Predicate Calculus

Formal Methods
Lecture 6
Farn Wang
Dept. of Electrical Engineering
National Taiwan University

Predicate Logic

„ Invented by Gottlob
Frege (1848–1925).
„ Predicate Logic is also
called “first-order logic”.

“Every good mathematician is at least half a philosopher,

and every good philosopher is at least half a
mathematician.”
2

1
Motivation

There are some kinds of human reasoning that

we can’t do in propositional logic.
„ For example:
Every person likes ice cream.
Billy is a person.
Therefore, Billy likes ice cream.
„ In propositional logic, the best we can do is
A∧BÆC, which isn’t a tautology.
‰ We’ve lost the internal structure.

Motivation

„ We need to be able to refer to objects.

„ We want to symbolize both a claim and the object
about which the claim is made.
„ We also need to refer to relations between objects,
‰ as in “Waterloo is west of Toronto”.
„ If we can refer to objects, we also want to be able to
capture the meaning of every and some of.
„ The predicates and quantifiers of predicate logic
allow us to capture these concepts.

2
Apt-pet

„ An apartment pet is a pet ∀x small ( x) ∧ pet ( x) ⊃ aptPet ( x)

that is small ∀x dog ( x) ⊃ pet ( x)
„ Dog is a pet
∀x cat ( x) ⊃ pet ( x)
„ Cat is a pet
∀x elephant ( x) ⊃ pet ( x)
„ Elephant is a pet
∀x dog ( x) ⊃ small ( x)
„ Dogs and cats are small.
∀x cat ( x) ⊃ small ( x)
„ Some dogs are cute
∃x dog ( x) ∧ cute( x)
„ Each dog hates some cat
„ Fido is a dog ∀x dog ( x) ⊃ ∃y cat ( y ) ∧ hates ( x, y )
dog ( fido)

Quantifiers

„ Universal quantification (∀) corresponds to

finite or infinite conjunction of the application
of the predicate to all elements of the domain.
„ Existential quantification (∃) corresponds to
finite or infinite disjunction of the application
of the predicate to all elements of the domain.
„ Relationship between ∀ and ∃ :
‰ ∃x.P(x) is the same as ¬∀x. ¬P(x)
‰ ∀x.P(x) is the same as ¬∃x. ¬P(x)

3
Functions

„ Consider how to formalize:

Mary’s father likes music
One possible way: ∃x(f(x, Mary)∧Likes(x,Music))
which means: Mary has at least one father and he
likes music.
„ We’d like to capture the idea that Mary only has one
father.
‰ We use functions to capture the single object that can be in
relation to another object.
‰ Example: Likes(father(Mary),Music)
„ We can also have n-ary functions.

Predicate Logic

„ syntax (well-formed formulas)

„ semantics
„ proof theory
‰ axiom systems
‰ natural deduction
‰ sequent calculus
‰ resolution principle

4
Predicate Logic: Syntax

The syntax of predicate logic consists of:

„ constants

„ variables x, y, …

„ functions f(), g(), …

„ predicates P(), Q(), …

„ logical connectives ∧, ∨, ¬, Æ, ↔

„ quantifiers ∀, ∃

„ punctuations: , . ( )

Predicate Logic: Syntax

Definition. Terms are defined inductively as

follows:
„ Base cases
‰ Every constant is a term.
‰ Every variable is a term.
„ inductive cases
‰ If t1,t2,t3,…,tn are terms then f(t1,t2,t3,…,tn) is a term,
where f is an n-ary function.
„ Nothing else is a term.
10

5
 Predicate Logic
- syntax
Definition. Well-formed formulas (wffs) are defined
inductively as follows:
„ Base cases:
‰ P(t1,t2,t3,…,tn) is a wff, where ti is a term, and P is an n-ary
predicate. These are called atomic formulas.
„inductive cases:
‰ If A and B are wffs, then so are
¬A, A∧B, A∨B, A⇒B, A⇔B
‰ If A is a wff, so is ∃x. A
‰ If A is a wff, so is ∀x. A
„Nothing else is a wff.
We often omit the brackets using the same
precedence rules as propositional logic for the logical
connectives.
11

Scope and Binding of Variables (I)

„ Variables occur both in nodes next to quantifiers

and as leaf nodes in the parse tree.
„ A variable x is bound if starting at the leaf of x,
we walk up the tree and run into a node with a
quantifier and x.
„ A variable x is free if starting at the leaf of x, we
walk up the tree and don’t run into a node with a
quantifier and x.

∀x.(∀x.( P( x) ∧ Q( x))) ⇒ (¬P( x) ∨ Q( y ))

12

6
 Scope and Binding of Variables (I)

The scope of a variable x is the subtree starting at

the node with the variable and its quantifier
(where it is bound) minus any subtrees with ∀x or ∃x
at their root.

Example:
A wff is closed if it contains no free occurrences of
any variable.
∀x.(∀x.( P( x) ∧ Q( x))) ⇒ (¬P( x) ∨ Q( y ))
scope
scope of this xof this x 13

Scope and Binding of Variables

∀x((P(x) ⇒ Q(x))∧S(x,y)) scope of ∀x

Parsing tree:
∀x

This is an open formula! interpreted with ∧

interpreted with
⇒ S
interpreted with

x y

P Q

x x free variable

bound variables 14

7
Scope and Binding of Variables

∀x((∃x(P(x) ⇒ Q(x)))∧S(x,y)) scope of ∀x

Parsing tree:
∀x

This is an open formula! interpreted with ∧

interpreted with ∃x S

interpreted with ⇒ x y

P Q

x x free variable

scope of ∃x
bound variables 15

Substitution

Variables are place holders.

„ Given a variable x, a term t and a formula P,
we define P[t / x] to be the formula obtained
by replacing each free occurrence of variable
x in P with t.
„ We have to watch out for variable captures in
substitution.

