A[x/a]. In particular vM (∀x (Aφ )) = vM ((∀x A)φ ) = T. ∃x A . In particular vM (A[x/a]φ ) = vM (Aφ [x/φ(a)]) = T. it has at least one satisﬁable path. . Thus M and φ satisfy at least one of θ. If S = θ. . Case 3. Since a is new. so vM (Aφ [x/c]) = T for all c ∈ UM . Since τ is satisﬁable. . then S is a path of τ ′ . If S = θ. A and θ. Suppose that τ ′ is obtained from τ by applying the rule . A∨B . If S = θ. and φ′ (b) = φ(b) for ′ ′ all b = a. where a is a new parameter. Case 2. so we have at least one of vM (Aφ ) = T and vM (B φ ) = T. it has at least one satisﬁable path.

.4. . A[x/a] ≡ B[x/a]. In particular τn = τ is satisﬁable. Since this is a path of τ ′ . . Thus vM (B φ ) = vM (B φ ) = T for all ′ B ∈ θ.3. Let X1 . ∀x A ≡ ∀x B and ∃x A ≡ ∃x B. Exercises 2. and let a be a parameter not occurring in A. . . ak are parameters not occurring in A. . . Thus M and φ′ satisfy θ. . Show that A is logically valid if and only if ¬ A is not satisﬁable. and vM (A[x/a]φ ) = vM (Aφ [x/c]) = T. written A ≡ B. If there exists a ﬁnite closed tableau starting with X1 . Note that A′ has no free variables. . Show that A is satisﬁable if and only if ∃x A is satisﬁable. Let A be a formula. Xk . . i. Proof. 37 .4.11. xk /ak ]. . . . . . . . Xk and each τi+1 is an immediate extension of τi . . Let x be a variable.3. . τ0 = . Xk is satisﬁable. Show that for any variable x and parameter a.11. Show that for any variable x. . Thus there is a ﬁnite sequence of tableaux τ0 .12. Suppose X1 . Given a formula A. .4 Logical Equivalence Deﬁnition 2. .3. it follows that τ ′ is satisﬁable.hence A[x/a]φ = Aφ [x/φ′ (a)] = Aφ [x/c]. but this is impossible since τ is closed. Show that A is logically valid if and only if A[x/a] is logically valid. . 2. Show that A is satisﬁable if and only if ¬ A is not logically valid. .3. . Theorem 2. Deﬁnition 2. 3. Let τ be a closed tableau starting with X1 . . Xk is not satisﬁable. it is a sentence. A and B are said to be logically equivalent. Let x be a variable.2. and by induction on i using Lemma 2. and a1 .e. ′ ′ ′ 2. . Xk be a ﬁnite set of (signed or unsigned) sentences with parameters. in the sense of Deﬁnition 2. if A ⇔ B is logically valid. .13 (the Soundness Theorem).3. it follows that all of the τi are satisﬁable. in the sense of Deﬁnition 2. where x1 . . 1.4.1.. . τ1 . . τn = τ such that X1 .4. Then τ0 is satisﬁable. Let A and B be formulas. We deﬁne A to be logically valid if and only if A′ is logically valid. .4. . . Show that A is logically valid if and only if ∀x A is logically valid. let A′ = A[x1 /a1 . We deﬁne A to be satisﬁable if and only if A′ is satisﬁable. xk are the variables which occur freely in A. Xk . Assume A ≡ B. then X1 . . . A[x/a]. Exercise 2. . Show that A is satisﬁable if and only if A[x/a] is satisﬁable.

. 2. and each bound variable is bound by at most one quantiﬁer. and ∀x (A ⇒ B) ≡ (∃x A) ⇒ (∀x B). where x1 . . Remark 2. then for any formula C containing A1 as a part. Show that A ≡ B if only if A′ ≡ B ′ . (a) ∃x ¬ A ≡ ¬ ∀x A 38 . provided x does not occur freely in A (f) ∀x (A ⇒ B) ≡ A ⇒ (∀x B).4. (a) ∀x A ≡ A. xk are the variables occurring freely in A and B. . . provided x does not occur freely in B (c) ∀x (A ∨ B) ≡ A ∨ (∀x B).Exercise 2. provided x does not occur freely in B (h) ∀x (A ⇒ B) ≡ (∃x A) ⇒ B. (a) ∃x (A ∧ B) ≡ A ∧ (∃x B). . it is not in general true that A ≡ A′ . also hold for formulas of the predicate calculus.4 and Remark 1. put A′ = A[x1 /a1 . . (a) ∀x (A ∧ B) ≡ (∀x A) ∧ (∀x B) (b) ∃x (A ∨ B) ≡ (∃x A) ∨ (∃x B) (c) ∃x (A ⇒ B) ≡ (∀x A) ⇒ (∃x B) Note however that.4.3. provided x does not occur freely in B (e) ∃x (A ⇒ B) ≡ A ⇒ (∃x B). Give examples illustrating these remarks. or that A ≡ A[x/a]. we have: 3. . . ∃x (A ∧ B) ≡ (∃x A) ∧ (∃x B).6. xk /ak ].5 for formulas of the propositional calculus. ak are parameters not occurring in A or in B. Remark 2. Also. The results of Exercises 1. .5. provided x does not occur freely in A (g) ∃x (A ⇒ B) ≡ (∀x A) ⇒ B. . . . provided y does not occur in A Note that the last two equivalences provide for “change of bound variables”. . and a1 . provided x does not occur freely in B 4. provided x does not occur freely in A (c) ∀x A ≡ ∀y A[x/y].1. . xk /ak ] and B ′ = B[x1 /a1 . it is not in general true that A ≡ ∀x A. . then the resulting formula is logically equivalent to C. provided x does not occur freely in A (b) ∃x (A ∧ B) ≡ (∃x A) ∧ B. In particular. . provided x does not occur freely in A (d) ∀x (A ∨ B) ≡ (∀x A) ∨ B.4.4. provided y does not occur in A (d) ∃x A ≡ ∃y A[x/y]. Some useful logical equivalences are: 1.3 and 1.7. if we replace one or more occurrences of the part A1 by A2 . On the other hand. provided x does not occur freely in A (b) ∃x A ≡ A. in general. where A′ is as in Deﬁnition 2. For a formula A. if A1 ≡ A2 . In this way.4. or that A ≡ ∃x A. and ∀x (A ∨ B) ≡ (∀x A) ∨ (∀x B). we can convert any formula into a logically equivalent formula where no variable occurs both free and bound.3.8. If A and B are formulas.3. Exercise 2.

we have (∃x P x) ∧ (∃x Qx) ≡ ≡ ≡ and this is in prenex form.15. We wish to put this into prenex form. and let A be ∀u ∃v Ruvz.8.12. Exercises 2.8. Note that A has the same free variables as A and is logically equivalent to A.13. First replace the given formula by a variant in which each bound variable is quantiﬁed only once and does not occur freely. Let A be ∀x ∃y Rxyz. A formula is said to be quantiﬁer-free if it contains no quantiﬁers.4. A variant of A is any formula A obtained from A by replacing all the bound variables of A by distinct new variables. 1. and B is quantiﬁer-free.4. Applying the equivalences of Remark 2. Example 2.16. A formula is said to be in prenex form if it is of the form Q1 x1 · · · Qn xn B.14.4. parts 1(c) and 1(d).9 (variants).4. Put the following into prenex form. Find a sentence in prenex normal form which is logically equivalent to (∀x ∃y Rxy) ⇒ ¬ ∃x P x.10. Then A and A are variants of each other. Exercise 2. ∃x ∀y ∀z (Rxy ⇒ ¬ P z). hence logically equivalent to each other. Let A be a formula.4. Example 2. where each Qi is a quantiﬁer (∀ or ∃).8.4.11 (prenex form). (∀x A) ⇔ (∀x B) 39 (∃x P x) ∧ (∃y Qy) ∃x (P x ∧ (∃y Qy)) ∃x ∃y (P x ∧ Qy) . Solution. The sentence ∀x ∀y ∃z ∀w (Rxy ⇒ (Rxz ∧ Rzy ∧ ¬ (Rzw ∧ Rwy))) is in prenex form.(b) ∀x ¬ A ≡ ¬ ∃x A (c) ∀x A ≡ ¬ ∃x ¬ A (d) ∃x A ≡ ¬ ∀x ¬ A Deﬁnition 2. Deﬁnition 2.) Example 2. (∃x A) ∧ (∃x B) 2.4. Let A and B be quantiﬁer-free formulas. in view of Remark 2. Exercise 2. each xi is a variable. Then use parts 3 and 4 to move all quantiﬁers to the front.4. (Hint: Use the equivalences of Remark 2.4.4. Consider the sentence (∃x P x) ∧ (∃x Qx).4. Show that every formula is logically equivalent to a formula in prenex form.

then S contains both A and B.3. It is not true in general that A ≡ A∗ . (∀x A) ⇔ (∃x B) Deﬁnition 2. Give an example illustrating this. It is not true in general that A is satisﬁable if and only if A∗ is satisﬁable. If S contains F ¬ A. A and B are logically equivalent if and only if each is a logical consequence of the other. Let A1 . . 3. S is U -replete if S “obeys the tableau rules” with respect to U . . then S contains A.4.18. S is open if it is not closed. If S contains F A ∧ B. Ak . Exercises 2. F A in the signed case. Let A be a formula. B be formulas. A is logically valid if and only if A is a logical consequence of the empty set.19. . .5. Deﬁnition 2.2.17 (universal closure). For formulas A and B. 40 . The universal closure of A is the sentence A∗ = ∀x1 · · · ∀xk A. S contains T A. Ak if (A1 ∧ · · · ∧ Ak ) ⇒ B is logically valid.4.4. Deﬁnition 2. 2. . 1. . . If S contains T ¬ A. then S contains both T A and T B. Show that A is logically valid if and only if A∗ . . This is equivalent to saying that the universal closure of (A1 ∧ · · · ∧ Ak ) ⇒ B is logically valid. A. it is not true in general that A ≡ B if and only if A∗ ≡ B ∗ . We list some representative clauses of the deﬁnition. ¬ A in the unsigned case. ∃x A is a logical consequence of A[x/a]. A[x/a] is a logical consequence of ∀x A. 4. Note that A∗∗ ≡ A∗ . then S contains at least one of ¬ A and ¬ B. If S contains T A ∧ B. Give an example illustrating this.5 The Completeness Theorem Let U be a nonempty set. . If S contains ¬ (A ∧ B).1. and let S be a set of (signed or unsigned) L-U sentences. where x1 . S is closed if S contains a conjugate pair of L-U -sentences. In other words. If S contains A ∧ B. is logically valid. Give an example illustrating this. Deﬁnition 2. 2. then S contains at least one of F A and F B. For completeness we state the following deﬁnition.4. We say that B is a logical consequence of A1 . Remark 2. 2. .5. the universal closure of A. but the converse does not hold in general. 1. but the converse does not hold in general. xk are the variables which occur freely in A. If S contains ¬ ¬ A. then S contains T A. . Let A be a formula. . then S contains F A. for some L-U -sentence A.20.

. an ∈ U n | P a1 · · · an ∈ S} in the unsigned case. If S contains ∃x A. then S is satisﬁable.7. for each n-ary predicate P of L. (c) If S contains P a1 · · · an . . Proof. If S contains ∀x A. . then vM (A) = T (d) if S contains ¬ A. If S contains ¬ ∃x A. 1. then by deﬁnition of M we have (a1 . . . so vM (P a1 · · · an ) = T. .3 (Hintikka’s Lemma). then S contains A[x/a] for all a ∈ U . then S contains F A[x/a] for at least one a ∈ U . We deﬁne an L-structure M by putting UM = U and. .5. We give the proof for some representative cases. then vM (A) = F in the unsigned case. 7 See also Exercise 2. If S contains ¬ ∀x A. say A = P a1 · · · an . (b) If S contains F P a1 · · · an . so / vM (P a1 · · · an ) = F. (a) if S contains T A. and PM = { a1 . . . . then S contains ¬ A[x/a] for all a ∈ U . then S contains T A[x/a] for all a ∈ U .3. In fact. . If S contains T ∃x A. then S contains T A[x/a] for at least one a ∈ U . then S contains ¬ A[x/a] for at least one a ∈ U . If S contains T ∀x A. Thus by deﬁnition of M we have (a1 . the claim is easily proved by induction on the degree of A. . PM = { a1 . then vM (A) = F in the signed case. an ) ∈ PM . an ) ∈ PM . Lemma 2. Assume S is U -replete and open. 41 . . In this case A is atomic. then S contains A[x/a] for at least one a ∈ U . . and (c) if S contains A. an ∈ U n | T P a1 · · · an ∈ S} in the signed case. (a) If S contains T P a1 · · · an . . We claim that for all L-U -sentences A. If S is U -replete and open7 . . so vM (P a1 · · · an ) = T. S is satisﬁable in the domain U . deg(A) = 0. then by deﬁnition of M we have (a1 . then vM (A) = T (b) if S contains F A. If S contains F ∃x A. In both cases. 4. then S does not contain T P a1 · · · an since S is open.5. . If S contains F ∀x A. . an ) ∈ PM . . . then S contains F A[x/a] for all a ∈ U .

Recall that a tableau is a tree whose nodes carry L-V -sentences. an ) ∈ PM . Note that for all a ∈ U we have deg(B[x/a]) < deg(A). . . . Hence by inductive hypothesis we have vM (B) = vM (C) = T. deg(A) > 0 and A = B ∧ C. . so / vM (P a1 · · · an ) = F.e.. Hence vM (B ∧ C) = F. Thus by deﬁnition of M we have (a1 .} be the set of parameters. . we apply tableau rules systematically to ensure that τ∞ will have the desired properties. . . then the construction halts. Call a node X of τi quasiuniversal if it is of the form T ∀x A or F ∃x A or ∀x A or ¬ ∃x A. Note that for all a ∈ U we have deg(B[x/a]) < deg(A). we set ∞ τ = τ∞ = i=0 τi . an .} is countably inﬁnite. we can extend τ0 to a (possibly inﬁnite) tableau τ with the following properties: every closed path of τ is ﬁnite. then by repleteness of S we see that S contains both TB and TC.(d) If S contains ¬ P a1 · · · an . As in Section 2. (d) If S contains ¬ (B ∧ C). . then by repleteness of S we see that S contains both B and C. τ1 . . . . then by repleteness of S we see that S contains at least one of ¬ B and ¬ C. . .. . then by repleteness of S we see that S contains at least one of FB and FC. (c) If S contains B ∧ C. deg(A) > 0 and A = ∃x B.3. . a2 . . We shall now use Hintikka’s Lemma to prove the completeness of the tableau method. (a) If S contains T B ∧ C. . Proof. . . deg(A) > 0 and A = ¬ B. Hence by inductive hypothesis we have at least one of vM (B) = F and vM (C) = F. By applying tableau rules. . (b) If S contains F B ∧ C. Let τ0 be a ﬁnite tableau. 2. and every open path of τ is V -replete.5. using the fact that V = {a1 . Lemma 2. and we set τ = τi . Hence by inductive hypothesis we have at least one of vM (B) = F and vM (C) = F. 5. τj = τi for all j ≥ i. Hence by inductive hypothesis we have vM (B) = vM (C) = T. . In any case. so the inductive hypothesis applies to B[x/a]. τi . . an . . so the inductive hypothesis applies to B[x/a]. The idea is to start with τ0 and use tableau rules to construct a sequence of ﬁnite extensions τ0 . Hence vM (B ∧ C) = F. deg(A) > 0 and A = ∀x B. i. 4. Hence vM (B ∧ C) = T. Here are the details of the construction. Note that deg(B) < deg(A) so the inductive hypothesis applies to B. Note that deg(B) and deg(C) are < deg(A) so the inductive hypothesis applies to B and C.4. . In the course of the construction. Hence vM (B ∧ C) = T. then S does not contain P a1 · · · an since S is open. 3. . Our construction begins 42 . If some τi is closed. Let V = {a1 .

then by K¨nig’s Lemma (Theorem 1. Suppose we have constructed τ2i . Corollaries 1. . Xk is not satisﬁable. In addition.5. .7. . Exercise 2.5. . Let X1 . In this construction.5. so all closed paths of τ∞ are ﬁnite. then there exists a ﬁnite closed tableau starting with X1 . Xk such that every closed path of τ is ﬁnite. . S is satisﬁable in V .9.4 there exists a (possibly inﬁnite) tableau τ starting with X1 . Let τ2i+1 be the ﬁnite tableau so obtained. . This proves our lemma. . the construction ensures that each open path of τ∞ is V -replete.8. . .3. . for some n-ary L-predicate P and a1 .with τ0 .5.6. Show that Lemmas 2. . Xk . Theorem 2. and P a1 · · · an . Next. In other words. . In particular. Let L.5. Xk is satisﬁable in V . In 2. let τ2i+2 be the ﬁnite tableau so obtained. S contains T P a1 · · · an . . . o If τ is open. Xk be a ﬁnite set of (signed or unsigned) sentences with parameters. Note however that the test is only partially eﬀective. For each quasiuniversal node X of τ2i and each n ≤ 2i.5. .6. 1. let S be an open path of τ . . .5. . 1. . Then S is V -replete. S is atomically open if it is not atomically closed. . It is 43 . Hence X1 . τ is ﬁnite. S is said to be atomically closed if S contains a conjugate pair of atomic L-U -sentences. the tableau method provides a test for logical validity of sentences of the predicate calculus. Again.5 (the Completeness Theorem).5. .5. By Lemma 2.9. we will not necessarily ﬁnd a ﬁnite tableau which demonstrates this. If a sentence A is logically valid. Remark 2. If X1 . . . extend each open path containing X by applying the appropriate tableau rule.5 continue to hold with “closed” (“open”) replaced by “atomically closed” (“atomically open”). and S be as in Deﬁnition 2. U . a closed path is never extended. . . ¬ P a1 · · · an in the unsigned case.5. Example 2. apply the appropriate tableau rule to extend each open path of τ2i containing X by T A[x/an ] or F A[x/an ] or A[x/an ] or ¬ A[x/an ] as the case may be.3 and 2.4 and Theorem 2.2. we will certainly ﬁnd a ﬁnite closed tableau starting with ¬ A. and every open path of τ is V -replete.10. Deﬁnition 2.12 we have seen an example of a sentence A∞ which is satisﬁable in a countably inﬁnite domain but not in any ﬁnite domain. . Proof. . . for each non-quasiuniversal node X of τ2i+1 . F P a1 · · · an in the signed case. If X1 . .1.11 carry over from the propositional calculus to the predicate calculus.5. .6).5. then X1 . Xk is satisﬁable. . Thus τ∞ has the desired properties. Xk is satisﬁable in the domain V .5. If τ is closed. . By Hintikka’s Lemma 2. an ∈ U . See the following example.5. But if A is not logically valid.

∀x ∀y ∀z ((Rxy ∧ Ryz) ⇒ Rxz) ∀x ∀y (Rxy ⇒ ¬ Ryx) ∀x ∃y Rxy ∃y Ra1 y Ra1 a2 ∀y (Ra1 y ⇒ ¬ Rya1 ) Ra1 a2 ⇒ ¬ Ra2 a1 / \ ¬ Ra1 a2 ¬ Ra2 a1 ∃y Ra2 y Ra2 a3 . . . ¬ Ra1 a2 ¬ Ra2 a3 ¬ Ra3 a1 ∃y Ra3 y Ra3 a4 . Clearly M |= A∞ . an | 1 ≤ m < n}. we will sometimes ﬁnd a ﬁnite open path which is U -replete for some ﬁnite set of parameters U ⊆ V .5. In the course of applying a tableau test. . a2 . . In this case.}. . ¬ Ra3 a2 ∀y ∀z ((Ra1 y ∧ Ryz) ⇒ Ra1 z) ∀z ((Ra1 a2 ∧ Ra2 z) ⇒ Ra1 z) (Ra1 a2 ∧ Ra2 a3 ) ⇒ Ra1 a3 ) / \ ¬ (Ra1 a2 ∧ Ra2 a3 ) Ra1 a3 .instructive to generate a tableau starting with A∞ . . . A∞ . 44 . .5. RM = { am . . Remark 2. .10. An inﬁnite open path gives rise (via the proof of Hintikka’s Lemma) to an inﬁnite L-structure M with UM = {a1 .11. an . the proof of Hintikka’s Lemma provides a ﬁnite L-structure with domain U . . . / \ . . Let A be the sentence (∀x (P x ∨ Qx)) ⇒ ((∀x P x) ∨ (∀x Qx)). . Example 2.

13.e. QM = {a}. PM = {b}. Clearly M falsiﬁes A. i. b}. Therefore. A tableau starting with the negation of this sentence (left to the reader) fails to close oﬀ. All men are mortal. translate the following argument into a sentence of the predicate calculus. Solution. Using the predicates Sx (“x is a set”) and Exy (“x is a member of y”).12. Use an unsigned tableau to test your sentence for consistency. M x (“x is mortal”). Every ancestor of an ancestor of an individual is an ancestor of the same individual.5.14. we have: ¬A ∀x (P x ∨ Qx) ¬ ((∀x P x) ∨ (∀x Qx)) ¬ ∀x P x ¬ ∀x Qx ¬Pa ¬ Qb P a ∨ Qa P b ∨ Qb / \ Pa Qa / \ P b Qb This tableau has a unique open path. Hx (“x is a man”). Socrates is mortal. Exercise 2. translate the following argument into a sentence of the predicate calculus. Socrates is a man. Exercise 2..5. translate the following into a sentence of the predicate calculus. which gives rise (via the proof of Hintikka’s Lemma) to a ﬁnite L-structure M with UM = {a. Therefore. Exercise 2. Use an unsigned tableau to test whether the argument is valid. ((∀x ∀y ((∃z (Rxz ∧ Rzy)) ⇒ Rxy)) ∧ ¬ ∃x Rxx) ⇒ ∃x ¬ ∃y Ryx.5. Using the predicates Sx (“x is Socrates”). No individual is his own ancestor. >N ) falsiﬁes the sentence. Is this argument valid? Justify your answer by means of an appropriate structure or tableau. The structure (N. There exists a set whose members are exactly those sets which are not members of themselves. thus showing that it is not logically valid. 45 . there is an individual who has no ancestor. satisﬁability.Testing A for logical validity. Using the predicate Rxy (“x is an ancestor of y”).

. nobody can solve it. Assume that S is ﬁnitely satisﬁable. Let US be the set of parameters so introduced.2 (the Compactness Theorem. Proof. Let S ′ be S together with the sentences B[x/cA ] or ¬ B[x/cA ] 46 .6 The Compactness Theorem Theorem 2.}. Use an unsigned tableau to test whether the argument is valid. . Proof. uncountable case). . introduce a new parameter cA .6). . Extend τ2i to τ2i by appending Ai ′ to each open path of τ2i . Suppose we have constructed τ2i .4 we have that every closed path of τ is ﬁnite. translate the following argument into a sentence of the predicate calculus.5. let S ′ be an inﬁnite path in τ .7. Finally put τ = τ∞ = i=1 τi .15. Therefore.5.3). Note also that τ is an inﬁnite. S ′ is satisﬁable. Use an unsigned tableau to test whether the argument is valid. Ai . . Therefore. nobody can solve it. Since {A0 . countable case). Let S be an uncountable set of sentences of the predicate calculus.5. Jx (“x is Joe”). 2. Theorem 2.Exercises 2. S is satisﬁable if and only if each ﬁnite subset of S is satisﬁable. . Details: Let S = {A0 .5). Using the predicates Sx (“x can solve this problem”). and every open path of τ is V -replete. As in Lemma 2. Let S be a countably inﬁnite set of sentences of the predicate calculus.6. . M x (“x is a mathematician”). .1) and the Completeness Theorem for predicate calculus (Theorem 2. . Joe is a mathematician and cannot solve it.6.4. By Hintikka’s Lemma for the predicate calculus (Lemma 2. If anyone can solve this problem.1 (the Compactness Theorem. Ai } is satisﬁable. . translate the following argument into a sentence of the predicate calculus. A1 .5. Hence S is satisﬁable. We combine the ideas of the proofs of the Countable Compactness Theorem for propositional calculus (Theorem 1. some mathematician can solve it. A1 . τ2i has at least ′ one open path. S is satisﬁable if and only if each ﬁnite subset of S is satisﬁable. Using the same predicates as above. Any mathematician can solve this problem if anyone can.6. By K¨nig’s Lemma (Theorem o 1. For each sentence A ∈ S of the form ∃x B or ¬ ∀x B. ﬁnitely branching tree.5. Then S ′ is a V -replete open set which contains S. Start by letting τ0 be the empty ′ tableau. 2. Now extend τ2i to τ2i+1 and then to τ2i+2 as in the proof of ∞ Lemma 2. Joe is a mathematician and cannot solve it. . 1.

