"
Y
A X
t h e branch has a member i n each i n t e r v a l o f
(remembering K6nig
order,
Y i s an i n f i n i t e branch i n
..Am]
Then f o r some t r u e l y f i n i t e
)
,
€aN.Suppose
t r u e l y f i n i t e m, q
m(*)
J,
"says" t h a t
(whose o r d e r i s i n c l u d e d i n t h e n a t u r a l o r d e r o f & N ) .
By t h e d e f i n i t i o n o f
A
j,
€aR a,, f i r s t
<
y
A
(X < 2 <
y)]
s a t i s f i e s " f o r every
e(-,-,p,q,A,
we can r e p l a c e
lemma) t h e s e t
Y)
Hence
m(*)+i'
i n the interval
Am(*)+i t* ( i n t h e t r e e
( 3 2 E
-t
t* such t h a t f o r e v e r y
,
(F2)
i t i s easy t o see t h a t
m(*)+l
x < y < a (x,y)
there i s
,...,Am)". by
m+l,
hence c l e a r l y
153
On Logical Sentences in PA
ip
€ f i N:
{p
E
a k ( 3 Y ) *(Y,
aN:
f o r every a
P, ill = E
y
...,Am )
BR a s
which belongs t o 2.7.
t* ( i n t h e t r e e
(x,y)
B(-,-,p,q,
l.
required.
m: For every n a t u r a l number
3 n CPQh ( n ,
in t h e i n t e r v a l
Am+1
E
t h e r e i s an element below Al,
t* such t h a t f o r every
there i s
A,+,
x < y < a , X E Am + l ,
k we can prove i n PA(Q
MM
) t h e statement
k).
Proof: F i r s t we prove t h a t t h e conclusion i s t r u e , i . e . t r u e i n t h e universe
6?)if
i t i s a model of say second o r d e r Peano a r i t h m e t i c .
Then f o r every n , t h e r e i s a p a i r
Suppose t h a t the conclusion f a i l s f o r k.
(Fn, H ) which forms acounterexample t o CP'
of functions
eh
n
We now d e f i n e by induction on i For
5
k i n f i n i t e sets
so t h a t R i + ,
Ai,
5 Ai.
i = 0 t h e r e i s no problem. Let A, = B, = t h e set of a l l natural numbers.
So suppose we have defined
A
for j s i
j
and we s h a l l define
By t h e i n f i n i t e Ramsey theorem we can get an i n f i n i t e ( a ) f o r every y Ai n y )
Now
(n,k).
z
<
in
A:,
f
j
for j = i+l.
Af 5 Ai such t h a t : F ( A , n y , A, n y , ...,
i s t h e function
Y¶Z
A
Z
restricted t o y.
f
does not depend on
z, i . e . i f
y
z,
< z,,
E
A
then
YSZ
SO l e t
f
= f
Y
for y
Y,Z
( b ) For
Y,
<
<
A' i '
z
Y, < y 2 i n
f
= YlZ,
f
YJ,
.
1
fy,ry, = fy,ry0 [Simply apply Ramsey theorem t o t h e natural t h r e e place c o l o u r i n g , f i r s t f o r ( a ) then f o r ( b ) l So
f
=
U If
Y
rz
are i n A i l
: z < y
i s a k-colouring ( o f t h e natural num-
b e r s ) . By t h e i n f i n i t e Ramsey theorem t h e r e i s i n
&3,
an i n f i n i t e s e t A:
A:,
which i s f-homogenous. Now we s h a l l deal with
H.
Again t h e r e is an i n f i n i t e
A: 5 A f
(A;
of course) such t h a t :
(c) i f H Z ( p , A, n y ,
y
<
z
. .., Ai
are i n
A:,
ny,
9)
< z
a r e in
95 Ai
n
y,
p
does not depend on
<
y
t h e n t h e value of
z.
Moreover (d) i f
x
<
y
A;,
9 5 Ai
nx,
p
<
x then t h e value of
BR,
S. SHELAH
154
..., Ai
HZ(p, A, n y.
