Anda di halaman 1dari 28

7

9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


Recursive Logic Frames

Saharon Shelah

Institute of Mathematics
Hebrew University
Jerusalem, Israel
Jouko V a an anen

Department of Mathematics
University of Helsinki
Helsinki, Finland
October 21, 2004
Abstract
We dene the concept of a logic frame, which extends the concept
of an abstract logic by adding the concept of a syntax and an axiom
system. In a recursive logic frame the syntax and the set of axioms
are recursively coded. A recursive logic frame is called complete (re-
cursively compact, countably compact), if every nite (respectively:
recursive, countable) consistent theory has a model. We show that
for logic frames built from the cardinality quantiers there exists at
least completeness always implies countable compactness. On the
other hand we show that a recursively compact logic frame need not
be countably compact.
1 Introduction
For the denition of an abstract logic and a generalized quantier the reader
is refereed to [4], [13], and [14]. Undoubtedly the most important among
abstract logics are the ones that have a complete axiomatization of validity.

This is an extended version of Preprint # 593 of the Centre de Recerca Matem` atica,
Barcelona. The second author is grateful for the hospitality of the CRM in April 2004.

This research was supported by The Israel Science Foundation. Publication number
[ShVa:790]

Research partially supported by grant 40734 of the Academy of Finland.


1
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


In many cases, most notably when we combine even the simplest generalized
quantiers, completeness of an axiomatization cannot be proved in ZFC alone
but depends of principles like CH or .
Examples of logics that have a complete axiomatization are:
The innitary language L

[10].
Logic with the generalized quantier
1

1
x(x, y) [x : (x, y)[
1
[27].
Logic with the conality quantier
Q
cof

0
xy(x, y, z) x, y) : (x, y, z) is a linear order
of conality
0
[24].
Logic with the cub-quantier
Q
cub

1
xy(x, y, z) x, y) : (x, y, z) is an
1
-like linear order
in which a cub of initial segments have a sup [24].
Logic with the Magidor-Malitz quantier, assuming
Q
MM

1
xy(x, y, z) X([X[
1
x, y X(x, y, z)) [15].
The extension L(

) of rst order logic was introduced by Andrzei


Mostowski in 1957. Here

is the generalized quantier


M[=

x(x, a) [b M : M[= (b, a)[ .


Mostowski asked whether L(

) is
0
-compact (i.e. every countable set
of sentences, every nite subset of which has a model, has itself a model)
and observed that L(

0
) is not. In 1963 Gerhard Fuhrken [7] proved that
L(

) is
0
-compact if
0
is small for (i.e. if
n
< for n < , then

n<

n
< ). His proof was based on the observation that the usual Los
Lemma

n<
M
n
/F [= n < : M
n
[= F
1
This quantier is usually denoted by Q
1
.
2
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


for ultralters F on and rst order sentences can be proved for
L(

) if
0
is small for . The
0
-compactness follows from the Los Lemma
immediately.
Vaught [27] proved
0
-compactness of L(

1
) by proving what is now
known as Vaughts Two-Cardinal Theorem and Chang [5] extended this to
L(

+
) by proving (
1
, ) (
+
, ), when
<
= . Jensen [9] extended
this to all under the assumption GCH+2

, which he showed to follow


from V = L. Keisler [11] proved with a dierent method
0
-compactness of
L(

) when is a singular strong limit cardinal. This led to the important


observation that if there are no inaccessible cardinals in L, then L(

) is
recursively axiomatizable and
0
-compact for all > . Also, if GCH holds,
then L(

) is
0
-compact for all > . We still do not know if this is
provable in ZFC:
Open Problem: Is it provable in ZFC that L(

) is
0
-compact for all
> ? In particular, is it provable in ZFC that L(

2
) is
0
-compact?
The best result today towards solving this problem is:
Theorem 1 ([26]) It is consistent, relative to the consistency of ZF that
L(

1
,

2
) is not
0
-compact.
and every regular cardinal is a successor cardinal (i.e. there are no weakly
inaccessible cardinals),
Our approach is to look for ZFC-provable relationships between complete-
ness, recursive compactness and countable compactness in the context of a
particular logic in the hope that such relationships would reveal important
features of the logic even if we cannot settle any one of these properties per se.
For example, the countable compactness of the logic L

1
,

2
,

3
, . . .)
cannot be decided in ZFC, but we prove in ZFC that if this logic is recur-
sively compact, it is countably compact. We show by example that recursive
compactness does not in general imply countable compactness.
2 Logic Frames
Our concept of a logic frame captures the combination of syntax, semantics
and proof theory of an extension of rst order logic. This is a very gen-
eral concept and is not dened here with mathematical exactness, as we do
3
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


not prove any general results about logic frames. All our results are about
concrete examples.
Denition 2 1. A logic frame is a triple L

= L, [=
L
, /), where
(a) L, [=
L
) is a logic in the sense of Denition 1.1.1 in [4].
(b) / is a class of L

-axioms and L

-inference rules. We write


A

if is derivable using the axioms and rules in /, and call a set
T of L

-sentences /-consistent if no sentence together with its


negation is derivable from T.
2. A logic frame L

= (L, [=
L
, /) is recursive if
(a) There is an eective algorithm which gives for each nite vocabu-
lary the set L[] and for each L[] a second order formula
2
which denes the semantics of .
(b) There is an eective algorithm which gives the axioms and rules
of /.
3. A logic frame L

= L, [=
L
, /) is a , )-logic frame, if each sen-
tence contains less than predicate, function and constant symbols,
and [L[][ whenever the vocabulary has less that symbols alto-
gether.
4. A logic frame L

= L, [=
L
, /) is:
(a) complete if every nite /-consistent L

-theory has a model.


