CHAPTER 8

Classical Predicate Semantics and Proof

ch8 Systems

1 Formal Predicate Languages

firstlan
Propositional languages are also called zero order languages, as opposed to pred-
icate languages that are called first order languages. The same applies to the
use of terms propositional and predicate logic; they are often called zero order
and first order logics and we will use both terms equally.
We define a predicate language L following the pattern established by the propo-
sitional languages definitions. The predicate language L is more complicated in
its structure and hence its alphabet A is much richer. The definition of its set
of formulas F is more complicated. In order to define the set F we introduce
an additional set T, called a set of terms of the predicate language L. We
single out this set not only because we need it for the definition of formulas,
but also because of its role in the development of other notions of predicate logic.

We will work with different predicate languages, depending on what applications

we have in mind. All of these languages have some common features, and we
begin with a following general definition.

def:predlan Definition 1
By a predicate language L we understand a triple
L = (A, T, F), (1) pl

where A is a predicate alphabet, T, is the set of terms, and F is a set of

formulas.

The components of L are as follows.

1. Alphabet A is the set

A = V AR ∪ CON ∪ P AR ∪ Q ∪ P ∪ F ∪ C, (2) alphabet

where V AR is set of predicate variables, CON is a set of propositional connec-

tives, P AR a set of parenthesis, Q a set of quantifiers, P, a set of predicate

1
symbols, F a set of functions symbols, and C a set of constant symbols. We
assume that all of the sets defining the alphabet are disjoint.

Predicate Variables V AR
We assume that we always have a countably infinite set V AR of predicate
variables, called usually variables. We denote variables by x, y, z, ..., with
indices, if necessary, what we often express by writing

V AR = {x1 , x2 , ....}.

Propositional connectives CON

We define the set of propositional connectives CON in the same way as in the
case of the propositional languages. It means that we assume that CON is
non-empty and finite set and that consider only the connectives with one or
two arguments, i.e.
CON = C1 ∪ C2
where C1 is a finite set (possibly empty) of unary connectives, C2 is a finite set
(possibly empty) of binary connectives of the language L.

Parenthesis P AR
As in the propositional case, we adopt the signs ( and ) for our parenthesis., i.e.
we define the set P AR as
P AR = {(, )}.

The set of propositional connectives CON defines a propositional part of the

predicate logic language. What really differ one predicate language from the
other is the choice of additional symbols to the symbols described above. These
are called quantifiers symbols, predicate symbols, function symbols, and con-
stant symbols. I.e. a particular predicate language is determined by specifying
the following sets of symbols.

Quantifiers Q
We adopt two quantifiers; ∀ (for all, the universal quantifier) and ∃ (there exists,
the existential quantifier), i.e. we have the following set of quantifiers

Q = {∀, ∃}.

In a case of the classical logic and the logics that extend it, it is possible to adopt
only one quantifier and to define the other in terms of it and propositional con-
nectives. It is impossible in a case of many non-classical logics, for example the
intuitionistic logic. But even in the case of classical logic two quantifiers express

2
better the common intuition, so we assume that we have two of them.

Predicate symbols P
Predicate symbols represent relations. We assume that we have an non empty,
finite or countably infinite set bf P of predicate, or relation symbols. We denote
predicate symbols by P, Q, R, ..., with indices, if necessary, what we often express
by writing
P = {P1 , P2 , ...}.

Each predicate symbol P ∈ P has a positive integer #P assigned to it; if

#P = n then say P is called an n-ary (n - place) predicate (relation) symbol.

Function symbols F
We assume that we have a finite (may be empty) or countably infinite set F of
function symbols. When the set F is empty we say that we deal with a language
without functional symbols. We denote functional symbols by f, g, h, ..., with
indices, if necessary, what we often express by writing

F = {f1 , f2 , ...}.

Similarly, as in the case of predicate symbols, each function symbol f ∈ F has

a positive integer #f assigned to it; if #f = n then say f is called an n-ary (n
- place) function symbol.

Constant symbols C
We also assume that we have a finite (may be empty) or countably infinite set
C of constant symbols. The elements of C are denoted by c, d, e..., with indices,
if necessary, what we often express by writing

C = {c1 , c2 , ...}.

When the set C is empty we say that we deal with a language without constant
symbols.

Sometimes the constant symbols are defined as 0-ary function symbols, i.e.
C ⊆ F. We single them out as a separate set for our convenience.

Observe that what distinguishes now one predicate language L form the pother
is the choice of the components CON , and P, F, C of its alphabet A. We
hence will write

LCON (P, F, C) (3) pl1

3
 to denote the predicate language L determined by P, F, C and the set of propo-
sitional connectives CON .

Once the set of propositional connectives is fixed, the predicate language is

determined by the sets P, F and C and we l write
L(P, F, C) (4) pl2

for the predicate language L determined by P, F and C (with a fixed set of

propositional connectives). If there is no danger of confusion, we may abbreviate
L(P, F, C) to just L.
We sometimes allow the same symbol to be used as an n-place relation symbol,
and also as an m-place one; no confusion should arise because the different uses
can be told apart easily. Similarly for function symbols.
Having defined the basic elements of syntax, the alphabet, we can now complete
the formal definition of the predicate language by defining two more complex
components: the set T of all terms and the set F of all well formed formulas of
the language L(P, F, C).

2. Terms T
The set T of terms of a predicate language L(P, F, C) is defined as follows.

terms Definition 2 (Terms)

Given a predicate language L(P, F, C) with an alphabet A. The set T of terms
of L is the smallest set T ⊆ A∗ meeting the conditions:
(i) any variable is a term, i.e. V AR ⊆ T;
(ii) any constant symbol is a term, i.e. C ⊆ T;
(iii) if f is an nplace function symbol, i.e. f ∈ F and #f = n and t1 , t2 , ..., tn ∈
T, then f (t1 , t2 , ..., tn ) ∈ T.

Example 1
Let f ∈ F, #f = 1, i.e. f is a one place function symbol. Let x, y be predicate
variables, c, d constants, i.e. x, y ∈ V AR, c, d ∈ C. The following expressions
are terms:
x, y, f (x), f (y), f (c), f (d), f (f (x), ) f (f (y)), f f ((c)), f (f (d)), ...etc.

Example 2
If F = ∅, C = ∅, then the set T of terms consists of variables only, i.e.
T = V AR = {x1 , x2 , ....}.

4
From the above we get the following observation.

Remark 1
For any predicate language L(P, F, C), the set T of its terms is always non-
empty.

Example 3
If f ∈ F, #f = 1, g ∈ F, #g = 2, x, y ∈ V AR, c, d ∈ C, then some of the
terms are the following:

f (g(x, y)), f (g(c, x)), g(f (f (c)), g(x, y)), g(c, g(x, f (c))).

From time to time, the logicians are and we may be informal about how we
write terms. For instance, if we denote a two place function symbol g by +, we
may write x + y instead +(x, y). Because in this case we can think of x + y as
an unofficial way of designating the ”real” term +(x, y), or even g(x, y).

2. Formulas F
Before we define the set of formulas, we need to define one more set; the set of
atomic, or elementary formulas. They are the ”smallest” formulas as were the
propositional variables in the case of propositional languages.

Atomic formulas
An atomic formula of a predicate language L(P, F, C) is any element of the
alphabet A∗ of the form
R(t1 , t2 , ..., tn ),
where R ∈ P, #R = n, i.e. R is n-ary relational symbol and t1 , t2 , ..., tn are
terms. The set of all atomic formulas is denoted by AF and is defines as

AF = {R(t1 , t2 , ..., tn ) ∈ A∗ : R ∈ P, t1 , t2 , ..., tn ∈ T, #R = n, n ≥ 1}. (5) aform

Example 4
Consider a language
L(∅, {P }, ∅),
for #P = 1, i.e. a language without neither functional, nor constant symbols,
and with one, one-place predicate symbol P . The set of atomic formulas contains
all formulas of the form P (x), for x any variable, i.e.

AF = {P (x) : x ∈ V AR}.

5
 Example 5
Let now
L = L({f, g}, {R}, {c, d}),
for #f = 1, #g = 2 , #R = 2, i.e. L has two functional symbols: one -place
symbol f and two-place symbol g; one two-place predicate symbol R, and two
constants: c,d. Some of the atomic formulas in this case are the following.

R(c, d), R(x, f (c)), R(f (g(x, y)), f (g(c, x))), R(y, g(c, g(x, f (c)))).

Given a predicate language

L = LCON (P, F, C),

where CON is non-empty, finite set of propositional connectives such that

CON = C1 ∪ C2 for C1 a finite set (possibly empty) of unary connectives,
C2 a finite set (possibly empty) of binary connectives of the language L. We
define the set F of all well formed formulas of the predicate language L as
follows.

def:form Definition 3 (Formulas)

The set F of all well formed formulas, called shortly set of formulas, of the
language LCON P, F, C) is the smallest set meeting the following conditions:

1. any atomic formula of LCON (P, F, C) is a formula, i.e.

AF ⊆ F;

2. if A is a formula of LCON (P, F, C), 5 is an one argument propositional

connective, then 5A is a formula of LCON (P, F, C), i.e. if the following
recursive condition holds

if A ∈ F, 5 ∈ C1 , then 5 A ∈ F;

3. if A, B are formulas of L(P, F, C), ◦ is a two argument propositional

connective, then (A ◦ B) is a formula of L(P, F, C), i.e. if the following
recursive condition holds

if A ∈ F, 5 ∈ C2 , then (A ◦ B) ∈ F;

4. if A is a formula of L(P, F, C) and x is a variable, then ∀xA, ∃xA are

formulas of L(P, F, C), i.e. if the following recursive condition holds

if A ∈ F, x ∈ V AR, ∀, ∃ ∈ Q, then ∀xA, ∃xA ∈ F.

In formulas ∀xA, ∃xA, the formula A is in the scope of the quantifier ∀, ∃,

respectively.

6
 Example 6
Let L be a language with with the set {∩, ∪, ⇒, ¬} of connectives and with
two functional symbols: one -place and one two-place, one two-place predicate
symbol, and two constants. We write L as

L = L{¬,∩,∪,⇒} ({R}, {f, g}, {c, d}, )

where #f = 1, #g = 2 , #R = 2. Some of the formulas of L are the following.

R(c, f (d)), ∃xR(x, f (c)), ¬R(x, y), ∀z(∃xR(x, f (c)) ⇒ ¬R(x, y)),

(R(c, d) ∩ ∃xR(x, f (c))), ∀yR(y, g(c, g(x, f (c)))), ∀y¬∃xR(x, y).

The formula R(x, f (c)) is in a scope of the quantifier ∃x in ∃xR(x, f (c)).

The formula (∃xR(x, f (c)) ⇒ ¬R(x, y)) isn’t in a scope of any quantifier.
The formula (∃xR(x, f (c)) ⇒ ¬R(x, y)) is in the scope of ∀ in ∀z(∃xR(x, f (c)) ⇒
¬R(x, y)).

modal Example 7
Let L be a language with with the set {¬, , ♦, ∩, ∪, ⇒} of connectives and P,
F, and C the same as in previous exercise, i.e.

L = L{¬,,♦,∩,∪,⇒} ({R}, {f, g}, {c, d}, )

where #f = 1, #g = 2 , #R = 2.
L is now a language of some first order modal logic. Some of the formulas of
L are the following.

♦¬R(c, f (d)), ♦∃xR(x, f (c)), ¬♦R(x, y), ∀z(∃xR(x, f (c)) ⇒ ¬R(x, y)),

(R(c, d) ∩ ∃xR(x, f (c))), ∀yR(y, g(c, g(x, f (c)))), ∀y¬♦∃xR(x, y).

The formula R(x, f (c)) is in a scope of the quantifier ∃x in ♦∃xR(x, f (c)).

The formula (∃xR(x, f (c)) ⇒ ¬R(x, y)) isn’t in a scope of any quantifier.
The formula (∃xR(x, f (c)) ⇒ ¬R(x, y)) is in the scope of ∀z in ∀z(∃xR(x, f (c)) ⇒
¬R(x, y)). Formula ¬♦∃xR(x, y) is in the scope of ∀y in ∀y¬♦∃xR(x, y).

Given a predicate language L = (A, T, F), we must distinguish between formulas

like P (x, y), ∀xP (x, y) and ∀x∃yP (x, y).
This is done by introducing the notion of free and bound variables, open
and closed formulas (sentences). Before we formulate proper definitions, here
are some simple observations.

7
1. Some formulas are without quantifiers.
For example formulas R(c1 , c2 ), R(x, y), (R(y, d) ⇒ R(a, z)). A formula with-
out quantifiers is called an open formula.
Variables x, y in R(x, y) are called free variables. The variables y in R(y, d)
and z in R(a,z) are also free.
2. Quantifiers bind variables within formulas.
The variable x is bounded by ∃x in the formula ∃xR(x, y), the variable y is
free. The variable y is bounded by ∀y in the formula ∀yR(x, y), the variable
y is free.
3. The formula ∃x∀yR(x, y) does not contain any free variables, neither does
the formula R(c1 , c2 ). A formula without any free variables is called a closed
formula or a sentence.
Sometimes in order to distinguish more easily which variable is free and which
is bound in the formula we might use the bold face type for the quantifier
bound variables and write the formulas as follows.

