Professional Documents
Culture Documents
is re
ursive.
R3. (Minimization) For G : !n ! ! re
ursive, su
h that for all a 2 !n there
exists some x 2 ! su
h that G(a; x) = 0, F : !n ! !, dened by
+1
F (a) = x(G(a; x) = 0)
is re
ursive. (Re
all that xP (x) for a relation P is the minimal x 2 ! su
h
that x 2 P obtains.)
Denition. R( !k ) is alled re ursive, or omputable (R is a re ursive relation) if R is a re ursive fun tion.
Proofs in this note are adaptation of those in [Sh into the dedu
tion system des
ribed in [E.
Many thanks to Peter Ahumada and Mi
hael Brewer who wrote up this note.
LECTURES BY B. KIM
+1
+1
+1
+1
COMPLETE PROOFS OF GODEL'S
INCOMPLETENESS THEOREMS
The rst is
learly satised if some x < f (a) satises P (x), while the se
ond is
satisifed if all x < f (a) satisfy P (x).
P6. For P
!n
+1
a re ursive relation, F : !n
+1
! !, dened by
_ x = a.
+1
su h that
are re
ursive.
Proof. Note that P is dened by
omposition of re
ursive fun
tions and
predi
ates, hen
e re
ursive by P1, and Q is dened by
omposition of re
ursive fun
tions, re
ursive predi
ates, and negation, hen
e re
ursive by P1
and P4.
P8. _ : ! ! ! !, dened by
a_b=
a b if a b,
0
otherwise,
is re
ursive.
Proof. Note that
is re
ursive.
Proof. Note that
F = G :R1 + + Gk :Rk :
1
LECTURES BY B. KIM
P10. If Q ; : : : ; Qk !n are re
ursive relations, and R ; : : : ; Rk !n are re
ursive relations partitioning !n , then P !n , dened by
8
>
>
<Q (a) if R (a),
..
P (a) i ...
.
>
>
:Q (a) if R (a),
k
k
is re
ursive.
Proof. Note that
8
>
>
<Q1 (a) if R (a),
..
P (a) = ...
.
>
>
: (a) if R (a),
k
Qk
is re
ursive by P9.
Denition. A relation P !n is re
ursively enumerable (r.e.) if there exists
some re
ursive relation Q !n su
h that
P (a) i 9xQ(a; x):
Remark If a relation R !n is re
ursive, then it is re
ursively enumerable, sin
e
R(a) i 9x(R(a) ^ x = x).
Negation Theorem. A relation R !n is re
ursive if and only if R and :R are
1
+1
re ursively enumerable.
Proof. If R is re ursive, then :R is re ursive. Hen e by above remark, both are r.e.
COMPLETE PROOFS OF GODEL'S
INCOMPLETENESS THEOREMS
Proof. Note that for p prime, pjz implies that p=j 1 + jz . But if pj1 + (j + k )z and
pj1 + jz , then pjkz , implying pjkjz or pjz , and thus pjz , a
ontradi
tion.
Lemma 3. J : !
! !, dened by J (a; b) = (a + b)
+ (a + 1), is one-to-one.
(a; i) = x<a _ 1 (9y <a (9z <a (a = J (y; z ) ^ Div(1 + (J (x; i) + 1) z; y))));
where Div(x; y) 9z < y + 1 (y = z x) (satised i xjy) is re
ursive. It is
lear
that is re
ursive, and that (a; i) a _ 1.
Given a ; : : : ; an 2 !, we want to nd a 2 ! su
h that (a; i) = ai for all
i < n. Let
= max
fJ (ai ; i) + 1g;
i<n
1
LECTURES BY B. KIM
Note that the map <> is dened on all sequen
es due to the properties of
proved above. Further, sin
e is re
ursive, <> is re
ursive, and <> is one-to-one,
sin
e
<a ; : : : ; an >= <b ; : : : ; bm >
implies that n = m and ai = bi for ea
h i. Note, too, that the sequen
e number of
the empty sequen
e is
<>= x( (x; 0) = 0) = 0:
An important feature of our
oding is that we
an re
over a given sequen
e from
its sequen
e number:
Denition. For ea
h i 2 !, we have a fun
tion ()i : ! ! !, given by
(a)i = (a; i):
Clearly ()i is re
ursive for ea
h i. () will be
alled the length and denoted lh.