(b) recursively compact if every /-consistent L

-theory, which is re-


cursive in the set of axioms and rules, has a model.
(c) (, )-compact if every L

-theory of cardinality , every subset


of cardinality < of which is /-consistent, has a model.
(d) countably compact, if it is (, )-compact.
2
Second order logic represents a strong logic with an eectively dened syntax. It is
not essential, which logic is used here as long as it is powerful enough.
4
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


Note that
countable compactness

recursive compactness

completeness
The weakest condition is thus completeness. We work in this paper al-
most exclusively with complete logic frames investigating their compactness
properties.
Denition 3 A logic frame L

= L, [=
L
, /) has
1. nite recursive character if for every possible universe
3
V

[= (L

is complete L

is recursively compact).
2. nite character if for every possible universe V

V

[= (L

is complete L

is countably compact).
3. recursive character if for every possible universe V

V

[= (L

is recursively compact L

is countably compact).
Finite (recursive) (, )-character means nite (respectively recursive) char-
acter with countably compact replaced by (, )-compact. Strong char-
acter, means (, )-character for all .
Example 4 Let
L(

+
) = L(

+
), [=
L(

+
)
, /(

+
)),
where
M[=

x(x, y) [x : M[= (x, y)[


3
I.e. inner model or forcing extension.
5
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


and /(

+
) has as axioms the basic axioms of rst order logic and

+
(x = y x = z)
x( ) (

+
x

+
x)

+
x(x)

+
y(y), where (x, ...) is a formula of
L(

+
) in which y does not occur

+
yx x

+
y

+
xy,
and Modus Ponens as the only rule. The logic L(

+
) was introduced by
Mostowski [18] and the above frame for =
0
by Keisler [12]. The logic
frame L(

+
) is an eective , )-logic frame. The logic frame L(

1
) is
countably compact, hence has nite character for a trivial reason. If =
<
,
then by Changs Two-Cardinal Theorem, L(

+
) is countably compact, in
fact (, )-compact ([22]). If V=L, then L(

+
) is (, )-compact for all
(Jensen [9]).
Example 5 Suppose is a singular strong limit cardinal. Let
L(

) = L(

), [=
L(

)
, /(

)),
where /(

) has as axioms the basic axioms of rst order logic, a rather


complicated set of special axioms (no simple set of axioms is known at present),
and Modus Ponens as the only rule. The logic frame L(

) is (, )-compact
for each < ([11],[22]).
Example 6 Suppose is strong limit -Mahlo
4
cardinal. Let
L(

) = L(

), [=
L(

)
, /(

)),
where /(

) has as axioms the basic axioms of rst order logic, axioms


given in [20], and Modus Ponens as the only rule. The logic frame L(

)
is (, )-compact for each < ([21],[22]).
Example 7 Suppose is a regular cardinal. Let
L(Q
cof

) = L(Q
cof

), [=
L(Q
cof

)
, /(Q
cof

)),
4
is 0-Mahlo if it is regular, (n + 1)-Mahlo, if there is a stationary set of n-Mahlo
cardinals below , and -Mahlo if it is n-Mahlo fo all n < .
6
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


where M[= Q
cof

xy(x, y, z) if and only if x, y) : M[= (x, y, z) is a linear


order of conality , and /(Q
cof

) has as axioms the basic axioms of rst order


logic, the axioms from [24], and Modus Ponens as the only rule. The logic
frame L(Q
cof

) is fully compact i.e. (, )-compact for all , hence has nite


character for a trivial reason. ([24]).
Example 8 Let
L(Q
cub

1
) = L(Q
cub

1
), [=
L(Q
cub

1
)
, /(Q
cub

1
)),
where M[= Q
cub

1
xy(x, y, z) if and only if x, y) : M[= (x, y, z) is an
1
-
like linear order in which a cub of initial segments have a sup, and /(Q
cub

1
)
has as axioms the basic axioms of rst order logic, and axioms from [3]. The
logic frame L(Q
cub

1
) is countably compact ([24]), hence has nite character
for a trivial reason. We shall give explicit axioms for the logic frame later.
Example 9 Magidor-Malitz quantier logic frame is
L(Q
MM

) = L(Q
MM

), [=, /
MM

),
where
Q
MM

xy(x, y, z) X([X[ X X x, y) : (x, yz))


and /
MM

is the set of axioms and rules introduced by Magidor and Malitz in


[15]. The logic frame L(Q
MM

+
) is an eective , )-logic frame. The logic
frame L(Q
MM

+
) is complete, if we assume 3, 3

and 3

+, but there is a
forcing extension in which L(Q
MM

1
) is not countably compact [1].
Example 10 Let
L

= L

, [=
L

, /

),
where /

has as axioms the obvious axioms and Changs Distributive Laws,


and as rules Modus Ponens, Conjunction Rule, Generalization Rule and the
Rule of Dependent Choices from [10]. This an old example of a logic frame
introduced by Tarski in the late 50s and studied intensively, e.g. by Karp
[10]. The logic frame L

is a

, )-logic frame. It is eective and (, )-


compact for all , if = = . It is complete, if =
1
, = . The logic
7
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


frame L

is complete also if
1. =
+
and
<
= , or
2. is strongly inaccessible, or
3. is weakly inaccessible, is regular and
( < )( < )(

< )[10],
although in these cases the completeness is not as useful as in the case as L

and L

. L

is not complete if = is a successor cardinal (D.Scott, see


[10]). L

is not (, )-compact unless is weakly compact, and then also


L

is (, )-compact. L

is not strongly compact unless is and then also


L

is. The logic frame L

is not of nite (, )-character, unless = ,


since it is in some possible universes complete, but not (, )-compact.
Example 11 Let
0
< < be compact cardinals. A sublogic L
1
of L

,
extending L(

), is dened in [8]. This logic is like L

in that it allows
quantication over sequences of variables of length < , but instead of con-
junctions and disjunctions of length < , the logic L
1
allows conjunctions and
disjunctions over sets of formulas indexed by a a set in a -complete ultral-
ter on a cardinal < . The logic L
1
is [, < )-compact, [, < )-compact,
has the Interpolation property and other nice properties.
The denition of logic frames leaves many details vague, e.g. the exact
form of axioms and rules. Also the conditions of a recursive logic frame would
have to be formulated more exactly for any general results. Going into such
details would take us too much astray from the main purpose of this paper.
3 Logics with recursive character
We now investigate the following quite general question involving an innite
sequence (
n
)
n<
of uncountable cardinals:
Question: For which sequences (
n
)
n<
of uncountable cardinals is the logic
L(
n
)
n<

0
-compact?
As the preceding discussion indicates we cannot expect a general solution
in ZFC. Extreme cases are
1.
n
=
1
for all n < .
8
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


2.
0
is small for each
n
.
3. Some
n
is the supremum of a subset of the others.
in which case we have a trivial solution (in case 2 we have Los Lemma and
therefore
0
-compactness, and in case 3 we have an easy counter-example to