) be an L∗ -structure satisfying S ∗ . Let (U∞ . .6. .6. 2. Then S ′ is a set of L-US -sentences. then S is satisﬁable in a countably inﬁnite domain. Prove that there there exists an L-structure (U∞ . . for all cA ∈ US . are new unary predicates. .3. Since S ⊆ S∞ . Show that M ′ can be obtained such that (UM ′ .9.} where P1 .7 Satisﬁability in a Domain The notion of satisﬁability in a domain was introduced in Deﬁnition 2.7. . If S is satisﬁable. . It follows by the Compactness Theorem that S ∗ is satisﬁable. . . Let S ∗ be S plus ∃x Pi x plus ∀x ∀y ((Pi x ∧ Pj y) ⇒ Rxy). . Exercise 2. . a′ . Let L∗ = L ∪ {P1 . . Sn+1 = Sn . . Let L = {R. 47 . RM ) is isomorphic to (N. 2. i = 1. Clearly the L-structure (U∞ . <N ). such that a∞i . let S ′′ be a maximal ﬁnitely satisﬁable set of L-US -sentences extending S ′ . . It is straightforward to verify that S∞ is U -replete and open. .) be an L-structure such that (UM . .6. Theorem 2. P∞2 . ann such that ani . . . RM ′ ) is a linear ordering which contains an inﬁnite descending sequence. . . Exercise 2. . and it is easy to verify that S ′ is ﬁnitely satisﬁable. ∈ UM ′ such that a′ .4. Exercise 2. Assume that for each n ≥ 1 there exists an L-structure (Un . a′ . . a∞j ∈ R∞ for all i and j with 1 ≤ i < j. In other words. .as the case may be. Let L be a language consisting of a binary predicate R and some additional predicates. S∞ is satisﬁable in the domain U . . . there exist elements a′ . . . . .} be a language which includes a binary predicate R. Rn . . . . . . M and M ′ satisfy the same L-sentences. a′ ∈ RM ′ for all 1 2 n n+1 n n = 1. .3 replacing (N. Clearly each ﬁnite subset of S ∗ is satisﬁed by all but ﬁnitely many of these structures. P2 . 1. Assume that S is ﬁnite or countably inﬁnite. . it follows that S is satisﬁable in U . RM . M ′ contains an inﬁnite R-descending sequence. . . . Pnn . ∞ ∞ ′′ Now inductively deﬁne S0 = S. Let S be a set of L-sentences. . .. Let S be a set of L-sentences.2. Note that M contains no inﬁnite R-descending sequence. . . . and Pni = {} for i > n. . . Rn .1. n ≥ 1. Hint: Use the Compactness Theorem. Pn1 . . By Zorn’s Lemma. . . P2 . 1 ≤ i < j. S∞ = n=0 Sn . . Show that there exists an L-structure M ′ such that: 1. Hence. Let M = (UM . .). . . . . R∞ . <N ) by an arbitrary inﬁnite linear ordering with no inﬁnite descending sequence.) satisfying S and containing elements a∞i . Generalize Exercise 2. P∞1 .) satisfying S and containing elements an1 . R∞ . anj ∈ Rn for all i and j with 1 ≤ i < j ≤ n. 2.) has the desired properties.5.6. Consider the L∗ -structures (Un . . . where Pni = {ani } for 1 ≤ i ≤ n. U = n=0 USn . 2. R∞ . by Hintikka’s Lemma. Solution.

if M |= S. Assume that there exists an onto mapping φ : UM → UM ′ such that for all n-ary predicates P of L and all n-tuples a1 .3. this holds if and only if vM ′ (B[x/a]′ ) = T for all a ∈ UM . we have: Theorem 2. then S is satisﬁable in a domain of cardinality κ. By the arithmetic of inﬁnite cardinal numbers. Then by deﬁnition of vM we have that vM (A) = T if and only if vM (B[x/a]) = T for all a ∈ UM . . Suppose for example that A = ∀x B. In the countable case we have that S is satisﬁable in V .6 we have vM (A) = vM ′ (A′ ) for all L-UM -sentences A.2.1 and 2.4. Corollary 2. Let M and M ′ be L-structures. which is countably inﬁnite. . . Let S be a set of L-sentences. . an ∈ (UM )n . The proof is by induction on the degree of A. then M ′ |= S. then S is satisﬁable in any domain of the same or larger cardinality.2.6. Thus our condition is equivalent to vM ′ (B ′ [x/φ(a)]) = T for all a ∈ UM . Our sentence An is (1) ∧ (2) ∧ (3) with (1) ∀x ∀y ∀z ((Rxy ∧ Ryz) ⇒ Ryz) (2) ∀x ∀y (Rxy ⇒ ¬ Ryx) (3) ∃x1 · · · ∃xn (Rx1 x2 ∧ Rx2 x3 ∧ · · · ∧ Rxn−1 xn ) On the other hand. Proof. ak /φ(ak )]. the cardinality of U is κ · ℵ0 = κ. Suppose S is satisﬁable in domain U . . . where U is as in the proof of 2.6.12 we have seen a sentence A∞ which is satisﬁable in a countably inﬁnite domain but not in any ﬁnite domain. Let U ′ be a set of cardinality greater than or equal to that of U . By Theorem 2. . If S is satisﬁable in a domain U .2.2. . an | φ(a1 ). . Since φ : UM → UM ′ is onto. . Given a positive integer n. In the uncountable case we have that S is satisﬁable in U . . we exhibit a sentence An which is satisﬁable in a domain of cardinality n but not in any domain of smaller cardinality. . . By deﬁnition of vM ′ this is equivalent to vM ′ (∀x B ′ ) = T. But for all a ∈ UM we have B[x/a]′ = B ′ [x/φ(a)]. In Example 2. φ(an ) ∈ PM ′ . . a1 . where A′ = A[a1 /φ(a1 ).7.7. Proof. . If S is satisﬁable.3. Regarding satisﬁability in ﬁnite domains. respectively. . . In particular. Assume that S is of cardinality κ ≥ ℵ0 . .7. But ∀x B ′ = A′ . φ(an ) ∈ PM } for all n-ary predicates P of L. Parts 1 and 2 follow easily from the proofs of Compactness Theorems 2. .7. .6. so our condition is equivalent to vM ′ (A′ ) = T. By inductive hypothesis. . an ∈ PM if and only if φ(a1 ). . . M and M ′ satisfy the same L-sentences. Let φ : U ′ → U be onto. Proof. Then as in Theorem 2. If M is any L-structure with UM = U . we can deﬁne an L-structure M ′ with UM ′ = U ′ by putting PM ′ = { a1 .2. M and M ′ satisfy the same L-sentences. this is equivalent to vM ′ (B ′ [x/b]) = T for all b ∈ UM ′ .2. we have: Example 2. In particular. . 48 .

5. 8 See Section 4. 49 .3 and Corollary 2.7.7.7.1.Remark 2. We shall see later8 that Theorem 2.4 fail for normal satisﬁability.

1.Chapter 3 Proof Systems for Predicate Calculus 3.1.2. . For k ≥ 1 it is customary to write X1 · · · Xk Y indicating that X1 . . For each rule of inference A1 · · · Ak B 50 . Recall that V is the set of parameters. . The sequence X1 . . A Hilbert-style proof system for L is a proof system with the following properties: 1. .1. . jk < i such that Xj1 . Notation 3. . Elements of R are called rules of inference. X = {A | A is an L-V -sentence} . An object X ∈ X is said to be derivable. Let L be a language. . . Xjk . . Y ∈ R. An abstract proof system consists of a set X together with a ∞ relation R ⊆ k=0 Xk+1 . Xi ∈ R. . . or a proof of X. Deﬁnition 3. indicating that Y ∈ R. . . there exist j1 . For k = 0 we may write Y or simply Y . . Xk . or provable. for each i ≤ n. In other words.1 Introduction to Proof Systems Deﬁnition 3. . Xk we may immediately infer the conclusion Y ”.3. . . Xn such that Xn = X and. This is to be understood as “we may immediately infer Y from no premises”. Xn is called a derivation of X. .1. . This is to be understood as “from the premises X1 . Elements of X are called objects. . if there exists a ﬁnite sequence of objects X1 . or “we may assume Y ”. . . The objects are sentences with parameters. . 2.

A A⇒B . . . This implies that. 1. for all L-V -sentences A and B. . 1 In other words. Ak are L-V -sentences.e. 51 . . . This property is known as soundness. Ak . pk are the atoms occurring in F . For example. . Ak . given an L-V -sentence A as input. . 3. Deﬁnition 3.2. we have a rule of inference A. . there is a Turing algorithm which. 1. p ⇒ (q ⇒ p) is a tautology.(i. . where F is a tautology.1. Remarks 3. we have that B is a logical consequence of A1 . and a quantiﬁcational component (companions). Obviously. . pk /Ak ]. A1 .4. For all L-V -sentences A. which is also of interest in its own right. A ⇒ (B ⇒ A) is a quasitautology. B ∈ R). from A and A ⇒ B we immediately infer B. An L-V -sentence A is logically valid if and only if A is derivable. p1 . . .. It implies that every L-V -sentence which is derivable is logically valid. .2. . This collection of inference rules is known as modus ponens. . 2.1 (quasitautologies). B. i. There is a decision procedure1 for quasitautologies. 4. . B ∈ R.3 we shall exhibit a particular Hilbert-style proof system. In Section 3. A quasitautology is an L-V -sentence of the form F [p1 /A1 . Remark 3. B In other words. . One such decision procedure is based on truth tables. We shall analyze logical validity into two components: a propositional component (quasitautologies). 0 if A is not a quasitautology. . In order to verify the completeness of LH . 2. every quasitautology is logically valid. and A1 . Another is based on propositional tableaux. A ⇒ B. we shall ﬁrst prove a result known as the Companion Theorem. . .2. The soundness of LH will be obvious..2 The Companion Theorem In this section we shall comment on the notion of logical validity for sentences of the predicate calculus.e. 3. This property is known as completeness. LH . will eventually halt with output 1 if A is a quasitautology. A tautology is a propositional formula which is logically valid.

Note that (C1 ∧ · · · ∧ Cn ) ⇒ A is quasitautologically equivalent to Cn ⇒ (Cn−1 ⇒ · · · ⇒ (C1 ⇒ A)). The case when C is of the form 3. Let C be a companion of A. Let A be an L-V -sentence.3(2). hence ′ vM (B[x/a]φ ) = vM (B φ [x/c]) = F. Since a does not occur in A. assume that A is satisﬁable. Also. . consider the case when C is of the form 3. If M.3. so let c ∈ UM be such that vM (B φ [x/c]) = F. φ be an L-V -structure satisfying A. Thus M.2. 52 . 2. For part 2 note that. φ′ satisﬁes A. let M.) Therefore. A companion sequence of A is a ﬁnite sequence C1 . Let C be a companion of A. .2. then A is logically valid if and only if (C1 ∧ · · · ∧ Cn ) ⇒ A is logically valid. φ′ satisﬁes C ∧ A. . In accordance with Deﬁnition 2. . then C is logically valid.2. M. φ satisﬁes ∀x B. Lemma 3. we regard the quasitautologies as trivial. (This result is known as Church’s Theorem. the parameter a may not occur in A or in B. Cn such that. Thus we have that A is logically valid if and only if ¬ A is not satisﬁable. If C1 . Ci+1 is a companion of (C1 ∧ · · · ∧ Ci ) ⇒ A. we have B[x/a]φ = B φ [x/c] = B φ [x/c]. A companion of A is any L-V -sentence of one of the forms (1) (∀x B) ⇒ B[x/a] (2) (3) (4) B[x/a] ⇒ (∀x B) (∃x B) ⇒ B[x/a] B[x/a] ⇒ (∃x B) where. in relation to the problem of characterizing logical validity. Proof. A is logically valid if and only if C ⇒ A is logically valid. i. Cn is a companion sequence of A. A is satisﬁable if and only if C ∧ A is satisﬁable.5 (companion sequences).3(3) is handled similarly. i.3 (companions). If C is of the form 3. we have that M.2. C ⇒ A is logically valid. It can be shown that there is no decision procedure for logical validity.6. .3..2. φ′ (b) = φ(b) for b = a. Deﬁnition 3. 1. then M. if and only if ¬ (C ∧ ¬ A) is logically valid.2. ′ ′ since a does not occur in B.3(1) or 3. Next. Lemma 3. φ satisﬁes C ∧ A. .2. for each i < n. C is a companion of ¬ A. . Deﬁne φ′ : V → UM by putting φ′ (a) = c. since C is a companion of A. if and only if C ∧ ¬ A is not satisﬁable (by part 1). Otherwise we have vM (∀x B φ ) = F. Proof.e.e. in (2) and (3).4. Deﬁnition 3. For part 1. hence M.3(4).2.11.. φ satisﬁes C. . φ′ satisﬁes ¬ B[x/a].

A is logically valid if and only if there exists a companion sequence C1 . . Thus we have a ﬁnite sequence of tableaux τ0 . . .5 let τ be a ﬁnite closed unsigned tableau starting with ¬ A. For the “only if” part. . The “if” part is immediate from Lemma 3. and each τi+1 is obtained by applying a tableau rule Ri to τi . Cn is a companion sequence for A. . . . If Ri is a quantiﬁer rule.2. ∀x B . and replace the application of (∗) by . . Proof. . let Ci be an appropriate companion. τ1 . This proves our theorem. Cn in which only propositional tableau rules are applied. . .4. 53 . if Ri is the tableau rule . .2. | B[x/a] . . Cn of A such that (C1 ∧ · · · ∧ Cn ) ⇒ A is a quasitautology. For instance. ¬ ∀x B / \ B[x/a] (∗) noting that the left-hand path is closed. . let Ci be the companion (∀x B) ⇒ B[x/a].Our lemma follows by n applications of part 2 of Lemma 3. .6. Theorem 3. . (∀x B) ⇒ B[x/a] .7 (the Companion Theorem). and we can easily transform τ into a closed tableau τ ′ starting with ¬A C1 . τn where τ0 = ¬ A. where a is an arbitrary parameter. . .2. . assume that A is logically valid.5. Thus C1 . . . . Thus (C1 ∧ · · · ∧ Cn ) ⇒ A is a quasitautology. . τn = τ . ∀x B . . By Theorem 2.

we obtain the companion sequence C1 . C3 for A. . . C2 is P a ⇒ ∃x P x. Exercise 3. . Let A be the logically valid sentence ∃x (P x ⇒ ∀y P y). .7 and its proof. . Let τ be the closed tableau ¬A ∃x(P x ∨ Qx) ¬ ((∃x P x) ∨ (∃x Qx)) ¬ ∃x P x ¬ ∃x Qx P a ∨ Qa ¬Pa ¬ Qa / \ Pa Qa which shows that A is logically valid. .2. let Ci be the companion B[x/a] ⇒ (∀x B).2. let A be the sentence (∃x (P x ∨ Qx)) ⇒ ((∃x P x) ∨ (∃x Qx)). C3 is Qa ⇒ ∃x Qx. . Cn for A such that (C1 ∧ · · · ∧ Cn ) ⇒ A is a quasitautology.9. . .2. C2 . where C1 is (∃x (P x ∨ Qx)) ⇒ (P a ∨ Qa). / \ ¬ B[x/a] ∀x B noting that the right-hand path is closed.Similarly. Clearly (C1 ∧ C2 ∧ C3 ) ⇒ A is a quasitautology. . As an example illustrating Theorem 3. if Ri is the tableau rule . and replace the application of (∗∗) by . . ¬ ∀x B . . Examining the applications of quantiﬁer rules in τ . Find a companion sequence C1 . . | ¬ B[x/a] where a is a new parameter. Example 3.8. ¬ ∀x B . B[x/a] ⇒ (∀x B) . (∗∗) 54 . .

. we ﬁrst construct a closed unsigned tableau starting with ¬ A. (∃y P y) ⇒ P b. where C1 C2 C3 is is is (P a ⇒ ∀y P y) ⇒ ∃x (P x ⇒ ∀y P y). ′ ′ A simpler companion sequence for A is C1 . (P a ⇒ ∀y P y) ⇒ ∃x (P x ⇒ ∀y P y). where C1 C2 C3 C4 is is is is (P a ⇔ ∃y P y) ⇒ ∃x (P x ⇔ ∃y P y). Let A be the logically valid sentence ∃x (P x ⇔ ∀y P y). Cn for A such that (C1 ∧ · · · ∧ Cn ) ⇒ A is a quasitautology. C4 . From this tableau we read oﬀ the companion sequence C1 .10. (P b ⇒ ∀y P y) ⇒ ∃x (P x ⇒ ∀y P y). C3 . . P a ⇒ ∃y P y. C2 . Find a companion sequence C1 .2. To ﬁnd a companion sequence for A. . . C2 where ′ C1 ′ C2 is is P a ⇒ ∀y P y. 55 . we read oﬀ the companion sequence C1 . ¬ ∃x (P x ⇔ ∃y P y) ¬ (P a ⇔ ∃y P y) / \ Pa ¬Pa ¬ ∃y P y ∃y P y ¬Pa Pb ¬ (P b ⇔ ∃y P y) / \ Pb ¬Pb ¬ ∃y P y ∃y P y From this tableau. Solution.Solution. (P b ⇔ ∃y P y) ⇒ ∃x (P x ⇔ ∃y P y). C3 . C2 . Exercise 3. P b ⇒ ∀y P y. The unsigned tableau ¬ ∃x (P x ⇒ ∀y P y) ¬ (P a ⇒ ∀y P y) Pa ¬ ∀y P y ¬Pb ¬ (P b ⇒ ∀y P y) Pb is closed and shows that A is logically valid.

where A is any quasitautology 2. (∀x B) ⇒ B[x/a] and B[x/a] ⇒ (∃x B) are rules of inference. C3 where ′ C1 ′ C2 ′ C3 is is is (∃y P y) ⇒ P a. (a) Lemma 3. In LH we have the following derivation: 1.3. A ⇒ B.3 Hilbert-Style Proof Systems Let L be a language.2 (soundness of LH ). C2 . (∀x A) ⇒ A[x/a] (by universal instantiation) 56 . 2. A ⇒ (∀x B) and B[x/a] ⇒ A. B is a rule of inference. then A is logically valid. The proof is straightforward by induction on the length of a derivation. LH consists of: 1. The induction step is similar to the proof of Lemma 3. A. 3. provided the parameter a does not occur in A or in B. B. Our Hilbert-style proof system LH for the predicate calculus is as follows: 1. A is a rule of inference. for all L-V sentences A. Proof. (P a ⇔ ∃y P y) ⇒ ∃x (P x ⇔ ∃y P y).3. (∃x B) ⇒ A where a does not occur in A. 4.3. Recall that V is the set of parameters. (∃x B) ⇒ A are rules of inference. P a ⇒ ∃y P y. Example 3. Deﬁnition 3. 4. In other words. (a) (∀x B) ⇒ B[x/a] (universal instantiation) (b) B[x/a] ⇒ (∃x B) (existential instantiation) 3. For each quasitautology A.2. A. A ⇒ B[x/a]. A A⇒B (modus ponens) B A ⇒ B[x/a] (universal generalization) A ⇒ (∀x B) B[x/a] ⇒ A (b) (existential generalization).′ ′ ′ A simpler companion sequence for A is C1 . LH is sound. Schematically. The objects are L-V -sentences.3.4. 5.1 (the system LH ). if A is derivable. 3.

We are assuming that C ⇒ A is derivable. Thus we obtain B by k applications of modus ponens. we are assuming that C ⇒ A is derivable. First. (∀x A) ⇒ (∃x A) is logically valid. B. ((∀x A) ⇒ A[x/a]) ⇒ ((A[x/a] ⇒ (∃x A)) ⇒ ((∀x A) ⇒ (∃x A))) (This is a quasitautology. we see that (¬ A) ⇒ (∀x B) is derivable. LH is closed under quasitautological consequence.5 that A is derivable. 57 . Lemma 3. by modus ponens.3(1). . . namely B[x/a] ⇒ (∀x B). suppose C is of the form 3.2. Thus A1 ⇒ (A2 ⇒ · · · ⇒ (Ak ⇒ B)) is a quasitautology.2. . hence derivable. We are assuming that B is a quasitautological consequence of A1 .2. Applying universal generalization to (i). i.2. .4. if A1 . .3.5 that both (i) (¬ A) ⇒ B[x/a] and (ii) (¬ A) ⇒ (¬ ∀x B) are derivable. then B is derivable. A[x/a] ⇒ (∃x A) (by existential instantiation) 3.. Lemma 3. (B[x/a] ⇒ (∀x B)) ⇒ A is derivable. by Lemma 3. . where a does not occur in A.3. We are also assuming that A1 . .3.3(4). B[x/a] ⇒ (∃x B) (by existential instantiation) 2. From this plus (ii).3(2). 4. A is derivable. (A[x/a] ⇒ (∃x A)) ⇒ ((∀x A) ⇒ (∃x A)) (from 1.3. and if C ⇒ A is derivable in LH . and modus ponens) Thus.3. then A is derivable in LH .3(3) or 3. where C is of the form 3. and modus ponens) 5. 2. (∀x A) ⇒ (∃x A) (from 2.2. suppose C is of the form 3. Proof. By universal instantiation. It follows by Lemma 3. . obtained from the tautology (p ⇒ q) ⇒ ((q ⇒ r) ⇒ (p ⇒ r)). Example 3. In addition. C is derivable. . Ak are derivable. In LH we have the following derivation: 1. (∃x (A ∧ B)) ⇒ (∃x B) is logically valid. The other cases. and modus ponens) 4. .3. namely (∀x B) ⇒ B[x/a]. Hence. We now turn to the proof that LH is complete.6. If C is a companion of A. .5. (A ∧ B)[x/a] ⇒ (∃x B) (from 1. Ak . Ak are derivable. it follows by Lemma 3. . In other words. and if B is a quasitautological consequence of A1 . (∃x (A ∧ B)) ⇒ (∃x B) (from 3 and existential generalization) Thus. Next.2.3. Proof.2. . are handled similarly. .e.) 4. (B[x/a] ⇒ (∃x B)) ⇒ ((A ∧ B)[x/a] ⇒ (∃x B)) (a quasitautology) 3. by Lemma 3. . 3. . Ak .

A is derivable if and only if A is logically valid. For the “if” part.5. Construct a Hilbert-style proof of ¬ ∃x ∀y (Eyx ⇔ ¬ Eyy). existential generalization.5. . Cn for A such that (C1 ∧ · · · ∧ Cn ) ⇒ A is a quasitautology.QT Exercise 3. assume that A is logically valid.QT 8.7. Exercise 3.QT 2. 3.7.2. 6. Ak . For convenience in writing proofs.7 (completeness of LH ). . By Theorem 3. 5.3. 8.6.10. existential instantiation. This is justiﬁed by Lemma 3. universal generalization.EG Exercise 3. . A proof in LH is 1. 5.2. . for all L-V -sentences A. EG to indicate universal instantiation. Similarly we use UI.8. .3. 7. (∀y Ray) ⇒ Rab Rab ⇒ (∃x Rxb) (∀y Ray) ⇒ (∃x Rxb) (∀y Ray) ⇒ (∀y ∃x Rxy) (∃x ∀y Rxy) ⇒ (∀y ∃x Rxy) UI EI 1. From this and n applications of Lemma 3. (∀x (P x ∧ Qx)) ⇒ (P a ∧ Qa) (∀x (P x ∧ Qx)) ⇒ P a (∀x (P x ∧ Qx)) ⇒ ∀x P x (∀x (P x ∧ Qx)) ⇒ Qa (∀x (P x ∧ Qx)) ⇒ ∀x Qx (∀x P x) ⇒ P a (∀x Qx) ⇒ Qa ((∀x P x) ∧ (∀x Qx)) ⇒ (P a ∧ Qa) ((∀x P x) ∧ (∀x Qx)) ⇒ ∀x (P x ∧ Qx) (∀x (P x ∧ Qx)) ⇔ ((∀x P x) ∧ (∀x Qx)) UI 1. hence derivable. 2. Remark 3.3. 4.11. A proof in LH is 1. there exists a companion sequence C1 .Theorem 3.QT 4. .UG 1.UG 3. . respectively. Hence Cn ⇒ (Cn−1 ⇒ · · · ⇒ (C1 ⇒ A)) is a quasitautology. 3. In other words. Proof. 4.9. UG.3.UG 4. 58 .3. LH is sound and complete. 10.2. Construct a Hilbert-style proof of the sentence (∃x ∀y Rxy) ⇒ (∀y ∃x Rxy).3. for quasitautology.QT 3. EI. we supplement the rules of LH with A1 · · · Ak B whenever B is a quasitautological consequence of A1 . Solution.3. we obtain derivability of A. 2. 9. The “only if” part is Lemma 3.9. We indicate applications of this rule by QT.3.UG UI UI 6. Construct a Hilbert-style proof of the sentence (∀x (P x ∧ Qx)) ⇔ ((∀x P x) ∧ (∀x Qx)) Solution. .