For every p
9)
n y,
63,
E
and i n f i n i t e s e t
zero i f f f o r a r b i t r a r i l y large q HZ(p, A,
..., Ai
y,
n
<
. 9):
z in A i / q
i t is
the value of
i s zero. Clearly a f i n i t e change i n
HZ, and "for a r b i t r a r i l y large q
"for every large enough q "
z )
we define T ( p ,
5 A:
f o r every y
E ( % ~
n y, D n y - q )
n o t change the value of
(and on
does not depend on y
3
does
can be replaced by
"
(see 2 . 2 ) .
Now we want t o apply the Galvin Prikry theorem t o T, more exactly t o a parametrized version of i t
( p as the parameter). Simply i t e r a t e the usual Galvin-
Prikry theorem on the natural numbers, and take the diagonal intersection. What we get we call
.
A;
Again remembering 2.2, and the conclusion of the Galvin-Prikry theorem, we can
so t h a t f o r every p,
find Aitl 5 A:, Ai/q
the value of
i f f o r some q > p ,
..., A 1.
HZ(p,A, n y ,
n Y, Ai+,nY-q)
f o r every y
i s zero,
<
z in
then 9 = P
will serve. If
..., Ak
A,, n
A,,
9
Aktl
are defined, choose y
y > contradicts thechoice of
Fz, HZ
<
z in Aktl
and < A , n y,
.
However we want t o prove t h i s in PA(QMM) (and not in ZFC o r i n second order number theory). The proof i s the same, replacing "a s e t of natural numbers" by "a definable s e t of natural numbers". Why i s < < F , H > : n < W > definable? We canchoose for each n , a minimal n n pair < F n , H n > (by a simple enough coding). We can apply the i n f i n i t e Ramsey theoMM
rem as Macintyre [MI proves theparallel statement holds (from PA(Q ), of course the proof depends on the formula defining the colouring and the i n f i n i t e s e t ) . Wore exactly he proves that i f
,...,xn,
~(x,
-
p, z)
A
i s i n f i n i t e and definable,
i s such that
then there i s a definable i n f i n i t e
V x1
B c A and co
<
,...,xn
( 3 2 < c ) ~ ( x ,..., , xn,p,z)
c such that
We are l e f t with the "parametrized Galvin-Prikry". Let a < * b mean:
<
a , b code f i n i t e increasing sequences, ,
e ( a ) > respectively, and ca(m) : m < a ( a ) > i s a proper i n i t i a l seg-
ment of < b ( m ) : m
<
e(b)>.
155
On Logical Sentences in PA We d e f i n e
A:
A:
(after
has been d e f i n e d )
by d e f i n i n g by i n d u c t i o n on
I, kk, 6, such t h a t k,
(a)
<
... <
k
2- 1
L o , C1,
and
..., sI-l
{0,11
E
> < * a < b and f o r e v e r y p < k f o r some q, (QMMa,b) C < k o , ..., k f i - l y, z i f q < y < z, y E A?, z E A? t h e n HZ(p, A, n y, A . n y.
(b) f o r every
...,
1
Ia(m) : m < z ( a ) > / q ) =
1
cPl.
c ~ =- 0~ . Now we can c a r r y t h e d e f i n i t i o n (as i n [ M I n o t i n g t h e f o r m u l a s we use have (c) i f compatible w i t h
(a) t (b),
bounded complexity).So we f i n i s h t h e p r o o f o f 2 . 7 .
5 3. A t r u e
ny-sentence o f P A n o t p r o v a b l e i n PA
I n summer '80 Friedman and H a r r i n g t o n o f f e r e d h o t l y t h e i r view t h a t i t i s one o f t h e main problems o f contemporary l o g i c t o f i n d mathematical sentences as ment i o n e d i n t h e t y t l e , as w e l l as t o f i n d n a t u r a l t h e o r i e s w i t h incomparable c o n s i stency s t r e n g t h . The " t e c h n i c a l d i f f i c u l t y " i s i m m a t e r i a l ; i n f a c t t h e easiness o f t h e p r o o f may i n d i c a t e t h e profoundness and n a t u r a l i t y o f t h e sentence. Now an answer t o such q u e s t i o n i s n a t u r a l l y more open t o debate t h a n t h e usual mathematic a l problem.