0
-compactness).
Let us call a logic recursively compact if every recursive set of sen-
tences, every nite subset of which has a model, itself has a model. Naturally
this concept is meaningful only for logics which possess a canonical G odel
numbering of its sentences. Let us call a logic recursively axiomatizable
if the set of (G odel numbers of) valid sentences of the logic is recursively
enumerable. By a result of Per Lindstr om [14] any recursively axiomatizable
logic of the form L(
n
)
nm
is actually recursively compact. This raises the
question:
Question: For which sequences (
n
)
n<
of uncountable cardinals is the logic
L(
n
)
n<
recursively axiomatizable?
We give an axiomatization / of L(
n
)
n<
. We do not know in general
whether this / is recursive (or r.e.). We give a combinatorial characteri-
zation of sequences (
n
)
n<
for which the logic frame (L(
n
)
n<
, [=, /) is
complete.
In the presence of an axiomatization / we can redene our compactness
properties. Rather than requiring that every nite subtheory has a model we
can require that every nite subtheory is /-consistent in the sense that no
contradiction can be derived from it by means of the axioms and rules of /.
It turns out that this change is not signicant in the sense that in our main
result we could use either. However, this modied concept of compactness
reveals an interesting connection between completeness and compactness:
we can think of completeness (every consistent sentence has a model) as a
compactness property of one-element theories. In this sense recursive com-
pactness is a strengthening of completeness.
For example, if A

is the Keisler axiomatization (from [12]) for L(

1
),
then it is consistent that L(

2
), A

) is complete (follows from GCH), and


it is also consistent that L(

2
), A

) is incomplete (follows from (


1
,
0
)
(
+
, ) which is consistent by [16]). However, we know it has provably nite
character (Proposition 28).
The main result of this paper (proved in Corollary 33) is the following:
9
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


Theorem 12 Suppose (
n
)
n<
is a sequence of uncountable cardinals. There
is an axiomatization / of L(
n
)
n<
such that the logic frame L(
n
)
n<
, [=
, /) has recursive character.
It is noteworthy that the above theorem is a result in ZFC. The proof
is based on formulating a partition theoretic equivalent condition for the

0
-compactness (equivalently recursive compactness) of L(
n
)
n<
.
There is a basic reduction of generalized quantiers of the form

to rst
order logic. This was established by Fuhrken [6]. A model M, . . . , A, <, . . .)
is called -like if A, <) is a -like linear order (i.e. of cardinality with
all initial segments of cardinality < ). Fuhrken established a canonical
translation
+
of L(

) to rst order logic so that


has a model
+
has a -like model.
Thus the questions of axiomatization and
0
-compactness of L(

) were
reduced to questions of axiomatization and
0
-compactness of rst order
logic restricted to -like models.
If =
+
the reduction is slightly simpler. Then we can use (, )-
models, i.e. models M, . . . , A, . . .), where [M[ = and [A[ = . The study
of model theory of (, )-models makes, of course, sense also if ,=
+
even
if this more general case does not arise from a reduction of L(

).
There is an immediate translation of the logic L(
n
)
n<
to rst order
logic on models that have for each n < a unary predicate P
n
and a
n
-like
linear order <
n
on P
n
. Let us call such models (
n
)
n<
-like models. Mutatis
mutandis, our approach applies also to logics of the form L(
n
)
n<m
.
For easier notation we x A
n
, <
n
) such that the sets A
n
are disjoint and
for each n the structure A
n
, <
n
) is a well-order of order type
n
. We say that
a
0
, . . . , a
n
) [

n<
A
n
]
<
is increasing if its restriction to any A
m
, <
m
)
is increasing in A
m
, <
m
).
Denition 13 A triple
T = E
a
: a

n<
A
n
), A
n
, <
n
) : n < ), h
n
: n < )),
where
(E1)

Each E
a
is an equivalence relation on [

n<
A
n
]
<
such that equiva-
lent sets have the same cardinality.
10
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


(E2)

If a A
n
, the number of equivalence classes of E
a
is <
n
.
(E3)

h
n
: [

n<
A
n
]
<
A
n
.
is called a (
n
)
n<
-pattern.
Let us now try to use the pattern to construct a (
n
)
n<
-like model. Let
us assume that our starting theory T has the property that every nite subset
has a (
n
)
n<
-like model. We assume the vocabulary L of T has cardinality
< min
n
: n < . Let L

be the Skolem-expansion of L and T

the
Skolem-closure of T. Let c
a
, a

n<
A
n
, be new constant symbols. Let <
n
be the predicate symbol the interpretation of which we want to be
n
-like.
Consider the axioms
(T1)

(Skolem-closure of T).
(T2)

<
n
c

for <
n
in A
n
.
(T3)

P
n
(c
a
) for a A
n
(T4)

P
m
(t(c
a
0
, . . . , c
an
)) t(c
a
0
, . . . , c
an
) <
m
c
hm({a
0
,...,an})
,
where a
0
, . . . , a
n
) [

n<
A
n
]
<
is increasing and t is a Skolem-term.
(T5)

t(c
a
0
, . . . , c
an
) = t(c
b
0
, . . . , c
bn
) ((t(c
a
0
, . . . , c
an
) <
m
c
a
)
(t(c
b
0
, . . . , c
bn
) <
m
c
a
)) for all Skolem-terms t and all increasing
a
0
, . . . , a
n
), b
0
, . . . , b
n
) [

n<
A
n
]
<
such that
a
0
, . . . , a
n
E
a
b
0
, . . . , b
n
, whenever a A
m
.
Let be an arbitrary nite subset of (T1)

-(T5)

. Let M be a (
n
)
n<
-
like model of T

. Let D
m
be the set of a A
m
such that c
a
occurs
in . Let us expand M to a model M

by adding interpretations to all the


constants c
a
, a

n<
A
n
, in such a way that they increase in P
M
m
, <
M
m
)
with a P
M
m
and are conal in <
M
m
. The model M

and induce in a
canonical way a (
n
)
n<
-pattern
T

= E

a
: a

n<
A
n
), A
n
, <
n
) : n < ), h

n
: n < )) (1)
as follows: If a A
m
, then dene for increasing a
0
, . . . , a
n
), b
0
, . . . , b
n
)
[

n<
A
n
]
<
11
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


a
0
, . . . , a
n
E

a
b
0
, . . . , b
n

M

[= t(c
a
0
, . . . , c
an
) = t(c
b
0
, . . . , c
bn
)
((t(c
a
0
, . . . , c
an
) <
m
c
a
) (t(c
b
0
, . . . , c
bn
) <
m
c
a
))
for all Skolem-terms t occurring in .
and
h

m
(a
0
, . . . , a
n
) = minb A
m
: t(c
a
0
, . . . , c
an
)
M

<
m
c
M

b
for all Skolem-terms t occurring in
An -cardinal identity is a triple
I = E
a
: a

n<
D
n
), D
n
, <
n
) : n < ), h
n
: n < )) (2)
where
(I1) The D
m
, <
m
) are disjoint nite linear orders, D
m
= for all but nitely
many m. The cardinality of

n<
D
n
is called the size of I.
(I2) Each E
a
, a D
m
, is an equivalence relation on T(D
m
) such that equiv-
alent sets have the same cardinality.
(I3) h
m
: [

n<
D
n
]
<
D
m
is a partial function.
An example of an -cardinal identity is the restriction
TD = E
a
D : a D

n<
D
n
), D
n
, <
n
)D : n < ), h
n
D : n < ))
of (
n
)
n<
-pattern to a nite D. An -cardinal identity
I = E
a
: a

n<
D
n
), D
n
, <
n
) : n < ), h
n
: n < ))
is a subidentity of another -cardinal identity
I