3. ⇒ . The objects of LH ′ are L-V -sentences containing only ∀. where a does not occur in B.EG 4.QT 2.QT Exercise 3. 3. (P a ⇒ ∀y P y) ⇒ ∃x (P x ⇒ ∀y P y) (¬ ∃x (P x ⇒ ∀y P y)) ⇒ P a (¬ ∃x (P x ⇒ ∀y P y)) ⇒ ∀y P y ∃x (P x ⇒ ∀y P y) EI 1.12. Solution..EG 3.3. Solution.3. A proof in LH is 1. Construct a Hilbert-style proof of ∃x (P x ⇒ ∀y P y). 4.e. The rules of LH ′ are: (a) quasitautologies (b) (∀x B) ⇒ B[x/a] (c) (∀x (A ⇒ B)) ⇒ (A ⇒ ∀x B) (d) (e) A A⇒B (modus ponens) B B[x/a] (generalization). Exercise 3.QT 3.13. A proof in LH is 1.QT 2.QT Exercise 3.14. 2.UG 1. Construct a Hilbert-style proof of ∃x (P x ⇔ ∀y P y). Construct a Hilbert-style proof of the sentence ¬ ∃x (Sx ∧ ∀y (Eyx ⇔ (Sy ∧ ¬ Eyy))).3. 4.15. ∀y (Eya ⇔ ¬ Eyy) ⇒ (Eaa ⇔ ¬ Eaa) ¬ ∀y (Eya ⇔ ¬ Eyy) (∀y (Eya ⇔ ¬ Eyy)) ⇒ ¬ ∃x ∀y (Eyx ⇔ ¬ Eyy) (∃x ∀y (Eyx ⇔ ¬ Eyy)) ⇒ ¬ ∃x ∀y (Eyx ⇔ ¬ Eyy) ¬ ∃x ∀y (Eyx ⇔ ¬ Eyy) UI 1. ∨ ). 2. A proof in LH is 1. ⇔ . Consider the following proof system LH ′ . 59 . ∀x B Show that LH ′ is sound and complete. 4.QT Exercise 3.QT 2. 5. ∧ . ¬ (i. 3. which is a “stripped down” version of LH . not containing ∃.3. (∀y (Eya ⇔ (Sy ∧ ¬ Eyy))) ⇒ (Eaa ⇔ (Sa ∧ ¬ Eaa)) (Sa ∧ (∀y (Eya ⇔ (Sy ∧ ¬ Eyy)))) ⇒ (Sb ∧ ¬ Sb) (∃x (Sx ∧ (∀y (Eyx ⇔ (Sy ∧ ¬ Eyy))))) ⇒ (Sb ∧ ¬ Sb) ¬ ∃x (Sx ∧ (∀y (Eyx ⇔ (Sy ∧ ¬ Eyy)))) UI 1. 2. 3.Solution.

by modus ponens. Hence. Since LH (S) includes LH . 60 . we can prove soundness and completeness of the restricted tableau method with ∀. Show that an L-V sentence B is derivable in LH (S) if and only if B is a logical consequence of S. it follows that B is a logical consequence of a ﬁnite subset of S.16. Also. B. . where a does not occur in A. An are derivable in LH (S). It remains to prove the following lemma: If C is a companion of A. ¬ . say A1 . (A1 ∧ · · · ∧ An ) ⇒ B is derivable in LH . . This completes the proof. (∀x ((¬ A) ⇒ B)) ⇒ ((¬ A) ⇒ ∀x B) is derivable in LH ′ . we see that ∀x ((¬ A) ⇒ B) is derivable in LH ′ . A ∈ S. (¬ A) ⇒ ∀x B is derivable in LH ′ . because we need to know that the parameter a does not occur in S. Let S be a set of L-sentences. the ones involving ∀. and from this we obtain the restricted Companion Theorem. Hence. we have that (A1 ∧ · · · ∧ An ) ⇒ B is derivable in LH (S). 1. Consider a proof system LH (S) consisting of LH with additional rules of inference A . ⇒ . Hence (A1 ∧ · · · ∧ An ) ⇒ B is logically valid. By induction on the length of derivations. by rule (c) of LH ′ . It follows quasitautologically from this and (2) that A is derivable in LH ′ . assume B is a logical consequence of S. then A is derivable in LH ′ . 1. . But A1 . . This completes the proof. The assumption that S is a set of L-sentences (not L-V -sentences) is used in the inductive steps corresponding to rules 4(a) and 4(b). From (1) and the generalization rule (e) of LH ′ . 2. There are now only two kinds of companions. . It follows quasitautologically that B is derivable in LH (S). An .3. Exercise 3. Indicate the modiﬁcations needed when S is a set of L-V -sentences. Solution. Soundness is proved just as for LH . and if C ⇒ A is derivable in LH ′ . universal and existential generalization. . . Just as for the full tableau method. Conversely. By the Compactness Theorem. Assume that (B[x/a] ⇒ (∀x B)) ⇒ A is derivable in LH ′ . by completeness of LH .Solution. it is straightforward to prove that each sentence derivable in LH (S) is a logical consequence of S. Consider a companion of the form B[x/a] ⇒ (∀x B). . It follows quasitautologically that both (1) (¬ A) ⇒ B[x/a] and (2) (¬ A) ⇒ ¬ ∀x B are derivable in LH ′ .

.. . . B is true in all L-structures satisfying S) if and only if there exists a ﬁnite set of L-sentences A1 . . . . Corresponding to signed tableau rules of the form . Ak ∈ S such that A1 . but allowing parameters from V ∪ V ′ . Ak ).2. . . .3. . etc. Let V ′ be a countably inﬁnite set of new parameters. .e. disjoint from V . . a block tableau starting with S is a block tableau generated from S by means of block tableau rules. .4. Notation 3. . . In rules 4(a) and 4(b). . . . Ak ⊢ B.2. Ak } ⊂ S such that A1 . Before presenting our Gentzen-style proof system for L.17. The block tableau rules are obtained from the signed tableau rules (pages 13 and 31) as follows. . S ∪ {¬ B} is not satisﬁable. The “if” part follows from the soundness of our proof system. .4 Gentzen-Style Proof Systems Throughout this section. Exercise 3. . we ﬁrst discuss the block tableau method. Notation 3. . X .18. X instead of S ∪ {X}. . Ak . The objects are L-V ∪V ′ -sentences.. . . one must impose the restriction that a does not occur in A.e.3. . we write S. B is provable from A1 . Prove that S |= B (i. X . . If S is a set of L-V -sentences.4. X . . . S. It follows by the Compactness Theorem that there exists a ﬁnite set {A1 . . We write S ⊢ B to indicate that B is derivable in LH (S). . . A block is a ﬁnite set of signed L-V -sentences. . Ak ⊢ B (i. It follows by completeness of our proof system that A1 . a trivial variant of the signed tableau method. . . and let B be an Lsentence. Deﬁnition 3.. Deﬁnition 3. Solution. . .3. i. X . 3.1. As usual. Thus B is a logical consequence of A1 . . . Given a block S. | Y . A block tableau is said to be closed if each of its end nodes is closed. With this modiﬁcation. Ak .e. . Let S be an inﬁnite set of L-sentences. assume that S |= B. V is the set of parameters. . | Y1 Y2 61 Y . A block is said to be closed if it contains T A and F A for some L-V -sentence A. / \ Y1 Z1 Y2 Z2 . Deﬁne LH (S) as before. . For the “only if” part. ¬ B is not satisﬁable. . . B. let L be a language. / \ Z .4. everything goes through as before. A block tableau is a rooted dyadic tree where each node carries a block. we modify our system as follows. If S is a block and X is a signed L-V -sentence.

F ∀x A. X. F Bn is a block. If S = T A1 . F B S. X | S. For example. . TB S.5. . We exhibit a closed block tableau demonstrating that ∃x A is a logical consequence of ∀x A. . we have the following block tableau rules: S. F B S. T A ⇒ B. . T ∀x A. T A[x/a] S. T A ∧ B | S. let |S| be the sequent A1 . Y1 . .we have block tableau rules S. FA S. F A ⇒ B. 62 . We now deﬁne our Gentzen-style system. Bn . F A ∧ B. Deﬁnition 3. Z2 S. . Example 3. A sequent is an expression of the form Γ → ∆ where Γ and ∆ are ﬁnite sets of L-V -sentences. . T A ⇒ B. Z1 . T A ⇒ B / \ S. F A[x/a] This block tableau is of course similar to the signed tableau T ∀x A F ∃x A T A[x/a] F A[x/a] which demonstrates the same thing. F A[x/a] where a is new. LG. T A. . This gives a one-to-one correspondence between blocks and sequents. F A ⇒ B | S. Z S. F A ∧ B. F A ∧ B / \ S. X. X | S. T A. T A ∧ B. . F ∃x A | T ∀x A. X. T Am . . X. T ∀x A | S.4. . F ∃x A. F ∀x A | S. Y2 S. T ∀x A. . T A[x/a] | T ∀x A. X. F A S. Y S. T B S. . . Y respectively. F B1 . Y1 . . X / \ S.4. X / \ S.4. Am → B1 . . F ∃x A. X. Y2 S. T A[x/a].

Deﬁnition 3.4.6 (the system LG). Our Gentzen-style proof system LG for the predicate calculus is as follows. 1. The objects of LG are sequents.2 2. For each closed block S, we have a rule of inference |S| . In other words, for all ﬁnite sets of L-V -sentences Γ and ∆ and all L-V -sentences A, we assume the sequent Γ, A → A, ∆. 3. For each non-branching block tableau rule S | S′ we have a rule of inference |S ′ |, |S| , i.e., 4. For each branching block tableau rule S / \ S ′ S ′′ we have a rule of inference |S ′ |, |S ′′ |, |S| , i.e., Thus LG includes the following rules of inference: Γ, A → A, ∆ Γ, ¬ A → A, ∆ Γ, ¬ A → ∆ Γ, A ∧ B, A, B → ∆ Γ, A ∧ B → ∆ Γ, A → ¬ A, ∆ Γ → ¬ A, ∆ Γ → A ∧ B, A, ∆ Γ → A ∧ B, B, ∆ Γ → A ∧ B, ∆ Γ, A → A ⇒ B, B, ∆ Γ → A ⇒ B, ∆ |S ′ | |S ′′ | . |S| |S ′ | . |S|

Γ, A ⇒ B → A, ∆ Γ, A ⇒ B, B → ∆ Γ, A ⇒ B → ∆ Γ, ∀x A, A[x/a] → ∆ Γ, ∀x A → ∆

Γ → ∀x A, A[x/a], ∆ Γ → ∀x A, ∆ where a does not occur in the conclusion.

2 For

this reason, LG is sometimes called a sequent calculus.

63

Exercise 3.4.7. Explicitly display the remaining inference rules of LG. Deﬁnition 3.4.8. A sequent A1 , . . . , Am → B1 , . . . , Bn is said to be logically valid if and only if the L-V -sentence (A1 ∧ · · · ∧ Am ) ⇒ (B1 ∨ · · · ∨ Bn ) is logically valid.3 Theorem 3.4.9 (soundness and completeness of LG). LG is sound and complete. In other words, a sequent Γ → ∆ is logically valid if and only if it is derivable in LG. In particular, an L-V -sentence A is logically valid if and only if the sequent →A is derivable in LG. Proof. Note that the sequent A1 , . . . , Am → B1 , . . . , Bn is logically valid if and only if the block T A1 , . . . , T Am , F B1 , . . . , F Bn is not satisﬁable. Thus, soundness and completeness of LG is equivalent to soundness and completeness of the block tableau method. The latter is in turn easily seen to be equivalent to soundness and completeness of the signed tableau method, as presented in Theorems 2.3.13 and 2.5.5. Exercise 3.4.10. Construct a Gentzen-style proof of the sequent ∃x ∀y Rxy → ∀y ∃x Rxy. Solution. A proof in LG is 1. 2. 3. 4. 5. ∃x ∀y Rxy, ∀y Ray, Rab → Rab, ∃x Rxb, ∀y ∃x Rxy ∃x ∀y Rxy, ∀y Ray, Rab → ∃x Rxb, ∀y ∃x Rxy ∃x ∀y Rxy, ∀y Ray → ∃x Rxb, ∀y ∃x Rxy ∃x ∀y Rxy → ∃x Rxb, ∀y ∃x Rxy ∃x ∀y Rxy → ∀y ∃x Rxy

Deﬁnition 3.4.11. In order to simplify the writing of Gentzen-style proofs, let LG + be LG augmented with the so-called weakening rules or padding rules: Γ→∆ Γ, A → ∆ Γ→∆ Γ → A, ∆

where A is an L-V -sentence. Clearly LG + is sound and complete. Remark 3.4.12. Any proof in LG is a proof in LG + , and any proof in LG + may be straightforwardly “padded out” to a proof in LG. Thus LG + diﬀers only slightly from LG. However, proofs in LG+ are easier. For example, patterned on the above proof in LG, we have the following proof in LG + :

3 In particular, the sequents A , . . . , A m → and → B1 , . . . , Bn are said to be logically 1 valid if and only if the L-V -sentences ¬ (A1 ∧ · · · ∧ Am ) and B1 ∨ · · · ∨ Bn are logically valid, respectively. The empty sequent → is deemed not logically valid.

64

1. 1.5. 2. 2.5. 3. 3.5. 4. 4.5. 5.

Rab → Rab ∀y Ray, Rab → Rab ∀y Ray → Rab ∀y Ray → Rab, ∃x Rxb ∀y Ray → ∃x Rxb ∃x ∀y Rxy, ∀y Ray → ∃x Rxb ∃x ∀y Rxy → ∃x Rxb ∃x ∀y Rxy → ∃x Rxb, ∀y ∃x Rxy ∃x ∀y Rxy → ∀y ∃x Rxy

or, omitting the applications of the padding rules, 1. 2. 3. 4. 5. Rab → Rab ∀y Ray → Rab ∀y Ray → ∃x Rxb ∃x ∀y Rxy → ∃x Rxb ∃x ∀y Rxy → ∀y ∃x Rxy

Exercise 3.4.13. Construct a Gentzen-style proof of the sequent → (∃x ∀y Rxy) ⇒ (∀y ∃x Rxy). Solution. A proof in LG + consists of the previous proof followed by 5.5. 6. ∃x ∀y Rxy → ∀y ∃x Rxy, (∃x ∀y Rxy) ⇒ (∀y ∃x Rxy) → (∃x ∀y Rxy) ⇒ (∀y ∃x Rxy)

Exercise 3.4.14. Construct a Gentzen-style proof of the sequent → ¬ ∃x (Sx ∧ ∀y (Eyx ⇔ (Sy ∧ ¬ Eyy))). Solution. A proof in LG + with padding rules omitted is 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. Eaa → Eaa Eaa, ¬ Eaa → Eaa, Sa ∧ ¬ Eaa → Sa → Sa → Eaa, ¬ Eaa Sa → Eaa, Sa ∧ ¬ Eaa Sa, Eaa ⇔ (Sa ∧ ¬ Eaa) → Sa, ∀y (Eya ⇔ (Sy ∧ ¬ Eyy)) → Sa ∧ ∀y (Eya ⇔ (Sy ∧ ¬ Eyy)) → ∃x (Sx ∧ ∀y (Eyx ⇔ (Sy ∧ ¬ Eyy)) → → ¬ ∃x (Sx ∧ ∀y (Eyx ⇔ (Sy ∧ ¬ Eyy)) axiom from 1 from 2 axiom from 1 from 4 and 5 from 4 and 6 from 7 from 8 from 9 from 10

Exercise 3.4.15. Construct a Gentzen-style proof of ∃x (P x ⇒ ∀y P y). Solution. A proof in LG + with padding rules omitted is 65

6. X | S. ∃x (P x ⇒ ∀y P y) → P a ⇒ ∀y P y.1 (the Interpolation Theorem). Let A and B be L-V -sentences. Write out all the rules of LG ′ explicitly.5.4. X | S. Y2 by a pair of rules S. P a ⇒ ∀y P y → P a. Show that LG(atomic) is sound and complete. We indicate this by writing I A ⇒ B. ∆ is assumed only for atomic L-V -sentences A. X. If A ⇒ B is logically valid. we can ﬁnd an L-V -sentence I such that: 1. 1. A → A.3. Y1 .17. 2. Construct a Gentzen-style proof of ∃x (P x ⇔ ∀y P y).1. Y1 S.5. 3.16. X. 4.7. Let LG ′ be the variant of LG corresponding to the modiﬁed block tableau rules. Exercise 3. Pa → Pa → P a. A ⇒ I and I ⇒ B are logically valid. Let LG(atomic) be a variant of LG in which Γ. 3.5 The Interpolation Theorem As usual. Each predicate and parameter occurring in I occurs in both A and B.4. Show that LG ′ is sound and complete. ∃x (P x ⇒ ∀y P y) → ∀y P y. The modiﬁed block tableau rules are a variant of the block tableau rules of Deﬁnition 3. Such an I is called an interpolant for A ⇒ B. 5. 2. X. replacing each non-branching rule of the form S. Y2 Show that the modiﬁed block tableau rules are sound and complete. 66 . let L be a language and let V be the set of parameters.) Exercise 3.4. ∃x (P x ⇒ ∀y P y) → ∃x (P x ⇒ ∀y P y) axiom from 1 from 2 from 3 from 4 from 5 Exercise 3.4. (Hint: Use the result of Exercise 2. X | S.18. Theorem 3. 2.

T Dn . . we modify the conclusion of the theorem to say that at least one of ¬ A and B is logically valid.5. we introduce a “symmetric” variant of LG.4 The conclusion is obvious in this case. . .Remark 3.e. We have M. . A signed variant of C1 . Cm → D1 . changing signs. . In order to prove the Interpolation Theorem. X → X. The objects are signed sequents. T Cm → T D1 . . N for all X. A signed sequent is an expression of the form M → N where M and N are ﬁnite sets of signed L-V -sentences. LG(symmetric) is the following proof system. We deﬁne a signed sequent to be logically valid if and only if the corresponding unsigned sequent is logically valid. .. . . X. wherein sentences do not move from one side of → to the other. . . .5. 5 An unsigned sequent is just what we have previously called a sequent. T A = F A. A variant of M → N is a signed sequent obtained from M → N by transferring sentences from one side of → to the other. . then obviously the theorem is incorrect as stated. 2. Deﬁnition 3. 4 This 67 .5. . If A and B have no predicates in common. because all L-V -sentences necessarily contain at least one predicate. Cm → D1 . 1. N and M. . where we use an overline to denote conjugation.3. . N are variants of each other. .2. F A = T A. . amounts to saying that at least one of the truth values T and F is an interpolant for A ⇒ B. M. . Let C1 . X. . X → N and M → X.5. In this case. Deﬁnition 3.4. . . . X → N and M → X. . Deﬁnition 3. Note that each signed sequent is a signed variant of one and only one unsigned sequent. i. Dn be an unsigned sequent5 . In particular. Dn is any variant of the signed sequent T C1 .5.

. Soundness and completeness of LG(symmetric) follows from Theorem 3. Proof. X. / \ Z . X. N 68 . In other words.7. then so are all signed variants of the conclusion. Z → N M. N M → X.3.5. . An unsigned sequent is derivable in LG if and only if all of its signed variants are derivable in LG(symmetric). X → N M. The proof is by induction on the length of derivations in LG. X . Y1 . . X . soundness and completeness of LG. LG(symmetric) is sound and complete. if all signed variants of the premises are derivable in LG(symmetric). hence derivable in LG(symmetric). Y . Z1 . . Z2 . . Z2 → N M. We now prove the Interpolation Theorem. N M → X. Y → N M. . Y1 .6. Z1 . .4. X → N or M → X. . X. Y1 . For each signed tableau rule of the form . Y → N M. for each rule of inference of LG. Lemma 3. Y2 . X → N respectively. | Y1 Y2 Y .9. N M → X. .5. Y2 → N M. Theorem 3.5. . A → A. . X . In particular. | Y . . N M → X. N M → X. . N M → X. Proof. X. Y . X. X . N M → X. The base step consists of noting that all signed variants of Γ. N M → X. using Lemma 3. . N M → X. an L-V -sentence A ⇒ B is logically valid if and only if the signed sequent T A → T B is derivable in LG(symmetric). X. N or M. X. Y2 . X → N M.6. N . X. X → X. X → N M. The inductive step consists of checking that. . / \ Y1 Z1 Y2 Z2 we have a corresponding pair of signed sequent rules M. a signed sequent is logically valid if and only if it is derivable in LG(symmetric). . Y2 → N M. ∆ are of the form M. Y1 . Z. This is straightforward. M → X.

X. for each rule of LG(symmetric).Deﬁnition 3. FB. F A. I I I 69 . F A ∧ B. F A → N M. and X that M. N M → F A ∧ B. we require instead that at least one of the signed sequents M → and → N be logically valid. N 6 In the special case when M and N have no predicates in common. F A.5. X.8. F ¬ A. An interpolant for M → N is an L-V -sentence I such that the signed sequents M → T I and T I → N are logically valid. F A ∧ B. F A[x/a] → N M. N I ∧J I M → T A ∧ B. F A ∧ B −→ N M → T A ∧ B. For the induction step we show that. T ¬ A. F B → N I ∨J J I I M. and all predicates and parameters occurring in I occur in both I M and N . we can ﬁnd an interpolant for the conclusion. T A[x/a]. N M → F ¬ A. N M → T ¬ A. This amounts to requiring that at least one of T. N F T and M. T A.6 We indicate this by writing M → N . We present some representative special cases. we note that X is an interpolant for M. For the base step. given interpolants for the premises of the rule. M. it suﬃces by Theorem 3. X → X. X → and → X. N M. T A ∧ B → N M. F ¬ A → N I I I I M → F ¬ A. T B → N M. N I I I I I M. N M → T ¬ A. In order to prove the Interpolation Theorem. T A → N M.7 to prove that every signed sequent derivable in LG(symmetric) has an interpolant. Let M → N be a signed sequent. N . F A → N I I I M → F A ∧ B. Thus we have M. N J M −→ T A ∧ B. T A ∧ B. T A. N M. F ∀x A → N where a does not occur in the conclusion. X → X. N M → T ∀x A. X → N and M → X. T A. N . M → T ∀x A. N are logically valid. T ¬ A → N M. F ∀x A. We prove this by induction on the length of derivations. F be an interpolant for M → N . X.5. X. T B.

T A[x/a] → N M. otherwise K = ∀z I[a/z] where z is a new variable. R is symmetric.. The obvious interpolant I says there exists a clique of size n. A contains a unary predicate Q denoting a clique. T ∀x A. Let n be a large positive integer. ∀x ∀y (Rxy ⇒ ¬ Exy). N . Clearly A ⇒ B is logically valid. Example 3. T ∀x A → N where K = I if a occurs in M. . A is: R and G are irreﬂexive relations on the universe. . . . This completes the proof. . It appears that there is no interpolant of length O(n). and there exist x1 . proportional to n2 . say n = 1000. G is transitive. B contains a binary predicate E saying that two vertices get the same color. xn−1 such that ∀y (Ex1 y ∨ · · · ∨ Exn−1 y). . and there exist x1 .. M. N where K = I if a occurs in F ∀x A. Let B say that the graph is not (n − 1)-colorable. T ∀x A. . i. there exist x1 . Note that the length of I is O(n2 ). i.e..9. . F A[x/a]. ∀x ∀y ((Qx ∧ Qy ∧ Gxy) ⇒ Rxy). undirected graph with a clique of size n. otherwise K = ∃z I[a/z] where z is a new variable. . Let A say that the universe consists of the vertices of a simple. N M → F ∀x A. . Note that the lengths of A and B are O(n). xn such that Rx1 x2 ∧ Rx1 x3 ∧ · · · ∧ Rxn−1 xn . K I K I 70 .where a does not occur in the conclusion.e. proportional to n.e. B is the negation of: E is an equivalence relation on the universe. . We give an example illustrating the Interpolation Theorem. M → F ∀x A. i. . xn such that Qx1 ∧ · · · ∧ Qxn ∧ Gx1 x2 ∧ · · · ∧ Gxn−1 xn . Both A and B contain a predicate R denoting adjacency in the graph.5.

IM = { a. For each n-ary predicate P of L. An L-structure M is said to be normal if the identity predicate denotes the identity relation. These extensions may be regarded as inessential features or “bells and whistles” which are introduced solely in order to make the predicate calculus more user-friendly. we have an axiom ∀x1 · · · ∀xn ∀y1 · · · ∀yn ((Ix1 y1 ∧ · · · ∧ Ixn yn ) ⇒ (P x1 · · · xn ⇔ P y1 · · · yn )) (congruence). Deﬁnition 4. The identity axioms for L are the following sentences: 1.Chapter 4 Extensions of Predicate Calculus In this chapter we consider various extensions of the predicate calculus. Let L be a language with identity. then ∀x ∀y (I1 xy ⇔ I2 xy) is a logical consequence of the identity axioms for I1 and I2 . If L contains two identity predicates I1 and I2 . Let L be a language with identity.2.1.1..e. ∀x ∀y (Ixy ⇔ Iyx) (symmetry) 3. 4. Exercise 4.1. I. a | a ∈ UM }. designated as the identity predicate.1.1 Predicate Calculus with Identity Deﬁnition 4. ∀x Ixx (reﬂexivity) 2.3. ∀x ∀y ∀z ((Ixy ∧ Iyz) ⇒ Ixz) (transitivity) 4. Deﬁnition 4. A language with identity consists of a language L with a particular binary predicate.1. i. Show that the identity predicate is unique in the following sense.4. 71 .

Theorem 4. Proof. applied to the set S ∪ {identity axioms for L}. we exhibit a sentence En which is normally satisﬁable in domains of cardinality n but not in domains of any other cardinality. then we have a normal L-structure M satisfying the same sentences as M . . Given a positive integer n. 4}. Let L be the language consisting of f . Let M be an L-structure satisfying the identity axioms for L.7. gM are the binary operations of addition and multiplication modulo 5. Conversely. For each a ∈ UM put a = {b ∈ UM | vM (Iab) = T}. Exercise 4. Write an L-sentence A such that for all normal L-structures M ′ . Let M = (UM . P ∈ L. Corollary 4.1.1. an ∈ PM } for all n-ary predicates P .5. 3. En says that there exist exactly n things. 1. Deﬁnition 4. Thus M is essentially just the ring of integers modulo 5. Let S be a set of L-sentences.Note that any normal L-structure automatically satisﬁes the identity axioms for L.7. Then we have a normal L-structure M and an onto mapping φ : UM → UM as in Theorem 2. using the fact that IM is a congruence with respect to each of the relations PM . gM . 2. Proof. S is normally satisﬁable if and only if S ∪ {identity axioms for L} is satisﬁable. S is normally satisﬁable if and only if each ﬁnite subset of S is normally satisﬁable. IM ) where UM = {0. . Proof.1.1. and fM . .1.5 and 2. I.11. This is immediate from Theorems 4. an | a1 . We also have the Compactness Theorem for normal satisﬁability: Corollary 4. IM is the identity relation on UM . g. This is straightforward.9.1.6. The sentence ∃x1 · · · ∃xn (∀y (Ix1 y ∨ · · · ∨ Ixn y) ∧ ¬ (Ix1 x2 ∨ Ix1 x3 ∨ · · · ∨ Ixn−1 xn )) has this property.3. .3.10.7.8 plus the Compactness Theorem for predicate calculus without identity (Theorems 2.1. Intuitively.1.6. we have: Theorem 4.1.1 and 2. If M is an L-structure satisfying the identity axioms for L. Regarding normal satisﬁability in particular domains.8. This is immediate from Corollary 4. . Note that M is a normal L-structure. . 72 . deﬁned by putting UM = {a | a ∈ UM }. S is normally satisﬁable if there exists a normal L-structure which satisﬁes S. . and PM = { a1 . we have: Example 4.6. . fM .2). M ′ satisﬁes A if and only if M ′ is isomorphic to M .