A s t h e a u t o r d i d n o t want t o go i n t o such d i s c u s s i o n , and H a r r i n g t o n wanted a s o l u t i o n , an agreement was reached: i f t h e a u t h o r c o u l d f i n d a s o l u t i o n which H a r r i n g t o n would t h i n k i s O.K.,
he would w r i t e i t up, d i s c u s s i t and p u b l i s h i t .
The c o n t e n t o f t h i s s e c t i o n was done i n s p r i n g '81, H a r r i n g t o n O.K.ed i t , as w e l l as I 4 ( w h i c h was done i n summer '80) b u t was t o o l a z y t o f u l f i l l h i s promise. 3.1. C o n t e x t :
N be a n a t u r a l number
1) Let
r
= < r : I < I(;)> 9"
a finitesequence
o f n a t u r a l numbers.-
2) L e t
K = K i = {(A,
<,
R)
: A
6 R
a subset o f
N,
<
a l i n e a r order o f
A
a sequence o f r e l a t i o n s o v e r an
A
,
,
r -place
Ik e ( R ) = I(;)}.
3 ) Members o f
K
a r e denoted by
A, B
,
letting
power o f t h e s e t o f t h e s e t o f elements o f 4) I n (4a)
K
A =
/ ~ ,l
so
i s the
llAll
A.
we d e f i n e
A
( B an end e x t e n s i o n o f
A) i f A
i s a submodel o f
B
and
S. SHELAH
156 x
IBI
E
,
- 1.4
(4b) A < B
y
satisfying
implies
y < x.
of
A'
into G
IIb/l 5 IlAll
A',
A <
en B over
i n t o @(N)
Let
+
1,
A < B and f o r any en t h e r e i s an embedding
.
A
f r o m @(N)
3.2. E ( N , r , k , n ) - p r i n c i p l e : F
PI
( B an u n i v e r s a l end e x t e n s i o n o f A) i f
A'
5) A function
1 ) Domain:
E
F
is
be a k - p l a c e f u n c t i o n s a t i s f y i n g
i s d e f i n e d on i n c r e a s i n g sequences o f l e n g t h
2) Choice F u n c t i o n :
..., A k )
F(A,,
...
3 ) Isomorphism I n v a r i a n c y : i f A 1 < i s an isomorphism
g
from A k
< Ak
onto
k
K
from
.
IA,[
E
if / G ( A ) I z f ( 1 A I ) .
f-small
B
k
E
K,
B1 <
mapping
... < B k
At
onto
E
Bil
K , and F
then
there and
g
commute, i .e. F(B~.
..., B k ) =
4 ) Weak H e r e d i t a r i t y : F o r e v e r y such t h a t : i f
f
B
-1
=
0)
then
Then - there
A,
+)
A2 <
<
i s a submodel o f
B~
F(A,
...,
g(F(Al,
,...,A k )
= F(Bl
...
. k t h e r e i s an x -small f u n c t i o n
< Ak,
f(Bil-l)
Ai,l
n Ail
5 Bil
(stipulating
,..., B k ) .
i s an i n c r e a s i n g sequence < A :~ il < n > on which
F
depends on
the f i r s t structure only. 3.3.
Fact
F to
F'
if
N > 22 k+n+'(r(i)+l),
F a s i n 3.2 and
N' >
a k - p l a c e f u n c t i o n s a t i s f y i n g 1 ) - 4 ) o f 3.2 f o r
N,
t h e n we can e x t e n d
N ' , i n one and o n l y one
way. __ P r o o f . By 3.2 ( 4 ) ( 3 ) (as i n t h e p r o o f o f 1.3 ( 1 ) ( b ) ) . 3.4.
Claim: I n
PA+PH
$* = ( V
r,
we can p r o v e (the
k, il)
r,
(N,
k , n ) - p r i n c i p l e h o l d s f o r e.g.
,k+n+ f C r ( i ) + l l
).