= E

a
: a

n<
D

n
), D

n
, <
n
) : n < ), h

n
: n < )),
in symbols I I

, if there is an order-preserving mapping :

n<
D
n

n<
D

n
such that
12
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


(S1) D
m
: D
m
, <
m
) D

m
, <

m
) is order-preserving.
(S2) d
0
, . . . , d
n
E
a
d

0
, . . . , d

n
d
0
, . . . , d
n
E
a
d

0
, . . . , d

n
holds
for d
0
, . . . , d
n
, d

0
, . . . , d

n
[

n<
D
n
]
n
.
(S3) h
m
(d
0
, . . . , d
n
)

m
h

m
(d
0
, . . . , d
n
) if d
0
, . . . , d
n
[

n<
D
n
]
n
.
Let I(T) be the set of all subidentities of TD for nite D. We write
(
n
)
n<
(I),
if I belongs to I(T) for every (
n
)
n<
-pattern T. Let I((
n
)
n<
) be the set
of all I such that (
n
)
n<
(I), i.e.
I((
n
)
n<
) =

I(T) : T is a (
n
)
n<
-pattern.
Denition 14 A (
n
)
n<
-pattern T is fundamental if I(T) = I((
n
)
n<
).
Suppose now that there is a fundamental (
n
)
n<
-pattern T. Let us see
how we can nish the constructions of a -like model for T. We built up a
(
n
)
n<
-pattern T

from the model M

. Since T is fundamental, there is a


nite set D

such that TD T

. Thus M can be expanded to a model


of .
To sum up, we have proved the following result:
Theorem 15 If there is a fundamental (
n
)
n<
-pattern then rst order logic
on (
n
)
n<
-models is -compact for all < min
n
: n < . In particular,
then L(
n
)
n<
is -compact for all < min
n
: n < .
The question of existence of fundamental (
n
)
n<
-patterns is, of course,
quite dicult. Let us recall some earlier results obtained by means of a
construction of a fundamental pattern:
Theorem 16 1. If
0
is small for , then L(

) is -compact for all


< . [22]
2. If

= and , then rst order logic on (, )-models is -


compact. In particular, then L(

+
) is -compact. [22]
3. If

() , then rst order logic on (, )-models is -compact. [28]


13
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


4. If cf() < , singular, singular strong limit, then rst order
logic on (, )-models is recursively axiomatizable and -compact. [23]
5. If is singular strong limit, L(

) is -compact and recursively ax-


iomatizable for each < . [23] (see [19] for details)
6. If is -Mahlo, then L(

)is -compact and recursively axiomatizable


for each < . [21]
If
0
is small for each
n
, then a simple enumeration argument gives a
fundamental (
n
)
n<
-pattern.
Corollary 17 [22] If
0
is small for each
n
, then rst order logic on (
n
)
n<
-
like models is -compact for all < min
n
: n < . In particular, then
L(
n
)
n<
is -compact for all < min
n
: n < .
If each
n
is singular strong limit and no
n
is a supremum of some of
the others, then there is a fundamental (
n
)
n<
-pattern c, and I((
n
)
n<
) is
recursive and independent of the cardinals
n
[23] (see [19] for details). Thus
we have:
Corollary 18 [23] If each
n
is singular strong limit and no
n
is a supre-
mum of some of the others, then L(
n
)
n<
is -compact and recursively
axiomatizable for each < min
n
: n < .
Example 19 L(
n
)
0<n<
is -compact and recursively axiomatizable for
all <

.
Example 20 The logic L(
n
)
0<n
fails to be
0
-compact for trivial rea-
sons. Still every fragment containing only nitely many generalized quanti-
ers is
0
-compact.
If each
n
is -Mahlo, then any -pattern is fundamental.
Corollary 21 [21] If each
n
is -Mahlo, then L(
n
)
n<
is -compact and
recursively axiomatizable for each < min
n
: n < .
The results of this Section could have been proved also for a nite se-
quence (
n
)
n<m
of uncountable cardinals, with obvious modications.
14
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


3.1 The character of L(

n
)
n<
Our goal in this Section is to give the axioms / of L(
n
)
n<
and prove that
L(
n
)
n<
/) has recursive character. Since L(
n
)
n<
is the union of its
fragments L(
n
)
n<m
, where n < , we rst introduce an axiomatization of
L(
n
)
n<m
and discuss its completeness.
3.1.1 Logic with nitely many quantiers
Keisler gave a simple and elegant complete axiomatization for L(

1
) based
on a formalization of the principle that if an uncountable set is divided into
non-empty parts, then either there are uncountably many parts or one part
is uncountable. If =
<
, this works also for L(

+
), but it certainly
does not work for L(

) if is singular. Keisler gave a dierent axiom-


atization for L(

) when is a singular strong limit cardinal. We give


a general axiomatization /
m
for L(
n
)
n<m
, whatever (
n
)
n<m
is, plus a
criterion when this is complete. The question whether /
m
is a recursive ax-
iomatization remains open. In certain cases we can assert its recursiveness.
We use this axiomatization to prove the nite character of the logic frame
L(
n
)
n<m
, /
m
).
In fact, we do not give the axioms of /
m
explicitly but only give a crite-
rion for their choice. Because of the nature of this criterion the set of G odel
numbers of the axioms is recursively enumerable. The method of straight-
ening Henkin-formulas introduced by Barwise [2], could be used to turn our
criterion into an explicit, albeit probably very complicated, set of axioms.
We dened above what it means for a (
n
)
n<m
-like model to induce a
(
n
)
n<m
-pattern. If we have a model that is not necessarily (
n
)
n<m
-like,
it may fail to induce a (
n
)
n<
-pattern but it still induces some -cardinal
identities. The concept of inducing an identity is dened as follows: The
model M of a vocabulary L