As intended, it follows from these denitions that ( < a : : : an >)i = ai and
lh( <a : : : an >) = n.
Note also that whenever a > 0; we have lh(a) < a and (a)i < a.
Denition. The relation Seq ! is given by
Seq(a) i 8x < a(lh(x) 6= lh(a) _ 9i < lh(a)((x)i 6= (a)i ):
That Seq is re
ursive is evident from properties enumerated above. From our
denition, it is
lear that Seq(a) if and only if a is the sequen
e number for some
sequen
e (in parti
ular, a = < (a) ; : : : ; (a)lh a >). Note that
:Seq(a) i 9x < a(lh(x) = lh(a) ^ 8i < lh(a)((x)i = (a)i ):
Denition. The initial sequen
e fun
tion Init : ! ! ! is given by
Init(a; i) = x(lh(x) = i ^ 8j < i((x)j = (a)j ):
Again, Init is evidently re
ursive. Note that for 1 i n,
Init( <a ; : : : ; an >) = <a ; : : : ; ai >;
as intended.
Denition. The
on
atenation fun
tion : ! ! ! is given by
1
+1
+1
( )
+1
+1
+1
+1
+1
( )+ +1
+1
+1
+1
COMPLETE PROOFS OF GODEL'S
INCOMPLETENESS THEOREMS
+1
F (a; b) =
is re
ursive.
Proof. Note that when G(a; b) < a, we have
)+1
LECTURES BY B. KIM
+1
+1
+1
+1
COMPLETE PROOFS OF GODEL'S
INCOMPLETENESS THEOREMS
Note that the natural numbers, N, are a model of the theory Q. If we add to
this theory the set of all generalizations of formulas of the form
('x ^ 8x(' ! 'xSx )) ! ';
providing the
apability for indu
tion, we
all this theory Peano Arithmeti
, or PA.
Thus Q PA, and PA ` Q.
Notation. We dene, for a natural number n,
n SS
: : : S} 0:
| {z
0
10
LECTURES BY B. KIM
!n , P
! (x < k ! '))
Q ` (x < 0 ! '):
This is (va
uously) true by axiom Q7. Now, assume that
Q ` 'x ! : : : ! ('xk ! (x < k ! ')):
We must show that
Q ` 'x ! ! ('xk ! (x < k + 1 ! ')):
Equivalently, we want to show that ` ' where = Q [ f'x ; :::; 'xk ; x < k + 1g.
By Q8, ` x < k _ x = k. In the rst
ase, the indu
tive hypothesis implies that
` ', while in the latter
ase, j= x = k ! ('xk ! '), and hen
e ` '. By either
route, proves '.
0
COMPLETE PROOFS OF GODEL'S
INCOMPLETENESS THEOREMS
11
Lemma 7. If (a) Q ` :'xk for k < n, and (b) Q ` 'xn , then for z 6= x not appearing
in ',
Q ` (' ^ 8z (z < x ! :'xz )) ! x = n:
Proof. We dene
Now, we obtain
()
()
Q ` 8z (z < n ! :'xz ):
Q`x=n! :
For the reverse impli
ation, we note that
j= 8z (z < x ! :'xz ) ! (n < x ! :'xn );
and thus (b) implies Q ` ! :(n < x). Now Q [f ; x < ng ` ' ^ :' by () and
the denition of . Therefore Q ` ! :(x < n) and by Axiom Q9 we
on
lude
Q ` ! x = n.
Q ` m + n = m + n;
Q ` y = m + n ! y = m + n;
Q ` '(m; n; y) ! y = m + n; and hen
e
Q ` 8y('(m; n; y) ! y = m + n);
as required. and < are similar (with < making additional use of Lemma
5).
Iin is representable by '(x ; : : : ; xn ; y) xi = y. In parti
ular, for any
k ; : : : ; kn 2 !, Iin (k ; : : : ; kn ) = ki , and hen
e
1
Q ` '(k ; : : : ; kn ; y)
1
! y = ki ! y = Iin (k ; : : : ; kn );
1
12
LECTURES BY B. KIM
R2. F (a) = G(H (a); : : : ; Hk (a)), where G and ea
h of the Hi are representable.