.1 and 2. On the other hand. . by Corollary 4. by the Compactness Theorems 2. Namely. For the second part. But any normal L-structure satisfying S ∗ satisﬁes S and has an inﬁnite domain. Let S ∗ = S ∪ {∃x Qi x | i ∈ X} ∪ {¬ ∃x (Qi x ∧ Qj x) | i. Another solution is to let A be a sentence describing a ﬁeld consisting of 5 elements.6. Solution. DP is the conjunction of P xi1 · · · xin for each n-tuple ai1 . up to isomorphism. then S is normally satisﬁable in all inﬁnite domains of cardinality ≥ the cardinality of S. . there is exactly one ﬁeld of 5 elements. Proof.1. . . It follows that each ﬁnite subset of S ∗ is satisﬁable. Hence. M ′ |= A if and only if M ′ is isomorphic to M . . j. Since S is normally satisﬁable in arbitrarily large ﬁnite domains.1. . j ∈ X. ain ∈ PM . and let P. 1.2.1. and X is a set of cardinality κ. let κ be a cardinal number ≥ the cardinality of S.Solution. and let M be a ﬁnite normal L-structure. . i ∈ X. Let a1 . .6. .1. . . Exercise 4. 2.9. any domain in which S ∗ is satisﬁable will contain pairwise distinct elements ai . . . Thus S ∗ is a set of L∗ -sentences of cardinality κ. we have: Theorem 4. . Construct an L-sentence A such that. for all normal L-structures M ′ . 3. For the ﬁrst part. If S is normally satisﬁable in some inﬁnite domain. . 4. 2. o 1. Let L be a ﬁnite language with identity. Furthermore. and will therefore have cardinality ≥ κ. By assumption. S ∗ is satisﬁable in a domain of 73 .7. S ∗ is satisﬁable. A brute force solution is to let A be ∃x0 ∃x1 ∃x2 ∃x3 ∃x4 B. As A we may take ∃x1 · · · ∃xk (DP ∧ · · · ∧ DQ ∧ ∀y (Ix1 y ∨ · · · ∨ Ixk y)) where for each n-ary predicate P of L. Hence. We are using the algebraic fact that. where each Qi is a new 1-ary predicate. and ¬ P xi1 · · · xin for each n-tuple / ai1 . k ranging over 0. S is normally satisﬁable in some inﬁnite domain. each ﬁnite subset of S ∗ is normally satisﬁable. ak be the elements of UM . by part 2 of Theorem 2. If S is normally satisﬁable in arbitrarily large ﬁnite domains. i = j} ∪ {identity axioms for L∗ }. .12. S ∗ is normally satisﬁable. let S ∗ = S ∪ {Hn | n = 1. . 2. Let L∗ = L ∪ {Qi | i ∈ X}. where B is the conjunction of {∀y (Ix0 y ∨ Ix1 y ∨ Ix2 y ∨ Ix3 y ∨ Ix4 y)} ∪ {¬ Ixi xj : i = j} ∪ {If xi xj xk | i + j = k mod 5} ∪ {Igxi xj xk | ij = k mod 5} with i. Hence.} where Hn is the sentence ∃x1 · · · ∃xn ¬ (Ix1 x2 ∨ Ix1 x3 ∨ · · · ∨ Ixn−1 xn ) saying that there exist at least n things. then S is normally satisﬁable in some inﬁnite domain.13 (L¨wenheim/Skolem Theorem). . let A be the conjunction of the ﬁeld axioms plus “there exist exactly 5 things”. ain ∈ PM . Q be the predicates of L.

let Mn be the normal L-structure Mn = (Un . Solution. . In ) where Un = {1. k ∈ (Un )3 | i + j = k} Qn = { i. Then M normally satisﬁes S and UM is of cardinality κ.6. Let M ∗ be a normal L∗ -structure with UM ∗ of cardinality κ.14.cardinality κ..1. i. k ∈ (Un )3 | i × j = k} Rn = { i.e. . Therefore. Pn . As Z we may take the conjunction of the following clauses. j ∈ (Un )2 | i < j} Sn = { i. (a) ∀x ∀y (Rxy ∨ Ryx ∨ Ixy) (b) ∀x ∀y (Rxy ⇒ ¬ Ryx) (c) ∀x ∀y ∀z ((Rxy ∧ Ryz) ⇒ Rxz) (d) ∀x ∀z (Sxz ⇔ (Rxz ∧ ¬ ∃y (Rxy ∧ Ryz))) (e) ∀u (Ou ⇔ ¬ ∃x Rxu) 74 . Exercise 4. j ∈ (Un )2 | i + 1 = j} In = { i. On . M is the L-structure with UM = UM ∗ and PM = PM ∗ for each predicate P in L. hence = κ. for all ﬁnite normal L-structures M ′ . .1.5 and 4. n} On = {1} Pn = { i. Let M be the reduct of M ∗ to L. Sn . by Theorems 4. j ∈ (Un )2 | i = j} Exhibit an L-sentence Z such that. Qn . j. j. .1. S ∗ is normally satisﬁable in a domain of cardinality ≤ κ. M ′ |= Z if and only if M ′ is isomorphic to Mn for some n. Rn . Let L be the following language: Ox: x = 1 P xyz: x + y = z Qxyz: x × y = z Rxy: x < y Sxy: x + 1 = y Ixy: x = y (identity predicate) For each positive integer n.

Let S be the set of L-sentences A with the following property: there exists n = nA such that for all primes p > nA . Let A be a sentence of the predicate calculus with identity.14. by induction along R.2. in terms of S. . Then Mp satisﬁes A1 . Mp satisﬁes A. let S0 = {A1 . for each positive integer k. To see this.9). The spectrum of A is the set of positive integers n such that A is normally satisﬁable in a domain of cardinality n. in terms of S and P . . among all sets of positive integers. Exercise 4. Let p be any prime > n. Among the sentences of S are those asserting that the universe has at least k elements. Clauses (f) and (g) deﬁne the addition predicate P . . . Ak .1. . Let L and Mn be as in Exercise 4. M∞ is inﬁnite. Put n = max(nA1 . it is unknown whether the complement of a spectrum is necessarily a spectrum. We show that the set {n ≥ 1 | n is even} is a spectrum. as usual in computational number theory. This proves our claim. We claim that every ﬁnite subset of S is normally satisﬁable. .2. (b) and (c) say that R is an irreﬂexive linear ordering of the universe. The spectrum problem is the problem of characterizing the spectra. . Example 4. it follows that S is normally satisﬁable.3.1. Clauses (h) and (i) deﬁne the multiplication predicate Q.) Solution. Let M∞ be a normal L-structure satisfying S. if Mp |= A for all suﬃciently large primes p. . we have k = [log2 n]. by induction along R. By the Compactness Theorem for normal satisﬁability (Corollary 4. Clause (d) says that S is the immediate successor relation.1. A spectrum is a set X of positive integers. . Clause (e) says that 1 is the ﬁrst element of the universe.2. such that X = spectrum(A) for some A. 75 .(f) ∀u (Ou ⇒ ∀x ∀z (P uxz ⇔ Sxz)) (g) ∀v ∀w (Svw ⇒ ∀x ∀z (P wxz ⇔ ∃y (Syz ∧ P vxy))) (h) ∀u (Ou ⇒ ∀x ∀z (Quxz ⇔ Ixz)) (i) ∀v ∀w (Svw ⇒ ∀x ∀z (Qwxz ⇔ ∃y (Qvxy ∧ P xyz))) Clauses (a). where k is the length of the input. This is a famous and apparently diﬃcult open problem.2 The Spectrum Problem Deﬁnition 4. 1 Jones/Selman [1] show that X is a spectrum if and only if there exists a nondeterministic Turing machine which accepts X in time 2ck . then M∞ |= A.2.1 In particular.15. Ak } be a ﬁnite subset of S.1. (Hint: Use the Compactness Theorem. . Remark 4. . nAk ). Show that there exists an inﬁnite normal L-structure M = M∞ with the following property: for all L-sentences A. Since the input is a positive integer n. with respect to R. with respect to R. Since M∞ satisﬁes these sentences. . 4.

Let U be a nonempty set. A binary relation R ⊆ U 2 is said to be an equivalence relation on U if it is reﬂexive, symmetric, and transitive, i.e., if the structure (U, R) satisﬁes (1) ∧ (2) ∧ (3): (1) (2) (3) ∀x Rxx ∀x ∀y (Rxy ⇔ Ryx) ∀x ∀y ∀z ((Rxy ∧ Ryz) ⇒ Rxz)

In this situation, the equivalence classes [a]R = {b ∈ U | a, b ∈ R}, a ∈ U , form a partition of U , i.e., a decomposition of the set U into pairwise disjoint, nonempty subsets. Let A be the following sentence of the predicate calculus with identity: (1) ∧ (2) ∧ (3) ∧ ∀x ∃y ((¬ Ixy) ∧ ∀z (Rxz ⇔ (Ixz ∨ Iyz))) Intuitively, A says that R is an equivalence relation with the property that each equivalence class consists of exactly two elements. Obviously, a ﬁnite set U admits an equivalence relation with this property if and only if the cardinality of U is even. Thus the spectrum of A is the set of even numbers. Exercises 4.2.4. Prove the following. 1. If X is a ﬁnite or coﬁnite2 set of positive integers, then X is a spectrum. 2. The set of even numbers is a spectrum. 3. The set of odd numbers is a spectrum. 4. If r and m are positive integers, {n ≥ 1 | n ≡ r mod m} is a spectrum. 5. If X and Y are spectra, X ∪ Y and X ∩ Y are spectra. Solution. 1. Let En be sentence in the language with only the identity predicate I, saying that the universe consists of exactly n elements (Exercise 4.1.10). If X = {n1 , . . . , nk }, then X is the spectrum of En1 ∨ · · · ∨ Enk , and the complement of X is spectrum of ¬ (En1 ∨ · · · ∨ Enk ). 2. The even numbers are the spectrum of a sentence which says: R is an equivalence relation on the universe, such that each equivalence class consists of exactly two elements. For more details, see Example 4.2.3. 3. The odd numbers are the spectrum of a sentence which says: R is an equivalence relation on the universe, such that each equivalence class consists of exactly two elements, except for one equivalence class, which consists of exactly one element. 4. We may assume that 0 ≤ r < m. If r = 0, the set

2A

set of positive integers is said to be coﬁnite if its complement is ﬁnite.

76

{n ≥ 1 : n ≡ 0 mod m} = {n ≥ 1 : m divides n with no remainder} is the spectrum of a sentence which says: R is an equivalence relation on the universe, such that each equivalence class consists of exactly m elements. If r > 0, the set {n ≥ 1 : n ≡ r mod m} = {n ≥ 1 : m divides n with remainder r} is the spectrum of a sentence which says: R is an equivalence relation on the universe, such that each equivalence class consists of exactly m elements, except for one equivalence class, which consists of exactly r elements. 5. Assume that X is the spectrum of A and Y is the spectrum of B. Then X ∪ Y is the spectrum of A ∨ B. Also, X ∩Y is the spectrum of A ∧ B, provided A and B have no predicates in common except the identity predicate. To arrange for this, replace B by an analogous sentence in a diﬀerent language. Exercise 4.2.5. Prove that, for any sentence A of the predicate calculus with identity, at least one of spectrum(A) and spectrum(¬ A) is coﬁnite. (Hint: Use part 1 of Theorem 4.1.13.) Example 4.2.6. We show that the set of composite numbers3 is a spectrum. Let L be a language consisting of two binary predicates, R and S, as well as the identity predicate, I. Let A be an L-sentence saying that R and S are equivalence relations, each with more than one equivalence class, and ∀x ∀y (∃ exactly one z)(Rxz ∧ Syz). Thus, for any normal L-structure M = (UM , RM , SM , IM ) satisfying A, we have that RM and SM partition UM into “rows” and “columns”, respectively, in such a way that the intersection of any “row” with any “column” consists of exactly one element of UM . Thus, if UM is ﬁnite, the elements of UM are arranged in an m × n “matrix”, where m, n ≥ 2. Therefore, the number of elements in UM is mn, a composite number. Conversely, for any m, n ≥ 2, there is an L-structure M as above, which satisﬁes A. Thus spectrum(A) is the set of composite numbers. Exercise 4.2.7. Use the result of Exercise 4.1.14 to prove the following: 1. The set of prime numbers and its complement are spectra. 2. The set of squares {1, 4, 9, . . .} and its complement are spectra. 3. The set of powers of 2, {2n | n = 1, 2, 3, . . .}, and its complement, are spectra.

3A

composite number is an integer greater than 1 which is not prime.

77

4. The set of prime powers {pn | p prime, n = 1, 2, . . .} and its complement are spectra. Solution. Let Z be as in Exercise 4.1.14 above. For each of the given sets X, we exhibit a sentence A with the following properties: X is the spectrum of Z ∧ A, and the complement of X is the spectrum of Z ∧ ¬ A. 1. ∃z ((¬ ∃w Rzw) ∧ (¬ ∃x ∃y (Rxz ∧ Ryz ∧ Qxyz)) ∧ (¬ Oz)). 2. ∃z ((¬ ∃w Rzw) ∧ ∃x Qxxz). 3. ∃z ∃v ((¬ ∃w Rzw) ∧ (∃u (Ou ∧ Suv)) ∧ ∀x ((¬ Ox ∧ ∃y Qxyz) ⇒ ∃w Qvwx)). 4. ∃z ∃v ((¬ ∃w Rzw) ∧ (¬ ∃x ∃y (Rxv ∧ Ryv ∧ Qxyv)) ∧ (¬ Ov) ∧ ∀x ((¬ Ox ∧ ∃y Qxyz) ⇒ ∃w Qvwx)). Exercise 4.2.8. 1. The Fibonacci numbers are deﬁned recursively by F1 = 1, F2 = 2, Fn = Fn−1 + Fn−2 for n ≥ 3. Show that the set of Fibonacci numbers {Fn | n = 1, 2, . . .} = {1, 2, 3, 5, 8, 13, 21, 34, 55, . . .} and its complement are spectra. 2. Show that {xy | x, y ≥ 2} and its complement are spectra. Exercise 4.2.9. Let X be a subset of {1, 2, 3, . . .}. Prove that if X is a spectrum then {n2 | n ∈ X} is a spectrum.

4.3

Predicate Calculus With Operations

In this section we extend the syntax and semantics of the predicate calculus, to encompass operations. As examples of operations, we may cite the familiar mathematical operations of addition (+) and multiplication (×). Such operations are considered binary, because they take two arguments. More generally, we consider n-ary operations. Deﬁnition 4.3.1 (languages). A language is a set of predicates P, Q, R, . . . and operations f, g, h, . . .. Each predicate and each operation is designated as n-ary for some nonnegative4 integer n. Deﬁnition 4.3.2 (terms, formulas, sentences). Let L be a language, and let U be a set. The set of L-U -terms is generated as follows. 1. Each variable is an L-U -term.

4A

0-ary operation is known as a constant. Syntactically, constants behave as parameters.

78

An atomic L-U -formula is an expression of the form P t1 · · · tn where P is an n-ary predicate of L. Let M be an L-structure. . 3. . If f is an n-ary operation of L. 4 and 5 of Deﬁnition 2. . Lemma 4. . vM (A) is deﬁned as in clauses 2 through 8 of Lemma 2. Deﬁnition 4. for all n-ary predicates P of L and all n-tuples a1 . . . .3. vM (tn ) ∈ PM . 3.3. for all n-ary operations f of L and all n-tuples a1 . . Deﬁnition 4. . The notions of substitution.. i. . φ(an )). An L-structure M consists of a nonempty set UM . . . we have vM (P t1 · · · tn ) = T F if vM (t1 ). / For non-atomic L-UM -sentences. . vM (tn )) for all n-ary operations f of L and all variable-free L-UM -terms t1 . . 1. . .3. Note that P t1 · · · tn is a sentence if and only if it is variable-free. Each element of U is an L-U -term. and if t1 . 2. . and L-U -sentences are deﬁned as in Section 2. . Two L-structures M and M ′ are said to be isomorphic if there exists an isomorphism of M onto M ′ . . . . . . . . .2. For atomic L-U -sentences. . . an n-ary relation PM ⊆ (UM )n for each n-ary predicate P of L. . There is a unique valuation vM : {t | t is a variable-free L-UM -term} → UM deﬁned as follows: (a) vM (a) = a for all a ∈ UM . tn are L-U -terms. and an n-ary function fM : (UM )n → UM for each n-ary operation f of L. an ∈ (UM )n . . . . . and t1 . . (b) vM (f t1 · · · tn ) = fM (vM (t1 ). φ(fM (a1 . . φ(an ) ∈ PM ′ . 2. . tn . . The set of L-U -formulas is generated as in clauses 2. an ∈ (UM )n . . .4. . . .3. an ∈ PM if and only if φ(a1 ). . a one-to-one ∼ correspondence φ : UM = UM ′ such that: 1. if vM (t1 ). An L-U -term is said to be variable-free if no variables occur in it.4 (isomorphism). .3 (structures).1. . There is a unique valuation vM : {A | A is an L-UM -sentence} → {T. . tn are L-U -terms. a1 .1. an ) = fM ′ (φ(a1 ). then f t1 · · · tn is an L-U -term. free variables.2. vM (tn ) ∈ PM . 79 . .e. . F} deﬁned as follows.5 (valuations).

. ∀x A . . . . | A[x/a] . . . . | T A[x/t] . | ¬ A[x/t] where t is a variable-free term . T ∃x A . ∃x A . . . . | T A[x/a] . . .Proof. The signed and unsigned tableau methods carry over to predicate calculus with operations. . . . Signed: . . The proof is as for Lemma 2. . . .3. ¬ ∀x A . . We modify the tableau rules as follows. F ∀x A . | ¬ A[x/a] where a is a new parameter 80 . . Deﬁnition 4. . . . | F A[x/t] where t is a variable-free term . . | A[x/t] . . . . .4. T ∀x A .2. | F A[x/a] where a is a new parameter Unsigned: . ¬ ∃x A .6 (tableau method). F ∃x A . . . .

13 carries over unchanged to the context of predicate calculus with operations.9 and Exercise 4. The Compactness Theorems 2. . also carry over. M and M ′ satisfy the same L-sentences.4 on logical equivalence also carry over. .1 on satisﬁability in inﬁnite domains. Strengthen the Interpolation Theorem 3. The notion of satisﬁability in a domain carries over unchanged to the context of predicate calculus with operations. .5.4 hold. In our Hilbert-style proof system LH .10. if S contains ∀x A then S contains A[x/t] for all variable-free L-U -terms t.5.3.3. our Gentzen-style proof system LG is modiﬁed in accordance with the modiﬁed tableau rules. Exercise 4.3.Remark 4.4 carries over to the context of predicate calculus with operations.11 (companions and proof systems). where A′ = A[a1 /φ(a1 ).12 (the Interpolation Theorem).13. With these changes.2. .10 below. the instantiation rules are modiﬁed as follows: (a) (∀x B) ⇒ B[x/t] (universal instantiation) (b) B[x/t] ⇒ (∃x B) (existential instantiation) where t is any variable-free term. Theorem 2.2.7.3.7 (soundness and completeness). Let M and M ′ be L-structures. 81 . With the tableau rules as above.3. The results of Section 2. . the Soundness Theorem 2. The conclusion of Hintikka’s Lemma 2.7.9 to show that Corollary 2. (Hint: The version with operations can be deduced from the version without operations.3. Remark 4.3 is modiﬁed to say that S is satisﬁable in the domain of variable-free L-U -terms.1 to say that each operation.2) is modiﬁed to say that. ak /φ(ak )].8 (satisﬁability in a domain). clauses (1) and (4) are modiﬁed as follows: (1) (4) (∀x B) ⇒ B[x/t] B[x/t] ⇒ (∃x B) where t is any variable-free term. Theorem 4.3.9.2 carry over unchanged.5. Remark 4.1 and 2.2.3.3.2. . See Theorem 4. In our notion of companion (Deﬁnition 3. Also.3.6.3.6. The conclusion of the Completeness Theorem 2. Proof. Xk is satisﬁable in the domain of variable-free L-V -terms.5 is modiﬁed to say that X1 . The proof is similar to that of Theorem 2. and Theorem 2.5. Theorems 2. .3. In particular.) Exercise 4. for example.3).3 carries over in an appropriately modiﬁed form.3. the soundness and completeness of LG and LH carry over.7.6 and 2. Use Theorem 4. predicate and parameter occurring in I occurs in both A and B.6 we have vM (A) = vM ′ (A′ ) for all L-UM -sentences A. Exercise 4.7. The notion of U -repleteness (Deﬁnition 2. Skolemization.11 on isomorphism. . Then as in Theorem 2. Assume that φ : UM → UM ′ is an onto mapping such that conditions 1 and 2 of Deﬁnition 4.

M |= A if and only if M o |= Ao .. The notions of normal structure and normal satisﬁability are deﬁned as before. Let L be a language with identity and operations. 1. For each n-ary operation f of L.4.4.1.1 (predicate calculus with identity and operations). (Hint: Use the result of Exercise 4. an ) = b}. we have an axiom ∀x1 · · · ∀xn ∀y1 · · · ∀yn ((Ix1 y1 ∧ · · · ∧ Ixn yn ) ⇒ If x1 · · · xn f y1 · · · yn ).e. Show that a normal Lo -structure satisﬁes the sentences Gf . f in L. given a sentence A involving some operations. Exercise 4. . For each n-ary operation f of L. construct a sentence Aoo involving no operations.2.2) as follows: 5. Note that Aoo will not be the same as the Ao of Exercise 4. 2. The predicate calculus with identity and operations is well suited for the study of algebraic structures such as number systems. See also Exercise 4.4. let Gf be the Lo -sentence ∀x1 · · · ∀xn ∃y ∀z (Iyz ⇔ Pf x1 · · · xn z). if and only if it is of the form M o for some L-structure M . 82 . Let Lo be the language with identity and without operations.2.2. as previously discussed in Section 4. We augment the identity axioms (Deﬁnition 4. obtained by replacing each n-ary operation f belonging to L by a new (n + 1)ary predicate Pf belonging to Lo . . . . for all L-structures M . one often writes t1 = t2 instead of It1 t2 . Show that to each L-sentence A we may associate an Lo -sentence Ao such that.) Remark 4. In other words. .4.4.4 Predicate Calculus with Identity and Operations Remark 4. such that spectrum(A) = spectrum(Aoo ).4. rings. Each normal L-structure M gives rise to a normal Lo -structure M o where (Pf )M ′ = { a1 . the equality axioms (i.1 on the predicate calculus with identity carry over unchanged to the predicate calculus with identity and operations. etc.3.2 (elimination of operations). Show that the spectrum problem for predicate calculus with identity and operations is equivalent to the spectrum problem for predicate calculus with identity and without operations. . and one refers to predicate calculus with equality rather than predicate calculus with identity. an . In such a context. the identity axioms) read as follows: ∀x (x = x). b ∈ (UM )n+1 | fM (a1 . The results of Section 4.4 (predicate calculus with equality). Exercise 4. In this notation. groups. .2 below.4.4. .

for each n-ary operation f . Thus G = (G. ∀x (1x = x1 = x). Show that the class of torsion-free groups can be characterized by a set of sentences.e. a group may be viewed as a normal L-structure G = (UG . iG . there is a set of sentences S such that.. 0R . =R ) where + and · are binary operations. 1R . −. ∀x ∀y (x + y = y + x). t−1 instead of it. Let S = {An : n ≥ 2}. To avoid ambiguity. For a ∈ G write an = a · · · · · a (n times).4. One also uses customary algebraic notation. 1. ∀x1 ∀y1 · · · ∀xn ∀yn ((x1 = y1 ∧ · · · ∧ xn = yn ) ⇒ f x1 · · · xn = f y1 · · · yn ). ∀x (x · 1 = 1 · x = x).4.. for each n-ary predicate P . and I is the identity predicate (a binary predicate). a constant). t1 + t2 instead of +t1 t2 . Exercise 4. Here UG is the underlying set of the group.5 (groups and rings). ∀x1 ∀y1 · · · ∀xn ∀yn ((x1 = y1 ∧ · · · ∧ xn = yn ) ⇒ (P x1 · · · xn ⇔ P y1 · · · yn )). ∀x ∀y ∀z ((x = y ∧ y = z) ⇒ x = z). R is required to satisfy the ring axioms. consisting of the identity axioms plus ∀x ∀y ∀y (x + (y + z) = (x + y) + z). t1 × t2 or t1 t2 instead of ×t1 t2 . e is the group identity element (a 0-ary operation. Thus a1 = a and an+1 = an · a. We could refer to the language {+. IG ). 1. i. I. =} as the language of rings. plus ∀x ∀y ∀z ((xy)z = x(yz)). 0 = 1. 1 instead of e. I}. +R .e. 0 and 1 are constants. for all groups G. if a = 1 then an = 1 for all positive integers n. ·R . Solution. ∀x ∀y ∀z (x · (y + z) = (x · y) + (x · z)). Examples 4. ·G . G is torsion-free if and only if G |= S. − is a unary operation. ∀x ∀y ∀z ((x + y) · z = (x · z) + (y · z)). parentheses are used. i is group inversion (a unary operation).. e. We say that G is torsion-free if for all a ∈ G. 0. ·. i. a ring may be viewed as a normal structure R = (R. t1 · t2 or t1 t2 instead of f t1 t2 . where f is the group composition law (a binary operation). It is customary to write G instead of UG . We could refer to L as the language of groups.6. ∀x (x−1 x = xx−1 = 1). consisting of the identity axioms for L. We say that G is a torsion group if for all a ∈ G there exists a positive integer n such that an = 1. Let G be a group. ∀x (x + (−x) = 0). and = is the equality predicate. Using predicate calculus with identity and operations. ∀x ∀y ∀z (x · (y · z) = (x · y) · z). =G ) and G is required to satisfy the group axioms. ∀x (x + 0 = x).g. −1 G . e. where An is the sentence 83 . fG . and L is the language {f. etc. eG .∀x ∀y (x = y ⇔ y = x). and t1 = t2 instead of It1 t2 . 1G . −R . Similarly.