(∀xQ(x, y), ∃yP (y), ∀yR(y, g(c, g(x, f (c)))),

(∀xP (x) ⇒ ∃yQ(x, y)), (∀x(P (x) ⇒ ∃yQ(x, y)))

Observe that the formulas ∃yP (y), (∀x(P (x) ⇒ ∃yQ(x, y))) are closed. We
call a close formula a sentence.

Example 8
Consider atomic formulas: P (y), Q(x, c), R(z), P1 (g(x, y), z). Here are some
non atomic formulas formed out of them.
1. (P (y) ∪ ¬Q(x, c)) ∈ F. This is an open formula A with two free variables
x,y. We denote A this as formula A(x, y).
2. ∃x(P (y) ∪ ¬Q(x, c)) ∈ F. We write x to denote that x is a bound variable.
The variable y is free. This is a formula B with one free variable y. We denote
B as a formula B(y).
3. ∀y(P (y) ∪ ¬Q(x, c)) ∈ F. The variable y is bound, the variable x is free.
We denote this formula by for example A1 (x).
4. ∀y∃x(P (y) ∪ ¬Q(x, c)) ∈ F has no free variables. It is a closed formula
called also a sentence.

Exercise 1
Given the following formulas of L:

P (x, f (c, y)), ∃cP (x, f (c, y)), ∀xf (x, P (c, y)), ∃xP (x, f (c, y)) ⇒ ∀yP (x, f (c, y)).

8
 1. Indicate whether they are, or are not well formed formulas of F. For those
which are not in F write a correct formula.
2. For each correct, or corrected formula identify all components: connectives,
quantifiers, predicate and function symbols, and list all its terms.
3. For each formula identify its s free and bound variables. State which are open
and which are closed formulas (sentences), if any.

Solution
Formula A1 = P (x, f (c, y)).
It is a correct atomic formula. P is a 2 argument predicate symbol, f is a
2 argument function symbol, c is a constant. We write it symbolically: P ∈
P, f ∈ F, c ∈ C. It is an open formula with two free variables x,y. We denote
it by A1 (x, y). It has no bound variables.
Formula A2 = ∃cP (x, f (c, y)).
It is a not a correct formula, i.e. ∃cP (x, f (c, y)) 6∈ F. The expression ∃c has no
meaning because c is a constant, not a variable.
The corrected formulas are: B1 = ∃xP (x, f (c, y)), B2 = ∃yP (x, f (c, y)), and
formulas B = ∃zP (z, f (c, y)) for any variable z different then x and y.
None of the correct formulas are open. Variable y is free in B1 = B1 (y), variable
x is free in B2 = B2 (x), both variables x and y are free in all formulas B =
B(x, y). All formulas are nether close, nor open. The terms appearing in any
of them are the same as in A1 = P (x, f (c, y)) and are: x, y, c, f (c, y).
Formula A3 = ∀xf (x, P (c, y)).
It is a not a correct formula, i.e. ∀xf (x, P (c, y)) 6∈ F. The function symbol f in
front f (x, P (c, y)) indicate a term and terms are not formulas. Moreover, the
atomic formula P (c, y) can’t be put inside a term!
Formula A4 = ∃xP (x, f (c, y)) ⇒ ∀yP (x, f (c, y)).
It is a not a correct formula. The correct formula is A = (∃xP (x, f (c, y)) ⇒
∀yP (x, f (c, y))). It has two free variables x and y and we write it as A = A(x, y).
Informally, in the formula P (x, y) both variables x and y are called free vari-
ables. They are not in the scope of any quantifier. The formula of that type
(without quantifiers) is an open formula.
The formal definition of the set of free variables of a formula is the following.

bfvar Definition 4 (Free and Bound Variables)

The set F V (A) of free variables of a formula A is defined by the induction of
the degree of the formula as follows.

1. If A is an atomic formula, i.e. A ∈ AF, then F V (A) is just the set of

variables appearing in the expression A;

9
 2. for any unary propositional connective, i.e any 5 ∈ C1 ,
F V (5A)= F V (A),
i.e. the free variables of 5A are the free variables of A;
3. for any binary propositional connective, i.e any ◦ ∈ C2 ,
F V (A ◦ B)= F V (A) ∪ F V (B),
i.e. the free variables of (A ◦ B) are the free variables of A together with
the free variables of B;
4. F V (∀xA) = F V (∃xA) = F V (A) − {x},
i.e. the free variables of ∀xA and ∃xA are the free variables of A, except
for x.

A formula with no free variables is called a sentence.

A variable is called bound if it is not free.
A formula with no bound variables is called an open formula.

Example 9 The formulas ∃xQ(c, g(x, d)), ¬∀x(P (x) ⇒ ∃y(R(f (x), y)∩¬P (c)))
are sentences. The formulas Q(c, g(x, d)), ¬(P (x) ⇒ (R(f (x), y) ∩ ¬P (c)))
are open formulas. The formulas ∃xQ(c, g(x, y)), ¬(P (x) ⇒ ∃y(R(f (x), y) ∩
¬P (c))) are neither sentences nor open formulas. They contain some free and
some bound variables; the variable y is free in the first formula, the variable x
is free in the second.

The definition 1 defines a predicate language L = LCON (P, F, C) (3) with its
sets of predicate, function and constant symbol possibly countably infinite sets.
We use its most general case with sets of predicate, function and constant symbol
all countably infinite sets for defining all relevant notions concerning provability
and semantics. In particular, we will define in detail the classical semantics for
this most general form of L and prove the completeness theorem for classical
predicate logic based on it.
When we deal with formal theory Th(SA) with a set SA of specific axioms
we restrict the language LCON (P, F, C) to the symbols characteristic for that
theory. We hence introduce the following definition.

langtheory Definition 5
Given a language LCON (P, F, C) = (A, T, F).
Let F0 ⊆ F be a non-empty, finite subset of formulas of L. Denote by P0 , F0 , C0
the sets of all predicate, function, and constant symbols appearing in the formu-
las from the set F0 . The language

LCON (P0 , F0 , C0 )

10
 is called a language defined by the set F0 of formulas.

Example 10 Consider a language L = L{¬,⇒,∪,∩} (P, F, C) and a following

set F0 of formulas of L
F0 = {∃xQ(c, g(x, d)), ¬∀x(P (x) ⇒ ∃y(R(f (x), y)∩¬P (e))), ¬(F (a)∩R(y, h(c))}.
A language defined by the set F0 of formulas is
L{¬,⇒,∪,∩} ({P, R, Q, F }, {g, f, h}, {a, c, d, e}),
where # Q = #R = 2, #P =# F = 1, # g = 2, #f = # h = 1.

It is common practice to use the notation

A(x1 , x2 , ..., xn ) (6) fvar

to indicate that F V (A) ⊆ {x1 , x2 , ..., xn } without implying that all of x1 , x2 , ..., xn
are actually free in A. This is similar to the practice in algebra of writing
p(x1 , x2 , ..., xn ) for a polynomial p in the variables x1 , x2 , ..., xn without imply-
ing that all of them have nonzero coefficients.

x/t Definition 6 (Replacing x by t in A)

If A(x) is a formula, and t is a term then A(x/t) or, more simply, A(t) de-
notes the result of replacing all occurrences of the free variable x by the term t
throughout. When using the notation A(t) we always assume that none of the
variables in t occur as bound variables in A.

The assumption that none of the variables in t occur as bound variables in A

is essential, otherwise by substituting t on the place of x we would distort the
meaning of A(t). Let t = y and A(x) is ∃y(x 6= y), i.e. the variable y in t is
bound in A. The substitution of t for x produces a formula A(t) of the form
∃y(y 6= y), which has a different meaning than ∃y(x 6= y).
But if t = z, i.e. the variable z in t is not bound in A, then A(x/t) = A(t) is
∃y(z 6= y) and express the same meaning as A(x).
Remark that if for example t = f (z, x) we obtain ∃y(f (z, x) 6= y) as a result of
substitution of t = f (z, x) for x in ∃y(x 6= y).
This notation is convenient because we can agree to write as
A(t1 , t2 , ..., tn ) or A(x1 /t1 , x2 /t2 , . . . , xn /tn )
a result of substituting in A the terms t1 , t2 , . . . , tn for all free occurrences (if
any) of x1 , x2 , . . . , xn , respectively. But when using this notation we always
assume that none of the variables in t1 , t2 , ..., tn occur as bound variables in A.
The above assumption that none of the variables in t1 , t2 , ..., tn occur as bound
variables in A is often expressed using the notion: t1 , t2 , ,̇tn are free for all theirs
variables in A which is defined formally as follows.

11
termfree Definition 7 (Term t free for y in A)
If A ∈ F and t is a term, then t is said to be free for y if no free occurrence of
y lies within the scope of any quantifier bounding variables in t.

Example 11 Let A , B be the formulas

∀yP (f (x, y), y), ∀yP (f (x, z), y),

respectively. The term t = f (x, y) is free for x and is not free for y in A. The
term t = f (x, z) is free for x and z in B. The term t = y is not free neither for
x nor for z in A, B.

Example 12
Let A be a formula

(∃xQ(f (x), g(x, z)) ∩ P (h(x, y), y)).

The term t1 = f (x) is not free for x in A; the term t2 = g(x, z) is free for z
only, term t3 = h(x, y) is free for y only because x occurs as a bound variable
in A; term t4 .

Definition 8 (Replacement)
If A(x), A(x1 , x2 , ..., xn ) ∈ F and t, t1 , t2 , ..., tn ∈ T , then A(x/t), A(x1 /t1 , x2 /t2 , . . . , xn /tn )
or, more simply just
A(t), A(t1 , t2 , ..., tn )
denotes the result of replacing all occurrences of the free variables x, x1 , x2 , ..., xn ,
by the terms t, t, t1 , t2 , ..., tn , respectively, assuming that t, t1 , t2 , ..., tn are free
for all theirs variables in A.

Classical Restricted Domain Quantifiers

We often use logic symbols, while writing mathematical statements. For exam-
ple mathematicians in order to say ”all natural numbers are greater then zero
and some integers are equal 1” often write it as

x ≥ 0, ∀x∈N and ∃y∈Z , y = 1.

Some of them, who are more ”logic oriented”, would also write it as

∀x∈N x ≥ 0 ∩ ∃y∈Z y = 1,

or even as
(∀x∈N x ≥ 0 ∩ ∃y∈Z y = 1).

12
 None of the above symbolic statements are formulas of the predicate language
L. These are mathematical statement written with mathematical and logic
symbols. They are written with different degree of ”logical precision”, the last
being, from a logician point of view the most precise.
Observe that the quantifiers in ∀x∈N and ∃y∈Z used in all of them are not
the one used in the predicate language L, which admits only quantifiers ∀x
and ∃y, for any variables x, y ∈ V AR. The quantifiers ∀x∈N , ∃y∈Z are called
quantifiers with restricted domain. The first is restricted to the domain of
natural numbers, the second to the integers. The restriction of the quantifier
domain can, and often is given by more complicated statements. For example
we say ”for all x > 2” and write ∀x>2 , or we say ”exists x > 2 and at same time
x + 2 < 8” and write symbolically ∃(x>2∩x+2<8) .
Our goal now is to correctly ”translate ” mathematical and natural language
statement into formulas of the predicate language L of the classical predicate
logic with the the set {¬.∩, ∪, ⇒} of propositional connectives. We say ”classical
predicate logic” to express that we define all notions for the classical semantics
to be defined formally in the next section 2. One can extend these notions to
non-classical logics, but we describe and will talk only about classical case. We
introduce the quantifiers with restricted domain into the classical predicate logic
language by expressing them within the language L as follows.

def:rq Definition 9
Given a classical predicate logic language

L = L{¬,∩,∪,⇒,¬} (P, F, C).

The quantifiers ∀A(x) , ∃A(x) are called quantifiers with restricted domain,
or restricted quantifiers, where A(x) ∈ F is any formula with any free vari-
able x ∈ V AR.
A formula ∀A(x) B(x) is an abbreviation of a formula ∀x(A(x) ⇒ B(x)) ∈ F.
We write it symbolically as

∀A(x) B(x) = ∀x(A(x) ⇒ B(x)). (7) resq1

A formula ∃A(x) B(x) stands for a formula ∃x(A(x) ∩ B(x)) ∈ F. We write it

symbolically as
∃A(x) B(x) = ∃x(A(x) ∩ B(x)) (8) resq2

The definition 9 of restricted quantifiers is obviously faithful to our intuitive

meaning of quantifiers. We use informally a symbol = to stress that we they
are in a sense equivalent with respect to classical semantics. We call (7) and (8)
transformations rules for restricted quantifiers.

Exercise 2

13
Given a mathematical statement S written with logical symbols
(∀x∈N x ≥ 0 ∩ ∃y∈Z y = 1)
1. Translate it into a proper logical formula with restricted domain quantifiers
i.e. into a formula of L that uses the restricted domain quantifiers.
2. Translate your restricted domain quantifiers logical formula into a correct
logical formula without restricted domain quantifiers, i.e. into a formula of L.

Solution
We proceed to write this and other similar problems solutions in a sequence of
steps.

STEP 1. We identify basic statements in S, i.e. mathematical statements that

involve only relations. They will be translated into atomic formulas. We iden-
tify the relations in the basic statements and choose the predicate symbols as
their names. We identify all functions and constants (if any) in the basic state-
ments and choose the function symbols and constant symbols as their names.