16

8
Substitution

In order not to mess up with the meaning of the

original formula, we have the following restrictions
on substitution.
„ Given a term t, a variable x and a formula P,

“t is not free for x in P”

if
‰ x in a scope of ∀y or ∃y in A; and
‰ t contains a free variable y.
„ Substitution P[t / x] is allows only if t is free for x in P.

17

Substitution [f(y)/x] not allowed since

meaning of formulas
messed up.
Example:
∀y(mom(x)∧dad(f(y))) ≡ ∀z(mom(x)∧dad(f(z)))
But
(∀y(mom(x)∧dad(y)))[f(y)/x] = ∀y(mom(f(y))∧dad(f(y)))

equivalent

(∀z(mom(x)∧dad(z)))[f(y)/x] = ∀z(mom(f(y))∧dad(f(z)))

18

9
Predicate Logic: Semantics

„ Recall that a semantics is a mapping

between two worlds.
„ A model for predicate logic consists of:
‰ a non-empty domain of objects: DI
‰ a mapping, called an interpretation that associates
the terms of the syntax with objects in a domain
„ It’s important that DI be non-empty,
otherwise some tautologies wouldn’t hold
such as (∀x. A( x)) ⇒ (∃x. A( x))

19

Interpretations (Models)

„ a fixed element c’ ∈ DI to each constant c of

the syntax
„ an n-ary function f’:DInÆ DI to each n-ary
function, f, of the syntax
„ an n-ary relation R’⊆ DIn to each n-ary
predicate, R, of the syntax

20

10
Example of a Model

„ Let’s say our syntax has a constant c, a function f

(unary), and two predicates P, and Q (both binary).

Example: P(c,f(c))
In our model, choose the domain to be the natural
numbers
„ I(c) is 0.

„ I(f) is suc, the successor function.

„ I(P) is `<‘

„ I(Q) is `=‘

21

Example of an Model

What’s the meaning of P(c,f(c)) in this model?

I ( P (c, f (c))) = I (c) < I ( f (c))
= 0 < suc( I (c))
= 0 < suc(0)
= 0 <1
Which is true.

22

11
Valuations

Definition.
A valuation v, in an interpretation I, is a function
from the terms to the domain DI such that:
„ ν(c) = I(c)

„ ν(f(t1,…,tn)) = f’(ν(t1),…, ν (tn))

„ ν(x)∈DI, i.e., each variable is mapped onto

some element in DI

23

Example of a Valuation
„ DI is the set of Natural Numbers
„ g is the function +
„ h is the function suc
„ c (constant) is 3
„ y (variable) is 1
v( g (h(c), y )) = v(h(c)) + v( y )
= suc(v(c)) + 1
= suc(3) + 1
=5
24

12
Workout
„ DI is the set of Natural Numbers
„ g is the function +
„ h is the function suc
„ c (constant) is 3
„ y (variable) is 1

ν(h(h(g(h(y),g(h(y),h(c)))))) = ?

25

On(A,B) False
Clear(B) True
On(C,Fl) True
On(C,Fl)∧ ¬On(A,B) True

26

13
Workout
Interpret the following formulas with respect to
the world (model) in the previous page.
On(A,Fl)⇒ Clear(B)
Clear(B)∧ Clear(C)⇒ On(A,Fl) B
Clear(B) ∨ Clear(A)
A
Clear(B)
Clear(C) C

27

Konwoledge
Does the following knowledge base (set of
formulae) have a model ?
On(A,Fl)⇒ Clear(B)
Clear(B)∧ Clear(C)⇒ On(A,Fl)
Clear(B) ∨ Clear(A)
Clear(B)
Clear(C)

28

14
An example

(∀x)[On(x,C)⇒ ¬Clear(C)]

29

Closed Formulas

„ Recall: A wff is closed if it contains no free

occurrences of any variable.
„ We will mostly restrict ourselves to closed
formulas.
„ For formulas with free variables, close the
formula by universally quantifying over all its
free variables.

30

15
Validity (Tautologies)

„ Definition. A predicate logic formula is satisfiable if

there is an interpretation and there is a valuation
that satisfies the formula (i.e., in which the formula
returns T).
„ Definition. A predicate logic formula is logically
valid (tautology) if it is true in every interpretation.
‰ It must be satisfied by every valuation in every
interpretation.
„ Definition. A wff, A, of predicate logic is a
contradiction if it is false in every interpretation.
‰ It must be false in every valuation in every interpretation.

31

Satisfiability, Tautologies, Contradictions

„ A closed predicate logic formula, is satisfiable

if there is an interpretation I in which the
formula returns true.
„ A closed predicate logic formula, A, is a
tautology if it is true in every interpretation.
BA
„ A closed predicate logic formula is a
contradiction if it is false in every
interpretation.
32

16
Tautologies

„ How can we check if a formula is a tautology?

„ If the domain is finite, then we can try all the
possible interpretations (all the possible functions
and predicates).
„ But if the domain is infinite? Intuitively, this is why a
computer cannot be programmed to determine if an
arbitrary formula in predicate logic is a tautology (for
all tautologies).
„ Our only alternative is proof procedures!
„ Therefore the soundness and completeness of our
proof procedures is very important!

33

Semantic Entailment

Semantic entailment has the same meaning as

it did for propositional logic.
φ1 , φ2 , φ3 B ψ
means that if v(φ1 ) = T and v(φ2 ) = T and v(φ3 ) = T
then v(ψ ) = T , which is equivalent to saying
(φ1 ∧ φ2 ∧ φ3 ) ⇒ ψ
is a tautology, i.e.,
(φ1 , φ2 , φ3 B ψ ) ≡ ((φ1 ∧ φ2 ∧ φ3 ) ⇒ ψ )

34

17
An Axiomatic System for Predicate Logic

FO_AL: An extension of the axiomatic system for

propositional logic. Use only: ⇒ .¬.∀
A ⇒ ( B ⇒ A)
( A ⇒ ( B ⇒ C )) ⇒ (( A ⇒ B ) ⇒ ( A ⇒ C ))
(¬A ⇒ ¬B ) ⇒ ( B ⇒ A)
∀x. A( x) ⇒ A(t ), where t is free for x in A
∀x.( A ⇒ B ) ⇒ ( A ⇒ (∀x.B)), where A contains
no free occurrences of x

35

FO_AL Rules of Inference

Two rules of inference:

„ (modus ponens - MP) From A and A ⇒ B , B
can be derived, where A and B are any well-
formed formulas.
„ (generalization) From A, ∀x. A can be derived,
where A is any well-formed formula and x is
any variable.