An }. Deﬁne a pseudo-ﬁnite group to be a group which satisﬁes S. Now let G be a torsion group satisfying {A2 .7. . . pseudo-ﬁnite group? Prove your answer. . σn . Let L be the language of groups. a set of predicates P. there is no set of sentences S with the property that.∀x (x = 1 ⇒ xn = 1). Let S be the set of Lsentences which are true in all ﬁnite groups. there is a ﬁxed n such that all of the sentences in S ′ are logical consequences of the group axioms plus {A2 . .. each designated as n-ary of type σ1 × · · · × σn for some nonnegative integer n and sorts σ1 . . each sentence in S ′ is a logical consequence of the group axioms plus {A2 . 2. . . G is a torsion group if and only if G |= S. . where p is a prime number greater than n. we may take G to be the additive group of integers modulo p. . for all groups G. . Suppose S ′ were a ﬁnite of sentences such that the groups satisfying S ′ are exactly the torsion-free groups. a set of sorts σ. contradicting our assumption on S ′ . . (For example. . . .. Exercise 4. σn . Q. each designated as n-ary of type σ1 × · · · × σn → σn+1 for some nonnegative integer n and sorts σ1 . a set of operations f. Note that every ﬁnite group is pseudo-ﬁnite. Does there exist an inﬁnite. An } for suﬃciently large n. A many-sorted language consists of 1. . . . . σn+1 . .4. Show that the class of torsion-free groups cannot be characterized by a ﬁnite set of sentences. 84 . . . τ. . An }. . . Show that the class of torsion groups cannot be characterized by a set of sentences.5. By the Compactness Theorem. .) Then G satisﬁes S ′ yet is not torsion-free. In particular. . I. 3. . Since S ′ is ﬁnite..e..5 Many-Sorted Predicate Calculus Deﬁnition 4. 4.1 (many-sorted languages). Clearly the groups satisfying S are exactly the torsion-free groups. . g. 2. 3. . each sentence in S ′ is a logical consequence of the group axioms plus S = {An : n ≥ 2} as above. Solution.

. . sentences.5. Naturally.5 (tableau method. carry over to the many-sorted context in the obvious way. . . Remark 4. .). formulas. 3. with clause 5 modiﬁed as follows: 5′ . generalizing Remark 4. In this way. Then the tableau method carries over in the obvious way. . Let U = (U σ . . . .6 on isomorphism. An atomic L-U -formula is an expression of the form P t1 . the substitution A[xσ /t] makes sense only when t is a term of sort σ. . Deﬁnition 4.2. Our notions of substitution. Each variable of sort σ is a term of sort σ.7. σn respectively. The L-U -formulas are generated as in Deﬁnition 2. countably inﬁnite set of variables of sort σ. . truth. . we obtain satisﬁability in the domain U = (U σ . U τ . are the sorts of L. . . etc. the notion of satisﬁability in a domain generalizes to the many-sorted context. . σ σ 3. .) consist of a set U σ for each sort σ of L. we assume a ﬁxed. . tn is a term of sort σn+1 .2 (terms.). tn . Each element of U σ is a term of sort σ. In the Completeness Theorem for the tableau method. then ∀xσ A and ∃xσ A are L-U -formulas. and if t1 . tn are terms of sort σ1 . . 85 . . where P is an n-ary predicate of type σ1 × · · · × σn .5. . For each sort σ. . U τ .4 (many-sorted domains).3.3 (many-sorted structures). . If f is an n-ary operaton of type σ1 × · · · × σn → σn+1 . If xσ is a variable of sort σ. . y σ . 2.3.3. . We deﬁne a domain or universe for L to be an indexed family of nonempty sets U = (U σ . an n-ary function fM : UM1 × · · · × UMn → UMn+1 for each n-ary operation of type σ1 × · · · × σn → σn+1 belonging to L. . . .5. An L-structure M consists of σ 1. . and if A is an L-U -formula. where σ. sentences). V τ . and results such as Theorem 2. . valuation. . . An L-formula is an L-U -formula where Uσ = ∅ for each sort σ. . .1. U τ .. . proof systems). tn are terms of sort σ1 . ﬁx a countably inﬁnite set V σ = {aσ .Deﬁnition 4.3. . an n-ary relation PM ⊆ UM1 × · · · × UMn for each n-ary predicate P of type σ1 × · · · × σn belonging to L. . Deﬁnition 4. denoted xσ . σ σ 2. are extended in the obvious way to the many-sorted context. where U σ is the set of variable-free L-V -terms of sort σ. .}. just as in Section 4. bσ . τ. 1. . with V = (V σ . For each sort σ of L. . z σ . then f t1 . σn respectively. the set of parameters of sort σ.5.9 on onto mappings.). σ Notions such as isomorphism. and Theorem 4. . The L-U -terms are generated as follows. Let L be a many-sorted language. satisﬁability. and t1 . . a nonempty set UM for each sort σ of L.. free and bound variables. The soundness and completeness of our proof systems LH and LG and the Interpolation Theorem also carry over.

This term is used for contrast with the many-sorted generalization which we are considering in this section.5. .1 concerning normal satisﬁability carry over to the many-sorted context. to be of speciﬁed distinct cardinalities.10. UM . σk are the sorts occurring in A. As identity axioms we may take the universal closures of all L-formulas of the form ∀xσ ∀y σ (I σ xy ⇒ (A ⇔ A[x/y])) σ where A is atomic. . It is natural to try o to generalize the L¨wenheim/Skolem Theorem 4. .6 (identity predicates). Generally speaking. A language or structure is said to be one-sorted if it has only one sort. The results of Section 4. the problem of characterizing many-sorted spectra has not been investigated thoroughly.Remark 4. This is straightforward provided we consider only normal structures σ τ M where all of the domains UM . . . .5. σ τ However. .5. 86 .8 (many-sorted spectrum problem). . Remark 4. Our reasons for including many-sorted predicate calculus in this course are as follows: 1. Remark 4. one-sorted logic tends to be a little simpler than many-sorted logic. the topic of so-called twocardinal theorems turns out to be rather delicate and complicated.11 (one-sorted languages). . nk ) σ such that there exists a normal L-structure M with UMi of cardinality ni .5. then this leads to diﬃcult issues. If A is an L-sentence and σ1 . For each sort σ of L. Let L be a many-sorted language with identity. the spectrum problem carries over to many-sorted predicate calculus. A many-sorted language with identity is a many-sorted language which contains an identity predicate for each sort. UM . . a | a ∈ σ σ UM } for all σ such that I belongs to L. . Deﬁnition 4. . are of the same inﬁnite cardinality. L may or may not contain a binary predicate I σ of type σ × σ designated as the identity predicate for σ. . . See for example the model theory textbook of Marker [2]. .9 (many-sorted L¨wenheim/Skolem theorems).7 (languages with identity). . . if we require UM . Even for two sorts. Remark 4. FIXME Remark 4. An L-structure M is said to be normal if IM = { a. the spectrum of A is the set of k-tuples of positive integers (n1 . . for i = 1. k. . So far as I know.13 to many-sorted predicate o calculus.1.5. In this way. it is more useful . .5. .

S) be a theory. T is consistent if and only if there exists an L-sentence A such that T ⊢ A. 1.1.1. completeness). Deﬁnition 5. Equivalently.3 (consistency. Deﬁnition 5. T is ﬁnitely axiomatized if S is ﬁnite.. we denote this by T ⊢ A.2. 3. A model of T is an L-structure M such that M satisﬁes S.. If A is a theorem of T . S) be a theory. 5. Two theories are equivalent if they have the same language and the same theorems. Thus T = (L. A is derivable in LH (S ∪ {identity axioms for L}). 1. 2. T is ﬁnitely axiomatizable if it is equivalent to a ﬁnitely axiomatized theory.e. together with a set of L-sentences called the axioms of T . categoricity. S). A theorem of T is an L-sentence A such that A is true in all models of T . T is consistent if and only if there is no L-sentence A such that both T ⊢ A and T ⊢ ¬ A. Deﬁnability 5.1 Theories and Models Deﬁnition 5. Models. Equivalently. Let T = (L. Equivalently. called the language of T . A theory T consists of a language L. i. then M is required to be normal with respect to these predicates. T is consistent if there exists at least one model of T . they have the same language and the same models. I.e. and S is the set of axioms of T . If L contains identity predicates. where L is the language of T .1.Chapter 5 Theories. 4. A is a logical consequence of the axioms of T . 87 .1. Let T = (L.

1.. Remark 5. Later in this chapter we shall present several interesting examples of theories.7.1. Loosely speaking. foundationally important model. Theorem 5. Nevertheless we shall see that. and (2) the language of T is a language with identity. 1. In the one-sorted case. we shall see in Chapter 6 that. Let T be a complete theory in a language with identity.6 (see also Deﬁnition 4. is intended as a precise explication of the informal notion of “deductive scientiﬁc theory”. On the other hand. Since the class is diverse. i. Completeness of T means that our theory is suﬃciently successful to decide the truth values of all statements expressible in the language of the theory. Assume that T is categorical.1.5 (mathematical theories. The theorems of T are the assertions of our theory.7.5.2. Then any two models of T are isomorphic. Equivalently. regrettably. we have: Exercise 5. The axioms of T are the basic assertions. See Section 5.2. T contains an identity predicate for each sort in the language of T . or at least a large part of it. Proof.2.3).6. 88 .6. One way to show that a theory is complete is to show that it is categorical. Categoricity of T means that our theory is fully successful in that it fully captures the structure of the underlying reality described by the theory. T is complete if and only if for all L-sentences A either T ⊢ A or T ⊢ ¬ A but not both. theories which are introduced in order to serve as a general axiomatic foundation for all of mathematics. then (1) T is complete. any two models of T are elementarily equivalent. The language of T is the vocabulary of our theory.9 and Deﬁnition 4. By a mathematical theory we mean a theory which is introduced in order to describe a certain class of mathematical structures. Consistency of T means that our theory is free of internal contradictions. The ﬁrst kind consists of mathematical theories.e. Also. If T is categorical. as deﬁned above. remarkably. show that T is categorical if and only if the universe UM is ﬁnite.1. See Sections 5.3 (see also Theorem 4. the theory is typically not intended to be complete. Although such theories are intended to be complete. T is complete if T is consistent and all models of T are elementarily equivalent. Hence by Theorem 2. Remark 5. The second kind consists of foundational theories. Our formal notion of theory. most of these theories turn out to be incomplete. Each such theory is intended to fully describe a speciﬁc.3).5. 3. foundational theories). from which all others are deduced. by Theorem 2.4.3. known as the intended model of the theory.5 and 5. Let M be a model of T . the examples are of two kinds. T is categorical if T is consistent and all models of T are isomorphic. several of these mathematical theories turn out to be complete. Hence T is complete.

Suppose T is not complete. M2 are not elementarily equivalent.13.1. .2.1. ∀x (x · 1 = 1 · x = x) (identity). . 89 . Then T is complete. By assumption (b). The theory of groups consists of the equality axioms plus ∀x ∀y ∀z (x · (y · z) = (x · y) · z) (associativity). Theorem 5. M1 and M2 are inﬁnite. Deﬁnition 5.2 Mathematical Theories In this section we give several examples of theories suggested by abstract algebra and other speciﬁc mathematical topics. Remark 5. Hence. Proof.1. σ Solution. Generalize Vaught’s Test to many-sorted theories. We point out that several of these mathematical theories are complete. 5..1. A similar but more useful test is provided by the following theorem. (b) all models of T are inﬁnite. except in very special circumstances. a constant 1 (the identity element).11.1 (groups). a unary operation −1 (inverse).2. If one of the universes UM is inﬁnite.13 to blow it up to an arbitrarily large uncountable cardinality.7 show that we cannot use categoricity as a test for completeness. . Theorem 5. Let T be a one-sorted theory. Clearly ′ ′ ′ ′ M1 . τ. there exist M1 and M2 which are models of T and not elementarily equivalent. ∀x (x · x−1 = x−1 · x = 1) (inverses).1.2. Exercise 5. A group is a model of the theory of groups. . show that T is categorical if σ τ and only if each of the universes UM .1. . In the many-sorted case with sorts σ. by Theorem 2. Assume that (a) T is consistent.9. Example 5. and (c) there exists an inﬁnite cardinal κ ≥ the cardinality of the language of T such that T is κ-categorical. . M2 respectively. we can use the L¨wenheim/Skolem o Theorem 4.8. M2 of cardinality κ elementarily equivalent to M1 .6 and Exercise 5. and a binary predicate = (equality). A group is said to be Abelian if it satisﬁes the additional axiom ∀x ∀y (x · y = y · x) (commutativity).6. Let κ be an inﬁnite cardinal ≥ the cardinality of the language of T .1. A one-sorted theory T is said to be κ-categorical if all models of T of cardinality κ are isomorphic. This contradicts assumption (c). there exist models o ′ ′ M1 . is ﬁnite.1. Since T is consistent. M1 and M2 are not isomorphic. UM . Let κ be an inﬁnite cardinal number. By the L¨wenheim/Skolem Theorem 4.10 (Vaught’s Test). The language of groups consists of a binary operation · (multiplication).

then T ∪ {∀x ∀y (x · y = y · x)} would be an L-theory whose models are just the Abelian torsion groups. hence is not a torsion group. Let DAG0 be the theory of inﬁnite torsion-free divisible Abelian groups.}. yet G2 is not a torsion group. . . . 3. By Compactness. Note that G is torsion-free if and only if it satisﬁes the axioms ∀x (xn = 1 ⇒ x = 1) for n = 2. Exercise 5. . Show that the theory of torsion-free Abelian groups is not ﬁnitely axiomatizable. . Exercise 5. Let G1 be a torsion group with elements of arbitrarily large ﬁnite order. By the Compactness Theorem for normal satisﬁability (Corollary 4. we see that any ﬁnite subset of S ∗ is normally satisﬁable. and let S be the set of all L-sentences true in G1 .9). (For example.2. Let p be a prime number greater than k. Then G2 is an Abelian group which is elementarily equivalent to G1 yet contains an element c of inﬁnite order. contradicting the fact that G1 and G2 are elementarily equivalent. Hence. Solution. . . and let S ∗ = S ∪ {cn = 1 | n = 1. . Let L∗ = L ∪ {c} where c is a new constant. . c) be a model of S ∗ . Deduce that there is no theory in the language of groups whose models are precisely the Abelian torsion groups. (This is because such groups may be viewed as vector spaces over the rational ﬁeld. By choosing c ∈ G1 appropriately. .3.2. Deduce that the theory of torsion-free groups is not ﬁnitely axiomatizable. A group G is torsion-free if for all a ∈ G. 3. it follows that S ∗ is normally satisﬁable. if a = 1 then an = 1 for all positive integers n. It can be shown that DAG0 is κ-categorical for all uncountable cardinals κ.. 2. Suppose that the theory of torsion-free Abelian groups were ﬁnitely axiomatizable.) Let L be the language of groups. . there is no theory in the language of groups whose models are precisely the torsion groups. . Solution. Show that there exist Abelian groups G1 and G2 such that G1 is a torsion group. we could take G1 to be the additive group of rational numbers modulo 1. G1 is elementarily equivalent to G2 . 90 . If the theory of torsion-free groups were ﬁnitely axiomatizable.) It follows by Vaught’s Test that DAG0 is complete..1.4. say n = 1.2. then G1 would be a model of T but G2 would not. If T were an L-theory whose models are just the torsion groups.A torsion group is a group G such that for all a ∈ G there exists a positive integer n such that an = 1. so let (G2 . for n = 2. by adjoining the single axiom ∀x ∀y (x · y = y · x). Q. then the theory of torsion-free Abelian groups would also be ﬁnitely axiomatizable. . Remark 5. the axioms would be logical consequences of the Abelian group axioms plus ﬁnitely many axioms of the form ∀x (xn = 1 ⇒ x = 1). Then the additive group of integers modulo p satisﬁes these axioms but is not torsion-free.2. If T were an L-theory whose models are just the Abelian torsion groups. This is a contradiction. k. A group is said to be divisible if it satisﬁes the axioms ∀x ∃y (y n = x).

The axioms for linear orderings are ∀x ∀y ∀z ((x < y ∧ y < z) ⇒ x < z). <). =. Remark 5. A linear ordering is said to be nontrivial if it satisﬁes ∃x ∃y (x < y).) Let L = {R. It follows by Vaught’s Test that DLO is complete. where Q is the set of rational numbers. up to isomorphism.Example 5. A path from a to b is a ﬁnite sequence of pairwise distinct vertices a = v0 . Deduce that there is no theory T in the language of graphs such that the models of T are exactly the connected graphs. Two vertices a. It is said to with end points if it satisﬁes ∃x ¬ ∃y (y < x) and ∃x ¬ ∃y (y > x). It is said to be dense if it is nontrivial and satisﬁes ∀x ∀y (x < y ⇒ ∃z (x < z < y)). irreﬂexive relation on VG . (Q. An example of a dense linear ordering without end points is (Q.8. b ∈ G1 appropriately. Thus a graph is essentially an ordered pair G = (VG .2. n + 1. we may take G1 = (N. for all partitions of VG into two disjoint nonempty sets X and Y . <) is the unique countable dense linear ordering without end points. b} where a. and let S be the set of L-sentences satisﬁed by G1 . . so let (G2 . G1 is elementarily equivalent to G2 . the distance from a to b. n | n ∈ N}. v1 is adjacent to v2 . The language of linear orderings consists of a binary predicate < plus the equality predicate =. if we let DLO be the theory of dense linear ordering without end points. . A graph is a model of the theory of graphs. The language of graphs consists of a binary predicate. vn = b such that a = v0 is adjacent to v1 . and ∀x ∀y (x < y ∨ x = y ∨ x > y). b ∈ VG are said to be adjacent if a.7 (graphs). It can be shown that. 2. Let G1 be a graph which is connected yet contains pairs of vertices which are at distance n for arbitrarily large n. =} be the language of graphs. n + 1 . If a and b are two vertices in a graph. or ∞ if there is no such path. Then G2 is a graph which is 91 . Hence by the Compactness Theorem S ∗ is normally satisﬁable. Equivalently. . The elements of VG are called vertices.} where An is an L∗ -sentence saying that there is no path of length n from a to b. Exercise 5. RG ). A linear ordering is a model of these axioms. By choosing a. we see that all ﬁnite subsets of S ∗ are normally satisﬁable. a. DLO is ℵ0 -categorical. v1 . b). there exist a ∈ X and b ∈ Y such that a and b are adjacent. It is said to be without end points if it satisﬁes ∀x ∃y (y < x) and ∀x ∃y (y > x). R. (For example. plus the equality predicate. . yet G2 is not connected.) Hence. b) be a model of S ∗ . . and let S ∗ = S ∪{An | n = 1.6. G is connected if and only if. and ∀x (¬ x < x). Let L∗ = L∪{a. b ∈ RG .5 (linear orderings). where VG is a nonempty set and RG is a symmetric. . G is said to be connected if for all a. b are new constants. we deﬁne d(a. .2. The theory of graphs consists of the equality axioms plus ∀x ∀y (Rxy ⇔ Ryx) and ∀x ¬ Rxx. Example 5. R1 ) where R1 = { n. b ∈ VG there exists a path from a to b. (This is proved by a backand-forth argument. Show that there exist graphs G1 and G2 such that G1 is connected. vn−1 is adjacent to vn = b.2. Solution. . . to be the smallest length of a path from a to b. . and < is the usual ordering of Q. .2.

then G1 would be a model of T but G2 would not. ∀x (x · 1 = 1 · x = x). . fn . ∀x (x + (−x) = 0). . ∀x ∀y ∀z ((x + y) · z = (x · z) + (y · z)) (right distributivity). Solution. ∀x ∀y ∀z (x · (y + z) = (x · y) + (x · z)) (left distributivity).9. ﬁnd c ∈ G such that for all a ∈ dom(fn ). The language of rings consists of binary operations + and · (addition and multiplication).2. ∀x ∀y ∀z (x · (y · z) = (x · y) · z). ∞ Finally. Example 5. . 0 = 1. . Suppose we have already constructed fn . b) = ∞. say G = {ak | k ∈ N} and G′ = {a′ | k ∈ N}. Exercise 5. k / k By randomness of G. . bn . am and not adjacent to b1 . constants 0 and 1 (the additive and multiplicative identity elements). which maps a ﬁnite subset of G isomorphically onto a ﬁnite subset of G′ . It follows by Vaught’s Test that this theory is complete. c′ is adjacent to fn (a) if and only if c is adjacent to a. let fn+1 = fn ∪ {(c. . . ∀x (x + 0 = x). bn there exists a vertex c such that c is adjacent to a1 . By randomness of G′ . a unary operation − (subtraction). b ∈ G2 are such that d(a. ∀x ∀y (x + y = y + x). At stage n of the construction. . we have a ﬁnite partial isomorphism. ﬁnd c′ ∈ G′ such / that for all a ∈ dom(fn ). yet a. We use a back-and-forth argument to construct k an isomorphism of G onto G′ . b1 . . am . . c is adjacent to a if and only if c′ is adjacent to fn (a). hence G2 is not connected. In either case. Start with f0 = ∅. . . let kn be the least k such that a′ ∈ ran(fn ). The theory of rings consists of the equality axioms plus ∀x ∀y ∀z (x + (y + z) = (x + y) + z). consider two cases.elementarily equivalent to G1 . A graph G is said to be random if for all ﬁnite sets of distinct vertices a1 .10 (rings). A ring is a model of the theory of rings. If n is even. Let G and G′ be two random graphs of cardinality ℵ0 . . c′ )}. and put c′ = a′ n . by construction. If there were an L-theory T whose models are exactly the connected graphs. ∀x (x · 0 = 0 · x = 0). . contradicting the fact that G1 and G2 are elementarily equivalent. f = n=0 fn is an isomorphism of G onto G′ . . If n is odd. and a binary predicate = (equality). . To construct fn+1 . and put c = akn . A ring is said to be commutative if it satisﬁes the additional axiom 92 . Show that the theory of random graphs is ℵ0 -categorical. let kn be the least k such that ak ∈ dom(fn ).2.