The basic statements in S are: x ∈ N, x ≥ 0, y ∈ Z, y = 1. The relations are:

∈ N, ∈ Z, ≥, =. We use one argument predicates symbols N, Z for ∈ N, ∈ Z,
respectively. We use two argument predicate symbols G for ≥, and E for =.
There are no functions. We have two constant symbols c1 , c2 for numbers 0 and
1, respectively.

STEP 2. We write the basic statements as atomic formulas of L.

We write N (x), Z(x) for x ∈ N, x ∈ Z, respectively. G(x, c1 ) for x ≥ 0 and

E(y, c2 ) for y = 1. These are all atomic formulas.

STEP 3. We re-write the statement S a logical formula with restricted domain

quantifiers.

The statement S becomes a restricted quantifiers formula:

(∀N (x) G(x, c1 ) ∩ ∃Z(y) E(y, c2 )).

STEP 4. We apply (7) and (8) to the formula from STEP 3. and obtain a
formula A of L as a representation of the given mathematical statement S.

A formula A ∈ F that corresponds to S is

(∀x (N (x) ⇒ G(x, c1 )) ∩ ∃y (Z(y) ∩ E(y, c2 ))).

14
 Here is a perfectly acceptable short solution. We presented the long solution in
order to explain all steps needed to be performed when one writes a solution.

Short Solution
The basic statements in S are: x ∈ N, x ≥ 0, y ∈ Z, y = 1. The corresponding
atomic formulas of L are: N (x), G(x, c1 ), Z(y), E(y, c2 ), respectively.
The statement S becomes becomes a restricted quantifiers formula
(∀N (x )G(x, c1 ) ∩ ∃Z(y) E(y, c2 )). Applying restricted quantifiers definition 9
and transformation rules (7), (8) we get a following formula A ∈ F

(∀x(N (x) ⇒ G(x, c1 )) ∩ ∃y(Z(y) ∩ E(y, c2 ))).

2 Classical Semantics
sec:clsem

The notion of predicate tautology is much more complicated then that of the
propositional. Predicate tautologies are also called valid formulas, or laws of
quantifiers to distinguish them from the propositional case. The formulas of a
predicate language L have meaning only when an interpretation is given for all
its symbols. We define an interpretation I by interpreting predicate, functional
symbols as a concrete relation, function defined in a certain set U 6= ∅, and
constants symbols as elements of the set U. The set U is called the universe
of the interpretation I. These two items specify a structure M = (U, I) for the
language L.
.
The semantics for a first order language L in general, and for the first order
classical logic in particular, is defined, after Tarski (1936) in terms of the struc-
ture M = [U, I], an assignment s of L, and a satisfaction relation (M, s) |= A
between structures, assignments and formulas of L.

The definition of a structure M = [U, I] and the assignment s of L is common

for different predicate languages and for different semantics and we define them
as follows.

structure Definition 10 (Structure)

Given a predicate language L = LCON (P, F, C). A structure for L is a pair

M = [U, I],

where U is a non empty set called a universe and I is an assignment called an

interpretation of the language L(P, F, C) in the universe U defined as follows.

15
 1. I assigns to any predicate symbol P ∈ P a relation PI defined in the
universe U . I.e. for any P ∈ P, if #P = n, then

PI ⊆ U n .

2. I assigns to any functional symbol f ∈ F a function fI defined in the

universe U . I.e. for any f ∈ F, if #f = n, then

fI : U n −→ U.

3. I assigns to any constant symbol c ∈ C an element cI of the universe. I.e

for any c ∈ C,
cI ∈ U.

Example 13
Let L be a language with one two-place predicate symbol, two functional symbols:
one -place and one two-place, and two constants, i.e.

L = L({R}, {f, g}, {c, d}, )

where #R = 2, #f = 1, #g = 2, and c, d ∈ C.
We define a structure M = [U, I] as follows. We take as the universe the set
U = {1, 3, 5, 6}. The predicate R is interpreted as ≤, what we write as RI : ≤.
We interpret f as a function fI : {1, 3, 5, 6} −→ {1, 3, 5, 6} such that fI (x) = 5
for all x ∈ {1, 3, 5, 6}, and we put gI : {1, 3, 5, 6} × {1, 3, 5, 6} −→ {1, 3, 5, 6}
such that gI (x, y) = 1 for all x ∈ {1, 3, 5, 6}. The constant c becomes cI = 3,
and dI = 6. We write the structure M as

M = [{1, 3, 5, 6} ≤, fI , gI , cI = 3, dI = 6]

Exercise 3
Given a language
L = L({R}, {g}, ∅, )
where #R = 2, #g = 2 . Define two structures for L, both with infinite universe:
one infinitely countable and one uncountable.

Solution
There are many such structures. Here are two of the very simple.
M1 = [N, , ≤, +], where N is the set of natural numbers, and for example
M2 = [R, , ≤, +], where R is the set of real numbers.

16
assign Definition 11 (Assignment)
Given a first order language L = L(P, F, C) with the set V AR of variables. Let
M = [U, I] be a structure for L with the universe U 6= ∅.
An assignment of L in M = [U, I] is any function
s : V AR −→ U (9) assing

The assignment s is also called an interpretation of variables VAR of L in a

structure M = [U, I].

Let M = [U, I] be a structure for L and s : V AR −→ U be an assignment of

variables VAR of L in a structure M.
Let T be the set of all terms of L. By definition 2, V AR ⊂ T. We use the
interpretation I to extend the assignment s to the set the set T of all terms of
L. Because of that we denote this extension by sI rather then by s∗ as we did
before. The extension sI of s is hence a mapping from T to U . It associates
with each t ∈ T an element I(t) ∈ U . We denote this element sI (t) by tI . We
define the extension sI (t) = tI of s by the induction of the length of the term
t ∈ T and call it an interpretation of terms of L in a structure M = [U, I].

i-terms Definition 12 (Interpretation of Terms)

Given a language L = L(P, F, C) and a structure M = [U, I] for L. Let an
assignment
s : V AR −→ U
be any interpretation of variables VAR of L (assignment) in M.
We extend s to a function
sI : T −→ U (10) sI
called an interpretation sI of terms of L in M.
The function sI is defined by induction on the complexity of terms as follows.

1. For any x ∈ V AR,

sI (x) = s(x);
2. for any c ∈ C,
sI (c) = cI ;
3. for any t1 , t2 , . . . , tn ∈ T, n ≥ 1, f ∈ F, such that #f = n, and for any
term t = f (t1 , t2 , . . . , tn ), we put
sI (t) = fI (sI (t1 ), sI (t2 ), . . . , sI (tn )),
i.e. we define
sI (f (t1 , t2 , . . . , tn )) = fI (sI (t1 ), sI (t2 ), . . . , sI (tn )),
for any t1 , t2 , . . . , tn ∈ T, f ∈ T, such that #f = n.

17
Exercise 4
Consider a language

L = L({P, R}, {f, g, h, }, {c, d})

for # P = # R = 2, #f = # g=1, # h= 2.
Let M = [{0, 1}, I], where the interpretation I is defined as follows.
PI = {(0, 0)}, RI = {(0, 0), (1, 1)},
fI (0) = 0, fI (1) = 0, gI (0) = 1, gI (1) = 1, hI is given by a formula
hI (x, y) = x for all (x, y) ∈ {0, 1} × {0, 1}, and cI = 1, dI = 0.

The assignment s : V AR −→ {0, 1} is such that s(x) = s(y) = s(z) = 1 = 1,

for x, y, z ∈ V AR (and any values for all other variables).
Given a set T0 of terms, evaluate sI (t) for all t ∈ T0 .
T0 = {z, y, x, c, f (c), f (x), g(z), f (g(d)), g(f (g(z))), h(c, f (g(d))), h(f (x), g(z))}.

Solution
First we evaluate terms that are variables and constants of L using the formulas
1. and 2. of definition 12: sI (x) = s(x), sI (c) = cI , respectively and obtain:
sI (z) = s(z) = 1, sI (y) = s(y) = 1, sI (x) = s(x) = 1, sI (c) = cI = 1, sI (d) =
dI = 2. We use the formula sI (f (t1 , t2 , . . . , tn )) = fI (sI (t1 ), sI (t2 ), . . . , sI (tn )
to evaluate the rest of terms in T0 and obtain:
sI f (c) = fI (sI (c)) = fI (cI )) = fI (1) = 0, sI f (x) = fI (sI (x)) = fI (1) = 0
sI g(z) = gI (sI (c)) = gI (1) = 1, sI (f (g(d)) = fI (sI (f (g(d))) = fI (gI (sI (d))) =
fI (gI (cI )) = fI (gI (1)) = fI (1) = 0,
sI g(f (g(z)) = gI (fI (gI (sI (z)))) = gI (fI (gI (1))) = gI (fI (1)) = gI (0) = 1,
sI (h(c, f (g(d)))) = hI (sI (c), sI (f (g(d)))) = hI (cI , fI (gI (sI (d))))
= hI (1, fI (gI (0))) = hI (1, fI (1)) = hI (1, 0) = 1,
sI (h(f (x), g(z))) = hI (fI (sI (x)), gI (sI (x))) = hI (fI (1), gI (1)) = hI (0, 1) = 0.

Observe that the interpretation of predicate symbols is irrelevant when eval-

uating an interpretation of terms, as the terms do not involve the predicate
symbols.

Example 14
Consider a language
L = L({P, R}, {f, h}, ∅ )

18
for # P = # R = 2, #f = 1, # h= 2.
Let M = [Z, I], where Z is the set on integers and the interpretation I for
elements of F and C is as follows.
fI : Z −→ Z is given by formula f (m) = m + 1 for all m ∈ Z.
hI : Z × Z −→ Z is given by formula f (m, n) = m + n for all m, n ∈ Z.
Let s : V AR −→ Z be any assignment such that s(x) = −5, s(y) = 2 and
t1 , t2 ∈ T be t1 = h(y, f (f (x))) and t2 = h(f (x), h(x, f (y)).
We evaluate:
sI (t1 ) = sI (h(y, f (x)) = hI (sI (y), fI (sI (x))) = +(2, fI (−5)) = 2 − 4 = −2,

sI (t2 ) = sI (h(f (x), h(x, f (y))) = +(fI (−5), +(−5, 3)) = −4 + (−5 + 3) = −6.

For any t ∈ T, let x1 , x2 , . . . , xn ∈ V AR be all variables appearing in t, we

write it, in a similar way as we did in (6) for variables in formulas, as

t(x1 , x2 , . . . , xn ).

Observation 1
For any term t(x1 , x2 , . . . , xn ) ∈ T, any structure M = [U, I] and any assign-
ments s, s0 of L in M, the following holds.
If s(x) = s0 (x) for all x ∈ {x1 , x2 , . . . , xn }, i.e the assignments s, s0 agree on
all variables appearing in t, then, sI (t) = s0 I (t).

Thus for any t ∈ T, the function sI : T −→ U defined by (10) depends on

only a finite number of values of s(x) for x ∈ V AR.
Given a structure M = [U, I] and an assignment s : V AR −→ U . We write
a
s(x) (11) forQ

to denote any assignment s0 : V AR −→ U such that s, s0 agree on all variables

except on x, such that s0 (x) = a, for certain a ∈ U .

Given a first order (predicate) language L = LCON (P, F, C). The satisfaction
relation (M, s) |= A between structures, assignments and formulas of L is de-
fined by induction on the complexity of formulas of L. It is the satisfaction
relation (M, s) |= A that allows us to distinguish one semantics for a given L
from the other, and consequently one logic from the other. We define now only
a classical satisfaction and the notion of classical predicate tautology.

19
def:sat Definition 13 (Classical Satisfaction)
Given a classical predicate (first order) language

L = L{¬,∩,∪,⇒,¬} (P, F, C). (12) classL

Let M = [U, I] be a structure for L, s be any assignment of L in M, i.e.

s : V AR −→ U . Let A ∈ F be any formula of L. We define a relation

(M, s) |= A

that reads: the assignment s satisfies the formula A in M, by induction on the

complexity of A as follows.

(i) A is atomic formula (5), i.e. A is P (t1 , t2 , . . . , tn ) for P ∈ P, #P = n.

(M, s) |= P (t1 , t2 , . . . , tn ) if and only if (sI (t1 ), sI (t2 ), . . . , sI (tn )) ∈ PI .

(ii) A is not atomic formula and has one of connectives of L as the main
connective.
(M, s) |= ¬A if and only if (M, s) 6|= A,
(M, s) |= (A ∩ B) if and only if (M, s) |= A and (M, s) |= B,

(M, s) |= (A ∪ B) if and only if (M, s) |= A or (M, s) |= B or both,

(M, s) |= (A ⇒ B) if and only if ether (M, s) 6|= A or else (M, s) |= B or
both.
(iii) A is not atomic formula and begins with one of the quantifiers.

(M, s) |= ∃xA if and only if there is s0 such that s, s0 agree on all variables
except on x, and (M, s0 ) |= A,
(M, s) |= ∀xA if and only if for all s0 such that s, s0 agree on all variables
except on x, and (M, s0 ) |= A.

Observe that that the truth or falsity of (M, s) |= A depends only on the values
of s(x) for variables x which are actually free in the formula A. This is why we
often write the condition (iii) as

(iii)’ A(x) is not atomic formula (with a free variable x) and begins with one
of the quantifiers.
(M, s) |= ∃xA(x) if and only if there is s0 such that s(y) = s0 (y) for all
y ∈ V AR − {x}, and (M, s0 ) |= A(x),
(M, s) |= ∀xA if and only if for all s0 such that s(y) = s0 (y) for all
y ∈ V AR − {x}, and (M, s0 ) |= A(x).