N = 2 __ P r o o f : D e f i n e by i n d u c t i o n on
m
e,
i s s m a l l e r enough t h a n
putation). Applying
PH
and i n d i s c e r n i b l e f o r F ' ( A ~, 1
e,
a model
Am<*A
e
we g e t a s e t (F'
as i n 3.2
..., A .
'k
) = F'(A. Ji
.
e
c
K
w i t h universe
il,
so t h a t i f
( j u s t t a k e c a r e o f 3.1 ( 4 ) ( b ) , easy com-
C
o f n a t u r a l numbers
for
,
A
N'
..., Ajk)
(V x < y ~ C ) C 2 2 ~ < y l ,
l a r g e enough)
,
and
ICI
>
4 Min C
157
On Logical Sentences in PA (we use an e q u i v a l e n t v a r i a n t o f P.H.).
: i
F
C>,
E
As i n 5 1
depends on t h e f i r s t v a r i a b l e o n l y . Now as i n t h e p r o o f o f 1.3
( 1 ) ( b ) we c o l l a p s e t h e s o l u t i o n below 3.5.
Claim: I n
PA+$*
N
.
we can p r o v e t h e c o n s i s t e n c y o f PA (hence
Proof. We can b u i l d a non-standard model o f dard,
N
l a r g e enough, and
..., Ak)
F(A,,
mal code f o r which
1
( v y)
e a s i l y f o r subsequences o f
r
PA,+$*,
<4> d e f i n e
F :
i s d e f i n e d as f o l l o w s :
let
{3Xcp(X.y)
=
..., A k > / = +
M, choose
cp(x,y)
3X[cp(X,y)
A
(v
z
< X)
and t h e n
( i f t h e r e i s one).
A,,
cp(x,y)
non s t a n -
be a f o r m u l a w i t h m i n i -
ic p ( Z , j ) l
(by the lexicographic order o f
E
k, n
"induction fails,i.e.
and t h e n t a k e minimal x
PAP$*).
1''
y.
y o , yl,
The r e s t i s as i n I 1
...>
)
the only
a d d i t i o n a l p o i n t i s why a r e a d d i t i o n and m u l t i p l i c a t i o n d e f i n a b l e ? T h i s i s by 3.1 ( 4 ) ( b ) .
5
4. On c o n s i s t e n c y s t r e n g t h
Let extending CON(T)
T PA
denote h e r e a ( r e c u r s i v e ) t h e o r y ( w i t h f i n i t e - s o r t , f i n i t e - l a n g u a g e ) b u t i t may a l s o "speak" on r e a l s and even a r b i t r a r y s e t s .
be t h e sentence ( i n
D e f i n i t i o n : We say
PA
T, scs T,
language) s a y i n g
T
i s consistent.
( t h e consistency strength o f
equal) than t h e consistency strength o f
T2)
if
PA
Let
t- CON(T,)
T,
i s smaller (or -f
CON(T,):
I t was observed t h a t e s s e n t i a l l y " l a r g e c a r d i n a l axioms a r e l i n e a r l y ordered"
(though i n some cases t h i s "has n o t y e t been proved").More e x a c t l y i t seems t h a t a l l s e t t h e o r i e s which has been c o n s i d e r e d so f a r , a r e l i n e a r l y o r d e r e d by Solovay
( I t h i n k ) has
found
T's
which a r e
5
-incomparable,
s
cs b u t t h e y were
.
cs " p a r a d o x i a l " (i .e. have s e l f - r e f e r e n t i a l sentences). We s h a l l t r y t o g e t more r e a sonable ones.
* * * * * * * Let
PA+
be
PA t C O N ( P A ) .