) This gives a contradiction. −2. there exists z ∈ F such that f (z) = 0. A ﬁeld is said to be of characteristic 0 if it satisﬁes 1 + ··· + 1 = 0 n for all positive integers n. Z = {. 2. 1. . . .}. an = 0. . 0. Exercise 5. 3. . A ﬁeld is a commutative ring satisfying the additional axiom ∀x (x = 0 ⇒ ∃y (x · y = 1)). our ﬁeld is said to be of characteristic p. n ≥ 1. Q.) Note that a ﬁeld is algebraically closed if and only if it satisﬁes the axioms ∀x0 ∀x1 · · · ∀xn (xn = 0 ⇒ ∃z (xn z n + · · · + x1 z + x0 = 0)) for n = 1. . say n = 1. a0 ∈ F . An example of a ﬁeld of characteristic p is the ring of integers modulo p. .11 (ﬁelds). an . An example of a non-commutative ring is the ring of 2 × 2 matrices. Example 5. Familiar ﬁelds such as the ﬁeld of rational numbers. a1 . . we can construct a ﬁeld which satisﬁes these axioms yet is not algebraically closed. the axioms would be logical consequences of the ﬁeld axioms plus ﬁnitely many of the ACF0 axioms as presented in Example 5. k.2. It is known that the complex ﬁeld C is algebraically closed. Show that the theory ACF0 of algebraically closed ﬁelds of characteristic 0 is not ﬁnitely axiomatizable. It can be shown that if a ﬁeld satisﬁes 1 + · · · + 1 = 0 for some positive integer n.2. Solution. . A ﬁeld F is said to be algebraically closed if for all nontrivial polynomials f (z) = an z n + · · · + a1 z + a0 . 2. −1. Let p be a prime number greater than k. . p.11 above. Suppose ACF or ACF0 were ﬁnitely axiomatizable. Show that the theory of ﬁelds of characteristic 0 is not ﬁnitely axiomatizable.12. for any ﬁnite subset of the ACF0 axioms. are of characteristic 0. . (The construction of such a ﬁeld is perhaps somewhat delicate. . then the least such n is a prime number. The ﬁeld of integers modulo p satisﬁes these axioms but is not of characteristic 0. Show that the theory ACF of algebraically closed ﬁelds is not ﬁnitely axiomatizable. 93 n n .. the axioms would be logical consequences of the ﬁeld axioms plus ﬁnitely many axioms of the form 1 + · · · + 1 = 0. By Compactness. . the ﬁeld of real numbers. But. . and the ﬁeld of complex numbers. (This theorem is known as the Fundamental Theorem of Algebra. By Compactness. Suppose that the theory of ﬁelds of characteristic 0 were ﬁnitely axiomatizable. C.2. . . a contradiction.∀x ∀y (x · y = y · x). R. . An example of a commutative ring is the ring of integers. In this case.

in the integers modulo p for all suﬃciently large primes p. for each prime p..2. i. It follows by Vaught’s Test that these theories are complete.9). Does there exist a ﬁeld of characteristic 0 satisfying S? Prove your answer.1. Remark 5. ∀v τ (0 · v = 0) for scalar multiplication. If V is a vector space. a constant 0 of type τ . By the Compactness Theorem for normal satisﬁability (Corollary 4. constants 0 and 1 of type σ. a unary operation − of type τ → τ . let ACFp be the theory of algebraically closed ﬁelds of characteristic p. Let ACF0 be the theory of algebraically closed ﬁelds of characteristic 0. In addition we have a binary operation · of “mixed” type σ × τ → τ .g. Let S0 = S ∪ {1 + · · · + 1 = 0 | n = 1. For the vectors we have a binary operation + of type τ × τ → τ .}. Let L be the language of rings. ∀xσ (x · 0τ = 0). a set of vectors S in V is said to span V if every vector v in V can be written as a linear combination of vectors in S.2.. The normal structures which satisfy S0 are ﬁelds of characteristic 0 satisfying S. ∀xσ ∀v τ ((−x) · v = −(x · v)). . the ring of integers modulo p is actually a ﬁeld. 2. . ∀xσ ∀y σ ∀v τ ((x · y) · v = x · (y · v)). and the axioms ∀xσ ∀v τ ∀wτ (x · (v + w) = (x · v) + (x · w)). and an equality predicate = of type τ × τ . It can be shown that the theories ACF0 and ACFp are κ-categorical for all uncountable cardinals κ. denoting scalars and vectors respectively.2. the Abelian group axioms for vectors.15 (vector spaces). ∀xσ ∀y σ ∀v τ ((x + y) · v = (x · v) + (y · v)). For the scalars we have binary operations + and · of type σ × σ → σ. A vector space is a model of these axioms. Note ﬁrst that. e. Solution. Let S be the set of L-sentences which are true in the ring of integers modulo p for all but ﬁnitely many prime numbers p.14. For each prime number p.13. ∀v τ (1 · v = v). The theory of vector spaces consists of the ﬁeld axioms for scalars. and an equality predicate = of type σ × σ. v = 94 n . a unary operation − of type σ → σ. .Exercise 5. The language of vector spaces is a 2-sorted language with sorts σ and τ . Any ﬁnite subset of S0 is normally satisﬁable. ∀xσ ∀v τ (x · (−v) = −(x · v)). it follows that S0 is normally satisﬁable. The answer is yes. denoting scalar multiplication. Example 5. Hence S includes the ﬁeld axioms.e.

Vector addition is given by ai | i < κ + b i | i < κ = a i + b i | i < κ .2.17 (elimination of quantiﬁers). A famous and important theorem of Tarski says that RCOF is complete. 3. . <) is said to be real-closed if for all polynomials f (x) = an xn + · · · + a1 x + a0 . we have that each sentence in the language of RCOF is equivalent over RCOF to a quantiﬁer-free sentence. An ordered ﬁeld (F. the unique vector space over F of dimension κ is the familiar space i<κ F . together with <.e. if f (b) < 0 < f (c) then there exists x ∈ F between b and c such that f (x) = 0. The language of ordered rings consists of the language of rings +. with ai ∈ F for all i.16 (ordered algebraic structures). ·. Clearly the ordered ﬁeld of real numbers (R. Remark 5. a0 ∈ F . =. plus the linear ordering axioms. 2. This holds despite the fact that RCOF is not κ-categorical for any κ. An example of an ordered ﬁeld is the ﬁeld of rational numbers (Q. It can be shown that. . <) is real-closed. . plus ∀x ∀y ∀z (x < y ⇔ x + z < y + z). an . 1. .2. . . The ordered ﬁeld axioms consist of the ﬁeld axioms. An important invariant of a vector space is its dimension. vn ∈ S and scalars a1 . <) with its usual ordering. a1 . . . −.. i. For example. Here the vectors are sequences ai | i < κ . ∀x ∀y ((x > 0 ∧ y > 0) ⇒ x · y > 0).. . S) is said to admit elimination of quantiﬁers if for all L-formulas A there exists a quantiﬁer-free Lformula A∗ such that T ⊢ the universal closure of A ⇔ A∗ . and scalar multiplication is given by c · ai | i < κ = c · ai | i < κ . An ordered ﬁeld is a model of these axioms. As a special case of quantiﬁer elimination. 0. . . Let RCOF be the theory of realclosed ordered ﬁelds. c ∈ F . Tarski’s method of proof is as follows. . it is evident that the quantiﬁer-free sentences of the language 95 . Note that an ordered ﬁeld is real-closed if and only if it satisﬁes the axioms ∀u ∀v ∀w0 ∀w1 · · · ∀wn ((u < v ∧ wn un + · · · + w1 u + w0 < 0 < wn v n + · · · + w1 v + w0 ) ⇒ ∃x (u < x < v ∧ wn xn + · · · + w1 x + w0 = 0)) for n = 1. the minimum cardinality of a spanning set. A theory T = (L. and ai = 0 for all but ﬁnitely many i. Example 5. up to isomorphism over a ﬁeld F . . Tarski uses algebraic methods to show that the theory RCOF admits elimination of quantiﬁers.a1 ·v1 +· · ·+an ·vn for some v1 . . On the other hand. and for all b. an . the formula ∃x (ax2 + bx + c = 0) is equivalent over RCOF to the quantiﬁer-free formula (a = b = c = 0) ∨ (a = 0 = b) ∨ (a = 0 ≤ b2 − 4ac). .

It follows immediately that the theory of dense linear orderings with end points is ℵ0 -categorical. torsion-free. q. where q is the cardinality of F .of RCOF are of a very simple nature. 3. The theory of inﬁnite-dimensional vector spaces over a ﬁeld of 5 elements. On the other hand. . as can be seen by considering the sentence ∃x (x + x = 1). any vector space of cardinality κ over a ﬁnite ﬁeld is of dimension κ. 1 + 0 = 1 ∧ (1 · 1) + 1 < 1 + 1 + 1. As noted in Example 5. it follows that RCOF is complete. Abelian. The theory of ﬁelds of characteristic 0. Therefore. as can be seen by considering the sentences ∃x (x · x = 1 + 1) and ∃z (z · z = −1). . 1. 5. the theory of ﬁnite-dimensional vector spaces over F is incomplete. the theory of ﬁelds of characteristic 0 is incomplete. 4. The vector spaces of dimension n = 0.15. . TF is complete. The ﬁelds Q. q 2 . 2. the theory of dense linear orderings without end points is ℵ0 -categorical.18. the theory of inﬁnite. Since the truth values of such sentences are decided by the axioms of RCOF. 96 . Therefore. .2. Solution. we see that for any particular ﬁnite ﬁeld F . The theory of ﬁnite-dimensional vector spaces over a ﬁeld of 5 elements. Hence. Hence.g. over a particular ﬁnite ﬁeld F are not elementarily equivalent. by Vaught’s Test. torsion-free groups is incomplete. the theory TF of inﬁnite-dimensional vector spaces over F is κ-categorical.. because they have diﬀerent ﬁnite cardinalities 1. 1.2. e. 4. and C are of characteristic 0. The theory of inﬁnite. 2. 2. As noted in Remark 5. Therefore. Combining these facts. Which of the following theories are complete? Justify your answers. yet they are not elementarily equivalent. 5. R. The additive groups of Z and Q are inﬁnite. by Vaught’s Test. Exercise 5. and torsion-free. . . 3.2. Abelian. 1. The theory of dense linear orderings with end points. Abelian groups. for any inﬁnite cardinal number κ. any two vector spaces over the same ﬁeld of the same dimension are isomorphic. each of these theories is complete.6. yet they are not elementary equivalent.

3. n. rk . . k. 97 .4. . <. We omit the proof of this well-known result.7 (G¨del’s beta-function). .6. For each k ∈ N we can ﬁnd inﬁnitely many a ∈ N such that the integers a + 1. . ·. . mk be pairwise relatively prime. we can ﬁnd r such that r ≡ ri mod mi for all i = 1. 0. nk .2. Suppose p is a prime number which divides both ai + 1 and aj + 1 where 1 ≤ i < j ≤ k. .3. We deﬁne β(a. . . a < b if and only if R |= ∃x (x = 0 ∧ a + x2 = b). r such that β(a. . . The exponential function (n. if there exists an L-formula D with n free variables x1 .3. 2a + 1. the 3-ary relation { m. . Let a be any multiple of k!. . ·.8. . Proof. Example 5.3. Let m1 . Given a ﬁnite sequence of natural numbers n0 . Note that the β-function is explicitly deﬁnable over the natural number system. . we see that for all a. or just deﬁnable over M . . Proof. let M be an Lstructure. It can be shown that the class of relations which are explicitly deﬁnable over N includes all relations which are computable in the sense of Turing. xn such that R = { a1 . i) = the remainder o of r upon division by a·(i+1)+1.1 (explicit deﬁnability). −. Lemma 5. . r. On the other hand.5. in view of Tarski’s theorem on elimination of quantiﬁers for RCOF (see Remark 5. 0. Since p divides ai + 1 and a is a multiple of k!. Lemma 5. From this it follows that. any subset of R which is deﬁnable over R consists of a union of ﬁnitely many points and intervals. b ∈ R. . . .3.3. =) by the formula ∃z (z = 0 ∧ x + z 2 = y) with free variables x. k) → nk is explicitly deﬁnable over the natural number system. we can ﬁnd natural numbers a. a contradiction since p divides ai + 1. . We say that R is explicitly deﬁnable over M . . .2. Deﬁnition 5. and j − i < k. . 1. . Thus < is explicitly deﬁnable over the commutative ring R = (R. Theorem 5. . . p divides a(j − i).3 Deﬁnability over a Model Deﬁnition 5. +. . xn /an ]}. . Lemma 5. .17).3. Given r1 . y. we clearly have p > k. i) = ni for all i = 0. for example. b ∈ R2 | a < b} on the set R of real numbers. . as follows. as we now show. r. ka + 1 are pairwise relatively prime. =) be the natural number system. Let N = (N.5 (the Chinese Remainder Theorem). 1. . . the set of integers is not deﬁnable over R. . +. and let R be an n-ary relation on UM .3. On the other hand. k ∈ N3 | m = nk } is deﬁnable over N. Viewing R as a commutative ring. . an | M |= D[x1 /a1 . The proof uses some number-theoretic lemmas.3. Example 5. Proof. k. Consider the binary relation < = { a. . In particular. 1. hence p divides a. Let L be a language. . n1 .

Z is not explicitly deﬁnable over R. Thus p0 = 2. an ∈ R} for all automorphisms f of M . r. show that the binary relation { k. Consider the set of integers. Now.11. r. Example 5.3. (This means. Let A be a formula expressing that x is prime.10 (implicit deﬁnability).13 (automorphisms). As noted above. pk | k ∈ N} is explicitly deﬁnable over N. p4 = 11. i + 1)) and C is ∀x (x ≤ n ⇒ (A ⇔ ∃i (i ≤ k ∧ β(a. =) be the ordered ﬁeld of real numbers.3. . 0. . such that for all n-ary relations R′ on UM . It can be shown that there exists a subset of N which is implicitly deﬁnable over the natural number system N = (N. However. i + 1))).3. r. we can ﬁnd r such that r ≡ ni mod a · (i + 1) + 1 for all i ≤ k. k) = n ∧ ∀i (i < k ⇒ β(a. r. note that m = nk holds if and only if ∃a ∃r (β(a.6 and Exercise 6. let M be an L-structure.4. Let L be a language. (M. it follows that β(a. An n-ary relation R on UM is said to be invariant if R = { f (a1 ). i) · n = β(a. Example 5. <. R′ ) |= D∗ if and only if R′ = R. and let R be an n-ary relation on UM . i) = ni . . r. i) = x))). i) < β(a. −. +. . ·. By the Chinese Remainder Theorem. <. ·. r.3.) Solution. 1.Proof. are pairwise relatively prime. An automorphism of M is an isomorphism of M onto itself. Let R = (R. . r. We then have pk = n if and only if ∃a ∃r (B ∧ C). p3 = 7.12. r. k) = m ∧ ∀i (i < k ⇒ β(a.9. Let pn be the nth prime number.4. Since 0 ≤ ni ≤ a < a · (i + 1) + 1. Let a ≥ max{ni | i ≤ k} be such that a · (i + 1) + 1. We say that R is implicitly deﬁnable over M if there exists a sentence D∗ in the language L∗ = L ∪ {R} with an additional n-ary predicate R.7 below. .4. Deﬁnition 5. for example 1 < x ∧ ¬ ∃u ∃v (u < x ∧ v < x ∧ u · v = x). We now turn to implicit deﬁnability. by the following sentence with an additional unary predicate Z: Z0 ∧ Z1 ∧ (¬ ∃x (Zx ∧ 0 < x < 1)) ∧ (∀x (Zx ⇔ Z(x + 1))). Show that the function k → pk is explicitly deﬁnable over the natural number system N. p2 = 5. Exercise 5. Let M be a structure.3. Z ⊂ R. 0) = 1 ∧ β(a. See Remark 6. Deﬁnition 5.3. where B is β(a. f (an ) | a1 . 0. to prove Theorem 5. . =) but is not explicitly deﬁnable over N. 1. r. p1 = 3. etc. i ≤ k. . Z is implicitly deﬁnable over R. +. 98 .

. . Prove that there exists a countable group G′ such that 1. xn . . R) onto (M. R′ ) |= D∗ .3. an ) is elementarily equivalent to (M. inasmuch as R has no automorphisms except the identity. Let us say that a1 .Exercise 5. . Since there are only countably many sentences. R is invariant under f . If R ⊆ (UM )n is implicitly deﬁnable over M . G′ is elementarily equivalent to G. Show that if UM is ﬁnite.15. Exercise 5. . . Exercise 5. an . If R is explicitly deﬁned over M by a formula D with free variables x1 . Example 5. f (an ) | a1 . all subsets of R are invariant under automorphisms of R. .3. −. . . an ∈ R}. Hence R′ = R. the n-types are just the orbits of (UM )n under the automorphism group of M .3. .12. Solution. i. . Let M be a structure. there are only ﬁnitely many n-types.16. .3. 99 . Then f is an isomorphism of (M. .14. it follows by Theorem 2. . However. Solution. . hence R is the union of some of the n-types. Show that any relation which is implicitly deﬁnable over M is invariant under all automorphisms of M . . . =) be the ordered ﬁeld of real numbers. 1. then R is implicitly deﬁned over M by the L∪{R}-sentence ∀x1 · · · ∀xn (Rx1 · · · xn ⇔ D). by Theorem 2. a1 . using the ǫ-δ-method. .6 and Exercise 4. (Hint: The relations y = ex and y = sin x are implicitly deﬁned by diﬀerential equations which can be expressed as formulas of the predicate calculus. b1 .17. Let R be an n-ary relation over UM . and 2.2. R) |= D∗ .1. and put R′ = { f (a1 ). . It can be shown that these relations are not explicitly deﬁnable over R. R′ ).e.) Show that if a structure M is saturated. hence R is explicitly deﬁnable over M . . b1 . then R is invariant under automorphisms of M . Now suppose R is implicitly deﬁned over M by an L∪{R}-sentence D∗ . bn ).11 shows that the subset Z of R is implicitly deﬁnable over R but not explicitly deﬁnable over R. Show that the relations y = ex and y = sin x are implicitly deﬁnable over R. then any relation which is implicitly deﬁnable over M is explicitly deﬁnable over M . Let R = (R. . bn ∈ (UM )n are of the same ntype if (M.. Exercise 5. and each n-type is explicitly deﬁnable over M . . Since (M. Consider the ordered ﬁeld R of real numbers.3. Since UM is ﬁnite. Thus (UM )n is partitioned into equivalence classes. the n-types. Let f be an automorphism of M . Give counterexamples showing that the converses of these assertions fail in general. . . <.) Exercise 5. only countably many subsets of R are implicitly deﬁnable over R. G′ has a subgroup which is not explicitly deﬁnable over G′ .2. then any relation which is implicitly deﬁnable over M is explicitly deﬁnable over M . 0.6 that (M. .3. Show that any relation which is explicitly deﬁnable over M is implicitly deﬁnable over M . . (In this exercise we assume familiarity with saturated models. Moreover. ·. Let G be a group which has inﬁnitely many distinct subgroups. . +.18. .

P ′ ) sato isfying S ′ .8.e. This method of extending a theory is known as deﬁnitional extension. . 5. T ⊢ ∀x1 · · · ∀xn ∃z ∀y (Iyz ⇔ D) where z is a new variable. and let D be an L-formula with free variables x1 . we shall consider the related but more restrictive notions of explicit and implicit deﬁnability over a theory. .1 (deﬁning a new predicate). xn . S) be a theory with an identity predicate I. S ′ ) where L′ = L ∪ {f } and S ′ = S ∪ {∀x1 · · · ∀xn ∀y (If x1 . .4. . Theorem 5. A pleasant surprise is that explicit deﬁnability over T is equivalent to implicit deﬁnability over T . an implicit deﬁnitional extension of a theory is always equivalent to an explicit deﬁnitional extension of the same theory. Let D be an L-formula with free variables x1 .4. Later. T . Introduce a new unary predicate P . implicit deﬁnability over M does not imply explicit deﬁnability over M . i. In this section we are considering only explicit deﬁnitional extensions. xn y ⇔ D)}.2 (deﬁning a new operation). xn ..Solution. It will be obvious that explicit deﬁnability over T implies explicit deﬁnability over any model of T . Remark 5. and let T ′ = (L′ . . y such that T ⊢ ∀x1 · · · ∀xn (∃ exactly one y) D. M . We have given examples showing that. S) be a theory. It follows by Compactness plus L¨wenheim/Skolem that we can ﬁnd a countable normal structure (G′ . and P ′ is a subgroup of G′ which is not explicitly deﬁnable over G′ . . Introduce a new n-ary predicate P . Let S be the set of sentences which are true in G. . Since G has inﬁnitely many subgroups. Deﬁnition 5. In Sections 5. in Section 5. Let T = (L. This is the content of Beth’s Deﬁnability Theorem. each ﬁnite subset of S ′ is normally satisﬁable in G by letting P be an appropriately chosen subgroup of G. Let T = (L. 100 . Deﬁnition 5.19.4 Deﬁnitional Extensions of Theories We now show how theories can be usefully extended by adding new predicates and operations which are explicitly deﬁnable in terms of old predicates and operations. in general. S ′ ) where L′ = L ∪ {P } and S ′ = S ∪ {∀x1 · · · ∀xn (P x1 · · · xn ⇔ D)}.4 and 5. we shall consider implicit deﬁnitional extensions. and implicit deﬁnability over T implies implicit deﬁnability over any model of T . Introduce a new n-ary operation f . .2 below. . in principle. Then G′ is a countable group which is elementarily equivalent to G. . It will turn out that. and let T ′ = (L′ . and let S ′ consist of S plus the sentences P 1 and ∀x (P x ⇒ P x−1 ) and ∀x ∀y ((P x ∧ P y) ⇒ P (x · y)) and ¬ ∀x (P x ⇔ D) where D is any formula with x as its only free variable.8 below.8.3. In this section we have considered explicit and implicit deﬁnability over a model.

Remark 5.4.3. In Deﬁnition 5.4.2 above, we have assumed for simplicity that T is one-sorted. In case T is many-sorted, we make the obvious modiﬁcations. Namely, letting σ1 , . . . , σn , σn+1 be the sorts of the variables x1 , . . . , xn , y respectively, we require that z is a new variable of sort σn+1 , I is an identity predicate of sort σn+1 , and f is an operation of type σ1 × · · · × σn → σn+1 . Remark 5.4.4. We are going to prove that these extensions of T are “trivial” or “harmless” or “inessential”, in the sense that each formula in the extended language L ∪ {P } or L ∪ {f } can be straightforwardly translated into an equivalent formula of the original language L. See Theorem 5.4.10 below. Lemma 5.4.5. Let T = (L, S) and T ′ = (L′ , S ′ ) be as in Deﬁnition 5.4.1 or 5.4.2. Then for all L-sentences A we have T ′ ⊢ A if and only if T ⊢ A. Proof. It suﬃces to show that, for each model M of T , there exists a model M ′ of T ′ such that M = M ′ ↾L, i.e., M is the reduct of M ′ to L. In the case of Deﬁnition 5.4.1, let M ′ = (M, PM ′ ) where PM ′ = { a1 , . . . , an | M |= D[x1 /a1 , . . . , xn /an ]}. In the case of Deﬁnition 5.4.2, let M ′ = (M, fM ′ ) where fM ′ (a1 , . . . , an ) = the unique b such that M |= D[x1 /a1 , . . . , xn /an , y/b]. Clearly M ′ is as desired. Deﬁnition 5.4.6 (translation). Let T = (L, S) and T ′ = (L′ , S ′ ) be as in Deﬁnition 5.4.1 or 5.4.2. To each L′ -formula A we associate an L-formula A′ , the translation of A into L. In order to deﬁne A′ , we ﬁrst modify D, replacing the free and bound variables of D by new variables which do not occur in A. (Compare Deﬁnition 2.4.9.) In the case of Deﬁnition 5.4.1, where we are adding a new predicate P , we obtain A′ from A by replacing each atomic formula of the form P t1 . . . tn by D[x1 /t1 , . . . , xn /tn ]. In the case of Deﬁnition 5.4.2, where we are adding a new operation f , the translation is more complicated. For non-atomic A we obtain A′ by induction on the degree of A, putting (¬ A)′ = ¬ A′ , (A ∧ B)′ = A′ ∧ B ′ , (∀x A)′ = ∀x A′ , etc. For atomic A we obtain A′ by induction on the number of occurrences of f in A. If there are no occurrences of f in A, let A′ be just A. Otherwise, write A in the form B[w/f t1 · · · tn ] where w is a new variable and f does not occur in t1 , . . . , tn , and let A′ be ∃w (D[x1 /t1 , . . . , xn /tn , y/w] ∧ B ′ ) or equivalently ∀w (D[x1 /t1 , . . . , xn /tn , y/w] ⇒ B ′ ). Lemma 5.4.7. Let T and T ′ be as in Deﬁnition 5.4.1 or 5.4.2, and let A → A′ be our translation of L′ -formulas to L-formulas as in Deﬁnition 5.4.6. Then: 1. A′ has the same free variables as A and is equivalent to it over T ′ , i.e., T ′ ⊢ A ⇔ A′ .

101

2. Propositional connectives and quantiﬁers are respected, i.e., we have (¬ A)′ ≡ ¬ A′ , (A ∧ B)′ ≡ A′ ∧ B ′ , (∀x A)′ ≡ ∀x A′ , etc. 3. If A happens to be an L-formula, then A′ is just A. Consequently, for all L′ -sentences A we have T ′ ⊢ A if and only if T ⊢ A′ . Proof. The ﬁrst part is straightforward. For the last part, since T ′ ⊢ A ⇔ A′ , we obviously have T ′ ⊢ A if and only if T ′ ⊢ A′ . But then, since A′ is an L-sentence, it follows by Lemma 5.4.5 that T ′ ⊢ A if and only if T ⊢ A′ . Deﬁnition 5.4.8 (deﬁnitional extensions). Let T be a theory. A deﬁnitional extension of T is a theory T ∗ which is a union of sequences of theories T0 , T1 , . . . , Tk beginning with T0 = T such that each Ti+1 is obtained by extending Ti as in Deﬁnition 5.4.1 or 5.4.2. Deﬁnition 5.4.9 (conservative extensions). Let T = (L, S) and T ∗ = (L∗ , S ∗ ) be theories such that L∗ ⊇ L. We say that T ∗ is a conservative extension of T if, for all L-sentences A, T ∗ ⊢ A if and only if T ⊢ A. Theorem 5.4.10. Let T ∗ be a deﬁnitional extension of T . Then T ∗ is a conservative extension of T . Moreover, there is a straightforward translation A → A∗ of L∗ -formulas to L-formulas, with the following properties. 1. A∗ has the same free variables as A and is equivalent to it over T ∗ , i.e., T ∗ ⊢ A ⇔ A∗ . 2. Propositional connectives and quantiﬁers are respected, i.e., we have (¬ A)∗ ≡ ¬ A∗ , (A ∧ B)∗ ≡ A∗ ∧ B ∗ , (∀x A)∗ ≡ ∀x A∗ , etc. 3. If A happens to be an L-formula, then A∗ is just A. Consequently, for all L∗ -sentences A, we have T∗ ⊢ A if and only if T ⊢ A∗ . Proof. This is clear from Lemmas 5.4.5 and 5.4.7. Remark 5.4.11. In practice, when it comes to working with speciﬁc theories T , the technique of deﬁnitional extensions is very useful. This is because formulas written in the extended language L∗ tend to be much shorter than their translations in L. This is particularly important for the foundational theories Z1 , Z2 , . . . , Zn , . . . , Z∞ , ZC, ZFC which are discussed in Sections 5.5 and 5.6 below. Exercise 5.4.12. Show that for any theory T there is a deﬁnitional extension of T which admits elimination of quantiﬁers. Solution. Let L be the language of T . For each L-formula A with free variables x1 , . . . , xn introduce a new n-ary predicate PA and a new axiom ∀x1 · · · ∀xn (PA x1 · · · xn ⇔ A). The resulting theory admits elimination of quantiﬁers. 102

Exercise 5.4.13. Let T = (L, S) be a theory. Let A be an L-formula with free variables x1 , . . . , xn , y. Let T ′ = (L′ , S ′ ) where L′ consists of L plus a new nary operation f , and S ′ consists of S plus ∀x1 · · · ∀xn (A[y/f x1 · · · xn ] ⇔ ∃y A). Show that T ′ is a conservative extension of T . Solution. Let B be an L-sentence such that T ′ ⊢ B. We must show that T ⊢ B. If T ⊢ B, let M be a model of T ∪ {¬ B}. Let M ′ = (M, fM ′ ) where fM ′ is an n-ary function on UM with the following property: for all a1 , . . . , an ∈ UM such that M |= ∃y A[x1 /a1 , . . . , xn /an ], we have M |= A[x1 /a1 , . . . , xn /an , y/b] where b = fM ′ (a1 , . . . , an ). (Note that fM ′ is not necessarily unique and is not necessarily deﬁnable over M . The existence of an fM ′ with the desired property follows from the axiom of choice.) Clearly M ′ is a model of T ′ ∪ {¬ B}, hence T ′ ⊢ B, a contradiction.