20
sat Exercise 5
For the structures Mi , find assignments si , s0 i (1 ≤ i ≤ 4), such that

(Mi , si ) |= Q(x, c), (Mi , s0 i ) 6|= Q(x, c) for Q ∈ P, c ∈ C.

The structures Mi are defined as follows (the interpretation I for each of them
is specified only for symbols in the formula Q(x, c), and N denotes the set of
natural numbers.

M1 = [{1}, QI :=, cI : 1], M2 = [{1, 2}, QI :≤, cI : 1],

M3 = [N, QI :≥, cI : 0], and M3 = [N, QI :≥, cI : 1.]

Solution
Consider M1 = [{1}, QI :=, cI : 1]. Observe that all s : V AR −→ {1} must
be defined by a formula s(x) = 1 for all x ∈ V AR. We evaluate (definition 12),
sI (x) = 1, sI (c) = cI = 1. By definition 13, (M1 , s) |= Q(x, c) if and only if
(sI (x), sI (c)) ∈ QI , i.e. (1, 1) ∈= what is true as 1 = 1. We have proved

(M1 , s) |= Q(x, c) for all assignments s : V AR −→ {1}.

Consider M2 = [{1, 2}, QI :≤, cI : 1]. Let s : V AR −→ {1, 2} be any

assignment, such that s(x) = 1. We evaluate sI (x) = 1, sI (c) = 1 and verify
whether (sI (x), sI (c)) ∈ QI , i.e. whether (1, 1) ∈≤ . This is true as 1 ≤ 1. We
have found s (in fact uncountably many such s) such that

(M2 , s) |= Q(x, c).

Let now s0 be any assignment s0 : V AR −→ {1, 2}, such that s0 (x) = 2. We

evaluate s0 I (x) = 1, s0 I (c) = 1 and verify whether s0 I (x), s0 I (c)) ∈ QI , i.e.
whether (2, 1) ∈≤ . This is not true as 2 6≤ 1. We have found s0 6= s (in fact
uncountably many such s’) such that

(M2 , s0 ) 6|= Q(x, c).

Consider M3 = [N, QI :≥, cI : 0]. Let s : V AR −→ N be any assignment, such

that s(x) = 5. We evaluate sI (x) = 5, sI (c) = 0. Observe that the condition
(sI (x), sI (c)) ∈ QI holds as 5 ≥ 0 and

(M3 , s) |= Q(x, c).

Let now s0 be any assignment s0 : V AR −→ N . By definition, s0 (x) = n, for

any n ∈ N , and s0 I (x), 0) ∈ QI holds for any s0 as n ≥ 0 for all n ∈ N . This
proves that there is no s0 , such that (M3 , s0 ) 6|= Q(x, c).

21
 Consider M4 = [N, QI :≥, cI : 1]. Let s : V AR −→ N be any assignment, such
that s(x) = 5. We evaluate sI (x) = 5, sI (c) = 1. Observe that the condition
(sI (x), sI (c)) ∈ QI holds as 5 ≥ 1 and hence
(M4 , s) |= Q(x, c).
Let now s0 be any assignment s0 : V AR −→ N , such that s0 (x) = 0. The the
condition (s0 I (x), s0 I (c)) ∈ QI does not holds as 0 6≥ 1 and
(M4 , s0 ) 6|= Q(x, c).

Directly from the definition13 we have that the following holds.

sat1 Example 15
Let Mi (1 ≤ i ≤ 4) be structures in defined the exercise 5 and let corresponding
assignments si be as defined as its solutions.
1. (M1 , s) |= Q(x, c), (M1 , s) |= ∀xQ(x, c), (M1 , s) |= ∃xQ(x, c).
2. (M2 , s) |= Q(x, c), (M2 , s) 6|= ∀xQ(x, c), (M1 , s) |= ∃xQ(x, c).
3. (M3 , s) |= Q(x, c), (M3 , s) |= ∀xQ(x, c), (M3 , s) |= ∃xQ(x, c).
4. (M4 , s) |= Q(x, c), (M4 , s) 6|= ∀xQ(x, c), (M4 , s) |= ∃xQ(x, c).

model Definition 14 (Model)

Given a language L, a formula A of L, and a structure M = [U, I] for L.
The structure M is a model for the formula A if and only if (M, s) |= A
for all s : V AR −→ U. We denote it as M |= A.
For any set Γ ⊆ F of formulas of L, M is a model for Γ if and only if
M |= A for all A ∈ Γ. We denote it as M |= Γ.

We define now a very important semantic notion. It has different names: logi-
cal consequence, logical implication, semantic consequence, logical (semantical)
entailment. We use a name logical consequence and define it as follows.

def:lcons Definition 15 (Logical Consequence)

For any A, B ∈ F and any set Γ ⊆ F of formulas of L, we say that a formula
B is a logical consequence of a set Γ and write it as Γ |= B, if and only if
all models of the set Γ are models of the formula B.

When Γ |= B we also say that Γ logically implies B. When Γ = {A} we write

it as A |= B and say A logically implies B.
We say that A and B are logically equivalent if and only if A |= B and
A |= B.

22
 Directly from the model definition 14 we get the following.

c-model Definition 16 (Counter Model)

Given a language L = L{¬,∩,∪,⇒,¬} (P, F, C), a formula A of L, and a structure
M = [U, I] for L.
The structure M = [U, I] is a counter model for the formula A if and only
if there is an assignment s : V AR −→ U , such that (M, s) 6|= A.
We denote it as M 6|= A.
For any set Γ ⊆ F of formulas of L, M is a counter model for Γ if and only
if there is A ∈ Γ, such that M 6|= A.
We denote it as M 6|= Γ.

Observe that if A is a sentence (definition 4) then the truth or falsity of

(M, s) |= A is completely independent of s. Hence if (M, s) |= A for some
s, it holds for all s and the following holds.

sentence Fact 1
For any formula A of L,
If A is a sentence, then if there s such that (M, s) |= A, then M is a model for
A, i.e. M |= A

We transform any formula A of L into a certain sentence by binding all its

free variables. The resulting sentence is called a closure of A and is defined as
follows.

cA Definition 17 (Closure)
Given a formula A of L.
By the closure of A we mean the formula obtained from A by prefixing in
universal quantifiers all variables the are free in A. If A does not have free
variables (i.e. is a sentence), the closure if A is defined to be A itself.

Obviously, a closure of any formula is always a sentence. For example, if A, B

are formulas
(P (x1 , x2 ) ⇒ ¬∃x2 Q(x1 , x2 , x3 )), (∀x1 P (x1 , x2 ) ⇒ ¬∃x2 Q(x1 , x2 , x3 )),
their respective closures are
∀x1 ∀x2 ∀x3 ((P (x1 , x2 ) ⇒ ¬∃x2 Q(x1 , x2 , x3 ))),
∀x1 ∀x2 ∀x3 ((∀x1 P (x1 , x2 ) ⇒ ¬∃x2 Q(x1 , x2 , x3 ))).

23
models Example 16
Let Q ∈ P, #Q = 2 and c ∈ C Consider formulas

Q(x, c), ∃xQ(x, c), ∀xQ(x, c)

and the structures from exercise 5 defined as follows.

M1 = [{1}, QI :=, cI : 1], M2 = [{1, 2}, QI :≤, cI : 1],

M3 = [N, QI :≥, cI : 0], and M4 = [N, QI :≥, cI : 1.]

Directly from example 15 and Fact 1, we get that:
1. M1 |= Q(x, c), M1 |= ∀xQ(x, c), M1 |= ∃xQ(x, c).
2. M2 6|= Q(x, c), M2 6|= ∀xQ(x, c), M2 |= ∃xQ(x, c).
3. M3 |= Q(x, c), M3 |= ∀xQ(x, c), M3 |= ∃xQ(x, c).
4. M4 6|= Q(x, c), M4 6|= ∀xQ(x, c), M4 |= ∃xQ(x, c).

trueM Definition 18 (True, False in M)

Given a structure M = [U, I] for L and a formula A of L. We say that:
A is true in M (written as M |= A) if and only if all assignments s of L in
M satisfy A, i.e. when M is a model for A.
A is false in M (written as M =| A) if and only if no assignment s of L in
M satisfies A.

By the definition 14 we have that A is true in M only when the structure M

is a model for A. This is why we use the notation M |= A in both cases.
Obviously, if A is not true in M, then it is false, and vice versa. This proves
correctness of our definition with respect to the intuitive understanding.
We get directly from definition 18 and the example 16 the following.

tf Example 17
Let M1 − M4 be structures defined in example 5.
1. Formulas Q(x, c), ∀xQ(x, c), ∃xQ(x, c) are all true in the structures M1
and M3 .
2. Formula ∃xQ(x, c) is also true in M2 and in M3 .
3. Formulas ¬Q(x, c), ¬∀xQ(x, c), ¬∃xQ(x, c) are all false in the structures
M1 and M3 .
4. Formula ¬∃xQ(x, c) is also false in M2 and in M3 .

24
 5. Formulas (Q(x, c) ∩ ¬Q(x, c)), (¬∀xQ(x, c) ∩ ∀xQ(x, c)), and the formula
(∃xQ(x, c) ∩ ¬∃xQ(x, c)) are all false in all structures M1 − M4 .
6. The formula ∀xQ(x, c) is false in a structure M5 = [N, QI :<, cI : 0].

Here are some properties of the notions ”A is true in M, written symbolically as

M |= A, and ”A is false in M, written symbolically as M=| A. They are obvious
under intuitive understanding of the notion of satisfaction. Their formal proofs
are left as exercise for the reader.

TFprop Property 1 (Truth, Falsity, Satisfaction)

Given a structure M = [U, I] for L and any formulas formula A, B of L. The
following properties hold.
P1. A is false in M if and only if ¬A is true in M, i.e.

M =| A if and only if M |= ¬A.

P2. A is true in M if and only if ¬A is false in M, i.e.

M |= A if and only if M =| ¬A.

P3. It is not the case that both M |= A and M |= ¬A, i.e. no formula of L
can be both true and false in M, i.e. there is no formula A, such that M |= A
and M =| A.
P4. If M |= A and M |= (A ⇒ B), then M |= B.
P5. (A ⇒ B) is false in M if and only if M |= A and M |= ¬B, i.e.

M =| (A ⇒ B) if and only if M |= A and M |= ¬B.

P6. M |= A if and only if M |= ∀xA.

P7. A formula A its true in M if and only if its closure (definition 17) is
true in M.

def:ptaut Definition 19 (Valid, Tautology)

Given a language L = L{¬,∩,∪,⇒,¬} (P, F, C), a formula A of L. .
A formula A is predicate tautology (is valid) if and only if M |= A for
all structures M = [U, I], i.e. when A is true in all structures M for L.
We write
|= A or |=p A,
to denote that a formula A is predicate tautology (is valid).

25
 We write
|=p A
when there is a need to stress a distinction between propositional and predicate
tautologies, otherwise we will use the symbol |= .
Predicate tautologies are also called laws of quantifiers.
Following the notation T for the set of all propositional tautologies (chapter ??)
we denote by Tp the set of all predicate tautologies, i.e.

Tp = {A of L{¬,∩,∪,⇒,¬} (P, F, C) : |=p A}. (13) Tp

Directly from the definition 18, the tautology definition 19 we get the following
basic properties of logical consequence as defined by definition 15.

logcons Property 2
For any A, B ∈ F and any set Γ ⊆ F of formulas of L,
P1. A |= B if and only if |= (A ⇒ B).
P2. If A |= B and A is true in M, then B is true in M.
P2. If Γ |= B and if all formulas in Γ are true in M, then B is true in M.

We get immediately from the above definition 19 of a following definition of a

notion ” A is not a predicate tautology”.

def:notaut Definition 20
For any formula A of predicate language L,
A is not a predicate tautology (6|= A) if and only if there is a structure
M = (U, I) for L, such that M 6|= A.
We call such structure M a counter-model for A.

The definition 20 says: to prove that a formula A is not a predicate tautology

one has to show a counter- model M = (U, I). It means one has to show a non-
empty set U, define an interpretation I, and an assignment s : V AR −→ U
such that (M, s) 6|= A.

We introduce, similarly as in a case of propositional semantic a notion of pred-

icate contradiction.

def:contr Definition 21 (Contradiction)

For any formula A of predicate a language L,
A is a predicate contradiction if and only if A is false in all structures M.

26
We denote it as =| A and write symbolically
=| A if and only if M=| A, for all structures M.
When there is a need to distinguish between propositional and predicate contra-
dictions we also use symbol
=|p A,
where ”p” stands for ”predicate”.

Following the notation C for the set of all propositional contradictions (chap-
ter ??) we denote by Cp the set of all predicate contradictions, i.e.
Cp = {A of L{¬,∩,∪,⇒,¬} (P, F, C) : =|p A}. (14) Cp

Directly from the definition 18 and Property 1 we have the following duality
property, the same as the one for propositional logic.

Fact 2
For any formula A of predicate a language L,
A ∈ Tp if and only if ¬A ∈ Cp ,
A ∈ Cp if and only if ¬A ∈ Tp .