We work i n s i d e
P A . A model w i l l mean one which
i s definable. Let
T,,
T,
be c o n s i s t e n t t h e o r i e s ( i n o u r " u n i v e r s e " which s a t i s f i e s PA).
s. SHELAH
158
T,
" s a y i n g " t h e r e i s a model o f
(think o f
PA+,
PA
+
CFMSI, o r o f course As T,
+
ZFC,
T, " s a y i n g " t h e r e i s a model o f
ATR, see Friedman, McAloon and Simpson
T,tCON(TI)
hence t h e r e i s MI
( 1 ) q Q ( n ) says
n
i s c o n s i s t e n t , hence ( b y Godel incompleteness has a model M,
M,
E
f ''$2(n)''
I=
such t h a t
(n
. By
t h e r e q u i r e m e n t on T,
i s a non-standard i n t e g e r )
where
i s a n a t u r a l number and t h e r e i s a p r o o f o f s i z e
n
PA
ZFC+large c a r d i n a l s ) .
T, +CON(T,) + iCON(T, +CONTI)
iCON(T,),
n
of
t 1 CON ( Te)
PA
f ( 2 ) +,(n)
says
$
Q
(n)
but
f ( 3 ) T~ = PA + ( 3 n ) r$,(n) i s consistent with
i s t h e f i r s t such number.
n
As we have assumed t h a t
with
TI,
o r use
or
i s consistent,
theorem) M,
I rrl-CAo;
T, says t h a t
has a model, c l e a r l y
+-,~,,(2~")1 f
+ (Vm) l$,(m)
PA + 3 n$,(n)
( a s even
PA+
PA
i s consistent
.
PA+)
By a theorem o f Friedman (based on a n a l y z i n g Godel incompletness theorem) f
( 4 ) Tb = PA + ( 3 n ) C$,(n)
'
( 5 ) PA+
+
Ta, Tb
are
s
cs
-incomparable.
L e t us p r o v e e.q.
*
Ta 1- CON(Ta)
rBecause f o r any model nition
11
PA
We s h a l l p r o v e t h a t csTb
217
+.
i s consistent with
Ta
$,(2
f
o f a mode:
phism from
No
Clearly
N,
of
N o of PA
(i.e.
PA++Ta, No
b e i n g a model o f
1: "N,
i n t o a p r o p e r i n i t i a l segment o f N,
satisfies
of bounded f o r m u l a s and
Ta -PA,
has a d e f i -
i s a model o f P A " ) and an isomorN,.
as end e x t e n s i o n s p r e s e r v e t h e s a t i s f a c t i o n
f
$,(x),
PA+
$,(x)
a r e such f o r m u l a s . 1
Note a l s o t h a t (6)
PA
+
.
Ta I- 7CON(Tb)
[Otherwise t h e r e i s a model a d e f i n i t i o n o f a model segment o f No I= T
a' formulas,
N, No
(i.e. f "$,(n)
and
f
N, o f No
No
of
PA+Ta+CON(Tb)
Tb a n d a n i s o m o r p h i s m g of
hence i n
No
No
there i s
onto a proper i n i t i a l
s a t i s f i e s t h e sentences s a y i n g t h o s e t h i n g s ) . AS 211 f and -I $,(2 ) " f o r some n, b u t as $, $1 a r e bounded f commute w i t h e x p o n e n t i a t i o n , c l e a r l y N, I. "$,(g(n)) and
159
On Logical Sentences in PA 1 $,(22g(n))".
So
f 2m N, I= Tb hence f o r some m, N , I= "$,(m) and $ , ( 2 ) " . f f and $,(g(n))",hence by q 2 ' s definition g ( n ) , m have t o be
But
f
N , k "$,(m)
equal. B u t
N, k "$,(2
2m
)
and 7 $ 1 ( 2 ' g ( m ) ) " hence
g(n),
rn
should be unequal,
contraddictionl. By ( 5 ) and ( 6 ) clearly
PA+
+
C O N ( T a ) I$
(as PA+ + T~
CON(Tb)
is
consistent (by ( 3 ) ) ; t h i s implies
PA I f CON(Ta)
(7)
+
So Ta $ csTb.