36

18
Soundness and Completeness of FO_AL

„ FO_AL is sound and complete.

„ Completeness was proven by Kurt Gödel in
1929 in his doctoral dissertation.
„ Predicate logic is not decidable

37

Deduction Theorem

„ Theorem. If H ∪ { A} Aph B by a deduction

containing no application of generalization to
a variable that occurs free in A, then H Aph A ⇒ B
„ Corollary. If A is closed and if H ∪ { A} A B then
ph
H A ( A ⇒ B)
ph

38

19
Proof by Refutation

„ A closed formula is a tautology (valid) iff its

negation is a contradiction.
i o s
nis
„ In other words, A closed e at
formula
t valid iff its
r pr
n y inte ?
negation is not satisfiable.
„ To prove
w m a
{P1,…, Pn}
t he⊨reS is equivalent to
Ho{P1,…, aPn,
prove re ¬S} ⊨ false
„ To prove {P1,…, Pn} ⊨ S becomes to check if
there is an interpretation for {P1,…, Pn, ¬S} .

39

Counterexamples

„ How can we show a formula is not a

tautology?
„ Provide a counterexample. A counterexample
for a closed formula is an interpretation in
which the formula does not have the truth
value T.

40

20
Example

Prove ∀x.∀y. A A ∀y.∀x. A

ph

41

Workout: Counterexamples

Show that (∀x.P( x) ∨ Q( x)) ⇔ ((∀x.P ( x)) ∨ (∀x.Q( x)))

is not a tautology by constructing a model
that makes the formula false.

42

21
What does ‘first-order’ mean?

„ We can only quantify over variables.

„ In higher-order logics, we can quantify over
functions, and predicates.
‰ For example, in second-order logic, we can
express the induction principle:
∀P.( P(0) ∧ (∀n.P(n) ⇒ P(n + 1))) ⇒ (∀n.P(n))
„ Propositional logic can also be thought of as
zero-order.

43

A rough timeline in ATP … (1/3)

450B.C. Stoics propositional logic (PL),
inference (maybe)
322B.C. Aristotle ``syllogisms“ (inference rules),
quantifiers
1565 Cardano probability theory (PL + undertainty)
1646 Leibniz research for a general decision procedure
-1716 to check the validty of formulas
1847 Boole PL (again)
1879 Frege first-order logic (FOL)
1889 Peano 9 axioms for natural numbers

44

22
 • To formalize all existing

A rough timeline in ATP … (2/3) theories to a finite,

complete, and
consistent set of axioms.
• decision procedures for
Resolve the all mathematical
1920‘s Hilbert
2nd Hilbert’s Hilbert‘s program theories
1922 Wittgenstein proof by truth tables • 23 open problems.
problem (in
1929 of
the theory Gödel completeness theorem of FOL
N) 1930 Herbrand a proof procedure for FOL based on
propositionalization
1931 Gödel incompleteness theorems for the consistency
of Peano axioms
1936 Gentzen a proof for the consisitency of Peano axioms
in set theory
1936 Church, undecidability Who is to
Turing of FOL prove the
Is type
consistency
1958 Gödel a method to prove the consistency
theoryof Peano
of set theory ?
axioms with type theory consistent ?
45

A rough timeline in ATP … (3/3)

1954 Davis First machine-generated proof
1955 Beth, Semantic Tableaus
Hintikka
1957 Newell, First machine-generated proof in
Simon Logic Calculus
1957 Kangar, Lazy substitution by free (dummy) Vars
Prawitz
1958 Prawitz First prover for FOL
1959 Gilmore More provers
Wang
1960 Davis Davis-Putnam Procedure
Putnam,
Longman
1963 Robinson Unification, resolution

46

23
 Kurt Gödel
American is in danger
I knew the because
of dictatorship
1906-1978I can
general
prove the
• Einstein, “his work no longer meant
much,anthat
• Born he relativity
contradiction
came to the in
Austro-Hugarian Institute
merely
• 12 Æ Czech… to was
have wrong.
American constitution.
the privilege of
walking
• refuse home withCzech
to learn Gödel.”
• thought
OnÆ
•• 23 someone
hisAustrian
citizen exam, … was to poison
him.
• established
• proved the completeness
a paradoxial solution to theof
The
g re a • general
ate1st-order
onlyrelativity
hislogic
wife’s cooking.
in his Ph.D. thesis
in th tes
e 20 t logicia ••• 25, established
1977, wifethe
his position,
wasincompleteness
ill and could
th ce n Permanent Princeton, 1946
ntury of N
notAlbert
t • 1st cook.Einstein Award, 1951
e g r eates • 32 Æ German
f th • Full
Jan.professor,
1978, died 1953of mal-nutrition.
One o evements . •• 34 Æ joined Princeton
achi h century •• 42 Æ American
National Science Medal, 1974
th e 20t
in • Emeritus professor, 1976
47

2007/04/03 stopped here.

48

24
Predicate Logic: Natural Deduction
Extend the set of rules we used for
propositional logic with ones to handle
quantifiers.