Obviously, the formulas (Q(x, c) ∩ ¬Q(x, c)), (¬∀xQ(x, c) ∩ ∀xQ(x, c)), and the
formula (∃xQ(x, c) ∩ ¬∃xQ(x, c)) defined in example 17 are not only false in
the structures M1 − M4 , but are false in all structures M for L. By definition
21 they all are predicate contradictions. Observe that they all are substitutions
of propositional contradictions (a ∩ ¬a) or (¬a ∩ a). By the same argument the
formulas (Q(x, c)∪¬Q(x, c)), (¬∀xQ(x, c)∪∀xQ(x, c)), (∃xQ(x, c)∩¬∃xQ(x, c))
are predicate tautologies as they are substitutions of propositional tautologies
(a ∪ ¬a) or (¬a ∪ a).
We put these examples and observations in a following theorems that establish
relationship between propositional and predicate tautologies and contradictions.
We write now |=, =| do denote respectively propositional tautologies and con-
tradiction, and |=p , =|p for predicate tautologies and contradictions. We first
formalize and prove (theorem 1) the intuitively obvious fact: if a formula A is
a propositional tautology (contradiction), then replacing propositional variables
in A by any formulas of a predicate language we obtain a formula which is a
predicate tautology (contradiction).

Example 18
Let consider the following example of a propositional tautology and a proposi-
tional contradiction.
|= ((a ⇒ b) ⇒ (¬a ∪ b)) and =| ((a ∪ ¬a) ⇒ (¬b ∩ b)).

27
 Substituting ∃xP (x, z) for a, and ∀yR(y, z) for b, we obtain, by theorem 1, that

|=p ((∃xP (x, z) ⇒ ∀yR(y, z)) ⇒ (¬∃xP (x, z) ∪ ∀yR(y, z))) and

=| ((∃xP (x, z) ∪ ¬∃xP (x, z)) ⇒ (¬∀yR(y, z) ∩ ∀yR(y, z))).

We put it all in a more formal and more general and precise language as follows.
Given a propositional language L0 = L{¬,∩,∪,⇒,¬} with the set F0 of formulas
and a predicate languageL = L{¬,∩,∪,⇒,¬} (P, F, C) with the set F of formulas.
Let A(a1 , a2 , . . . , an ) ∈ F0 and A1 , A2 , . . . , An ∈ F. We denote by

A(a1 /A1 , a2 /A2 , . . . , an /An ) (15) sfor

the result of replacing in A the free variables a1 , a2 , . . . , an by the formulas

A1 , A2 , . . . , An ∈ F. Of course A(a1 /A1 , a2 /A2 , . . . , an /An ) ∈ F.

thm:sub1 Theorem 1
Given a propositional language L0 with the set F0 of formulas and a predicate
language L with the set F of formulas.
For any A(a1 , a2 , . . . an ) ∈ F0 and any A1 , A2 , . . . An ∈ F the following holds.
1. If |= A(a1 , a2 , . . . , an ), then |=p A(a1 /A1 , a2 /A2 , . . . , an /An ).
2. If =| A(a1 , a2 , . . . , an ), then =|p A(a1 /A1 , a2 /A2 , . . . , an /An ).

Proof 1. follows directly from satisfaction definition 13. 2. follows from defini-
tion 13, property 1, and definition 21.
Some predicate tautologies are, by theorem 1, substitutions of propositional
formulas. Visibly a predicate formula (∀x A(x) ⇒ ∃x A(x)) can not be obtained
as a substitution of a propositional formula. We prove now that it is a predicate
tautology.

p1 Fact 3
For any formula A(x) of L,

|= (∀x A(x) ⇒ ∃x A(x)).

Proof
Assume that 6|= (∀x A(x) ⇒ ∃x A(x)). By definition 20 there is a structure
M = (U, I) and s : V AR −→ U, such that (M, s) 6|= (∀x A(x) ⇒ ∃x A(x)).
By definition 13, (M, s) |= ∀x A(x) and (M, s) 6|= ∃x A(x). It means that
(M, s0 ) |= A(x) for all s0 such that s, s0 agree on all variables except on x,
and it is not true that there is s0 such that s, s0 agree on all variables except

28
 on x, and (M, s0 ) |= A(x). This is impossible and this contradiction proves
|= (∀x A(x) ⇒ ∃x A(x)).
Given a set F of formulas of a predicate language L. We denote by OF set of all
open formulas of L, i.e. formulas without quantifiers. We prove that any open
formula in order to be predicate tautologies must be a substitution definied in
theorem 1 of a propositional tautology. I.e. we have the following substitution
theorem.

thm:sub2 Theorem 2
Any open formula A of a predicate language L is a predicate tautology if and
only if it is a substitution of a propositional tautology as defined in theorem 1.

Proof
Observe that every open formula from A ∈ OF is a form
B(a1 /A1 , a2 /A2 , . . . , an /An ) for certain propositional formula B(a1 , a2 , . . . an ),
where A1 , A2 , . . . , An are predicate atomic formulas from the set AF as defined
in (5). Theorem 2 follows directly from the following.

lemsub2 Lemma 1
Let σ be a one to one mapping from the set V0 of propositional variables of
propositional language L0 into the set AF of the atomic formulas of the predicate
language L. For any A(a1 , a2 , . . . an ) ∈ F0 ,

|= A(a1 , a2 , . . . , an ), if and only if |=p A(a1 /σ(a1 ), . . . , an /σ(an )).

Proof of lemma
The implication ”if |= A(a1 , a2 , . . . , an ), then
|=p A(a1 /σ(a1 ), . . . , an /σ(an ))” holds as a particular case of theorem 2. We
prove now the converse implication by proving its opposite

if 6|= A(a1 , a2 , . . . , an ), then 6|=p A(a1 /σ(a1 ), . . . , an /σ(an )). (16) opp

Assume 6|= A(a1 , a2 , . . . , an ). There exists a truth assignment v : V0 −→ {T, F }

such that v ∗ (A(a1 , a2 , . . . , an )) = F . We construct a counter model M for
A(a1 /σ(a1 ), . . . , an /σ(an )) as follows. Let M = [T, I], where T is the set of all
terms of L, and for any c ∈ C, f ∈ F, P ∈ P we put cI = c, fI (t1 , t2 , . . . , tn ) =
f (t1 , t2 , . . . tn ), PI ⊆ T#P .
Let now the s assignment of L in M be an identity, i.e. s : V AR −→ T is
such that s(x) = x for all x ∈ V AR. We extend s to the interpretation of terms
(definition 12) as follows.
sI (x) = s(x) = x, sI (c) = cI = c, sI (f (t1 , t2 , . . . tn )) = fI (sI (t1 ), . . . , sI (tn )) =
f (t1 , t2 , . . . tn ), i.e. we have that sI (t) = t for all t ∈ T.

29
 We have that for every atomic formula P (t1 , t2 , . . . tn ) there is exactly one propo-
sitional variable a, such that P (t1 , t2 , . . . tn ) = σ(a). We define now that PI as
follows.
(t1 , t2 , . . . tn ) ∈ PI if and only if P (t1 , t2 , . . . tn ) = σ(a) and v(a) = T .
(t1 , t2 , . . . tn ) 6∈ PI if and only if P (t1 , t2 , . . . tn ) = σ(a) and v(a) = F .
We assumed that v : V0 −→ {T, F } is such that v ∗ (A(a1 , a2 , . . . , an )) = F .
Directly form definition of the assignment s and the interpretation I we have
that ([T, I], s) 6|= A(a1 /σ(a1 ), . . . , an /σ(an )). It end the roof of lemma 1 and
hence the proof of theorem 2.

Fact 4
The converse implication to (3) is not a predicate tautology, i.e. there is a
formula A of L, such that
6|= (∃x A(x) ⇒ ∀x A(x)). (17) p2

Proof
Observe that to prove (17) we have to provide an example of an instance of a
formula A(x) and construct a counter-model M = (U, I) for it. Let A(x) be an
atomic formula P (x, c), for any P ∈ P, #P = 2. The instance is
(∃x P (x, c) ⇒ ∀x P (x, c)).
We take as M = (N, PI :<, cI : 3) for N set of natural numbers. Let s be any
assignment s : V AR −→ N. We show now (M, s) |= ∃x P (x, c). Take any s0
such that s0 (x) = 2 and s0 (y) = s(y) for all y ∈ V AR−{x}. We have (2, 3) ∈ PI ,
as 2 < 3 and hence there exists s0 that agrees with s on all variables except on x,
and (M, s0 ) |= P (x, c). But (M, s) 6|= ∀x P (x, c) as for example for s0 such that
s0 (x) = 5 and s0 (y) = s(y) for all y ∈ V AR − {x}, (2, 3) 6∈ PI , as 5 6< 3. This
proves that M = (N, PI :<, cI : 3) is a counter model for ∀x P (x, c). Hence
6|= (∃x A(x) ⇒ ∀x A(x)).
The ”shorthand” solution is: the formula (∃x P (x, c) ⇒ ∀x P (x, c)) becomes
in M = (N, PI :<, cI : 3) a mathematical statement (written with logical
symbols): ∃n n < 3 ⇒ ∀n n < 3. It is an obviously false statement in the set
N of natural numbers, as there is n ∈ N , such that n < 3, for example n = 2,
and it is not true that all natural numbers are smaller then 3.
We have to be very careful when we deal with quantifiers with restricted
domain (definition 9). We adopt the following definition for restricted domain
quantifiers.

rtaut Definition 22 (Restricted Quantifiers Tautology)

For any formulas
A(x), B(x) ∈ F with any free variable x ∈ V AR, and for the restricted domain

30
quantifies ∀B(x) , ∃B(x) we define

|= ∀B(x) A(x) if and only if |= ∀x (B(x) ⇒ A(x)),

|= ∃B(x) A(x) if and only if |= ∃x (B(x) ∩ A(x)).

The most basic predicate tautology (3) fails when we use the quantifiers with
restricted domain. We show now that

6|= (∀B(x) A(x) ⇒ ∃B(x) A(x)). (18) rq1

By definition 22 to prove (18) means to prove that corresponding proper formula

of L obtained by the restricted quantifiers transformations rules (48), (49) is
not a predicate tautology, i.e. to show that

6|= (∀x(B(x) ⇒ A(x)) ⇒ ∃x(B(x) ∩ A(x))). (19) rq2

In order to prove (19) we have to provide an example of particular formulas

A(x), B(x) and to construct a counter model for these particular formulas. We
take as B(x), A(x) atomic formulas Q(x, c), P (x, c) and we construct a counter
model a corresponding formula

(∀x(Q(x, c) ⇒ P (x, c)) ⇒ ∃x(Q(x, c) ∩ P (x, c))) (20) crq2

as follows. We take M = (N, I), where N is the set of real numbers and the
interpretation I is defined as QI :<, PI :>, cI : 0. The ”shorthand” solution is
as follows. The formula 19) becomes a mathematical statement

(∀n (n < 0 ⇒ n > 0) ⇒ ∃n∈N (n < 0 ∩ n > 0)).

This statement is a false in the set N of natural numbers because the statement
n < 0 is false for all natural numbers and F ⇒ B is a true implication for
any logical value of B, so ∀n (n < 0 ⇒ n > 0) is a true statement and
∃n (n < 0 ∩ n > 0) is obviously false in the set N of natural numbers.
The restricted quantifiers law corresponding to the predicate tautology (3) is:

|= (∀B(x) A(x) ⇒ (∃x B(x) ⇒ ∃B(x) A(x))). (21) rq3

By definition 22 and restricted quantifiers transformations rules (7), (8) proving

(19 ) is means proving

|= (∀x(B(x) ⇒ A(x)) ⇒ (∃x B(x) ⇒ ∃x (B(x) ∩ A(x)))).

We leave the proof and an exercise for the reader.

31
 3 Predicate Tautologies
sec:predtaut
We have already proved in Fact 3 the basic predicate tautology

|= (∀x A(x) ⇒ ∃x A(x)).

We are going to prove now the following.

Fact 5 (Dictum de Omni)

For any formula A(x) of L,

|= (∀x A(x) ⇒ A(t)), |= (∀x A(x) ⇒ A(x)), (22) p3

|= (A(t) ⇒ ∃x A(x)), (23) p4

where t is a term, A(t) is a result of substitution of t for all free occurrences
of x in A(x), and t is free for x in A(x) (definition 7), i.e. no occurrence of a
variable in t becomes a bound occurrence in A(t).

Proof of (22) is constructed in a sequence of steps. We leave details to the

reader to complete as an exercise. Here are the steps.
S1 Consider a structure M = [U, I] and s : V AR −→ U. Let t, u be two terms.
Denote by t0 a result of replacing in t all occurrences of a variable x by the term
u, i.e. t0 = t(x/u). Let s0 results from s by replacing s(x) by sI (u). We prove
by induction over the length of t that

sI (t(x/u)) = sI (t0 ) = s0 I (u). (24) st

S2 Let t be free for x in A(x). A(t) is a results from A(x) by replacing t for
all free occurrences of x in A(x), i.e. A(t) = A(x/t). Let s : V AR −→ U and
s0 be obtained from s by replacing s(x) by sI (u). We use (24) and induction on
the number of connectives and quantifiers in A(x) and prove

(M, s) |= A(x/t) if and only if (M, s0 ) |= A(x). (25) Mt

S3 Directly from definition 13 and (25) we get that for any M = (U, I) and
any s : V AR −→ U ,

if (M, s) |= ∀xA(x), then (M, s) |= A(t).