c
a
: a

nN
A
n
, L containing unary
predicates P
n
, n N, and the nite set of rst order sentences in the
vocabulary of M induce the -cardinal identity
I = E
a
: a

n<
D
n
), D
n
, <
n
) : n < ), h
n
: n < ))
dened as follows: Let D
n
be the set of a A
n
for which c
a
occurs in . If
a D
m
, then dene for increasing a
0
, . . . , a
n
), b
0
, . . . , b
n
) [

n<
D
n
]
<
15
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


a
0
, . . . , a
n
E

a
b
0
, . . . , b
n

M

[= t(c
a
0
, . . . , c
an
) = t(c
b
0
, . . . , c
bn
)
((t(c
a
0
, . . . , c
an
) <
m
c
a
) (t(c
b
0
, . . . , c
bn
) <
m
c
a
)
for all Skolem-terms t occurring in .
and
h

m
(a
0
, . . . , a
n
) = minb D
m
: t(c
a
0
, . . . , c
an
)
M

<
m
c
M

b
for all Skolem-terms t occurring in ,
(or undened).
This concept is the heart of our axiom system/
m
. Suppose is a sentence
in L(
n
)
n<m
. Fuhrken introduced a reduction method by means of which
there is a rst order sentence
+
in a larger vocabulary such that has a
model if and only if
+
has a (
n
)
n<m
-like model.
Denition 22 A sentence of L(
n
)
n<m
in the vocabulary L is said to
be /
m
-consistent, if for all I I((
n
)
n<m
) and all nite
+

there
is a model M of such that M and induce I. The set /
m
of axioms of
L(
n
)
n<m
consists of all sentences of L(
n
)
n<m
for which is not
/
m
-consistent.
The denition of the axioms /
m
may seem trivial as we seem to take
all valid sentences as axioms. However, whether all valid sentences are
actually axioms depends on whether we can prove the completeness of our
axioms. Also, while there is no obvious reason why the set of valid sen-
tences should be recursively enumerable in I((
n
)
n<
), the set /
m
of axioms
certainly is.
Lemma 23 Suppose is a sentence of L(
n
)
n<m
and has a model. Then
is /
m
-consistent.
Proof. Suppose I I((
n
)
n<m
) and
+

is nite. Suppose M is a
(
n
)
n<
-like model of . Then M and induce a (
n
)
n<
-pattern T. Since
I I((
n
)
n<
), there is a nite D such that I TD. Thus M and
induce I. 2
16
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


Lemma 24 If there is a fundamental (
n
)
n<m
-pattern, then every /
m
-consistent
sentence of L(
n
)
n<m
has a model.
Proof. Suppose is an /
m
-consistent sentence of L(
n
)
n<m
. Let T be
a fundamental (
n
)
n<m
-pattern. Let T =
+
. It suces to show that the
theory (T1)

-(T5)

constructed from T and T is nitely consistent. Let


be a nite part of (T1)

-(T5)

and let D be the set of a

n<
A
n
for which
c
a
occurs in . Note, that if we let I = TD, then I I((
n
)
n<m
). By
assumption, T

has a model M such that M and induce TD. Thus


M can be expanded to a model of . 2
Proposition 25 If every /
m
-consistent sentence of L(
n
)
n<m
has a model,
then there is a fundamental (
n
)
n<m
-pattern.
Proof. Let I be an arbitrary -cardinal identity, as in (2). Let the size of
I be k. Let D
i
= d
0
, . . . , d
k
. Below s ranges over sequences s
i
: i k)
of natural numbers k. We say that
0
, . . . ,
n
is of type s if the
intersection of
0
, . . . ,
n
with D
i
has size s(i) for each i k. Consider
the following sentences of L(
n
)
n<m
in a vocabulary consisting of a unary
predicate P
i
, a binary predicate <
i
and n-ary function symbols F
n
i
and H
n
i
for each i < m, and n k. Let
I
be the conjunction of
1. P
n
, <
n
) is a
n
-like linear order for n < m,
2. F
s
i
is a function mapping sets a
0
, . . . , a
n
of type s to P
i
for n < k
and i < m.
3. The range of F
s
i
is bounded in P
i
.
4. H
s
i
is a function mapping sets a
0
, . . . , a
n
of type s to P
i
for n < k
and i < m.
5. There are no x
0
. . . x
k
of type s which would satisfy
(a) F
s
i
(x
r
0
, . . . , x
rn
) = F
s
i
(x
s
0
, . . . , x
sn
)
(b) F
s
i
(x
r
0
, . . . , x
rn
) <
i
d
a
whenever r
0
, . . . , r
n
), r
0
, . . . , r
n
) [

n<
D
n
]
<
are increasing and of type s, d
r
0
, . . . , d
rn
E
a
d
s
0
, . . . , d
sn
, a D
i
,
and
17
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


(c) x
h({dr
0
,...,drn
})

i
H
n
i
(x
r
0
, . . . , x
rn
)
whenever h(d
r
0
, . . . , d
rn
) D
i
.
Any model Mof
I
and any choice of a conal suborder A

n
, <
n
) of P
n
, <
n
)
M
of type
n
(for n < ) gives rise to a (
n
)
n<m
-pattern T

as in (1), where for


a A

i
a
0
, . . . , a
n
E

a
b
0
, . . . , b
n

If (F
n
i
)
M
(a
0
, . . . , a
n
) <
i
a or (F
n
i
)
M
(b
0
, . . . , b
n
) <
i
a then
(F
n
i
)
M
(a
0
, . . . , a
n
) = (F
n
i
)
M
(b
0
, . . . , b
n
).
and
h

i
(a
0
, . . . , a
n
) = (H
n
i
)
M
(a
0
, . . . , a
n
).
We have written into the sentence
I
the condition that I is not in I(T

).
On the other hand, if I / I((
n
)
n<m
), it is easy to construct a model of
I
.
Moreover, if I
0
, . . . , I
n
/ I((
n
)
n<m
), it is not hard to construct a model of

I
0
. . .
In
.
Let I
n
, n < , be a list of all I / I((
n
)
n<m
). Without loss of generality,
this list is recursive in /
m
. Suppose the set of valid L(
n
)
n<m
-sentences
is r.e. in /
m
. Now we use an argument (due to Per Lindstr om [14]) from
abstract model theory. Let A be a set of natural numbers which is co-r.e. in
/
m
but not r.e. in /
m
. Say,
n A k((n, k) B),
where B is recursive in /
m
. Let P be a new unary predicate symbol and
n
the rst order sentence which says that P has exactly n elements. Let T be
the theory
n