49

Predicate Logic: Natural Deduction

50

25
Example

Show ∀x.P( x) ⇒ Q( x), ∀x.P( x) ND

A ∀x.Q( x)

51

Workout

„ Show P(a), ∀x.P( x) ⇒ ¬Q( x) ND

A ¬Q ( a )
„ Show ¬∀x.P( x) ND
A ∃x.¬P( x)

52

26
 Proof by Refutation

„ To prove {P1,…, Pn} ⊨ S is equivalent to

prove that there is no interpretation for
{P1,…, Pn, ¬S} .
„ But there are infinitely many interpretations!
„ Can we limit the range of interpretations ?
„ Yes, Herbrand interpretations!

53

Herbrand’s theorem
- Herbrand universe of a formula S
„ Let H0 be the set of constants appearing in S.
‰ If no constant appears in S, then H0 is to consist of a
single constant, H0={a}.
„ For i=0,1,2,…
Hi+1=Hi ∪ {f n(t1,…,tn)| f is an n-place function in S; t1,…,tn ∈ Hi }
„ Hi is called the i-level constant set of S.
„ H∞ is the Herbrand universe of S.

54

27
Herbrand’s theorem
- Herbrand universe of a formula S
Example 1: S={P(a),∼P(x)∨P(f(x))}
„ H0={a}

„ H1={a,f(a)}

„ H2={a,f(a),f(f(a))}

„ .

„ .

„ H∞={a,f(a),f(f(a)),f(f(f(a))),…}

55

Herbrand’s theorem
- Herbrand universe of a formula S
Example 2: S={P(x)∨Q(x),R(z),T(y)∨∼W(y)}
„ There is no constant in S, so we let H0={a}
„ There is no function symbol in S, hence
H=H0=H1=…={a}
Example 3: S={P(f(x),a,g(y),b)}
„ H0={a,b}
„ H1={a,b,f(a),f(b),g(a),g(b)}
„ H2={a,b,f(a),f(b),g(a),g(b),f(f(a)),f(f(b)),f(g(a)),f(g
(b)),g(f(a)),g(f(b)),g(g(a)),g(g(b))}
„ …
56

28
Herbrand’s theorem
- Herbrand universe of a formula S
Expression
‰ a term, a set of terms, an atom, a set of atoms, a
literal, a clause, or a set of clauses.
Ground expressions
‰ expressions without variables.
It is possible to use a ground term, a ground
atom, a ground literal, and a ground clause –
this means that no variable occurs in respective
expressions.
Subexpression of an expression E
‰ an expression that occurs in E.
57

Herbrand’s theorem
- Herbrand base of a formula S
„ Ground atoms Pn(t1,…,tn)
‰ Pn is an n-place predicate occurring in S,
‰ t1,…,tn ∈ H∞
„ Herbrand base of S (atom set)
‰ the set of all ground atoms of S
„ Ground instance of S
‰ obtained by replacing variables in S by members of
the Herbrand universe of S.

58

29
Herbrand’s theorem
- Herbrand universe & base of a formula S
Example
„ S={P(x),Q(f(y))∨R(y)}

„ C=P(x) is a clause in S

„ H={a,f(a),f(f(a)),…} is the Herbrand universe of

S.
„ P(a), Q(f(a)), Q(a), R(a), R(f(f(a))), and
P(f(f(a))) are ground atoms of C.

59

Workout

{P(x), Q(g(x,y),a)∨R(f(x))}
„ please construct the set of ground terms

„ please construct the set of ground atoms

60

30
Herbrand’s theorem
- Herbrand interpretation of a formula S
„ S, a set of clauses.
‰ i.e., a conjunction of the clauses
„ H, the Herbrand universe of S and
„ H-interpretation I of S
‰ I maps all constants in S to themselves.
‰ Forall n-place function symbol f and h1,…,hn
elements of H,
I (f (h1,…,hn) ) = f(h1,…,hn)

61

Herbrand’s theorem
- Herbrand interpretation of a formula S
„ There is no restriction on the assignment to
each n-place predicate symbol in S.
„ Let A={A1,A2,…,An,…} be the atom set of S.
„ An H-interpretation I can be conveniently
represented as a subset of A.
‰ If Aj ∈ I, then Aj is assigned “true”,
‰ otherwise Aj is assigned “false”.

62

31
Herbrand’s theorem
- Herbrand interpretation of a formula S
Example: S={P(x)∨Q(x),R(f(y))}
„ The Herbrand universe of S is
H={a,f(a),f(f(a)),…}.
„ Predicate symbols: P, Q, R.
„ The atom set of S:
A={P(a),Q(a),R(a),P(f(a)),Q(f(a)),R(f(a)),…}.
„ Some H-interpretations for S:
‰ I1={P(a),Q(a),R(a),P(f(a)),Q(f(a)),R(f(a)),…}
‰ I2= ∅
‰ I3={P(a),Q(a),P(f(a)),Q(f(a)),…}
63

Herbrand’s theorem
- Herbrand interpretation of a formula S
„ An interpretation of a set S of clauses does
not necessarily have to be defined over the
Herbrand universe of S.
„ Thus an interpretation may not be an
H-interpretation.

Example:
„ S={P(x),Q(y,f(y,a))}

„ D={1,2}
64

32
Herbrand’s theorem
- Herbrand interpretation of a formula S
„ But Herbrand is conceptually general enough.
Example (cont.) S={P(x),Q(y,f(y,a))}
„ D={1,2}

„ – an interpretation of S:

a f(1,1) f(1,2) f(2,1) f(2,2)

2 1 2 2 1

P(1) P(2) Q(1,1) Q(1,2) Q(2,1) Q(2,2)

T F F T F T
65

Herbrand’s theorem
- Herbrand interpretation of a formula S
„ But Herbrand is conceptually general enough.