This proves that (∀x A(x) ⇒ A(t)) is a predicate tautology. Observe that a
term x is free for x in A(x), so we also get as a particular case of t = x that
|= (∀x A(x) ⇒ A(x)).
Proof of (23) follows from (22), theorem 1, property 1, theorem 5, and defin-
ability law (49). We carry it as follows. First we observe that by theorem 1 we

32
 have that |= ((∀x¬A(x) ⇒ ¬A(t)) ⇒ (A(t) ⇒ ¬∀x¬A(x))) as a substitution of
propositional tautology ((a ⇒ ¬b) ⇒ (b ⇒ ¬a)). By just proved (22) we have
that |= (∀x¬A(x) ⇒ ¬A(t)) for A(x) being a formula ¬A(x). By P 4 in property
1, we get |= (A(t) ⇒ ¬∀x¬A(x)). We apply the existential quantifier definability
law (49) and equivalence substitution theorem 5 and get |= (A(t) ⇒ ∃x A(x)).
This ends the proof of (22).

Remark the restrictions in (22) and (23) are essential. Here is a simple example
explaining why they are needed in (22). The example for (23) is similar.
Let A(x) be a formula ¬∀y P (x, y), for P ∈ P. Notice that a term t = y is not
free for y in A(x). Consider (22) A(x) = ¬∀y P (x, y) and t = y.

(∀x¬∀y P (x, y) ⇒ ¬∀y P (y, y)), (26) Pp3

Take M = [N, I] for I such that PI : =. Obviously, M |= ∀x¬∀y P (x, y)

as ∀m¬∀n(m = n) is a true mathematical statement in the set N of natural
numbers. M 6|= ¬∀y P (y, y) as ¬∀n (n = n) is a false statement for n ∈
N . Hence M is a counter model for for (26) and we proved that without the
restriction (22) does not hold.
Here are some useful and easy to prove properties of the notion ”t free for x in
A(x)” (definition 7).

termprop Property 3 (t free for x in A(x))

For any formula A ∈ F and any term t ∈ T the following properties hold.
P1. A closed tern t, i.e. term with no variables is free for any variable x in A.
P2. A term t is free for any variable in A if none of the variables in t is bound
in A.
P3. Term t = x is free for x in any formula A.
P4. Any term is free for x in A if A contains no free occurrences of x.

Here are some more important predicate tautologies.

Generalization
For any formulas A(x), B(x), A, B of L, where A, B does not contain any free
occurrences of x,

|= ((B ⇒ A(x)) ⇒ (B ⇒ ∀x A(x))), (27) gen2

|= ((B(x) ⇒ A) ⇒ (∃xB(x) ⇒ A)), (28) gen3

Distributivity 1

33
For any formulas A(x), B(x), A, B of L, such that A , B does not contain any
free occurrences of x,

|= (∀x(A ⇒ B(x)) ⇒ (A ⇒ ∀x B(x))), (29) Mdistr

|= ∀x(A(x) ⇒ B) ⇒ (∃xA(x) ⇒ B) (30) 1dis1

|= ∃x(A(x) ⇒ B) ⇒ (∀xA(x) ⇒ B) (31) 1dis2

The restrictions that the formulas A, B do not contain any free occurrences of
x is essential for both Generalization and Distributivity 1 tautologies.
Here is a simple example explaining why they are needed in (29). The relaxation
of the assumption that A, B do not contain any free occurrences of x would lead
to the following disaster. Let A and B(x) be both atomic formula P(x). Thus x
is free in A and we have the following instance of (29).

(∀x(P (x) ⇒ P (x)) ⇒ (P (x) ⇒ ∀x P (x))).

Observe that ∀x(P (x) ⇒ P (x)) is a predicate tautology. Take M = [N, I] for
I such that PI = ODD, where ODD ⊆ N is the set of odd numbers. Let
s : V AR −→ N . By definition if I, sI (x) ∈ PI if and only if sI (x) ∈ ODD.
Then obviously (M, s) 6|= ∀x P (x) and M = [N, I] is a counter model for (29)
as (M, s) |= ∀x(P (x) ⇒ P (x)).
The examples for (30), (31), and (29) similar.
Distributivity 2
For any formulas A(x), B(x) of L,

|= (∃x (A(x) ∩ B(x)) ⇒ (∃xA(x) ∩ ∃xB(x))), (32) 2dis1

|= ((∀xA(x) ∪ ∀xB(x)) ⇒ ∀x (A(x) ∪ B(x))), (33) 2dis2

|= (∀x(A(x) ⇒ B(x)) ⇒ (∀xA(x) ⇒ ∀xB(x))) (34) 2dis3
The converse mplications to (32), (33), (34) are not a predicate tautologies , i.e.
there are formulas A(x), B(x), such that

6|= ((∃xA(x) ∩ ∃xB(x)) ⇒ ∃x (A(x) ∩ B(x))). (35) n2dis1

6|= (∀x (A(x) ∪ B(x)) ⇒ (∀xA(x) ∪ ∀xB(x))), (36) n2dis2

6|= ((∀xA(x) ⇒ ∀xB(x)) ⇒ ∀x(A(x) ⇒ B(x))) (37) n2dis3
To prove (35), (36) we have to find particular formulas A(x), B(x) ∈ F and
counter models M = [U, I] for these particular cases.
Consider (35). We take as A(x), B(x) atomic formulas Q(x, c), P (x, c). The
particular case of (35) is now a formula

((∃xP (x, c) ∩ ∃xQ(x, c)) ⇒ ∃x (P (x, c) ∩ Q(x, c))).

34
Take M = [R, I] where R is the set of real numbers, and the interpretation I is
QI :>, PI :<, cI : 0. The particular case formula becomes an obviously false
mathematical statement
((∃x∈R x > 0 ∩ ∃x∈R x < 0) ⇒ ∃x∈R (x > 0 ∩ x < 0)).

Consider (36). We take as Let A(x), B(x) be atomic formulas Q(x, c), R(x, c).
The particular case of (36 ) is now a formula
(∀xQ(x, c) ∪ R(x, c)) ⇒ (∀xQ(x, c) ∪ ∀xR(x, c))).
Take M = (R, I) where R is the set of real numbers and QI :≥, RI :<, cI : 0.
The particular formula becomes an obviously false mathematical statement
(∀x∈R (x ≥ 0 ∪ x < 0) ⇒ (∀x∈R x ≥ 0 ∪ ∀x∈R x < 0)).

De Morgan
For any formulas A(x), B(x) of L,
|= (¬∀xA(x) ⇒ ∃x¬A(x)), (38) 1mall
|= (¬∃xA(x) ⇒ ∀x¬A(x)), (39) 1mex
|= (∃x¬A(x) ⇒ ¬∀xA(x)), (40) 2mall
|= (∀x¬A(x)) ⇒ ¬∃xA(x)). (41) 1mex

We prove (38) as an example.The proofs of all other laws are similar. As-
sume that (38) does not hold. By definition 16 there is M = (U, I) and
s : V AR −→ U, such that (M, s) |= ¬∀xA(x)) and (M, s) 6|= ∃x¬A(x).
Consider (M, s) |= ¬∀xA(x). By satisfaction definition 13, (M, s) 6|= ∀xA(x).
This holds only if for all s0 , such that s, s0 agree on all variables except on x,
(M, s0 ) 6|= A(x).
Consider (M, s) 6|= ∃x¬A(x). This holds only if there is no s0 , such that
(M, s0 ) |= ¬A(x), i.e. there is no s0 , such that (M, s0 ) 6|= A(x). This means
that for all s0 , (M, s0 ) |= A(x). Contradiction with (M, s0 ) 6|= A(x).
Quantifiers Alternations
For any formula A(x, y) of L,
|= (∃x∀yA(x, y) ⇒ ∀y∃xA(x, y)). (42) qalt

The converse implications to (42) is not a predicate tautology. Take as A(x, y)

an atomic formulas R(x, y). Take M = (R, I) where R is the set of real numbers
and RI :< . The instance of (42) particular formula becomes a mathematical
statement
(∀y∃x(x < y) ⇒ ∃x∀y(x < y))
that obviously false in the set of real numbers. We proved

6|= (∀y∃xA(x, y) ⇒ ∃x∀yA(x, y)). (43) nqalt

35
 3.1 Equational Laws of Quantifiers

The most frequently used laws of quantifiers have a form of a logical equivalence,
symbolically written as ≡. This not a new logical connective. This is a very
useful symbol. It has the same properties as the equality = and can be used in
the same way we use the equality symbol =.
Note that we use the same equivalence symbol ≡ and the tautology symbol
|= for propositional and predicate languages and semantics when there is no
confusion. Formally we define the predicate equivalence as follows.

def1:peq Definition 23 (Logical equivalence)

For any formulas A, B ∈ F of the predicate language L,

A ≡ B if and only if |= (A ⇒ B) and |= (B ⇒ A).

Remark that our predicate language L = L{¬,∩,∪,⇒,¬} (P, F, C) we defined the

semantics for (definition 13 ) does not include the equivalence connective ⇔. If
it does we extend the satisfaction definition 13 in a natural way and can adopt
the following definition 24 of logical equivalence that is obviously equivalent
definition to the propositional one and to our definition 23.

def2:peq Definition 24
For any formulas A, B ∈ F of the predicate language L,

A ≡ B if and only if |= (A ⇔ B).

We re-write the basic theorem 1 establishing relationship between propositional

and some predicate tautologies as follows.

thm1 Theorem 3 (Tautologies)

If a formula A is a propositional tautology, then by substituting for propositional
variables in A any formula of the predicate language L we obtain a formula which
is a predicate tautology.

Directly from the theorem 3 and logical equivalence definition 23 we get that
the following is true.

thm2 Theorem 4 (Equivalences)

Given propositional formulas A, B.
If A ≡ B is a propositional equivalence, and A0 , B 0 are formulas of the predicate
language L obtained by a substitution of any formula of L for propositional vari-
ables in A and B, respectively, then A0 ≡ B 0 holds under predicate semantics.

36
 e2 Example 19
Consider the following propositional logical equivalence:

(a ⇒ b) ≡ (¬a ∪ b).

Substituting ∃xP (x, z) for a, and ∀yR(y, z) for b, we get from theorem 4 that
the following equivalence holds:

(∃xP (x, z) ⇒ ∀yR(y, z)) ≡ (¬∃xP (x, z) ∪ ∀yR(y, z)).

We prove in similar way as in the propositional case (chapter 3) the following.

thm:subeq Theorem 5 (Equivalence Substitution)

Let a formula B1 be obtained from a formula A1 by a substitution of a formula
B for one or more occurrences of a sub-formula A of A1 , what we denote as

B1 = A1 (A/B).

Then the following holds for any formulas A, A1 , B, B1 of L.

If A ≡ B, then A1 ≡ B1 . (44) thm:leqv

Directly from the Dictum de Omi (22) and the Generalization (27) tautologies
we get the proof of the following theorem 6 useful for building new logical
equivalences from the old, known ones.

thm3 Theorem 6
For any formulas A(x), B(x) of L.

if A(x) ≡ B(x), then ∀xA(x) ≡ ∀xB(x),

if A(x) ≡ B(x), then ∃xA(x) ≡ ∃xB(x).

e3 Example 20
We know from the example 19 that the formulas (∃xP (x, z) ⇒ ∀yR(y, z)) and
(¬∃xP (x, z) ∪ ∀yR(y, z)) are logically equivalent. We get, as the direct conse-
quence of the theorem 6 the following equivalences:

∀z(∃xP (x, z) ⇒ ∀yR(y, z)) ≡ ∀z(¬∃xP (x, z) ∪ ∀yR(y, z)),

Theorem 4 and theorem 6 allow us to use propositional tautologies and predicate

formulas to build predicate equivalences. Here is a simple example.

37
 Exercise 6
Prove that for any formulas A(x), B(x) of L

¬∀x¬(A(x) ∪ B(x)) ≡ ¬∀x(¬A(x) ∩ ¬B(x)). (45) exer

Solution
By the substituting A(x) for a, and any formula B(x) for b, in the propositional
de Morgan Law: ¬(a ∪ b) ≡ (¬a ∩ ¬b), we get via theorem 4 that

¬(A(x) ∪ B(x)) ≡ (¬A(x) ∩ ¬B(x)).

Applying the theorem 6 to the above we obtain that

∀x¬(A(x) ∪ B(x)) ≡ ∀x(¬A(x) ∩ ¬B(x)).

We know, from the propositional logic, that for any propositional variables a, b,
a ≡ b if and only if ¬a ≡ ¬b. Substituting ∀x¬(A(x) ∪ B(x)) and ∀x(¬A(x) ∩
¬B(x)) for a and b, respectively, we get that

∀x¬(A(x) ∪ B(x)) ≡ ∀x(¬A(x) ∩ ¬B(x))

if and only if
¬∀x¬(A(x) ∪ B(x)) ≡ ¬∀x(¬A(x) ∩ ¬B).
But we have proved that ∀x¬(A(x)∪B) ≡ ∀x(¬A(x)∩¬B) holds, so we conclude
that the equivalence (45) also holds.

e4 Exercise 7
Prove that for any formulas A(x), B of L

∀x¬(A(x) ∪ B) ≡ ∀x(¬A(x) ∩ ¬B)

Solution
By the substituting A(x) for a, and any formula B for b, in the propositional
de Morgan law: ¬(a ∪ b) ≡ (¬a ∩ ¬b), we get that

¬(A(x) ∪ B) ≡ (¬A(x) ∩ ¬B).