I
i
: k i((n, k) B), and C = n : T [=
n
. We
show that C A. Suppose T [=
n
. If n / A, then there is k such that
(n, k) / B. Let M be a model of
I
j
: i < k
m
. If
n

I
i
T,
then i < k, whence M [=
I
i
. So M [= T, a contradiction. Since C is r.e.
in /, there is n A C. Thus there is M [= T such that M [=
n
. Since
k((n, k) B), the sentence
n

I
i
is in T, and thereby true in M for
every i. Since M [=
n
, M [=
I
i
for all i. Let T be the (
n
)
n<m
-pattern
that M gives rise to. T is necessarily a fundamental (
n
)
n<m
-pattern. 2
18
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


Summing up:
Theorem 26 Suppose (
n
)
n<m
is a sequence of uncountable cardinals. The
following conditions are equivalent:
1. /
m
is a complete axiomatization of L(
n
)
n<m
.
2. L(
n
)
n<m
,/
m
) is recursively compact.
3. L(
n
)
n<m
is -compact for all < min
0
, . . . ,
m1
.
4. There is a fundamental (
n
)
n<m
-pattern.
Corollary 27 L(
n
)
n<m
,/
m
) has nite character.
We do not know if /
m
is recursive, except in such special cases as in
Corollaries 18 and 21.
Recall the denition of I(
+
, ) in [26] and [19].
Proposition 28 1. Suppose I(
+
, ) is recursive, and either A

is recur-
sive or there is a universe V

V in which L(

+
), A

) is recursively
compact, then L(

+
), A

) has nite character.


2. Suppose L(

1
), A

) is coherent (i.e. if a sentence has a model it is


consistent with A

). Then L(

+
), A

) has nite character.


Proof. 1. Suppose L(

+
), A

) is complete. Let L(

+
) say in
the language of set theory that
I
holds for all I / I(). Since I(
+
, ) is
recursive, this can be written in L(

+
). We show that is consistent with
the axioms A

: If A

is recursive, L(

+
), A

) is recursively compact and


there is a fundamental (
+
, )-pattern, whence is consistent with A

. On
the other hand, if there is a universe V

in which L(

+
), A

) is recursively
compact, then in V

there is a fundamental (
+
, )-pattern, and hence in V

the sentence is consistent with A

. Thus is consistent with A

also in V .
By completeness has a model. Thus there is a fundamental (
+
, )-pattern
and L(

+
), A

) is
0
-compact.
2. Completeness implies (
1
,
0
) (
+
, ). We know that I(
1
,
0
) is
recursive ([25]). Since (
1
,
0
) (
+
, ) implies I(
1
,
0
) = I(
+
, ), also
the latter is recursive. Now we use part 1. 2
Corollary 29 The logic frame L(

+
), A

), where A

is the Keisler ax-


iomatization for L(

1
) and is an arbitrary cardinal, has nite character.
19
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


3.1.2 Logic with innitely many quantiers
The axioms / are simply all the axioms /
m
m < , put together.
Proposition 30 If L(
n
)
n<
,/) is recursively compact, then there is a
fundamental (
n
)
n<
-pattern.
Proof. Let I
n
, n < , be a list of all I / I((
n
)
n<
). Without loss of
generality, this list is recursive in /. Note that if I / I((
n
)
n<
), then there
is an m such that I / I((
n
)
n<m
), so we can use the sentences
In
. Let T
be the set of all
In
, n < . This theory is recursive in / and it is nitely
consistent. Hence it has a model. The (
n
)
n<
-pattern the model M gives
rise to is clearly fundamental. 2
Theorem 31 Suppose (
n
)
n<
is a sequence of uncountable cardinals. The
following conditions are equivalent:
1. / is a complete axiomatization of L(
n
)
n<
.
2. For every m < there is a fundamental (
n
)
n<m
-pattern.
Theorem 32 Suppose (
n
)
n<
is a sequence of uncountable cardinals. The
following conditions are equivalent:
1. L(
n
)
n<
,/) is recursively compact.
2. L(
n
)
n<
is -compact for all < min
n
: n < .
3. There is a fundamental (
n
)
n<
-pattern.
Corollary 33 L(
n
)
n<
,/) has recursive character.
Example 34 If
n
=
n
for 0 < n , then L(
n
)
n<
,/) is complete
but not
0
-compact and thereby does not have nite character.
Above we investigated -like models and related them to logic frames
arising from generalized quantiers. Similar results can be proved for models
with predicates of given cardinality and also for models with a linear order in
which given predicates have given conalities, but these results do not have
natural formulations in terms of generalized quantiers.
20
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


4 A logic which does not have recursive char-
acter
We show that there is a logic frame L

which is recursively compact but not


countably compact. We make use of the quantier Q
St
from [24]. To recall
the denition of Q
St
we adopt the following notation:
Denition 35 Let A = (A, R) be an arbitrary
1
-like linearly ordered struc-
ture. We use H(A) to denote the set of all initial segments of A. A ltration
of A is subset X of H(A) such that A =

IX
I and X is closed under
unions of increasing sequences. Let D(A) be the lter on H(A) generated by
all ltrations of A.
Denition 36 The generalized quantier Q
St
is dened by
A [= Q
St
xy(x, y, a)
if and only if A = (A, R

), where R

= (b, c) : A [= (b, c, a), is an


1
-like
linear ordered structure such that
I H(A) : I does not have a sup in R

/ D(A).
Respectively, the generalized quantier Q
Cub
, denable in terms of Q
St
and