Example (cont.) – we can define an H-interpretation I*

corresponding to I.
„ First we find the atom set of S
‰ A={P(a),Q(a,a),P(f(a,a)),Q(a,f(a,a)),Q(f(a,a),a),Q(f(a,a),f(a,a)),…}
„ Next we evaluate each member of A by using the given table
‰ P(a)=P(2)=F a f(1,1) f(1,2) f(2,1) f(2,2)
‰ Q(a,a)=Q(2,2)=T 2 1 2 2 1
‰ P(f(a,a))=P(f(2,2))=P(1)=T
P(1) P(2) Q(1,1) Q(1,2) Q(2,1) Q(2,2)
‰ Q(a,f(a,a))=Q(2,f(2,2))=Q(2,1)=F
‰ Q(f(a,a),a)=Q(f(2,2),2)=Q(1,2)=T T F F T F T
‰ Q(f(a,a),f(a,a))=Q(f(2,2),f(2,2))=Q(1,1)=T
„ Then I*={Q(a,a),P(f(a,a)),Q(f(a,a),a),…}.

66

33
Herbrand’s theorem
- Herbrand interpretation of a formula S
„ If there is no constant in S, the element a used to
initiate the Herbrand universe of S can be mapped
into any element of the domain D.
„ If there is more than one element in D, then there is
more than one H-interpretation corresponding to I.

67

Herbrand’s theorem
- Herbrand interpretation of a formula S
„ Example: S={P(x),Q(y,f(y,z))}, D={1,2}

f(1,1) f(1,2) f(2,1) f(2,2)

1 2 2 1

P(1) P(2) Q(1,1) Q(1,2) Q(2,1) Q(2,2)

T F F T F T

„ Two H-interpretations corresponding to I are:

‰ I*={Q(a,a),P(f(a,a)),Q(f(a,a),a),…} if a=2,
‰ I*={P(a),P(f(a,a)),…} if a=1.
68

34
Herbrand’s theorem
- Herbrand interpretation of a formula S
Definition: Given an interpretation I over a domain D,
an H-interpretation I* corresponding to I is an H-
interpretation that satisfies the condition:
„ Let h1,…,hn be elements of H (the Herbrand
universe of S).
„ Let every hi be mapped to some di in D.

„ If P(d1,…,dn) is assigned T (F) by I, then P(h1,…,hn)

is also assigned T(F) in I*.
Lemma: If an interpretation I over some domain D
satisfies a set S of clauses, then any H-
interpretation I* corresponding to I also satisfies S.

69

Herbrand’s theorem

A set S of clauses is unsatisfiable if and only if S is

false under all the H-interpretations of S.

„ We need consider only H-interpretations for

checking whether or not a set of clauses is
unsatisfiable.
„ Thus, whenever the term “interpretation” is used, a
H-interpretation is meant.

70

35
Herbrand’s theorem

Let ∅ denote an empty set. Then:

„ A ground instance C’ of a clause C is satisfied by an
interpretation I if and only if there is a ground literal L’ in C’
such that L’ is also in I, i.e. C’∩I≠∅.
„ A clause C is satisfied by an interpretation I if and only if
every ground instance of C is satisfied by I.
„ A clause C is falsified by an interpretation I if and only if there
is at least one ground instance C’ of C such that C’ is not
satisfied by I.
„ A set S of clauses is unsatisfiable if and only if for every
interpretation I there is at least one ground instance C’ of
some clause C in S such that C’ is not satisfied by I.

71

Herbrand’s theorem
Example: Consider the clause C=∼P(x)∨Q(f(x)). Let I1, I2,
and I3 be defined as follows:
„ I 1= ∅
„ I2={P(a),Q(a),P(f(a)),Q(f(a)),P(f(f(a))),Q(f(f(a))),…}
„ I3={P(a),P(f(a)),P(f(f(a))),…}
C is satisfied by I1 and I2, but falsified by I3.

Example: S={P(x),∼P(a)}.
The only two H-interpretations are:
„ I1={P(a)},

„ I2= ∅.
S is falsified by both H-interpretations and therefore is
unsatisfiable.
72

36
Resolution Principle
- Clausal Forms
Clauses are universally quantified disjunctions
of literals;
all variables in a clause are universally
quantified
(∀x1 ,..., xn )(l1 ∨ ... ∨ ln )
written as
l1 ∨ ... ∨ ln
or
{l1 ,..., ln } 73

Resolution Principle
- Clausal forms
{Nat(s(A)),¬Nat(A)} {Nat(s(A)),¬Nat(A)}
{Nat(A)} {Nat(x)}
„ Examples: gives gives
{Nat(s(A))} {Nat(s(A))}
{Nat(s(s(x))),¬Nat(s(x))}
{Nat(s(A))}
gives
{Nat(s(s(A)))}

„ We need to be able to work with variables !

„ Unification of two expressions/literals

74

37
Resolution Principle
- Terms and instances
„ Consider following atoms
P(x,f(y),B)

P(z,f(w),B) alphabetic variant

P(x,f(A),B) instance
P(g(z),f(A),B) instance
P(C,f(A),A) not an instance
„ Ground expressions do not contain any variables

75

Resolution Principle
- Substitution
A substitution s = {t1 / v1 ,..., tn / vn } substitutes
variables vi for terms ti ( ti does NOT contain vi )

Applying a substitution s to an expression ω

yields the expression ω s which is ω
with all occurrences of vi replaced by ti
P(x,f(y),B)

P(z,f(w),B) s ={z/x,w/y}
P(x,f(A),B) s ={A/y}
P(g(z),f(A),B) s ={g(z)/x,A/y}
P(C,f(A),A) no substitution !
76

38
Workout

Calculate the substitutions for the resolution of

the two clauses and the result clauses after
the substitutions.
„ ¬P(x), P(f(a))∨Q(f(y),g(a,b))

„ ¬P(g(x,a)), P(y)∨Q(f(y),g(a,b))

„ ¬P(g(x,f(a))), P(g(b,y))∨Q(f(y),g(a,b))

„ ¬P(g(f(x),x)), P(g(y,f(f(y)))∨Q(f(y),g(a,b)))