Applying the theorem 4 to the above we obtain that

∀x¬(A(x) ∪ B) ≡ ∀x(¬A(x) ∩ ¬B).

As we can see, it is possible to obtain a fair amount of predicate tautologies from

the propositional tautologies and theorems 3, 4 and 6, but as we have proved
will for never obtain for example the most basic law: (∀xA(x) ⇒ ∃xA(x)), any
many the most important others.

38
We concentrate now only on these laws which have a form of a logical equiva-
lence.They are called the equational laws for quantifiers.
Directly from the definition 23 and the de Morgan tautologies (38)-(41) we get
one of the most important equational laws, called also De Morgan Laws.
De Morgan Laws
¬∀xA(x) ≡ ∃x¬A(x) (46) mall
¬∃xA(x) ≡ ∀x¬A(x) (47) mexists

Now we will apply them to show that the quantifiers can be defined one by the
other i.e. that the following Definability Laws hold.
Definability Laws
∀xA(x) ≡ ¬∃x¬A(x) (48) q1
∃xA(x) ≡ ¬∀x¬A(x) (49) q2
The law (48) is often used as a definition of the universal quantifier in terms of
the existential one (and negation), the law (49) is a definition of the existential
quantifier in terms of the universal one (and negation).
Proof of (48)
Substituting any formula A(x) for a variable a in the propositional equivalence
a 𠪪a we get by theorem 4 that A(x) 𠪪A(x). Applying the theorem 6
to the above we obtain ∃xA(x) ≡ ∃x¬¬A(x). By the de Morgan Law (46)
∃x¬¬A(x) ≡ ¬∀x¬A(x) and hence ∃xA(x) ≡ ¬∀x¬A(x), what ends the proof.
Proof of (49)
We obtain ∀xA(x) ≡ ∀¬¬A(x) in a similar way as above. By the de Morgan
Law (47), ∀¬¬A(x) ≡ ¬∃¬A(x) and hence ∀xA(x) ≡ ¬∃¬A(x), what ends the
proof.
Other important equational laws are the following introduction and elimination
laws. We prove later the first two of them. We show that the laws (52) - (57) can
be deduced from laws (50) and (51), the de Morgan laws (46), (47), definability
laws (48), (49), propositional tautologies and theorems 3, 4, and theorem 5.
Introduction and Elimination Laws
If B is a formula such that B does not contain any free occurrence of x,
then the following logical equivalences hold.

∀x(A(x) ∪ B) ≡ (∀xA(x) ∪ B) (50) 1

∀x(A(x) ∩ B) ≡ (∀xA(x) ∩ B) (51) 2
∃x(A(x) ∪ B) ≡ (∃xA(x) ∪ B) (52) 3
∃x(A(x) ∩ B) ≡ (∃xA(x) ∩ B) (53) 4
∀x(A(x) ⇒ B) ≡ (∃xA(x) ⇒ B) (54) 5

39
 ∃x(A(x) ⇒ B) ≡ (∀xA(x) ⇒ B) (55) 6
∀x(B ⇒ A(x)) ≡ (B ⇒ ∀xA(x)) (56) 7

∃x(B ⇒ A(x)) ≡ (B ⇒ ∃xA(x)) (57) 8

The equivalences (50)-(53) make it possible to introduce a quantifier that pre-

cedes a disjunction or a conjunction into one component on the condition that
the other component does not contain any free occurrence of a variable which
is bound by that quantifier. These equivalences also make possible to eliminate
a quantifier from a component of a disjunction or a conjunction and to place
it before that disjunction or conjunction as a whole, on the condition that the
other component does not contain any free occurrence of a variable which that
quantifier would then bind.
The equivalences (54)-(57 )make it possible to introduce a quantifier preced-
ing an implication into the consequent of that implication, on the condition
that that antecedent does not contain any free occurrence of a variable which
is bound by that quantifier; they also make it possible to introduce a univer-
sal quantifier preceding an implication into the consequent of that implication
while changing it into an existential quantifier in the process, on the condition
that the consequent of that implication does not contain any free occurrence of
a variable bound by that quantifier. Equivalences (54)-(57) further enable us
to eliminate quantifiers from the antecedent of an implication to the position
preceding the whole implication, while changing a universal quantifier into an
existential one, and vice versa, in the process, and also to eliminate quantifiers
from the consequent of an implication to the position preceding the whole impli-
cation; the conditions that the other component of the implication in question
does not contain any free occurrence of a variable which that quantifier would
then bind, must be satisfied, respectively.
As we said before, the equivalences (50)-(57) are not independent, some of them
are the consequences of the others. Assuming that we have already proved (50)
and (51), the proof of (52) is as follows.
Proof of (52) ∃x(A(x) ∪ B) is logically equivalent, by the definability law
(49) to ¬∀x¬(A(x) ∪ B). By the reasoning presented in the proof of (45 ) for
B instead of B(x), we have that ¬∀x¬(A(x) ∪ B) ≡ ¬∀x(¬A(x) ∩ ¬B). By
the introduction law (51), ¬∀x(¬A(x) ∩ ¬B) ≡ ¬(∀x¬A(x) ∩ ¬B). Substituting
∀x¬A(x) for a and ¬B for b in propositional equivalence ¬(a∩¬b) ≡ (¬a∪¬¬b),
we get, by the theorem 4 that ¬(∀x¬A(x) ∩ ¬B) ≡ (¬∀x¬A(x) ∪ ¬¬B). In a
similar way we prove that ¬¬B ≡ B, by the definability law (49) ¬∀x¬A(x) ≡
∃xA(x), hence by theorem 5 (¬∀x¬A(x) ∪ ¬¬B) ≡ (∃xA(x) ∪ B) and finally,
∃x(A(x) ∪ B) ≡ (∃xA(x) ∪ B), what end the proof.

40
We can write this proof in a shorter, symbolic way as follows:
law 49
∃x(A(x) ∪ B) ≡ ¬∀x¬(A(x) ∪ B)
thm 3, 4
≡ ¬∀x(¬A(x) ∩ ¬B)
law 51
≡ ¬(∀x¬A(x) ∩ ¬B)
(46), thm 5
≡ (¬∀x¬A(x) ∪ ¬¬B)
thm 5
≡ (∃xA(x) ∪ B)

Distributivity Laws
Let A(x), B(x) be any formulas with a free variable x.
Law of distributivity of universal quantifier over conjunction

∀x (A(x) ∩ B(x)) ≡ (∀xA(x) ∩ ∀xB(x)) (58) univ

Law of distributivity of existential quantifier over disjunction.

∃x (A(x) ∪ B(x)) ≡ (∃xA(x) ∪ ∃xB(x)) (59) exist

Alternations of Quantifiers Laws

Let A(x, y) be any formula with a free variables x,y.

∀x∀y (A(x, y) ≡ ∀y∀x (A(x, y) (60) alt1

∃x∃y (A(x, y) ≡ ∃y∃x (A(x, y) (61) alt2

Renaming the Variables

Let A(x) be any formula with a free variable x and let y be a variable that does
not occur in A(x).
Let A(y) be a result of replacement of each occurrence of x by y, then the
following holds.
∀xA(x) ≡ ∀yA(y), (62) v1
∃xA(x) ≡ ∃yA(y). (63) v2

Restricted De Morgan Laws

For any formulas A(x), B(x) ∈ F with a free variable x,

¬∀B(x) A(x) ≡ ∃B(x) ¬A(x), ¬∃B(x) A(x) ≡ ∀B(x) ¬A(x). (64) RdeMorgan

Here is a poof of first equality. The proof of the second one is similar and is left
as an exercise.

¬∀B(x) A(x) ≡ ¬∀x (B(x) ⇒ A(x)) ≡ ¬∀x (¬B(x)∪A(x)) ≡ ∃x ¬(¬B(x)∪A(x))

41
 ≡ ∃x (¬¬B(x) ∩ ¬A(x)) ≡ ∃x (B(x) ∩ ¬A(x)) ≡ ∃B(x) ¬A(x)).

Restricted Introduction and Elimination Laws

If B is a formula such that B does not contain any free occurrence of x, then
the following logical equivalences hold for any formulas A(x), B(x), C(x).

∀C(x) (A(x) ∪ B) ≡ (∀C(x) A(x) ∪ B), (65) r1

∃C(x) (A(x) ∩ B) ≡ (∃C(x) A(x) ∩ B), (66) r4

∀C(x) (A(x) ⇒ B) ≡ (∃C(x) A(x) ⇒ B), (67) r5

∀C(x) (B ⇒ A(x)) ≡ (B ⇒ ∀C(x) A(x)). (68) r7

The proofs are similar to the proof of the restricted de Morgan Laws.
The similar generalization of the other Introduction and Elimination Laws (51),
(52), (55), (57) for restricted domain quantifiers fails. We can easily follow the
proof of (18) and construct proper counter-models proving the following.

∃C(x) (A(x) ∪ B) 6≡ (∃C(x) A(x) ∪ B),

∀C(x) (A(x) ∩ B) 6≡ (∀C(x) A(x) ∩ B),

∃C(x) (A(x) ⇒ B) 6≡ (∀C(x) A(x) ⇒ B),
∃C(x) (B ⇒ A(x)) 6≡ (B ⇒ ∃xA(x)).

Nevertheless it is possible to correctly generalize them all as to cover quantifiers

with restricted domain. We show it in a case of (51) and leave the other cases
to the reader as an exercise.

res2 Example 21
The restricted quantifiers version of (51) is the following.

∃C(x) (A(x) ∪ B) ≡ (∃C(x) A(x) ∪ (∃x C(x) ∩ B)). (69) r2

We derive (74) as follows.

∃C(x) (A(x) ∪ B) ≡ ∃x(C(x) ∩ (A(x) ∪ B)) ≡ ∃x((C(x) ∩ A(x)) ∪ (C(x) ∩ B))

≡ (∃x(C(x) ∩ A(x)) ∪ ∃x(C(x) ∩ B)) ≡ (∃C(x) A(x) ∪ (∃x C(x) ∩ B)).

We leave it as an exercise to specify and write references to transformation or

equational laws used at each step of our computation.

42
 4 Proof Systems: Soundness and Completeness
We adopt now general definitions from chapter ?? concerning proof systems to
the case of classical first order (predicate) logic.
We refer the reader to chapters ?? and ?? for a great array of examples, exercises,
homework problems explaining in a great detail all notions we introduce here
for the predicate case. The examples and exercises we provide here are not
numerous and restricted to the laws of quantifiers.
Given a language L = L{¬,∩,∪,⇒,¬} (P, F, C). Any proof system

S = (L = L{¬,∩,∪,⇒,¬} (P, F, C), F, LA, R) (70) Psys

is a predicate (first order) proof system.

The predicate proof system (70) is a Hilbert proof system if the set R of its
rules contains the Modus Ponens rule
A ; (A ⇒ B)
(M P ) ,
B
where A, B ∈ F.

Semantic Link: Logical Axioms LA

We want the set LA of logical axioms to be a non-empty set of classical predicate

tautologies (13), i.e.
LA ⊆ Tp ,
where
Tp = {A of L{¬,∩,∪,⇒,¬} (P, F, C) : |=p A}.

Remark 2
We use symbols |=p , Tp to stress the fact that we talk about predicate lah=nguage
and classical predicate tautologies.

Semantic Link 2: Rules of Inference R

We want the the rules of inference r ∈ R of S to preserve truthfulness. Rules

that do so are called sound. We define it formally as follows.

d:sound Definition 25 (Sound Rule)

Given an inference rule r ∈ R of the form
P1 ; P2 ; .... ; Pm
(r) ,
C

43
 where P1 .P2 , . . . , Pm , C ∈ F.
(i) We say that the rule (r) is sound if and only if the following condition
holds for all structures M = [U, I] for L.

If M |= {P1 , P2 , .Pm } then M |= C. (71) r-sd

(ii) The rule (r) is not sound if and only if there is a structure M = [U, I],
such that
M |= {P1 , P2 , .Pm } and M |6 = C. (72) n-s

In order to prove that the rule (r) is sound we have to show the implication
(71). It means, by definitions 14, 18, we have to show that that if all premisses
of the rule (r) are true in M = [U, I], so is its conclusion. This also justifies
correctness of the definition 25; sound rules do preserve the truthfulness as it
is defined in our semantics.

r1 Exercise 8
Prove the soundness of the rule

¬∀xA(x)
(r1) . (73) r1
∃x¬A(x)

Proof
Assume that the soundness condition (71) does not hold for for all structures
M = [U, I]. It means we assume that there is a structure M = [U, I], such that
M |= ¬∀xA(x) and M 6|= ∃x¬A(x).
Let M |= ¬∀xA(x). By definition 14, for all s : V AR −→ U we have (M, s) |=
¬∀xA(x)). Hence by satisfaction definition 13, (M, s) 6|= ∀xA(x). This holds
only if for all s0 , such that s, s0 agree on all variables except on x, (M, s0 ) 6|= A(x).
Observe that M 6|= ∃x¬A(x) only if there is no s0 , such that (M, s0 ) |= ¬A(x),
i.e. there is no s0 , such that (M, s0 ) |= A(x). This means that for all s0 ,
(M, s0 ) |= A(x). Contradiction with (M, s0 ) 6|= A(x).

r2 Exercise 9
Prove the soundness of the rule

∀xA(x)
(r2) . (74) r2
∃xA(x)

Proof
Assume that the soundness condition (71) does not hold for for all structures
M = [U, I]. It means we assume that there is a structure M = [U, I], such that
M |= ∀xA(x) and M 6|= ∃xA(x).