1
, is dened by
A [= Q
Cub
xy(x, y, a)
if and only if A = (A, R

), where R

= (b, c) : A [= (b, c, a), is an


1
-like
linear ordered structure such that
I H(A) : I does not have a sup in R

D(A).
It follows from [24] and [3] that L

(Q
St
) equipped with some natural
axioms and rules is a complete countably compact logic frame.
Denition 37 If S
1
, then the generalized quantier Q
St
S
is dened by
A [= Q
St
S
xy(x, y, a)
if and only if R

is an
1
-like linear order of A with a ltration I

: <
1

such that
<
1
((I

has a sup in R

) S).
21
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


The syntax of the logic
L
St
is dened as follows: L
St
extends rst order logic by the quantiers Q
1
,
Q
St
and the innite number of new formal quantiers Q
St
Xn
(we leave X
n
unspecied).
If we x a sequence S
0
, S
1
, ...) and let Q
St
Xn
be interpreted as Q
St
Sn
, we
get a denition of semantics of L
St
. We call this semantics the S
0
, S
1
, ...)-
interpretation of L
St
. Thus L
St
has a xed syntax and xed axioms, given
below, but many dierent semantics, depending on our interpretation of
X
0
, X
1
, . . .) by various S
0
, S
1
, ...).
Denition 38 We call a nite sequence = S
0
, S
1
, ...S
n
) (or an innite
sequence S
0
, S
1
, ...)) of subsets of
1
stationary independent, if all nite
Boolean combinations of the sets S
i
are stationary.
Denition 39 The axioms of / are
(Ax1) The usual axioms and rules of L(Q
1
).
(Ax2) Q
St
xy(x, y, z) (, , z) is an
1
-like linear order.
(Ax3) Q
St
Xn
xy(x, y, z) Q
St
xy(x, y, z).
(Ax4) Independence Axiom Schema: Any non-trivial Boolean combination of
the set S
n
interpreting the X
n
is stationary, i.e. , where is
the conjunction of the formulas
(a) (, , z) is an
1
-like linear order.
(b) Q
St
X
i
xy((x, y, z)
i
(x, z)
i
(y, z)), i=0,...,l.
and is the conjunction of the formulas Q
St
xy((x, y, z)

i<l
((y, z))
(i)
),
for all : l 2.
(Ax4) Pressing Down Axiom Schema:
[Q
St
xy(x, y, u) xz((z, x, u) (x, z, u))]
zQ
St
xy((x, y, u) (x, z, u) (y, z, u))
22
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


The axioms of L
St

are the above added with the usual axioms and rules
of L

.
Denition 40 Suppose S
0
, S
1
, ...) is stationary independent. We dene a
new recursive logic frame
L
St
(S
0
, S
1
, ...) = L
St
(S
0
, S
1
, ...), [=, /),
where / is as in Denition 39. Let L
St

(S
0
, S
1
, ...) be the extension of
L
St
(S
0
, S
1
, ...) obtained by allowing countable conjunctions and disjunctions.
The standard proof (see e.g. [3]) shows:
Lemma 41 1. The logic frames L
St
(S
0
, S
1
, ...) and L
St

(S
0
, S
1
, ...) are
complete for all stationary independent S
0
, S
1
, ...).
2. L
St
(S
0
, S
1
, ...) has a model in an S
0
, S
1
, ...)-interpretation for some
stationary independent S
0
, S
1
, ...) if and only if has a model in an
S
0
, S
1
, ...)-interpretation for all stationary independent S
0
, S
1
, ...).
An immediate consequence of Lemma 41 is that the set V al(L
St
) of
sentences of L
St
which are valid under S
0
, S
1
, ...)-interpretation for some
(equivalently, all) stationary independent S
0
, S
1
, ...) is recursively enumer-
able, provably in ZFC, and the predicate has a model, where L
St

,
is a
ZFC
1
-denable property of .
By making dierent choices for the stationary independent S
0
, S
1
, ...),
we can get logics with dierent properties. Clearly there is a trivial choice
of S
0
, S
1
, ...) for which L
St
fails to have countable compactness. On the
other hand, CH fails if and only if there is a choice of S
0
, S
1
, ...) which will
make L
St
countably compact. We make now a choice of S
0
, S
1
, ...) which
will render L
St
(S
0
, S
1
, ...) recursively compact but not countably compact.
Let us x a countable vocabulary which contains innitely many sym-
bols of all arities. Let T
n
, n < , list all Ax(L
St
)-consistent recursive L
St
-
theories in the vocabulary . Let
n
be a new disjoint copy of for each
n < . Let

consist of the union of all the


n
, the new binary predicate
symbol <

, and new unary predicate symbols P


n
for n < . If is a formula
and d 2, let ()
d
be , if d = 0, and , if d = 1. If S
1
, then (S)
d
is
dened similarly. For any : 2 let

L
St

be the conjunction of the


following sentences of the vocabulary

:
23
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


(a) T
n
translated into the vocabulary
n
.
(b) <

is an
1
-like linear order of the universe.
(c) Q
St
Xn
xy(x <

y P
n
(x) P
n
(y)).
(d) x

n
(P
n
(x))
(n)
.
Lemma 42 There is : 2 such that

has a model.
Proof. Let consist of the sentences (a)-(c). By Lemma 41, has a model
M of cardinality
1
in the S
0
, S
1
, ...)-interpretation for some stationary in-
dependent S
0
, S
1
, ...). Get a new : 2 by Cohen-forcing. Then in the
extension V []

n
(S
n
)
(n)
= .
Thus V [] satises the
1
-sentence
(

has a model). (3)


By the Levy-Shoeneld Absoluteness Lemma and Proposition 41 there is
in V such that (3) holds in V . 2
Now let S

0
, S

1
, ...) be stationary independent such that

has a model
M

in the S

0
, S

1
, ...)-interpretation.
Theorem 43 The recursive logic frame L
St
(S

0
, S

1
, ...) is recursively compact
but not countably compact.
Proof. Suppose T is a consistent recursive theory in L
St
. W.l.o.g. T = T
m
for some m < . Thus M


n
gives immediately a model of T. To
prove that L
St
is not countably compact, let T be a theory consisting of the
following sentences:
(i) <

is an
1
-like linear order.
(ii) Q
St
S

n
xy(x <

y P
n
(x) P
n
(y)) for n < .
(iii) Q
St
xy(x <

y P(x) P(y)).
(iv) x(P(x) (P
n
(x))
(n)
) for n < .
24
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


Any nite subtheory of T contains only predicates P
0
, ..., P
m
for some m, and
has therefore a model: we let P
i
= S

i
for i = 0, ..., m and
P = (P
0
)
(0)
... (P
m
)
(m)
.
On the other hand, suppose A, <

, P, P
0
, P
1
, ...) [= T. By (ii) there are
ltrations D
n

: <
1
) of <

and clubs E
n
such that for all n and for all
E
n
<
1
: D
n

has a sup in A, <

) = S

n
.
By (iii) there is a ltration F

: <
1
) of <

such that
B = <
1
: F

has a sup in P
is stationary. Let E



n
E
n
be a club such that C

= D
n

= F

for
E

and n < . Let E

B and a = sup F

. Then a P. Hence
a

n
(P
n
)
(n)
by (iv). As a = sup D
n

for all n, we have a



n
(S

n
)
(n)
,
contrary to the choice of . We have proved that theory T has no models. 2
Thus L
St
does not have nite character. We end with an example of a
logic which, without being provably complete, has anyhow nite character:
Recall that
S
for S
1
is the statement that there are sets A

,
S, such that for any X
1
, the set S : X = A

is stationary.
Denition 44 Let L

be the extension of L

by Q
1
, Q
St
and Q
St
S
, where
o =

, if there is no bistationary S with


S

1
, if there is a bistationary S with
S
but no maximal one
S, if S is a maximal bistationary S with
S
We get a recursive logic frame L

= L

, [=, /) by adapting the set Ax(L


St
)
to the case of just one bistationary set.
Theorem 45 L

has nite character.