Assume that G is represented in Q by ' and the Hi are represented in
Q by i , respe
tively. We show that F is represented by
(x; y) 9z ; : : : ; zk ( (x; z ) ^ ^ k (x; zk ) ^ '(z ; : : : ; zk ; y)):
In other word we want to show, for any a ; :::; an 2 !,
(y)
Q ` (a ; : : : ; an ; y) ! y = G(H (a); : : : ; Hk (a))
where a = (a :::an ).
Now, for = Q [ f(a ; : : : ; an ; y)g, sin
e the i represent Hi , we have
that ` 9z ; : : : ; zk (z = H (a) ^ ^ zk = Hk (a) ^ '(z ; : : : ; zk ; y)):
Hen
e we have
j= 9z ; : : : ; zk ('(H (a); : : : ; Hk (a); y));
and sin
e the zi do not appear,
j= '(H (a); : : : ; Hk (a); y):
Sin
e ' represents G, we have
j= y = G(H (a); : : : ; Hk (a));
as required.
On the other hand, for = Q [ fy = G(H (a); : : : ; Hk (a))g,
` '(H (a); : : : ; Hk (a); y)
` 9z ; : : : ; zk (z = H (a) ^ zk = Hk (a) ^ '(z ; : : : ; zk ; y))
` 9z ; : : : ; zk ( (a; zi ) ^ k (a; zk ) ^ '(z ; : : : ; zk ; y))
` (a ; : : : ; an ; y)
Thus (y) is established.
R3. F (a) = x(G(a; x) = 0), where G is representable in Q and for all a there
exists x su
h that G(a; x) = 0, is representable in Q.
Assume G is represented in Q by '(x ; : : : ; xn ; x; y). Let
(x ; : : : ; xn ; x) 'y ^ 8z (z < x ! :'yxz ):
Let F (a) = b and ki = G(a; i) for i 2 !. Then
Q ` '(a ; : : : ; an ; i; y) ! y = ki ;
thus
Q ` '(a ; : : : ; an ; i; 0) ! 0 = ki ;
and if j < b, so that kj 6= 0, then
Q ` :'(a ; : : : ; an ; j; 0):
On the other hand, kb = 0, so
Q ` '(a ; : : : ; an ; b; 0):
Hen
e, by Lemma 7,
Q ` ('(a; x; y)y ^ 8z (z < x ! :'(a; x; y)y xz )) ! x = b;
and thus,
Q ` (a; x) ! x = b:
By generalization, we have that represents F in Q, as desired.
1
COMPLETE PROOFS OF GODEL'S
INCOMPLETENESS THEOREMS
13
1 2
+2
14
LECTURES BY B. KIM
+2
8
Form((a) )
>
>
>
<Form((a) ) ^ Form((a) )
Form(a) i
>
Vble((a) ) ^ Form((a) )
>
>
:
AtF (a)
2
(5) Sub : ! ! !, su
h that Sub(dte; dxe; due) = dtxu e and Sub(d'e; dxe; due) =
d'xu e for terms t and u, variable x, and formula '.
3
Proof. Dene
8
if Vble(a) ^ a = b,
>
>
>
>
>
< (a) ; Sub((a) ; b;
); : : :
if lh(a) > 1 ^ (a) 6= h(8)
>
>
>
<
: : : ; Sub((a)lh a ; b;
) >
^ Seq(a);
Sub(a; b;
) =
>
< (a) ; (a) ; Sub((a) ; b;
) > if a = <h(8); (a) ; (a) >,
>
>
>
>
>
^ (a) 6= b
>
>
:a
otherwise.
1
( )
Note that, if well-dened, the fun
tion has the properties desired above.
We show Sub is well-dened by indu
tion on a: a = 0 falls into the
rst or last
ategory sin
e lh(0) = 0, hen
e Sub(0; b;
) is well-dened for
all b;
2 !. If a 6= 0, then (a)i < a for all i lh(a), and thus we may
assume the values Sub((a)i ; b;
) are well-dened, showing Sub(a; b;
) to be
well-dened in all
ases.