77

Resolution Principle
- Composing substitutions
„ Composing substitutions s1 and s2 gives s1 s2
which is that substitution obtained by first
applying s2 to the terms in s1and adding
remaining term/vars pairs to s1
θ ={g(x,y)/z}{A/x,B/y,C/w,D/z}=
{g(A,B)/z,A/x,B/y,C/w}

„ Apply to P(x,y,z)θ
gives
P(A,B,g(A,B))
78

39
ωs

Resolution Principle
- Properties of substitutions
(ω s1 ) s2 = ω ( s1s2 )
( s1s2 ) s3 = s1 ( s2 s3 ) associativity
s1s2 ≠ s2 s1 not commutative

79

Resolution Principle
- Unification
„ Unifying a set of expressions {wi}
‰ Find substitution s such that wi s = w j s for all i, j
‰ Example
{P(x,f(y),B),P(x,f(B),B)}
s={B/y,A/x} not the simplest unifier
s={B/y} most general unifier (mgu)
„ The most general unifier, the mgu, g of {wi} has the
property that if s is any unifier of {wi} then there
exists a substitution s’ such that {wi}s={wi}gs’
„ The common instance produced is unique up to
alphabetic variants (variable renaming)
„ usually we assume there is no common variables 80
in the two atoms

40
Workout

P(B,f(x),g(A)) and P(y,z,f(w))

„ construct an mgu

„ construct a unifier that is not the most general.

81

Workout

Determine if each of the following sets is

unifiable. If yes, construct an mgu.
„ {Q(a), Q(b)}

„ {Q(a,x),Q(a,a)}

„ {Q(a,x,f(x)),Q(a,y,y)}

„ {Q(x,y,z),Q(u,h(v,v),u)}

„ {P(x1,g(x1),x2,h(x1,x2),x3,k(x1,x2,x3)),

P(y1,y2,e(y2),y3,f(y2,y3),y4)}

82

41
Resolution Principle
- Disagreement set in unification
The disagreement set of a set of expressions
{wi} is the set of subterms { ti } of {wi} at the
first position in {wi} for which the {wi} disagree

{P(x,A,f(y)),P(w,B,z)} gives {x,w}

{P(x,A,f(y)),P(x,B,z)} gives {A,B}
{P(x,y,f(y)),P(x,B,z)} gives {y,B}

83

Resolution Principle
- Unification algorithm
U nify( T erm s )
Initialize k ← 0;
Initialize T k = T erm s ;
Initialize σ k = {};
* If T k is a singleton, then output σ k . O therw ise, continue.
Let D k be the disagreem ent set of T k
If there exists a var v k and a term t k in D k such that v k
does not occur in t k , continue. O therw ise, exit w ith failure.
σ k + 1 ← σ k {t k / v k };
T k + 1 ← T k {t k / v k };
k ← k + 1;
G oto * 84

42
Predicate calculus Resolution

John Allan Robinson (1965)

Let C1 and C2 be two clauses with
literals l1 ∈ C1 and ¬l2 ∈ C2 such that
C1 and C2 do not contain common variables,
and mgu (l1 , l2 ) = θ
then C = [{ C1 − {l1}} ∪ {C2 − {¬l2 }}]θ
is a resolvent of C1 and C2
85

Predicate calculus Resolution

John Allan Robinson (1965)

no common
Given variables!
C: l1∨l2∨…∨lm
C: ¬k1∨k2∨…∨kn
θ=mgu(l1,k1)
the resolvent is
l2θ ∨…∨ lmθ ∨ k2θ ∨…∨ knθ
86

43
 Resolution Principle
- Example P(x)∨ Q(f(x)) and R(g(x))∨ ¬Q(f(A))
Standardizing the variables apart
Why P(x)∨ Q(f(x)) and R(g(y))∨ ¬Q(f(A))
can we Substitution θ ={A/x}
do this ? Resolvent P(A)∨ R(g(y))

Why we P(x)∨ Q(x,y) and ¬P(A)∨ ¬R(B,z)

think the Standardizing the variables apart
variables in
2 clauses
Substitution θ ={A/x}
are
irrelevant ? Resolvent Q(A,y)∨ ¬R(B,z)
87

Workout

Find all the possible resolvents (if any) of the

following pairs of clauses.
„ ¬P(x)∨Q(x,b),

88

44
Workout

Find all the possible resolvents (if any) of the

following pairs of clauses.
„ ¬P(x)∨Q(x,b), P(a)∨Q(a,b)

„ ¬P(x)∨Q(x,x), ¬Q(a,f(a))

„ ¬P(x,y,u)∨¬P(y,z,v)∨¬P(x,v,w)∨P(u,z,w),

P(g(x,y),x,y)
„ ¬P(v,z,v)∨P(w,z,w), P(w,h(x,x),w)

89

Resolution Principle
- A stronger version of resolution
Use more than one literal per clause

{P(u),P(v)} and {¬P(x),¬P(y)}

do not resolve to empty clause.
However, ground instances
{P(A)} and {¬P(A)} resolve to empty clause

90

45
 Resolution Principle
- Factors
Let C1 be a clause such that there exists
a substitution θ that is a mgu of a set of literals
in C1. Then C1θ is a factor of C1

Each clause is a factor of itself.