Proof. Suppose there is no bistationary S with
S
. Then the consistent
sentence < is an
1
-like linear order Q
St
xy(x < y) Q
St
S

(x < y) has no
model, so L is incomplete. Suppose there is a bistationary S with
S
but no
maximal one. Then the consistent sentence < is an
1
-like linear order
Q
St
xy(x < y P(x)) Q
St
S

(x < y P(x)) has no model, so L is again


incomplete. Finally, suppose there is a maximal bistationary S with
S
.
Now L is countably compact by an analog of Lemma 41. 2
25
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


Our results obviously do not aim to be optimal. We merely want to indi-
cate that the concept of a logic frame oers a way out of the plethora of inde-
pendence results about generalized quantiers. The logic L

n+1
)
n<
is
a good example. The results about its countable compactness under CH and
countable incompactness in another model of set theory leave us perplexed
about the nature of the logic. Having recursive character reveals something
conclusive and positive, and raises the question, do other problematic logics
also have recursive character. Our logic L
St
is the other extreme: it is al-
ways completely axiomatizable, but a judicious choice of S
0
, S
1
, ...) renders
it recursively compact without being countably compact.
Open Question: Does the Magidor-Malitz logic L(Q
MM
1
) have recursive
character?
L(Q
MM
1
) is countably compact whenever holds ([15]). But L(Q
MM
1
)
may fail to be countably compact ([1]). The question is whether L(Q
MM
1
) is
countably compact in every model in which it is recursively compact.
References
[1] Uri Abraham and Saharon Shelah. A
2
2
well-order of the reals and
incompactness of L(Q
MM
). Annals of Pure and Applied Logic, 59:132,
1993.
[2] Jon Barwise. Some applications of Henkin quantiers. Israel J. Math.,
25(1-2):4763, 1976.
[3] Jon Barwise, Matt Kaufmann, and Michael Makkai. Stationary logic.
Ann. Math. Logic, 13(2):171224, 1978.
[4] J. Barwise and S. Feferman, editors. Model-theoretic logics. Springer-
Verlag, New York, 1985.
[5] C. C. Chang. A note on the two cardinal problem. Proc. Amer. Math.
Soc., 16:11481155, 1965.
[6] G. Fuhrken. Skolem-type normal forms for rst-order languages with a
generalized quantier. Fund. Math., 54:291302, 1964.
26
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


[7] Gebhard Fuhrken. Languages with added quantier there exist at least

. In Theory of Models (Proc. 1963 Internat. Sympos. Berkeley), pages


121131. North-Holland, Amsterdam, 1965.
[8] Wilfrid Hodges and Saharon Shelah. There are reasonably nice logics.
J. Symbolic Logic, 56:300322, 1991.
[9] R. Bj orn Jensen. The ne structure of the constructible hierarchy. Ann.
Math. Logic, 4:229308, 1972. With a section by Jack Silver.
[10] Carol R. Karp. Languages with expressions of innite length. North
Holland Publishing Co., Amsterdam, 1964.
[11] H. J. Keisler. Models with orderings. In Logic, Methodology and Philos.
Sci. III (Proc. Third Internat. Congr., Amsterdam, 1967), pages 3562.
North-Holland, Amsterdam, 1968.
[12] H. Jerome Keisler. Logic with the quantier there exist uncountably
many. Ann. Math. Logic, 1:193, 1970.
[13] Per Lindstr om. First order predicate logic with generalized quantiers.
Theoria, 32:186195, 1966.
[14] Per Lindstr om. On extensions of elementary logic. Theoria, 35:111,
1969.
[15] Menachem Magidor and Jerome Malitz. Compact extensions of L(Q).
Ia. Ann. Math. Logic, 11(2):217261, 1977.
[16] William Mitchell. Aronszajn trees and the independence of the transfer
property. Ann. Math. Logic, 5:2146, 1972/73.
[17] Michael Morley. Partitions and models. In Proceedings of the Summer
School in Logic (Leeds, 1967), pages 109158, Berlin, 1968. Springer.
[18] A. Mostowski. On a generalization of quantiers. Fund. Math., 44:1236,
1957.
[19] J. H. Schmerl. Transfer theorems and their applications to logics. In
Model-theoretic logics, Perspect. Math. Logic, pages 177209. Springer,
New York, 1985.
27
7
9
0


r
e
v
i
s
i
o
n
:
2
0
0
4
-
1
0
-
2
1







m
o
d
i
f
i
e
d
:
2
0
0
4
-
1
0
-
2
1


[20] J. H. Schmerl. An axiomatization for a class of two-cardinal models. J.
Symbolic Logic 42:174178, 1977.
[21] James H. Schmerl and Saharon Shelah. On power-like models for hy-
perinaccessible cardinals. J. Symbolic Logic, 37:531537, 1972.
[22] Saharon Shelah. Two cardinal compactness. Israel J. Math., 9:193198,
1971.
[23] Saharon Shelah. On models with power-like orderings. J. Symbolic Logic,
37:247267, 1972.
[24] Saharon Shelah. Generalized quantiers and compact logic. Transac-
tions of the American Mathematical Society, 204:342364, 1975.
[25] Saharon Shelah. Appendix to: Models with second-order properties.
II. Trees with no undened branches (Ann. Math. Logic 14(1978), no.
1, 7387). Ann. Math. Logic, 14:223226, 1978.
[26] Saharon Shelah. The pair (
n
,
0
) may fail
0
compactness. In Proceed-
ings of LC2001, volume submitted of Lecture Notes in Logic. ASL.
[27] R. L. Vaught. The completeness of logic with the added quantier there
are uncountably many. Fund. Math., 54:303304, 1964.
[28] R. L. Vaught. A L owenheim-Skolem theorem for cardinals far apart.
In Theory of Models (Proc. 1963 Internat. Sympos. Berkeley), pages
390401. North-Holland, Amsterdam, 1965.
shelah@math.huji.ac.il
jouko.vaananen@helsinki.fi
28