5.5

Foundational Theories

In this section and the next, we present several speciﬁc theories which are of signiﬁcance in the foundations of mathematics. We point out that these theories are, in certain senses, “almost complete” or “practically complete.” On the other hand, we shall see in Chapter 6 that these theories are not complete. Deﬁnition 5.5.1 (ﬁrst-order arithmetic). One of the most famous and important foundational theories is ﬁrst-order arithmetic, Z1 , also known as Peano Arithmetic, PA. The language of ﬁrst-order arithmetic is L1 = {+, ·, 0, S, =} where + and · are binary operations, S is a unary operation, 0 is a constant, and = is the equality predicate. Let Q be the ﬁnitely axiomatizable L1 -theory with axioms ∀x (Sx = 0), ∀x ∀y (Sx = Sy ⇒ x = y), ∀x (x + 0 = x), ∀x ∀y (x + Sy = S(x + y)), ∀x (x · 0 = 0), ∀x ∀y (x · Sy = (x · y) + x). The axioms of Z1 consist of the axioms of Q plus the induction scheme, i.e., the universal closure of (A[x/0] ∧ ∀x (A ⇒ A[x/Sx])) ⇒ ∀x A where A is any L1 -formula. Note that the induction scheme consists of an inﬁnite set of axioms. It can be shown that Z1 is not ﬁnitely axiomatizable.

103

However. ·. the axioms of Z1 are intended to be complete. The idea behind the axioms of Z1 is that we are trying to write down a set S1 of L1 -sentences with the property that. of L1 with the number variables m. ·. = of L1 with the +. we can say that Z1 is “practically complete. . +. in practice.5 (second-order arithmetic). y. Thus L1 is a sublanguage of L2 . S. of L2 . See also my book [3]. . y τ . The language of Z2 is a 2-sorted language.3. Thus we have a deﬁnitional extension of Z1 in which basic properties such as ∀m ∀n ∀k (mk nk = (mn)k ) and ∀n ∀i ∀j (ni nj = ni+j ) can be stated and proved.5.5. ·. Here + and · are binary number operations. but the exceptions are obscure and marginal. 0. and 0 is of type σ. y σ . Unfortunately this intention cannot be fulﬁlled. .. and 0 is a numerical constant. Exercise 5. In addition. the intended model of Z1 .4. =) where N = {0. Z2 . This is because it has been found that. all or most of the theorems of number theory which can be written in (deﬁnitional extensions of) Z1 are either provable or refutable in Z1 . The intended model of Z1 is the natural number system N = (N. Similarly it can be shown that the associative laws ∀x ∀y ∀z (x + (y + z) = (x + y) + z) and ∀x ∀y ∀z (x · (y · z) = (x · y) · z) and the distributive law ∀x ∀y ∀z (x · (y + z) = (x · y) + (x · z)) are theorems of Z1 .9 that the exponential function (n. n. and S is the successor function. Another important foundational theory is second-order arithmetic. k. with sorts σ and τ designating numbers and sets respectively. Deﬁnition 5. . Z. . The number variables xσ . The predicates and operations of L2 are +. . m. are written as X. S. . in deﬁnitional extensions of Z1 . 0. and their properties proved. it is known o that Z1 “almost” fulﬁlls the intention. 104 . the numerical equality predicate = is a binary predicate of type σ × σ. In particular.2 (practical completeness of ﬁrst-order arithmetic). = of L2 . .} and +. 1. Thus + and · are of type σ × σ → σ. S is a unary number operation.3. . ·. Our remarks above concerning practical completeness of Z1 apply all the more to these deﬁnitional extensions of Z1 . we identify the operations and predicates +. . n. j.5. . We identify the variables x. =. and the membership predicate ∈ is a binary predicate of “mixed” type σ × τ . . . Also. . ∈.5. . S. . S is of type σ → σ. S1 ⊢ A if and only if N |= A.3 (deﬁnitional extensions of ﬁrst-order arithmetic). for all L1 -sentences A.Remark 5. Obviously the axioms of Z1 are true in this model. Similarly. as shown by G¨del’s First Incompleteness Theorem (see Chapter 6).. Y. 0.” Remark 5. 0. ·. S(n) = n + 1 for all n ∈ N. S. 0. Show that the commutative laws ∀x ∀y (x + y = y + x) and ∀x ∀y (x · y = y · x) are theorems of Z1 . . . . L2 . In this sense. It can be shown that the same deﬁnitions work abstractly over the theory Z1 .4 and Exercise 5. We have seen in Theorem 5. while the set variables xτ . l. . are written as i. k) → nk and the “kth prime” function k → pk are deﬁnable over N. = are as expected. . Exceptions are known. 2. many other number-theoretical operations can be introduced.

it is obvious that the axioms of Z2 are true in the 2-sorted structure P (N)..5. Note that the comprehension scheme consists of an inﬁnite set of axioms. plus the comprehension scheme.1)..5. descriptive set theory. The intended model of Z2 is the 2-sorted structure P (N) = (N. The foundational signiﬁcance of Z2 is that. Remark 5.2). +. 0.e. S. This includes diﬀerential equations.e. This intention cannot be fulﬁlled because of the G¨del Incompleteness Theorem.” Remark 5. P (N) = {X | X ⊆ N}.5. ∀m (m · 0 = 0). the universal closure of ∃X ∀n (n ∈ X ⇔ A) where A is any L2 -formula in which X does not occur.7 (deﬁnitional extensions of second-order arithmetic). +. =. ∈) where (N.5. plus the induction axiom ∀X ((0 ∈ X ∧ ∀n (n ∈ X ⇒ Sn ∈ X)) ⇒ ∀n (n ∈ X)). it is possible and convenient to develop the bulk of ordinary countable and separable mathematics. and ∈ is the membership relation between natural numbers and sets of natural numbers. ∀m (Sm = 0). within deﬁnitional extensions of Z2 . P (N) is the power set of N.2). =) is the natural number system (see Remark 5. i. It can be shown that Z2 is not ﬁnitely axiomatizable.6 (practical completeness of second-order arithmetic). analysis. ∈ = { n. P (N). ∀m (m + 0 = m). combinatorics. i. 0. S. ∀m ∀n (m · Sn = (m · n) + m). Z2 is “practically complete. the set of all subsets of N. For details of how this can be done. etc. but o again. ·.e. geometry. X ∈ N × P (N) | n ∈ X}. topology. and again.. ∀m ∀n (Sm = Sn ⇒ m = n). ·. see my book [3].. algebra. As in the case of Z1 (compare Remark 5. functional analysis.5. ∀m ∀n (m + Sn = S(m + n)). i.e. the idea behind Z2 is that we are trying to write down axioms for the complete theory of P (N).The axioms of Z2 consist of Q (Deﬁnition 5. 105 . i.

the membership predicate and the identity predicate. . . S. . Show that all of the axioms of Z1 are theorems of Z2 . nk . From the induction axiom of Z2 we have (0 ∈ X ∧ ∀n (n ∈ X ⇒ Sn ∈ X)) ⇒ ∀n (n ∈ X). . Deﬁnition 5. Lset . .5. X1 . an n-sorted theory whose intended model is the n-sorted structure P n−1 (N) = P (P · · · (P (N)) · · ·) n−1 where P is the power set operation. X1 . X1 . In this sense Z1 is a subtheory of Z2 . we shall prove this when A is any L2 -formula. More generally. =. . The language of set theory. for each positive integer n ≥ 1 we can deﬁne nth-order arithmetic. P (N). known as axiomatic set theory. From this it follows that (A[n/0] ∧ ∀n (A ⇒ A[n/Sn])) ⇒ ∀n A. Z3 .1 (the language of set theory). w. y. Since this holds for arbitrary n1 . Similarly we can deﬁne third-order arithmetic. . the universal closure of (A[n/0] ∧ ∀n (A ⇒ A[n/Sn])) ⇒ ∀n A is a theorem of Z2 . Xl . . sets of numbers. v. . The intended model of Z3 is the 3-sorted structure P 2 (N) = (N. . . More generally. are intended to range over 106 . . nk .Exercise 5. The union Z∞ = n=1 Zn is known as simple type theory. ∞ 5. Recall that Q is a subtheory of Z2 . ·. . . . we use the comprehension scheme of Z2 to prove the existence of a set X such that ∀n (n ∈ X ⇔ A). Remark 5. ∈2 ) where ∈1 = { n.6 Axiomatic Set Theory We now turn to another kind of foundational theory. ∈1 . The variables u. ∈ and =.5. Solution. for any L1 -formula A. Xl . . n1 . P (P (N)). x. Zn . . . A ∈ P (N) × P (P (N)) | X ∈ A}. Given n1 .8.5. +.5. . . and sets of sets of numbers. It remains to show that. we obtain the universal closure. . Thus we have a sequence of theories Z1 ⊆ Z2 ⊆ · · · ⊆ Zn ⊆ Zn+1 ⊆ · · · .9 (nth order arithmetic). . . . because it or something like it was oﬀered by Russell as a solution of the Russell Paradox. .6. z. . nk . Xl . Let the free variables of A be among n. X ∈ N × P (N) | n ∈ X} and ∈2 = { X. . This theory is historically important. As noted in Deﬁntion 5. L1 is a sublanguage of L2 . 0. a 3-sorted theory with variables intended to range over numbers. Within Z2 we reason as follows. consists of two binary predicates.

the ordered pair consisting of x and y. We also deﬁne a binary operation x∪y = {x. We also use notations such as x = {u. ZC. x is a subset of y. v. Note that u ∈ x means that u is an element of the set x. Using extensionality. The pairing axiom: ∀x ∀y ∃z ∀u (u ∈ z ⇔ (u = x ∨ u = y)). so we deﬁne a unary operation x = this z. the union of x and y. . We also deﬁne a binary operation (x.sets. which embodies our restriction to well-founded sets. Later we shall introduce another axiom. a member of x. we ﬁrst note that the axioms are intended to apply only to sets which are pure and well-founded. y) = {{x}. so we deﬁne a binary operation {x.} meaning that x is a set whose elements are all u such that . 107 . The axiom of extensionality: ∀x ∀y (x = y ⇔ ∀u (u ∈ x ⇔ u ∈ y)). i. The union axiom: ∀x ∃z ∀u (u ∈ z ⇔ ∃v (u ∈ v ∧ v ∈ x)). A set x is said to be pure if all elements of x are sets. Note that {x. . 2. i. {x.4.6. . By extensionality this z is unique. etc. the axiom of foundation. all elements of elements of elements of x are sets. i. we shall tacitly employ the technique of deﬁnitional extensions.. Remark 5. . y}}.e. and x = {u | . all elements of elements of x are sets.6. which has been discussed in Section 5. Remark 5. .6. . Note that {x} is the singleton set consisting of x. 1 Recall that L set is a one-sorted language with only set variables. v) ⇔ (x = u ∧ y = v)). y) = (u. . In order to motivate and clarify our presentation of the axioms of set theory. 4. . v.2 (pure well-founded sets). the union of all of the sets which are elements of x.. . y}.3. Deﬁnition 5.4 (Zermelo set theory). Zermelo set theory with the axiom of choice. The power set axiom: ∀x ∃z ∀y (y ∈ z ⇔ y ⊆ x).e.e. Extensionality1 says that x = y is equivalent to x ⊆ y ∧ y ⊆ x. 3. In addition. A set x is said to be well-founded if there is no inﬁnite descending ∈-chain · · · ∈ un+1 ∈ un ∈ · · · ∈ u2 ∈ u1 ∈ u0 = x. x}. . is a theory in the language Lset consisting of the following axioms. we can prove the basic property ∀x ∀y ∀u ∀v ((x. .} meaning that x is a set whose elements are u. y} is the unordered pair consisting of x and y. 1.. We deﬁne a binary predicate ⊆ by x ⊆ y ⇔ ∀u (u ∈ x ⇒ u ∈ y). By extensionality this z is unique. Note that x is the union of x. In order to state the axioms and theorems of set theory eﬃciently.. y} = this z. we deﬁne a unary operation {x} = {x. This fact together with the axiom of extensionality embodies our restriction to pure sets.

The comprehension scheme asserts the existence of a set z = {u ∈ x | A}. The idea here is that x is a given set.By extensionality this z is unique. the domain and range of f are deﬁned as unary operations f | ∃y ((x. y))) and ∀x ∀y ∀z (((x. i. We also deﬁne f (x).. and / (c) x × y = {w ∈ P (P (x ∪ y)) | ∃u ∃v (u ∈ x ∧ v ∈ y ∧ (u. z) ∈ f ) ⇒ y = z). the set of all subsets of x. we deﬁne binary operations (a) x ∩ y = {u ∈ x | u ∈ y}. We also deﬁne a constant ∅ = {} = x \ x. (b) x \ y = {u ∈ x | u ∈ y}. and A expresses a property of elements u ∈ x. y) ∈ f .. y) ∈ f ∧ (x. 5. i. v) = w)}. By extensionality. Note that the comprehension scheme consists of an inﬁnite set of axioms. P (x) = {y | y ⊆ x}. Using comprehension.e. the set-theoretic diﬀerence of x and y. this z is unique. a set f such that ∀w (w ∈ f ⇒ ∃x ∃y (w = (x. to be the unique y such that (x. It can be shown that ZC is not ﬁnitely axiomatizable. so we deﬁne a unary operation P (x) = this z.e. the Cartesian product of x and y. the value of f at x ∈ dom(f ). y) ∈ f )}. the empty set. Note that P (x) is the power set of x. Using comprehension. z is a subset of x consisting of all u ∈ x such that A holds. 108 . The comprehension scheme: the universal closure of ∀x ∃z ∀u (u ∈ z ⇔ (u ∈ x ∧ A)) where A is any Lset -formula in which z does not occur. We deﬁne a unary predicate Fcn(f ) saying that f is a function. y) ∈ f )} dom(f ) = {x ∈ and ran(f ) = {y ∈ f | ∃x ((x. the intersection of x and y.

we can prove the existence of a unique smallest set z as above. the set of natural numbers. Using comprehension. Similarly. Note that ω ⊆ HF. For m. 7. i. We deﬁne a constant ω = this w. . for all n ∈ ω. The axiom of inﬁnity: ∃z (∅ ∈ z ∧ ∀x ∀y ((x ∈ z ∧ y ∈ z) ⇒ x ∪ {y} ∈ z)).e. those pure well-founded sets x such that C(x) is ﬁnite. i. . On this basis. We deﬁne |x|. The axiom of choice says that. for all n. n − 1}. To see this. n + 1 = n ∪ {n}. namely the intersection of all such sets. . there exists a function which chooses one element from each of the sets.6. we can prove that there exists a unique smallest set w such that ∅ ∈ w ∧ ∀x (x ∈ w ⇒ x ∪ {x} ∈ w).. inductively identiﬁed with hereditarily ﬁnite sets via n = {0. We deﬁne a constant HF = this z. |n| = n. The elements of HF are the hereditarily ﬁnite sets. Note that HF is an inﬁnite set. n ∈ ω we deﬁne m + n = |(m × {0}) ∪ (n × {1})| and m · n = |m × n|. Thus we have a copy of the natural number system. elements of elements of elements of x. 1.. . The axiom of foundation amounts to saying that all sets are well-founded. and we have 0 = ∅ and. given an indexed family of nonempty sets. note that if there were an inﬁnite descending ∈-chain · · · ∈ un+1 ∈ un ∈ · · · ∈ u2 ∈ u1 ∈ u0 109 . we can prove the essential properties of + and · on ω. We can prove that. . Here C(x) = x ∪ x∪ x ∪···. The axiom of choice: ∀f ((Fcn(f ) ∧ ∀x (x ∈ dom(f ) ⇒ f (x) = ∅)) ⇒ ∃g (Fcn(g) ∧ dom(g) = dom(f ) ∧ ∀x (x ∈ dom(f ) ⇒ g(x) ∈ f (x)))). We deﬁne an inﬁnite sequence to be a function f such that dom(f ) = ω. from the deﬁnition of ω it follows that ∀x ((x ⊆ ω ∧ 0 ∈ x ∧ ∀n (n ∈ x ⇒ n ∪ {n} ∈ x)) ⇒ x = ω). to be the least such n. the cardinality of x. . . Moreover. A set x is said to be ﬁnite if ∃n ∃f (n ∈ ω ∧ Fcn(f ) ∧ dom(f ) = n ∧ ran(f ) = x).e. There is a history of controversy surrounding this axiom. . C(x) is the set consisting of all elements of x. Thus ω = N. 8. The elements of ω are just the natural numbers. elements of elements of x. The axiom of foundation: ∀x (x = ∅ ⇒ ∃u (u ∈ x ∧ u ∩ x = ∅)). We deﬁne a ﬁnite sequence to be a function f such that dom(f ) ∈ ω.

.5 (foundational signiﬁcance of set theory). well-founded sets. The intended model of Zermelo/Fraenkel set theory is the collection V of all pure. V is also known as the universe of set theory. The replacement scheme: universal closure of (∀u (∃ exactly one v) A) ⇒ ∀x ∃y ∀v (v ∈ y ⇔ ∃u (u ∈ x ∧ A)) where A is any Lset -formula in which y does not occur. Zermelo/Fraenkel set theory with the axiom of choice. It can be 110 . The existence of the set P ω (HF) cannot be proved in Zermelo set theory. . N.7 (practical completeness of set theory). We can also develop the theory of higher mathematical objects such as manifolds. As already noted.6. u1 . . etc. set-theoretical foundation for virtually all of mathematics. Under this assumption. . Thus P ω (HF) is the set of all sets of ﬁnite order over HF. Let A be a formula which associates to each n ∈ N the set P n (HF). . un . u3 ∈ u2 ∩ x. i. it is possible to follow the usual Dedekind construction of the integers Z. operators on Banach spaces. . and the real numbers R. .then we would have a counterexample to the axiom of foundation. Zermelo set theory with the axiom of choice. x = ∅ and ∀u (u ∈ x ⇒ u ∩ x = ∅). un+1 . .6 (Zermelo/Fraenkel set theory). consists of ZC. Conversely. However. By extensionality. this y is unique.6. within ZC we have the natural number system. the existence of P ω (HF) can be proved in Zermelo/Fraenkel set theory. and then the union axiom gives us {P n (HF) | n ∈ N} = P ω (HF).}. It can be shown that ZFC is not ﬁnitely axiomatizable. then we could use the axiom of choice to obtain an inﬁnite descending ∈-chain u1 ∈ x. plus 9. . namely x = {u0 .. all within (deﬁnitional extensions of) ZC. Since N is a set. topological spaces. u2 ∈ u1 ∩ x. Deﬁnition 5. The most obvious model of Zermelo set theory is the set P ω (HF) = n∈ω P n (HF) = n∈N P (P · · · (P (HF)) · · ·) n consisting of HF plus all subsets of HF plus all subsets of subsets of HF plus . The signiﬁcance of ZC and similar theories is that they can serve as an axiomatic. Remark 5. ZFC. the replacement scheme for A gives us the set {P n (HF) | n ∈ N}. On this basis. The axioms of ZC express evident properties of P ω (HF). the rational numbers Q. The idea here is that to each u is associated exactly one v such that A holds. if x were a counterexample to the axiom of foundation.e.6.. . Note that the replacement scheme consists of an inﬁnite set of axioms. Remark 5. . we assert that for all sets x there exists a set y consisting of all v associated to some u ∈ x. . . . as follows.

It is straightforward to show that the interpretability relation is transitive.7. T1 is interpretable in T2 and not vice versa.1.g. in some abstract sense. At the same time.7. If T1 is interpretable in T2 . it has been found that ZFC is “practically complete” in the sense that all Lset -sentences expressing evident properties of V are provable in ZFC. Deﬁnition 5. We sometimes write T1 ≤ T2 to mean that T1 is interpretable in T2 . We say that T1 is interpretable in T2 if T1 is a subtheory of some deﬁnitional extension of T2 .3.e. i. Let T1 and T2 be theories.7 Interpretability Deﬁnition 5. Results of this kind follow from G¨del’s Second Incompleteness Theorem. second-order arithmetic is not interpretable in ﬁrst-order arithmetic. Remark 5. i. partially ordered by the interpretability relation. the Continuum Hypothesis. e. First-order arithmetic is interpretable in second-order arithmetic. Examples 5. More generally. Let T1 and T2 be theories. It can be shown that (n+1)st-order arithmetic and set theory are not interpretable in nth order arithmetic.. In particular. In other words.7. This o hierarchy is of obvious foundational interest. and which are known to be neither provable nor refutable in ZFC.. Thus we have equivalence classes of theories under mutual interpretability. 111 .2 (interpretability).5 (the G¨del hierarchy).shown that V = α P α (HF). Thus V is the collection of all sets of all transﬁnite orders over HF. Intuitively. Remark 5. which are neither evidently true nor evidently false in V according to our current understanding. for all n ≥ 1. and second-order arithmetic is interpretable in set theory. this means that T2 is “at least as strong as” T1 . then: 1. Consistency of T2 implies consistency of T1 . Thus it appears that ZFC accurately reﬂects our current understanding of V . The foundational signiﬁcance of interpretability is highlighted by the following observations. We have o Z1 < Z2 < · · · < Zn < Zn+1 < · · · < Z∞ < ZC < ZFC where T1 < T2 means that T1 is “weaker than” T2 .4.7. there are many interesting Lset -sentences. T1 ≤ T2 and T2 ≤ T3 imply T1 ≤ T3 .7. The axioms of ZFC express evident properties of V . The partial ordering of foundational o theories under interpretability is sometimes known as the G¨del hierarchy. T2 is “stronger than” T1 .. where α ranges over transﬁnite ordinal numbers. Remark 5. Moreover.6. We say that T1 is a subtheory of T2 if the language of T1 is included in the language of T2 and the theorems of T1 are included in the theorems of T2 .7. nth-order arithmetic is interpretable in (n + 1)st-order arithmetic and in set theory.e. 5.

8. . but operations can be handled similarly. By the Compactness Theorem.1. which says that an implicit deﬁnitional extension of a theory T is always equivalent to an explicit deﬁnitional extension of the same theory..e. We say that T implicitly deﬁnes P if. T . T ∪ T ′ ⊢ ∀x1 · · · ∀xn (P x1 · · · xn ⇔ P ′ x1 · · · xn ). We state and prove Beth’s Deﬁnability Theorem. Hence T ∪ T ′ ⊢ ∀x1 · · · ∀xn (P x1 · · · xn ⇔ P ′ x1 · · · xn ). T explicitly deﬁnes P if and only if T implicitly deﬁnes P . xn such that T ⊢ ∀x1 · · · ∀xn (P x1 · · · xn ⇔ D). Eﬀective essential incompleteness of T1 implies eﬀective essential incompleteness of T2 . Ak ∈ S such that (A ∧ A′ ) ⇒ ∀x1 · · · ∀xn (P x1 · · · xn ⇔ P ′ x1 · · · xn ) 112 . we have T ∪ T ′ ⊢ ∀x1 · · · ∀xn (P x1 · · · xn ⇔ P ′ x1 · · · xn ).e. 3. Deﬁnition 5. We say that T explicitly deﬁnes P if there exists an L-formula D not involving P with free variables x1 .2 (Beth’s Deﬁnability Theorem). T implicitly deﬁnes P . there are ﬁnitely many axioms A1 . say T ⊢ ∀x1 · · · ∀xn (P x1 · · · xn ⇔ D). . 5. Assume ﬁrst that T explicitly deﬁnes P .2. . i. i. S) be a theory. Theorem 5. . .8 Beth’s Deﬁnability Theorem In this section we consider implicit deﬁnitional extensions of theories. 2. . . We consider only the case of predicates. assume that T implicitly deﬁnes P . Conversely. Let T = (L.. and let P be an n-ary predicate of L. Essential incompleteness of T1 implies essential incompleteness of T2 . letting P ′ be a new n-ary predicate and letting T ′ = T [P/P ′ ]. Proof. .8. It follows that T ′ ⊢ ∀x1 · · · ∀xn (P ′ x1 · · · xn ⇔ D). 1.