44
 Let M |= ∀xA(x). By definition 14, for all s : V AR −→ U we have (M, s) |=
∀xA(x)).
By definition 13, (M, s) |= ∀x A(x) and (M, s) 6|= ∃x A(x). It means that
(M, s0 ) |= A(x) for all s0 such that s, s0 agree on all variables except on x, and
it is not true that there is s0 such that s, s0 agree on all variables except on x,
and (M, s0 ) |= A(x). This is impossible and this contradiction proves soundness
of (r2).

er3 Exercise 10
Prove that the rule
∃xA(x)
(r3) . (75) r3
∀xA(x)
is not sound.

Proof
Observe that to prove that the rule (75) is not sound we have to provide an
example of an instance of a formula A(x) and prove (ii) of definition 25 for it.
Let A(x) be an atomic formula P (x, c), for any P ∈ P, #P = 2. We take as
M = (N, PI :<, cI : 3) for N set of natural numbers. Let s be any assignment
s : V AR −→ N. Obviously (M, s) |= ∃x P (x, c).
Take any s0 such that s0 (x) = 2 and s0 (y) = s(y) for all y ∈ V AR − {x}. We
have (2, 3) ∈ PI , as 2 < 3 and hence there exists s0 that agrees with s on all
variables except on x, and (M, s0 ) |= P (x, c). But (M, s) 6|= ∀x P (x, c) as for
example for s0 such that s0 (x) = 5 and s0 (y) = s(y) for all y ∈ V AR − {x},
(2, 3) 6∈ PI , as 5 6< 3.
This proves that M = (N, PI :<, cI : 3) is a model for (∃x P (x, c) and hence
6|= ∀x A(x)).
The ”shorthand” solution is: the formula (∃x P (x, c) becomes in M = (N, PI :<
, cI : 3) a true mathematical statement (written with logical symbols): ∃n n <
3. The formula (∀x P (x, c) becomes a mathematical frmula ∀n n < 3 which is
an obviously false statement in the set N of natural numbers, as there is n ∈ N ,
such that n < 3, for example n = 2, and it is not true that all natural numbers
are smaller then 3. So the rule (r3) is not sound.

d:ssound Definition 26 (Strongly Sound Rule)

An inference rule r ∈ R of the form
P1 ; P2 ; .... ; Pm
(r)
C
is strongly sound if the following condition holds for all structures M = [U, I]
for L.

45
 M |= {P1 , P2 , .Pm } if and only if M |=C. (76) sseq

We can, and we do state it informally as: ” an inference rule r ∈ R is strongly

sound when the conjunction of all its premisses is logically equivalent to its
conclusion”. We denote it informally as

P1 ∩ P2 ∩ . . . ∩ Pm ≡ C. (77) ss-equiv

Example 22
The sound rule (73)
¬∀xA(x)
(r1)
∃x¬A(x)
is strongly sound by De Morgan Law (46).
The sound rule (75)
∀xA(x)
(r2)
∃xA(x)
is not strongly sound by exercise 10.

def:sound Definition 27 (Sound Proof System)

Given the predicate (first order) proof system (70)

S = (L, F, LA, R).

We say that the proof system S is sound if the following conditions hold.
(1) LA ⊆ Tp ;
(2) Each rule of inference r ∈ R is sound.
The proof system S is strongly sound if the condition (2) is replaced by the
following condition (2’)
(2’) Each rule of inference r ∈ R is strongly sound under M.

The set of all provable expressions of S is denoted by PS and is defined as

follows.
PS = {A ∈ F : `S A}. (78) S-prov

When we define (develop) a proof system S our first goal is to make sure that it
a ”sound” one, i.e. that all we prove in it is true. Proving the following theorem
establishes this goal.

46
 thm:Ss Theorem 7 (Soundness Theorem for S)
Given a predicate proof system S.
For any A ∈ F, the following implication holds.

If `S A then |=p A. (79) s-impl

We write (79) it in a more concise form as

PS ⊆ Tp . (80) Ss

Proof
Observe that if we have already proven that S is sound as stated in the definition
27, the proof of the implication (79) is a straightforward application of the
mathematical induction over the length of the formal proof of the formula A.
It means that in order to prove the Soundness Theorem 7 for a proof system Sit
is enough to verify the two conditions of the definition 27 (1) LA ⊆ Tp and (2)
each rule of inference r ∈ R is sound.

We again refer the reader to chapter ?? for detailed examples, exercises and
problems.
As we can see, proving Soundness Theorem 7 for any proof system we develop
is indispensable and the proof is quite easy. The next step in developing a logic
(classical predicate logic in our case now) is to answer necessary and a difficult
question: Given a proof system S, about which we know that all it proves it
true (tautology). Can we prove all we know to be true (all tautologies)?
Proving the following theorem establishes this goal.

th:compl Theorem 8 (Completeness Theorem for S)

Given a predicate proof system S.
For any A ∈ F, the following holds.

`S A if and only if |=p A. (81) s-comp

We write (81) it in a more concise form as

PS = Tp . (82) Scomp

The Completeness Theorem consists of two parts:

Part 1: Soundness Theorem: PS ⊆ Tp .
Part 2: Completeness part of the Completeness Theorem: Tp ⊆ PS .
Proving the Soundness Theorem for S is usually a straightforward and not a very
difficult task. Proving the Completeness part of the Completeness Theorem is

47
always a crucial and very difficult task. There are many methods and techniques
for doing so, even for classical proof systems (logics) alone. Non-classical logics
often require new sometimes very sophisticated methods. We presented two
proofs of the Completeness Theorem for classical propositional Hilbert style
proof system in chapter ??, and a constructive proofs for automated theorem
proving systems for classical propositional logic the chapter ??.
We present a proof of the Completeness Theorem for predicate (first order) logic
in the next chapter ??.

5 Homework Problems
Predicate Languages

1. Given the following formulas A1 − A5 of a predicate language L.

A1 = R(x, y, g(c, x)), A2 = ∃xP (x, f (x, y)), A3 = ∃dR(x, y, g(c, d)),

A4 = ∀z(f (x, P (c, y)), A5 = ∃yP (x, f (c, y)) ∪ ∀yP (x, f (c, y)).
(a) Indicate whether they are, or are not well formed formulas of F. For
those which are not in F write a correct formula.
(b) For each correct, or corrected formula identify all components: con-
nectives, quantifiers, predicate and function symbols, and list all its
terms.
(c) For each formula identify its s free and bound variables. State which
are open and which are closed formulas (sentences), if any.
(d) Describe a language defined by the set F0 = {A1 , A2 , . . . A5 } of
formulas that are correct or corrected.
2. For the following mathematical statements write their corresponding for-
mulas of predicate language L.

(a) ∀n>1 (n + 3 < 8 ∪ ∃x∈R x + n > 8)

(b) ∀x∈R ∃n∈N (x + n > 0 ⇒ ∃m∈N (m = x + n))
(c) If all natural numbers are smaller then zero, then the sum of any two
integers is smaller then zero.
(d) For all natural numbers The following implication holds for all natural
numbers: if n > 0, then there is a real number x, such that n + x = 0
or there is an integer m, such that m > 0.

3. For each of the following formulas (some with restricted quantifiers) write
2 corresponding natural language sentences.

48
 (a) ∀x(P (x) ⇒ ∃yQ(x, y)).
(b) ∀x∃y(P (x) ∩ ¬Q(x, y)).
(c) ∀A(x) ∃A(y) B(y).
(d) ∃P (x) ∀N (x) R(x, y).

4. Is the term t = f (x, y) free for x in the following formulas?

(a) (P (x, y) ⇒ ∀yP (x, y).

(b) (∀yP (x, y) ∪ ∃yP (x, y)).
(c) ∀xP (x, y).
(d) ∀yP (x, y).
(e) (∀yQ(y) ⇒ P (x, y)).

5. Justify that for any formula A ∈ F and any term t ∈ T the following facts
hold.

(a) A closed tern t, i.e. term with no variables is free for any variable x
in A.
(b) A term t is free for any variable in A if none of the variables in t is
bound in A.
(c) Term t = x is free for x in any formula A.
(d) Any term is free for x in A if A contains no free occurrences of x.

6. Translate the following formulas in everyday English.

(a) ∀x(P (x) ∩ ∀y(¬L(x, y) ⇒ ¬H(x))), where P (x) means ” x is a per-

son”, L(x, y)means ”x likes y” , and H(x) means ”x is happy”.
(b) ∀x((E(x) ∩ P (x)) ⇒ E(x, b)), where E(x) means ” x is an even inte-
ger”, P(x) means ” x is prime”, Q(x,y) means ” x is equal to x”, and
b denotes 2.
(c) ¬∀y((P (y)∩∀x(P (x)) ⇒ E(x, y))), where P(x) means ”x is an integer,
and L(x, y)means ”x ≤ y”.

7. Use the restricted quantifiers to translate the following natural language

sentences into a proper formulas of a proper formal predicate language
L = L{¬,∩,∪,⇒,¬} (P, F, C). In each case specify the sets P, F, C.

(a) Some politician are honest, some are not.

(b) Any sets that have the same elements are equal.
(c) Somebody hates everyone who does not hate himself.
(d) Birds can fly and if anyone can fly Tweety can.
(e) Anyone who knows logic loves it.

49
Classical Semantics

1. Given a predicate language L(P, F, C) and a structure M = [U, I] such

that U = N and PI : =, fI : +, gI : ·, aI : 0, bI : 1 for N set on natural
numbers. For each of the following formula A decide whether M |= A
or not. Do so by examining the corresponding mathematical statement
defined by M.
(a) ∀x∃y(P (x, f (y, y)) ∪ P (x, f (f (y, y), b))).
(b) ∀x∃y(P (g(x, y), a) ⇒ (P (x, a) ∪ P (y, a))).
(c) ∃yP (f (y, y), b).
2. Let M = [U, I] be a structure such that U = Z and PI : =, fI : + for Z
set of integers. For each of the following formula A decide whether M |= A
or not. Do so by examining the corresponding mathematical statement
defined by M.
(a) ∀x∀yP (f (x, y), f (y, x)).
(b) ∀x∀y∀zP (f (x, f (y, z)), f (f (x, y), z)).
(c) ∀x∀y∃zP (f (x, z), y)
3. Let M = [U, I] be a structure such that U = N − {0} or N set of natural
numbers and PI : =, fI (x, y) is xy For each of the following formula A
decide whether M |= A or not. Do so by examining the corresponding
mathematical statement defined by M.
(a) ∀x∀yP (f (x, y), f (y, x)).
(b) ∀x∀y∀zP (f (x, f (y, z)), f (f (x, y), z)).
(c) ∀x∀y∃zP (f (x, z), y).
4. For each formula below, where P, Q ∈ P, find a structure M = [U, I] for
L = L{¬,∩,∪,⇒,¬} (P, F, C) that is its counter model. Justify its correct-
ness.

(a) ( (∀xP (x) ⇒ ∀yQ(y)) ⇒ (∀x(P (x) ⇒ Q(x)))).

(b) ( (∀xP (x) ∪ ∀yQ(y)) ⇒ (∀x(P (x) ∪ ∀xQ(x)))).

5. Show that the following formulas are predicate tautologies for any formulas
A, B in L.
(a) (∀x∀yA(x, y) ⇒ ∀y∀xA(x, y)).
(b) (∃x∃yA(x, y) ⇒ ∃y∃xA(x, y)).
(c) (∀x(A(x) ⇒ B(x)) ⇒ (∀xA(x) ⇒ ∀xB(x))).
6. Prove that the following formulas are not predicate tautologies by finding
their proper instances and constructing counter models for them.

50
 (a) ((∃xA(x) ∩ ∃xB(x)) ⇒ ∃x (A(x) ∩ B(x))).
(b) (∀x (A(x) ∪ B(x)) ⇒ (∀xA(x) ∪ ∀xB(x))).
(c) ((∀xA(x) ⇒ ∀xB(x)) ⇒ ∀x(A(x) ⇒ B(x))).

7. Prove that the following formulas are predicate tautologies for any for-
mulas A(x), B(x), A, B of L, such that A , B does not contain any free
occurrences of x.

(a) (∀x(A ⇒ B(x)) ⇒ (A ⇒ ∀x B(x))),

(b) (∃x(A(x) ⇒ B) ⇒ (∀xA(x) ⇒ B)).
(c) (∀x(A(x) ⇒ B) ⇒ (∃xA(x) ⇒ B)).
8. Prove that the restrictions: ”A , B does not contain any free occurrences
of x” are essential for all of the following tautologies, i.e. give examples of
formulas for which the laws without these restrictions fail and construct
counter models for them.

(a) (∀x(A(x) ⇒ B) ⇒ (∃xA(x) ⇒ B)).

(b) (∃x(A(x) ⇒ B) ⇒ (∀xA(x) ⇒ B)).

9. Prove that the converse implication to the formulas listed below are pred-
icate tautologies for any formulas A(x), B(x), A, B of L, such that A, B
does not contain any free occurrences of x.

(a) (∀x(A ⇒ B(x)) ⇒ (A ⇒ ∀x B(x))),

(b) (∃x(A(x) ⇒ B) ⇒ (∀xA(x) ⇒ B)).
(c) (∀x(A(x) ⇒ B) ⇒ (∃xA(x) ⇒ B)).

51