(6) Free ! , su
h that for formula ', term , and variable x, Free(d'e; dxe)
if and only if x o
urs free in ', and Free(d e; dxe) if and only if x o
urs
in
2
Proof. Dene
8
>
<9j < (lh(a) 1) (Free((a)j ; b)) if lh(a) > 1
Free(a; b) i Free((a) ; b) ^ (a) 6= b
if lh(a) > 1
>
:
a=b
otherwise.
+2
^ (a) =
6 h(8),
^ (a) = h(8),
1
Free learly has the desired property, and that it is well-dened follows by
COMPLETE PROOFS OF GODEL'S
INCOMPLETENESS THEOREMS
15
Proof. Note
Sent(a) i Form(a)
(8) Subst(a; b;
) ! su
h that for a given formula ', variable x, and term t,
Subst(d'e; dxe; dte) if and only if t is substitutable for x in '.
3
Proof. Dene
8
>
Subst((a) ; b;
)
>
>
>
>
>
Subst
((a) ; b;
) ^ Subst((a) ; b;
)
<
Subst(a; b;
) i :Free(a; b) _ (:Free(
; (a) )
>
>
>
^ Subst((a) ; b;
))
>
>
>
:0 = 0
2
otherwise.
Note that Subst has the desired property, and is well-dened by essentially
the same indu
tion used above.
(9) We dene
8
:False((a) ; b) ^ False((a) ; b) if a = <h(!); (a) ; (a) >
>
>
>
<
^ Form((a) ) ^ Form((a) );
False(a; b) i
>
:
False
((
a
)
;
b
)
if
a
= <h(:); (a) > ^ Form((a) ),
>
>
:
Form(a) ^ (b)a = 0
otherwise.
2
False is re
ursive by the same indu
tion as applied above. We note the
signi
an
e of False presently.
To ea
h b 2 !, we may asso
iate a truth assignment vb su
h that for a prime
formula (atomi
or of the form 8x'),
vb ( ) = F i (b)d e = 0:
Further, for any truth assignment v : A ! fT; Fg, where A is a nite set of prime
formulas, there exists a b su
h that v = vb : we may write A = f' ; : : : ; 'n g su
h
that d' e < d' e < < d'n e. For 1 j d'n e dene
j = 0 when j = d'i e
for some i n and v('i ) = F , and
j = 1 otherwise. Then b = <
; : : : ;
d'n e >
satises vb = v on A.
Then moreover, for any formula ' built up from A,
1
Proof. Re all bd : !
16
LECTURES BY B. KIM
x
t , with
Proof. Re all we take axiom group 3 to be the formulas having the following
form: 8x(
0 ) ! (8x
! 8x
0 ). Thus
! 8x
, where
^ Form(y)
^ :Free(y; x) ^ a = <h(!); y; <h(8); x; y >>)
^ a = <h(=); x; x>):
COMPLETE PROOFS OF GODEL'S
INCOMPLETENESS THEOREMS
8
>
<a = <h(8); (a) ; (a) >
Gen(a; b) i 0 = 0
>
:0 = 1
2
^ Vble((a) ) ^ Gen((a) ; b)
2
17
if a > b,
if a = b,
if a < b.
We have, to this point, dened three
odings: <> on sequen
es of natural numbers, h on the language and logi
al symbols, and de on the terms and formulas. We
presently dene a fourth
oding, of sequen
es of formulas:
d e : fsequen
es of L-formulasg ! !;
given by
d ' ; : : : ; 'n e = < d' e; : : : ; d'n e > :
This map is one-to-one, as it is derived from the established (inje
tive)
odings,
and in parti
ular, we
an determine, for a given number, if it lies in the image of
d e , and, if so, re
over the asso
iated sequen
e of formulas.
Denition. Given L, let T be a theory (a
olle
tion of senten
es) in L. Dene
T = fde j 2 T g:
We say that T is axiomatizable if there exists a theory S , axiomatizing T (that
is, su
h that Cn S = Cn T ), su
h that S is re
ursive. We say that T is de
idable
if Cn T is re
ursive.
We shall make use of the following relations:
DedT = fdd' ; : : : ; 'n e j ' ; : : : ; 'n is a dedu
tion from T g !.