Also, {P(f(y)),R(f(y),y)}is a factor of {P(x),P(f(y)),R(x,y)}
with θ = { f ( y ) / x}

91

Resolution Principle
- Example of refutation

92

46
Resolution Principle
- Example
Hypothesies Clausal Form
∀x (dog(x) ⇒ animal(x)) ¬dog(x) ∨ animal(x)
dog(fido) dog(fido)
∀y (animal(y) ⇒ die(y)) ¬animal(y) ∨ die(y)

Conclusion Negate the goal

die(fido) ¬die(fido)

93

Resolution Principle
- Example ¬dog(x) ∨ animal(x) ¬animal(y) ∨ die(y)

{x {x → y}

dog(fido) ¬dog(y) ∨ die(y)

{y → fido}

die(fido) ¬die(fido)

94

47
Workout (resolution)
- Proof with resolution principle
Hypotheses:
„ P(m(x),x) ∨ Q(m(x))
„ ¬P(y,z) ∨ R(y)
„ ¬Q(m(f(x,y))) ∨ ¬T(x,g(y))
„ S(a) ∨ T(f(a),g(x))
„ ¬R(m(y))
„ ¬S(x) ∨ W(x,f(x,y))
Conclusion
W(a, y)
95

Resolution

Properties
„ Resolution is sound

„ Incomplete

Given P(A)
Infer {P(A),P(B)}

„ But fortunately it is refutation complete

‰ If KB is unsatisfiable then KB |-

96

48
Resolution Principle
- Refutation Completeness
To decide whether a formula KB ⊨ w, do
„ Convert KB to clausal form KB’
„ Convert ¬w to clausal form ¬w’
„ Combine ¬w’ and KB’ to give Δ
„ Iteratively apply resolution to Δ and add the
results back to Δ until either no more
resolvents can be added, or until the empty
clause is produced.

97

Resolution Principle
- Converting to clausal form (1/2)
To convert a formula KB into clausal form
1. Eliminate implication signs*
( p ⇒ q ) becomes (¬p ∨ q)
2. Reduce scope of negation signs*
¬( p ∧ q) becomes (¬p ∨ ¬q )
3. Standardize variables
(∀x)[¬P(x)∨(∃x)Q(x)] becomes (∀x)[¬P(x)∨(∃y)Q(y)]
4. Eliminate existential quantifiers using Skolemization
* Same as in prop. logic

98

49
 Resolution Principle
- Converting to clausal form (2/2)
5. Convert to prenex form
‰ Move all universal quantifiers to the front
6. Put the matrix in conjunctive normal form*
‰ Use distribution rule
7. Eliminate universal quantifiers
8. Eliminate conjunction symbol *
9. Rename variables so that no variable occurs in
more than one clause.

99

Resolution Principle
- Skolemization
Consider(∀x)[(∃y)Height(x,y)]
The y depends on the x
Define this dependence explicitly using a skolem function h(x)
Formula becomes (∀x)[Height(x,h(x))]
General rule is that each occurrence of an existentially
quantified variable is replaced by a skolem function whose
arguments are those universally quantified variables
whose scopes includes the scope of the existentially
quantified one

Skolem functions do not yet occur elsewhere !

Resulting formula is not logically equivalent ! 100

50
 Resolution Principle
- Examples of Skolemization
[(∀w)Q(w)]⇒(∀x){(∀y){(∃z)[P(x,y,z)⇒(∀u)R(x,y,u,z)]}}
gives
[(∀w)Q(w)]⇒(∀x){(∀y)[P(x,y,g(x,y))⇒(∀u)R(x,y,u,g(x,y))]}
(∀x)[(∃y) F(x,y)] gives (∀x)F(x,h(x))
but
(∃y)[(∀x)F(x,y)] gives [(∀x)F(x,sk)] skolem constant
Not logically equivalent !
A well formed formula and its Skolem form are not logically
equivalent.
However, a set of formulae is (un)satisfiable if and only if
its skolem form is (un)satisfiable. 101

Resolution Principle
- Example of conversion to clausal form

102

51
 Workout

Convert the following formula to clausal form.

„ ∃x(P(x)∧∀y
((∃z.Q(x,y,s(z)))Æ(Q(x,s(y),x)∧R(y))))
„ ∀x∀y(S(x,y,z)Æ ∃z(S(x,z) ∧ S(z,x)))

103

Resolution Principle
- Example of refutation by resolution
1.¬P(x)∨ ¬P(y)∨ ¬I(x,27)∨ ¬I(y,28)∨ S(x,y)
all packages in room 27 are smaller than any of those in 28
2.P(A)
3.P(B)
4.I(A,27)∨ I(A,28)
5.I(B,27)
6.¬S(B,A)
Prove I(A,27)

104

52
Resolution Principle
- Search Strategies
„ Ordering strategies
‰ In what order to perform resolution ?
‰ Breadth-first, depth-first, iterative deepening ?
‰ Unit-preference strategy :
„ Prefer those resolution steps in which at least one
clause is a unit clause (containing a single literal)
„ Refinement strategies
‰ Unit resolution : allow only resolution with unit
clauses

105

Resolution Principle
- Input Resolution
„ at least one of the clauses being resolved is a
member of the original set of clauses
„ Input resolution is complete for Horn-clauses
but incomplete in general
„ E.g. {P, Q},{¬P, Q},{P, ¬Q},{¬P, ¬Q}

„ One of the parents of the empty clause

should belong to original set of clauses

106

53
Workout

Use input resolution to prove the theorem in

page workout(resolution)!

107

Resolution Principle
- Linear Resolution
„ Linear resolvent is one in which at least one
of the parents is either
‰ an initial clause or
‰ the resolvent of the previous resolution step.

„ Refutation complete
„ Many other resolution strategies exist

108

54
workout

Use linear resolution to prove the theorem in

page workout(resolution)!

109

Resolution Principle
- Set of support
„ Ancestor : c2 is a descendant of c1 iff c2 is a
resolvent of c1 (and another clause) or if c2 is a
resolvent of a descendant of c1 (and another
clause); c1 is an ancestor of c2
„ Set of support : the set of clauses coming from
the negation of the theorem (to be proven) and
their descendants
„ Set of support strategy : require that at least one
of the clauses in each resolution step belongs to
the set of support

110

55
workout

Use set of support to prove the theorem in

page workout(resolution)!