CON(Tb)
. Tb $ csTa
i s t o t a l l y analogous. n f NOW (Woodin suggests) wecan replace the sentences ( 3n ) r $ , ( n ) + - 1 $ ~ ( 2 ' ) I n f ( 3 n ) r $ , ( n ) + $ J 2 ' ) I by inequalities of the indicator functions corresponding t o
and
The proof of
f , , f,
T I and T, ( i . e . the function T , , T,).
t o the consistency of
exhibiting the
E.g. ( V n ) [ i f
f,(n)
ITsentences :
corresponding
i s defined then
so i s
f , ( f , ( f , ( n ) ) 1 and i t s negation. So i f we accept those functions as "mathematical"
( n o t j u s t reasonable methamathematical ) we get mathematical theories of incomparable consistency strength. (Originally we have used three t h e o r i e s ) . We a r r i v e t o the dangerous question o f which PA
function: on
see Paris and Harrington [ P H I ,
T's
have matheratical indicator
on many theories ( l i k e ZFC+large
cardinal) see Friedman rFl1 on ATR, see Friedman McAloon and Simpson rFMSl on 1
IT,-CA,
see 5 2. Alternatively f o r an indicator function
f ( f * ( n ) + l ) , and use
$,
$3
where
$,
f
f*
define
= "the f i r s t
by
f*(O) = 0
n f o r which f ( n ) i s
:Q
f*(n+l)=
mod 4".
+ * * * * * * * Notice the following two phenomena
( A ) For any two natural s e t theories, not only they are
5
cs
-comparable, b u t one
i s interpretable in the other ( o r expected t o be s o ) . ( 6 ) Similarly, f o r any theories, e.q. undecidab l i t y r e s u l t s are gotten by i n t e r -
pretation. Friedman rFr21 proved a theorem saying ( A ) i s really true. Concerning ( B ) however, in CSh1, (under C H )
the monadic theory of the re 1 order i s proven undecidable
without the usual interpretation. I n Gurevich and Shelah
CGSl1 t h i s
i s explained i t i s a Boolean-valued i n t e r p r e t a t i o n , and by CGS21 the usual i n t e r pretation i s impossible. Now we can t r a n s l a t e i t t o ( A ) : l i s t the reasonable axioms f o r the monadic theory of the real order (considered as a two-sort model).
S. SHELAH
160
REFERENCES
CEHMRl P. Erdos, A. H a j n a l , A. Mate and R. Rado, C o m b i n a t o r i a l s e t t h e o r y , N o r t h H o l l a n d P u b l . Co.
CF11
H. Friedman, On t h e necessary use o f A b s t r a c t s e t t h e o r y , Advances i n Mathem a t i c s 41 (1981), 209-280.
TF21
H. Friedman, T r a n s l a t a b i l i t y and r e l a t i v e c o n s i s t e n c y .
C FMS 1 H. Friedman, K. McAloon and S.G. Simpson, A f i n i t e c o m b i n a t o r i a l p r i n c i p l e which i s e q u i v a l e n t t o t h e I - c o n s i s t e n c y o f p r e d i c a t i v e a n a l y s i s . CGSll
Y. Gurevich and S. Shelah, The monadic t h e o r y and t h e n e x t w o r l d . I s r a e l J.
Math. CGS21 Y. Gurevich and S. Shelah, A r i t h m e t i c cannot be i n t e r p r a t e d i n monadic theory o f 8. [MI
A. M a c i n t y r e , Ramsey q u a n t i f i e r s i n a r i t h m e t i c , Proc. o f a L o g i c Symp. .(Karpacz 1979) ed. L. P a c h o l s k i and A. W i l k i e , S p r i n g e r V e r l a g L e c t u r e Notes i n Mathematics.
[PHI
J . P a r i s and L. H a r r i n q- t o n ,- A mathematical incompleteness i n Peano a r i t h m e t i c , Handbook o f Mathematical L o g i c , ed. Barwise, N o r t h - H o l l a n d Publ. Co.., 1977, 1133-1142.
CSh 1
S. Shelah, The monadic t h e o r y o f o r d e r , Annals of Math. 102 (1975), 379-419.
CSSl
S.G. Simpson and J.Schmer1, On t h e r o l e o f Ramsey q u a n t i f i e r s i n f i r s t o r d e r a r i t h m e t i c , J. Symb. L o g i c .