5. . . .3) and (5.3) and D[x1 /a1 . . . .6) 113 . so T ⊢ ∀x1 · · · ∀xn (P x1 · · · xn ⇔ D). . From the logical validity of (5. Hence A ⇒ ∀x1 · · · ∀xn (P x1 · · · xn ⇔ D) is logically valid. it follows quasitautologically that A ⇒ (P a1 · · · an ⇔ D[x1 /a1 .1. xn /an ] ⇒ (A′ ⇒ P ′ a1 · · · an ) ′ (5.1) is logically valid. Introducing parameters a1 . .is logically valid. xn /an ] ⇒ (A ⇒ P a1 · · · an ) (5. . we see that (A ∧ A′ ) ⇒ (P a1 · · · an ⇔ P ′ a1 · · · an ) is logically valid.2).4) are logically valid. . where A = A1 ∧ · · · ∧ Ak and A′ = A[P/P ′ ].5) is logically valid. It follows quasitautologically that (A ∧ P a1 · · · an ) ⇒ (A′ ⇒ P ′ a1 · · · an ) (5. . (5. xn /an ] (5. . . . . and P and P do not occur in D. we can ﬁnd an L-formula D with free variables x1 .5). an . . xn such that D[x1 /a1 . Thus we see that T explicitly deﬁnes P . . . . . . . . .4) is logically valid.2) (5. xn /an ]) is logically valid. it follows that D[x1 /a1 . Thus (A ∧ P a1 · · · an ) ⇒ D[x1 /a1 . . . By the Interpolation Theorem 3. . . xn /an ] is an interpolant for (5. Since (5.

predicates P . Let f be a k-ary primitive recursive function. Theorem 6. . Thus n = S · · · S 0.1 Primitive Recursive Arithmetic Deﬁnition 6.3 Interpretability of PRA in Z1 G¨del Numbers o Let L be a countable language. . n + 1 = S n.1. As usual. To each natural number n we associate a variable-free PRAterm n as follows: 0 = 0. Assume that to all the sorts σ. . Then for all k-tuples of natural numbers m1 . and operations f of L have been assigned distinct positive integers #(σ).1 (G¨del numbers).3. denoted #(t) or o 114 .1. n These terms are known as numerals. To each L-V -term t and L-V -formula A o we assign a positive integer. mk ) .1. let V be the set of parameters. . #(P ). . . Proof. 6. .2. .Chapter 6 Arithmetization of Predicate Calculus 6. Deﬁnition 6. mk we have PRA ⊢ f m1 · · · mk = f (m1 . the G¨del number of t or of A.2 6. #(f ) respectively.

The language L is said to be primitive recursive if the following items are primitive recursive. σ #(vi ) = = = = = = = = = = 2 · 3#(σ) · 5i 22 · 3#(σ) · 5i 23 · 3#(f ) · p2 2 ·3 2 ·3 5 4 #(P ) #(A) #(t1 ) n · · · pn+1 #(aσ ) i #(f t1 · · · tn ) #(¬ A) #(A ∧ B) #(A ∨ B) #(A ⇒ B) #(A ⇔ B) #(∀v A) #(∃v A) #(t ) if f is an n-ary operation if P is an n-ary predicate #(P t1 · · · tn ) = · #(t ) #(tn ) p2 1 · · · pn+1 26 · 3#(A) · 5#(B) 27 · 3#(A) · 5#(B) 28 · 3#(A) · 5#(B) 29 · 3#(A) · 5#(B) 210 · 3#(v) · 5#(A) 211 · 3#(v) · 5#(A) Deﬁnition 6. Sort(x) Pred(x) Op(x) arity(#(P )) ≡ ≡ ≡ = x = #(σ) x = #(P ) x = #(f ) n n #(σi ) #(σi ) for some sort σ for some predicate P for some operation f if P is an n-ary predicate if f is an n-ary operation if 1 ≤ i ≤ n and P is an n-ary predicate of type σ1 × · · · × σn if 1 ≤ i ≤ n + 1 and f is an n-ary operation of type σ1 × · · · × σn → σn+1 arity(#(f )) = sort(#(P ).3. If L is primitive recursive. i) = Lemma 6. Var(x) Param(x) Term(x) ClTerm(x) AtFml(x) Fml(x) sort(#(t)) Proof. We have Var(x) ≡ (x)0 = 1 ∧ x = 2(x)0 · 3(x)1 · 5(x)2 ∧ Sort((x)1 ) 115 ≡ ≡ ≡ ≡ ≡ ≡ = x = #(v) x = #(a) x = #(t) x = #(t) x = #(A) x = #(A) #(σ) for some variable v for some parameter a for some term t for some closed term t for some atomic formula A for some formula A if t is a term of sort σ . respectively. i) = sort(#(f ). then the following are primitive recursive.#(A).3.3.2.

#(v). i + 1)) . sub((x)1 .3. Proof. z) such that for any formula A and any variable v and any closed term t. We then have if Var(x) ∨ Param(x) . To show that the predicate Term(x) is primitive recursive.4 (substitution).z) · 3sub((x)1 . Lemma 6.z) pi ≡ ((x)0 = 4 ∧ Pred((x)1 ) ∧ (+)) if 3 ≤ (x)0 ≤ 4. sort(x) = 0 otherwise.y. For formulas we have AtFml(x) and Fml(x) ≡ AtFml(x) ∨ ((x)0 = 5 ∧ x = 2(x)0 · 3(x)1 ∧ Fml((x)1 )) ∨ (6 ≤ (x)0 ≤ 9 ∧ x = 2(x)0 · 3(x)1 · 5(x)2 ∧ Fml((x)1 ) ∧ Fml((x)2 )) ∨ (10 ≤ (x)0 ≤ 11 ∧ x = 2(x)0 · 3(x)1 · 5(x)2 ∧ Var((x)1 ) ∧ Fml((x)2 )) and this completes the proof. For closed terms. otherwise. we ﬁrst show that the function sort(x) is primitive recursive.y. z (x)0 2 (x)0 2 sub(x. lh(x) − sort(x) = 0 otherwise .y.z) if 10 ≤ (x)0 ≤ 11 ∧ (x)1 = y. sub(#(A). z) = 2(x)0 (x) 2 0 x if x = y. 116 . where (+) · arity((x)1 ) = lh(x) − 2 ∧ x = · lh(x)−1 i=0 pi (x)i · ∧ (∀i < lh(x) − 2) (sort((x)i+2 ) = sort((x)1 . · 3(x)1 · ·3 · lh(x)−1 i=2 sub((x)i . if 6 ≤ (x)0 ≤ 9. #(t)) = #(A[v/t]) . deﬁne clsort(x) like sort(x) replacing Var(x) ∨ Param(x) by Param(x).z) · 3(x)1 · 5sub((x)2 .z) · 5sub((x)2 .y. There is a primitive recursive function sub(x. y. We then have ClTerm(x) ≡ clsort(x) > 0 . y. sort((x)1 . where sort(#(t)) = #(σ) if t is a term of sort σ.y. (x)1 · 1) if (x)0 = 3 ∧ Op((x)1 ) ∧ (+) .and Param(x) ≡ (x)0 = 2 ∧ x = 2(x)0 · 3(x)1 · 5(x)2 ∧ Sort((x)1 ) . Put lh(x) = least w < x such that (x)w = 0. Then Term(x) ≡ sort(x) > 0 . if (x)0 = 5.

then the predicate Snt(x) ≡ x = #(A) for some sentence A is primitive recursive.4. Lemma 6. Then we can ﬁnd an L-formula B such that T ⊢ B ⇔ A[x/#(B) ] . For example. This completes the proof. #(x).6.3.. Lemma 6. then d(#(A)) = #(A[x/#(A) ]. by Exercise 2. Proof. Let L be the language of T . if x is the only free variable of A. let D be the formula A[x/d x].5.1. a formula A is a sentence if and only if σ A[v/a] = A for all variables v occurring in A. The free variables of B are those of A except for x.10.1.1 (Self-Reference Lemma). 23 · 3#(S) · 5num(x) . Note that d(#(D)) = #(B).2 that PRA ⊢ d #(D) = #(B). num(z)).3. Since T includes PRA. we could take T to be an appropriate deﬁnitional extension of Z1 or Z2 or ZFC.4 Undeﬁnability of Truth In this section.e. and let B be the formula A[x/d #(D) ]. Hence T ⊢ A[x/d #(D)] ⇔ A[x/#(B) ]. In particular. If L is primitive recursive. The recursion equations for num(x) are num(0) = num(x + 1) = #(0) . Put d(z) = sub(z. 6. i. it follows that T ⊢ d #(D) = #(B). Or. Recall that. 117 . T ⊢ B ⇔ A[x/#(B) ]. we could take T to be PRA itself. Thus d is a 1-ary primitive recursive function such that. Proof. D[x/#(D) ]. by Section 6. Now given A as in the hypothesis of the lemma. if A is any L-formula containing the number variable x as a free variable. 2y)) . In other words.2. let T be a theory which includes PRA. Thus we have Snt(x) ≡ Fml(x) ∧ (∀y < x) (Var(y) ⇒ x = sub(x. There is a primitive recursive function num(x) such that num(n) = #(n) for any nonnegative integer n. y.Lemma 6. It follows by Theorem 6. Let A be an L-formula with a free number variable x. Note also that if y = #(vi ) then σ 2y = #(ai ). then B is an L-sentence. Proof.

4.5.e.1. Remark 6. Choose a primitive recursive predicate AxT for the set of G¨del numbers of axioms of T . T could be PRA itself. 1. 5. it can be shown that TrueM is implicitly deﬁnable over M . +.Deﬁnition 6. =) as above.5 The Provability Predicate In this section. Show that SatM is implicitly deﬁnable over M . ·. For example. Let X be a subset of the number domain of M which is explicitly deﬁnable over M . Applying Lemma 6. Since M is a model of T . Thus each n ∈ ω is identiﬁed with the element of M that is denoted by n.e. Let M = (ω. If M is any model of T ..1 to the negation of A. TrueM = {#(B) | B is a sentence and M |= B}. Proof. More generally.4. 1. +.4. . i.e. .d. then the characteristic function of TrueM is is not included in the characteristic function of any subset of M that is explicitly deﬁnable over M . 118 .4. n = vM (n).4 (undeﬁnability of truth).3..7.5. Then TrueM is not explicitly deﬁnable over M . Let SatM be the satisfaction relation on M . 0.4. such that A explicitly deﬁnes X over M .7.4. ·. or T could be any of the mathematical or foundational theories discussed in Sections 5. Deﬁnition 6.6. This1 may be paraphrased by saying that arithmetical truth is not arithmetically deﬁnable. With M = (ω. i. if M is any model of T . we obtain an L-sentence B such that T ⊢ B ⇔ ¬ A[x/#(B) ]. Theorem 6. and let T be an Ltheory which is primitive recursively axiomatizable. In terms of AxT show that various predicates o 1 This result is due to Tarski [5].6. 5. 0.5. let TrueM be the set of G¨del o numbers of sentences that are true in M . Exercise 6.} and 0M = 0 and SM (n) = n + 1 for all n ∈ ω. Z1 . q. An ω-model of T is a model M of T such that the number domain of M is ω = {0. then TrueM is not explicitly deﬁnable over M .2. it follows that #(B) ∈ TrueM if and only if #(B) ∈ X. if M is any model of T . (See also Exercise 6. 2. More generally.4. 6. If M is an ω-model of T . =) be the standard model of ﬁrst-order arithmetic.) Thus the Beth’s Deﬁnability Theorem does not hold for deﬁnability over this particular model. Let A be an L-formula with a free number variable x and no other free variables.4. let L be a primitive recursive language.2. 1. . / Hence the characteristic function of TrueM is not included in the characteristic function of X. Let M be an ω-model of Z1 or Z2 or ZFC. Corollary 6. we may assume that ω is identiﬁed with a subset of the number domain of M in such a way that 0M = 0 and SM (n) = n + 1 for each n ∈ ω. Deﬁnition 6.

5. Proof. Proof. Note that S is self-referential and says “I am not provable in T . For any L-sentence A. Let A(x) be a PRA-formula with one free variable x. This is just Lemma 6. or it could be an appropriate deﬁnitional extension of Z1 or Z2 or ZFC. Since Prf T (x.1 specialized to PRA. let PvblT be a provability predicate for T . For any L-sentence A. Proof.6. we have PRA ⊢ PvblT (#(A)) ⇒ PvblPRA (#(PvblT (#(A)))) . let T be a theory which is primitive recursively axiomatizable and includes PRA.4 (derivability condition 3). Lemma 6. Lemma 6. Hence PRA ⊢ PvblT (#(A)).2.6. This is an instance of Lemma 6. Introduce the provability predicate PvblT by deﬁnition: PvblT (x) ⇔ ∃y Prf T (x.3 (derivability condition 2). for all L-sentences B. if T ⊢ A then PRA ⊢ PvblT (#(A)) .1) 119 .5. y) . Suppose T ⊢ A. T could be PRA itself. We can ﬁnd a PRA-sentence S such that PRA ⊢ S ⇔ ¬ PvblT (#(S)) .2 that PRA ⊢ Prf T (#(A). For example. 6. Lemma 6. y) is a primitive recursive predicate.e. The details of the formalization are in Section 6.1 with A(x) ≡ ¬ PvblT (x).5. This is the Self-Reference Lemma 6. This is a straightforward consequence of the fact that our rules of inference include modus ponens. Lemma 6.7.5. we have PRA ⊢ PvblT (#(A ⇒ B)) ⇒ (PvblT (#(A)) ⇒ PvblT (#(B))) . Note that. PvblT (#(B)) is true if and only if T ⊢ B. Then Prf T (#(A).5. q.1. Lemma 6.associated with T are primitive recursive. (6.4.1. As in Section 6.d. For any L-sentences A and B.6.” Proof.2 formalized in PRA.2 (derivability condition 1). Then we can ﬁnd a PRA-sentence B such that PRA ⊢ B ⇔ A(#(B)). Proof. Let p be a proof of A in T . #(p)).6 The Incompleteness Theorems In this section. #(p)) holds. it follows by Theorem 6.

6. we have T ⊢ ¬ S.2).4 and 6. By Lemma 6. By Lemma 6. By (6. As in Section 6. Proof.6.6. Exercise 6. By the results of Section 6.2.6. where A[x/#(B)] says that for any proof p of B in T there exists a proof q of ¬ B in T such that #(q) < #(p). Thus T is inconsistent.1) that S is true. For all PRA-sentences A.2. By Lemma 6. Theorem 6. By Lemma 6.2 we have PvblPRA (#(S ⇔ ¬ PvblT (#(S)))). Let S be as in Lemma 6.4 we have T ⊢ S.1 to obtain a sentence B such that PRA ⊢ B ⇔ A[x/#(B)]. from (6.6. It follows by (6. It follows that PRA ⊢ ∀x (PvblPRA (x) ⇒ PvblT (x)). Hence by Lemmas 6. ¬ ConT .6. Thus S ′ is also true. Hence T ⊢ S ′ . Deﬁnition 6.6.2 we have PRA ⊢ PvblT (#(S)).2 and 6. Therefore.6. If T is consistent. Let S be a PRA-sentence as in Lemma 6.6.5. Hence by (6. Show that if T is as in Theorems 6. T ⊢ S.2 and 6.5. ConT is deﬁned to be the sentence ¬ PvblT (#(0 = 0)). Proof.5. By Theorem 6. Note that ConT is a PRA-sentence which asserts the consistency of T .5. Theorem 6. Suppose for a contradiction that T ⊢ S. it suﬃces to show PRA ⊢ ConT ⇒ S.4 we have PvblT (#(0 = 0)). Show that PRA ⊢ S ⇔ ConT . Let S be as in Lemma 6..2.2.5.6.6. PRA ⊢ S ⇔ S ′ . to show T ⊢ ConT . Reasoning in PRA.7. Hence by Lemmas 6. suppose PvblT (#(S)).8. i.1) and Lemma 6.Lemma 6. Moreover.5. then T ⊢ ConT .e.3. Details: We need to prove (6. Hence T ⊢ S ⇔ S ′ . we have PRA ⊢ PvblPRA (#(A)) ⇒ PvblT (#(A)) .5.6.6.4. If T is consistent.5 it follows that PvblT (#(¬ S)). Hint: Use the Self-Reference Lemma 6.1) it suﬃces to show PRA ⊢ ConT ⇒ ¬ PvblT (#(S)) . In particular we have: Lemma 6. (6. Since T includes PRA. show that T ⊢ B and T ⊢ ¬ B.2. Proof. Using the assumption that T is consistent.7 (the Second Incompleteness Theorem).3 we have PvblPRA (#(PvblT (#(S)))). then T ⊢ S. let S ′ be the translation of S into the language of ﬁrst-order arithmetic. Assume now that the primitive recursive predicate AxT has been chosen in such a way that PRA ⊢ ∀x (AxPRA (x) ⇒ AxT (x)).4 we have PvblPRA (#(¬ S)). 120 .5. it suﬃces to show T ⊢ ConT ⇒ S. If T is consistent. then T is incomplete.3.2) But this is just Lemma 6. Since T includes PRA.4 (the First Incompleteness Theorem). Exercise 6. then we can ﬁnd a sentence S ′ in the language of ﬁrst-order arithmetic such that S ′ is true yet S ′ is not a theorem of T .1) it follows that PRA ⊢ ¬ S.3 formalized in PRA. This completes the proof.9 (Rosser’s Theorem).6.6.6.6.

6.6.10.5. Give an example of a T as in Theorems 6. or T = Z1 + ¬ ConZ1 .7 such that T ⊢ ¬ ConT .3 FIXME Write this section! 121 . Solution.7 Proof of Lemma 6. 6. etc.6.Exercise 6.4 and 6. We may take T = PRA + ¬ ConPRA .

[2] David Marker. Selman. 1974. 1999. Springer. XIV + 445 pages. Smullyan. 2002. XXII + 229 pages. [4] Raymond M. First-Order Logic. 1994. Model Theory: An Introduction. Springer-Verlag. Introduction to Logic and to the Methodology of Deductive Sciences. Turing machines and the spectra of ﬁrst-order formulas. Dover Publications. Oxford University Press.Bibliography [1] Neil D. 39:139–150. VIII + 342 pages. Jones and Alan L. New York. 4th edition. Simpson. 122 . Perspectives in Mathematical Logic. Journal of Symbolic Logic. Subsystems of Second Order Arithmetic. 1995. [5] Alfred Tarski. [3] Stephen G. XII + 158 pages.

94 ACFp . 43 atomic formula. 108 categoricity. 108 diﬀerential equations. 104 divisible group. 11 distance. 111 complete theories. 91 distributivity. 66 bound variable. 58. 51. 30. 103 complex numbers. 109 Beth’s Deﬁnability Theorem. 108 dyadic. 58 conservative extension. 88–92. 11 closed. 10. 93 choice. 76 commutativity. 108 domain. 100. 84 assignment. 20 element. 85 practical. 94–96. 38 characteristic. 90 DLO. 91 dom. 6 associativity. 52 completeness. 118 binary. 24 dense linear ordering. 3. 93 algebraically closed ﬁeld. 24. 15. 98 axiom of choice. 24. 51. 40. 104 atomically closed. 106 arity. 52 clause. 10. 60. Fundamental Theorem of. 92 Compactness Theorem. 22 arithmetic ﬁrst-order. 104 nth order. 77 comprehension. 93 Cartesian product. 109 axiom of foundation. 3. 43. 10. 88. 72 companion. 112. 78. 25 C (complex numbers). 91 diﬀerence. 81. 64. 103 second-order. 100. 105. 93 ACF0 . 10. 110 deﬁnability. 35. 40 atomically. 109 Church’s Theorem. 97 deﬁnitional extension. 102 consistency. 66. 89 change of bound variables. 27. 103–105. 87 quasitautological. 90 Dedekind. 3 block tableau. 92. 89. 43 coﬁnite. 94 Algebra. 79. 87 constant. 93 composite number. 93 antisymmetry. 107 123 . 19. 61 modiﬁed. 85. 102 degree. 8. 108 congruence.Index Abelian group. 71. 78 DAG0 . 99 disjunctive normal form. 89 ACF. 57. 85 automorphism. 26. 72 consequence logical.

45. 24. 85 natural numbers. 104 normal satisﬁability. 63 G¨del number. 3. 67 LH . 82 normal structure. 41 identity. 87 logical equivalence. 12. 78 ﬁeld. 59 linear ordering. 81. 19. 91 group. 112 exponential function. 4 formation tree. 82. 20 ﬁrst-order arithmetic. 115 LG. 20 immediate successor. 108 invariance. 22. 97. 40. 95 empty set. 35. 26. 4 formula. 85. 76 ex . 20 o L1 . 66. 60 LH ′ . 85 foundation. 97. 27. 66 LG(symmetric). 103 L2 . 79. 98 irreﬂexivity. 88–96 membership. 104–106. 94. 81. 84 n-ary predicate. 10. 60. 97 intended model. 111 intersection. 27. 91 logical consequence. 109 foundational theories. 104 n. 114 124 . 71. 99 explicit deﬁnability. 37 logical validity. 86 identity predicate. 37. 82. 25 function. 102–111 free variable. 71. 20 implicit deﬁnability. 85 n-ary operation. 24. 51 N (natural numbers). 78 primitive recursive. 35 immediate predecessor. 71. 56. 83 Hilbert-style proof system. 108 end node. 104 modus ponens. 70. 104–106 mixed type. 88 elimination of quantiﬁers. 8. 8. 91 equality. 85 LG + . 103 formation sequence. 86 identity axioms. 93 Gentzen-style proof system. 79 K¨nig’s Lemma. 20 end points. 78. 103 inﬁnity. 3. 108 Fibonacci numbers. 88. 10. 110 interpolation. 112 induction. 84 mathematical theories.elementary equivalence. 27 Fcn. 75 isomorphism. axiom of. 64 LG ′ . 66 LG(atomic). 109 ﬁnitely branching. 84 n-ary relation. 106 numeral. 67 many-sorted. 88. 93. 24. 26. 86 nth-order arithmetic. 69 interpretability. 35. 97 extensionality. 79. 104 language. 79. 114 n-ary function. 97. 64. 107 falsity. 71. 98. 27. 114 o G¨del hierarchy. 108 Fundamental Theorem of Algebra. 72. 85 LH (S). 63. 86 immediate extension. 93 ﬁnite. 111 o graph. 50 Hintikka’s Lemma. 109 integers. 82 equivalence relation.

Cartesian. 20 S (successor function). 18. 105–108 practical completeness. 27. 83. 40 operation. 82. 25. 51 term. 50 Gentzen-style. 95 quasitautological consequence. 103–105. 20 successor. 106. 97–99 ran. 108 one-sorted. 71 relation. 108 PA. 12 signed sequent. 111 structure. 64 pair ordered. 78. 26 sequent. 110 replete. 106 singleton. 39 prime numbers. 99 sort. 95 real-closed ﬁeld. 91 partial. 24 quantiﬁer-free. 92 range. 28. 15. 95 real numbers. 31. 95. 104 satisfaction. 8. 94 second-order arithmetic. 108 proof system. 26. 93 RCOF. 62 sequent calculus. 80 Tarski. 115 product. 26. 12. 57. 107. 93. 93 quantiﬁer. 84 ordered ﬁeld. 22 P (power set). 63 Hilbert-style. 31. 85 125 . 108 random graph. 86 strength. 51. 107 pairing. 22 partition. 107 unordered. 86 open. 95. 35 sub. 24. 71 tableau. 84 soundness. 20 predicate. 101 reﬂexivity. 93. 107 sin x. 8. 35. 95 ordered pair. 20 Peano arithmetic. 78. 39 quantiﬁer elimination. 116 subtheory. 67 symmetry. 107 Q. 107 ordering linear. 111 subtree. 104 sentence. 76 path. 37 scalar. 18. 56 quasiuniversal. 104 symmetric. 85 replacement. 103 power set. 110 signed formula. 98. 99 reduct. 8. 103 Q (rational numbers). 35 satisﬁability. 92 root. 42 R (real numbers). 40 ring. 75. 84 prenex form. 85 partial ordering.∅ (empty set). 103 padding. 53. 27. 111 predecessor. 118 tautology. 58 quasitautology. 77 primitive recursive language. 107 parameter. 67 simple type theory. 95. 108 rational numbers. 67 signed variant. 51 spectrum. 63 set theory. 15. 50 pure set. 20 successor function. 74.

27 truth values. 26. 103 foundational. 104 Zn . 26. 87 complete. 90 transitivity. 111 torsion. 107 unary. 12 unsigned sequent. 107 Zermelo/Fraenkel. 110 Zermelo. 88–96 practically complete. 94–96. 108 variable. 85 bound. 94 weakening. 103 Z2 . 106 Z∞ . 110 126 . 107 Z (integers). 20 truth. 67 valuation. 102–111 mathematical. 3 union. 71 tree. 25 free. 110 unordered pair. 88. 25 variable-free. 107 ZFC. 79 value. 79 variant. 89 vector space. 67 Vaught’s Test. 40 universe. 107 unsigned formula. 107 universal closure. 6 type theory. 106 (union). 83. 93 Z1 . 85. 64 well-founded set. 106 ZC.theories. 88–92. 6. 103–105. 24. 39.

- NCTM
- Problem-Oriented Course on Mathematical
- Goldrei.classic Set Theory for Guided Independent Study
- Kleene - Introduction to Metamathematics
- Set Theory Notes
- Aspects of Mathematical Logic
- 2000. Leary - A Friendly Introduction to Mathematical Logic
- Ellitot Mendelson - Introduction to Mathematical Logic (4th Ed)
- Handbook of Mathematical Logic
- Virginia Klenk - Understanding Symbolic Logic
- Quantitative Literacy
- logic
- Language_Proof and Logic
- ➢ I Failed Every Subject Except For
- Rachev S.T. Financial Eco No Metrics (LN, Karlsruhe 2006)(146s)_FL
- Infinity
- Geometry
- Transition to Advanced Mathematics Proof Techniques Number Theory Chapter 1 Outline
- The Divine Cosmos
- Complex Analysis
- 3642125972
- General Relativity Crowell
- Chap 01 Real Analysis
- Lecture Notes
- The Elements of Periodic Table
- Introduction to mathematical analysis
- 12th Psychology
- Best way to learn periodic table (quick and easy)
- Financial Mathematics (Lectures)

Sign up to vote on this title

UsefulNot usefulRead Free for 30 Days

Cancel anytime.

Close Dialog## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Loading