111

Resolution Principle
- Answer extraction
Suppose we wish to prove whether KB |=
(∃w)f(w)
We are probably interested in knowing the w for
which f(w) holds.
Add Ans(w) literal to each clause coming from
the negation of the theorem to be proven;
stop resolution process when there is a
clause containing only Ans literal

112

56
 1.¬P(x)∨ ¬P(y)∨ ¬I(x,27)∨ ¬I(y,28)∨ S(x,y)
Resolution all packages in room 27 are smaller than any of those in 28

Principle 2.P(A)
3.P(B)
- Example 4.I(A,27)∨ I(A,28)

of answer
5.I(B,27)
6.¬S(B,A)
extraction Prove (∃u)I(A,u), i.e. in which room is A?

113

Workout

„ Use answer extraction to prove the theorem

in page workout(resolution)!

114

57
Theory of Equality
„ Herbrand Theorem does not apply to FOL with equality.
„ So far we’ve looked at predicate logic from the point of view of
what is true in all interpretations.
‰ This is very open-ended.

„ Sometimes we want to assume at least something about our

interpretation to enrich the theory in what we can express and
prove.
„ The meaning of equality is something that is common to all
interpretations.
‰ Its interpretation is that of equivalence in the domain.

‰ If we add = as a predicate with special meaning in predicate logic,

we can also add rules to our various proof procedures.
„ Normal models are models in which the symbol = is interpreted
as designating the equality relation.

115

Theory of Equality
- An Axiomatic System with Equality
To the previous axioms and rules of inference,
we add:
EAx1 ∀x.x = x
EAx2 ∀x.∀y.x = y ⇒ ( A( x, x) ⇒ A( x, y ))
EAx3 ∀x.∀y.x = y ⇒ f ( x) = f ( y )

116

58
Theory of Equality
- Natural Deduction Rules for Equality

117

Theory of Equality
- Natural Deduction Rules for Equality

118

59
Theory of Equality
- Substitution
„ Recall: Given a variable x, a term t and a
formula P, we define P[t / x] to be the formula
obtained by replacing ALL free occurrence of
variable x in P with t.
„ But with equality, we sometimes don’t want to
substitute for all occurrences of a variable.
„ When we write P[t / x] above the line, we get to
choose what P is and therefore can choose
the occurrences of a term that we wish to
substitute for.
119

Theory of Equality
- Substitution
Recall from existential introduction:

„ Matching the top of our rule, P = Q( x0 , x), so line 3

of the proof is P[ x0 / x] , which is Q( x0 , x)
„ So we don’t have to substitute in for every
occurrence of a term.

120

60
Theory of Equality
- Examples
From these two inference rules, we can derive
two other properties that we expect equality
to have:
„ Symmetry : ND A ∀x, y.( x = y ) ⇒ ( y = x)
A ∀x, y, z.( x = y ) ∧ ( y = z ) ⇒ ( x = z )
„ Transitivity : ND

121

Theory of Equality
- Example

122

61
Theory of Equality
- Example

123

Theory of Equality
- Leibniz’s Law
„ The substitution inference rule is related to
Leibniz’s Law.
„ Leibniz’s Law:
if t1 = t2 is a theorem, then so is P[t1 / x] ⇔ P[t2 / x]
„ Leibniz’s Law is generally referred to as the
ability to substitute “equals for equals”.

124

62
Leibniz

Gottfried Wilhelm von Leibniz (1646-

1716)
„ The founder of differential and
integral calculus.
„ Another of Leibniz’s lifelong aims
was to collate all human knowledge.
“[He was] one of the last great
polymaths – not in the frivolous
sense of having a wide general
knowledge, but in the deeper
sense of one who is a citizen of
the whole world of intellectual
inquiry.” 125

Theory of Equality
- Example
„ From our natural deduction rules, we can
derive Leibniz’s Law:
t1 = t2 A P(t1 ) ⇔ P(t2 )
ND

126

63
Theory of Equality
- Equality: Semantics
„ The semantics of the equality symbol is
equality on the objects of the domain.
„ In ALL interpretations it means the same
thing.
„ Normal interpretations are interpretations in
which the symbol = is interpreted as
designating the equality relation on the
domain.
„ We will restrict ourselves to normal
interpretations from now on.
127

Theory of Equality
- Extensional Equality
„ Equality in the domain is extensional, meaning it is
equality in meaning rather than form.
„ This is in contrast to intensional equality which is
equality in form rather than meaning.
„ In logic, we are interested in whether two terms
represent the same object, not whether they are the
same symbols.
„ If two terms are intensionally equal then they are
also extensionally equal, but not necessarily the
other way around.

128

64
Theory of Equality
- Equality: Counterexamples
„ Show the following argument is not valid:
∃x.P ( x) ∧ Q ( x), P ( A), A = B B Q( B)
„ where A,B are constants

129

Theory of Arithmetic

„ Another commonly used theory is that of

arithmetic.
„ It was formalized by Dedekind in 1879 and
also by Peano in 1889.
„ It is generally referred to as Peano’s Axioms.
„ The model of the system is the natural
numbers with the constants 0 and 1, the
functions +, *, and the relation <.

130

65
Peano’s Axioms

131

Intuitionistic Logic

„ “A proof that something exists is constructive if it provides a

method for actually constructing it.”
„ In intuitionistic logic, only constructive proofs are allowed.
„ Therefore, they disallow proofs by contradiction. To show φ , you
can’t just show ¬φ is impossible.
„ They also disallow the law of the excluded middle arguing that
you have to actually show one of φ or ¬φ before you can
conclude φ ∨ ¬φ
„ Intuitionistic logic was invented by Brouwer. Theorem provers
that use intuitionistic logic are Nuprl, Coq, Elf, and Lego.
„ In this course, we will only be studying classical logic.

132

66
Summary

„ Predicate Logic (motivation, syntax and

terminology, semantics, axiom systems,
natural deduction)
„ Equality, Arithmetic
„ Mechanical theorem proving

133

67