Note that
1
18
LECTURES BY B. KIM
We
an see that if we say T is axiomatizable in wider sense when S axiomatizing T is re
ursively enumerable, then the above two theorems still hold with this
seemingly weaker notion. In fa
t, two notions are equivalent, whi
h is known as
Craig's Theorem.
where 9 and
In parti
ular, we note d(') is satisable pre
isely when ' is satisable by some
truth assignment taking v to the Godel number of ', and LN j= d(') pre
isely
when ' is satised by every truth assignment taking v to d'e.
0
!!
Then
COMPLETE PROOFS OF GODEL'S
INCOMPLETENESS THEOREMS
19
However, writing out what formula this en
odes and introdu
ing our usual abbreviations, we have
dg(d'e) = d:8v :(:(v = d'e ! :'))e
= d9v (v = d'e ^ ')e
= dd(')e;
as desired.
0
Fixed Point Theorem (Godel). For any LN -formula '(x) (i.e., either a senten
e
or a formula having x as the only free variable), there is some LN -senten
e su
h
that
Q`
! '(de):
20
LECTURES BY B. KIM
In fa
t, the above
orollary is true for any
ountable L
ontaining a k-ary predi
ate or fun
tion symbol, k 2, or at least two unary fun
tion symbols.
Godel-Rosser First In
ompleteness Theorem. If T is a theory in a
ountable
reasonable L LN , with T [ Q
onsistent and T axiomatizable, then T is not
omplete.
` ) Q ` PfT (de):
COMPLETE PROOFS OF GODEL'S
INCOMPLETENESS THEOREMS
21
In parti
ular, T ` implies that PrfT (de; m) for some m 2 !. Sin
e PrfT is
re
ursive, it is representable in Q, hen
e Q ` PrfT (de; m), and
Q ` 9xPrfT (de; x);
or
Q ` PfT (de):
By the xed point lemma, there exists a senten
e su
h that
T Q ` ! :PfT (de):
()
If T ` , then Q ` PfT (de), and thus Q ` :, and hen
e T ` :, a
ontradi
tion.
Thus T 0 .
On the other hand, if T is !-
onsistent (i.e., whenever T ` 9x'(x), then for
some n 2 !, T 0 :'(n)), then T 0 :. In parti
ular, if T ` :, then
T ` PfT (de);
by (). That is,
T ` 9xPrfT (de; x):
However, if PrfT (de; m) for some m 2 !, then T ` ,
ontradi
ting the
onsisten
y of T . Thus we must have :PrfT (de; m) for all m 2 !. Sin
e Q represents
PrfT ,
T Q ` :PrfT (de; m)
for all m 2 !,
ontradi
ting the !-
onsisten
y of T .
Rosser generalized Godel's proof by singling out for T a senten
e su
h that
T 0 and T 0 :, without the assumption of !-
onsisten
y.
We now begin our approa
h to Godel's Se
ond In
ompleteness Theorem. We x
T , a theory in a
ountable reasonable language L LN .
We note the following fa
t from Hilbert and Bernays' Grundlagen der Mathematik, 1934.
Fa
t. If T is
onsistent, T ` P A, and T is re
ursive, then for any senten
es and
in L,
I. T ` ) Q ` PfT (de)
II. PA ` (PfT (de) ^ PfT (d ! e)) ! PfT (de)
III. PA ` PfT (de) ! PfT
dPfT (de)e
:PfT (d0 6= 0e). Clearly ConT
and by (II),
22
LECTURES BY B. KIM
(z)
dPfT (de)e :
COMPLETE PROOFS OF GODEL'S
INCOMPLETENESS THEOREMS
23
as desired.
Now assume that T
Remark. Godel's Se
ond In
ompleteness Theorem in fa
t follows from Lob's Theorem. In parti
ular, given T as in the hypotheses of both theorems, if T ` ConT ,
then
T ` PfT (d0 6= 0e) ! 0 6= 0:
But by Lob's Theorem, this in turn implies that T ` 0 6= 0, showing that su
h a
theory, if
onsistent,
annot prove its own
onsisten
y.
Referen
es
[BJ G. S. Boolos and R. C. Jerey, Computability and logi
.
[E H. Enderton, A mathemati
al introdu
tion to logi
.
[Sh J. R. Shoeneld, Mathemati
al logi
.
[Sm R. M. Smullyan, Godel's in
ompleteness theorems.