Solution Hints to the Exercises

from
A Concise Introduction to
Mathematical Logic
by Wolfgang Rautenberg
Third Edition, Springer Science
+
Business Media
Last version September 2010
Hints to the exercises can as a rule easily be supplemented to a complete
solution. Exercises that are essential for the text are solved completely.
The reader may mail improved solutions to the author whose website is
www.math.fu-berlin.de/~raut.
1
2 Solution hints to the Exercises
Section 1.1
1. (a): x
k
is ctional in f i a
k
= 0. (b): Because of the uniqueness,
2
n+1
(= number of subsets of 0, . . . , n) is the number sought for. (c):
induction on formulas in , + and p
1
, . . . , p
n
.
2. Consider on F the property c: is prime or there are , F with
= or = ( ) where = or = . Formula induction
shows c for all F.
3. Verify by induction on the property c: no proper initial segment
of is a formula nor can be a proper initial segment of a formula.
Induction step: Case = . Then a proper initial segment of
either equals (hence is not a formula), or has the form where is
a proper initial segment of . Thus / F by the induction hypotheses,
hence also / F (since a formula starting with must have the form
for some formula by Exercise 2). Case = ( ). Let be a
proper initial segment of or conversely. Assume that is a formula
so that = (

), some

F (Exercise 2). Then ,=

, for
otherwise necessarily = . Hence

is a proper initial segment of

or conversely, a contradiction to the induction hypothesis c.
4. Assume that ( ) = (

), hence =

. If ,=

then
is a proper initial segment of

or conversely. This is impossible by

Exercise 3. Consequently =

, hence =

and =

.
Section 1.2
1. w((p q
1
) (p q
2
)) = 0 i wp = 1, wq
1
= 0 or wp = 0, wq
2
= 0,
and the same condition holds for w(pq
1
pq
2
) = 0. In a similar
way the second equivalence is treated.
2. p p + 1, 1 p +p, p q p +q, p +q p q (p q).
3. Induction on the F
n
0, 1, , (= set of formulas in 0, 1, , and
p
1
, . . . p
n
). If f, g B
n
are monotonic then so is a faga, where is
or . For simplicity, treat rst the case n = 1. Converse: Induction
on the arity n. Clear for n = 0, with the formulas 0 and 1 representing
the two constants. With f B
n+1
also f
k
: x f(x, k) is monotonic
(k = 0, 1). Let
k
F
n
0, 1, , represent f
k
(induction hypothesis).
Then
0
(
1
p
n+1
) represents f. Note that w
0
w
1
for all w.
Solution hints to the Exercises 3
4. By Exercise 3, a not representable f B
n+1
is not monotonic in
the last argument, say. Then f(a, 1) = 0 and f(a, 0) = 1 for some
a 0, 1
n
, hence g : x f(a, x) is negation. This proves the claim.
Section 1.3
1. (a): MP easily yields p q r, p q, p r. Apply (D) three times.
2. The deduction theorem yields ( ) ( ) ( ).
3. Assume that w X, . Then clearly w X, or w X, .
5. Let X / X Then X, for each . Thus, X by (T).
Section 1.4
1. X[ Y X
0
, . . . ,
n
, some
0
, . . . ,
n
Y .
Hence X (
_
in

i
) , or equivalently, X
_
in
a
i
. This all is
easily veried if is replaced by .
2. Supplement Lemma 4.4 by proving X X or X .
3. Choose X, such that X and X

. Let Y X be
maximally consistent in . Dene by p

= for p Y and p

=
otherwise. Induction on yields with the aid of ( ) and () page 28
() Y

; / Y

.
In proving (), , , (), and ()
are needed which easily follow from the rules. By (),

, hence

. Clearly Y

(i.e.,

for all Y ), and so

. But
Y

(substitution invariance). Thus,

. Therefore

for all
by (1), so that is maximal by denition.
4. There is a smallest consequence relation with the properties ( 1)(2),
namely the calculus of this section. Since and is already
maximal according to Exercise 3, and must coincide.
Section 1.5
1. For nite M easily shown by induction on the number of elements of
M. Note that M has a maximal element. General case: Add to the
formulas in Example 1 the set of formulas p
ab
[ a
0
b.
4 Solution hints to the Exercises
2. : Assume M, N / F. Then
\
(M N) =
\
M
\
N F, because
M, N F. Therefore M N / F. : M F implies M N F
by condition (b). For proving () from () observe that M
\
M F.
3. : Let U be trivial, i.e., E U for some nite E I. Induction on
the number of elements in E and Exercise 2 easily show that i
0
U
for some i
0
E. The converse is obvious.
Section 1.6
1. First verify the deduction theorem, which holds for each calculus with
MP as the only rule and A1, A2 among the axioms; cf. Lemma 6.3.
X is consistent i X , for X X ( ) = by
A1, hence X by A3. Now prove X i X X ,
provided X is maximally consistent. This allows one to proceed along
the lines of Lemma 4.5 and Theorem 4.6.
2. Apply Zorns lemma to H := Y X [ Y . Note that if K H is
a chain then

K H due to the nitarity of .
3. (a): Such a set X satises () : X for all . For otherwise
X, , hence X ( ) , and so X by Peirces
axiom. Suppose / X. Then X, , by (), and so X, .
(b): With (a) easily follows X i X X as in
Exercise 1. Proceed with an adaptation of Lemma 4.5.
4. Crucial for completeness is the proof of (m): by
induction on the rules of . (m) implies (M): X, X, ,
proving rst that a calculus based solely on unary rules satises
X for some X. E.g., .
Although () () and conversely, it is still tricky to show that
() (). (M) implies X, & X, X, ,
because X, X, and X, , therefore
X, . From this it follows []: X X or X ,
provided X is -maximal, for note that
X & X X, & X, X, X .
Having [] one may proceed with a slight modication of Lemma 4.5.
Solution hints to the Exercises 5
Section 2.1
1. There are 10 essentially binary Boolean functions f. The corresponding
algebras (0, 1, f) split into 5 pairs of isomorphic ones. For example,
(0, 1, ) (1, 0, ).
2. : Choose c = a in a b & a c b c to get a b b a.
3. For simplicity, treat rst the case n = 2 using transitivity.
5. For simplicity, let the signature contain only the symbols r, f, both
unary. Then ra ra
j
rha and hfa = h(fa
i
)
iI
= fa
j
= fha.
Section 2.2
1. Trivial if t is a prime term. A terminal segment of f

t either equals
f

t or has the form t

k
t
k+1
t
n
for some k n (t

k
t
k+1
t
n
means t

n
in case k = n), where t

k
a terminal segment of t
k
. By the induction
hypotheses, t

k
is a term concatenation, hence so is t

k
t
k+1
t
n
.
2. It suces to prove (a) t = t

t = t

, for all t, t

T , all ,

S
L
by induction on t. This is obvious for prime t. Let t = ft
1
t
n
and
t = t

with t

= f

1
t

m
. Then clearly f = f

and m = n, hence
t
1
t
n
= t

1
t

. Thus t
1
= t

1
and t
2
t
n
= t

2
t

by the
induction hypothesis for t
1
. Similarly, t
2
= t

2
. . . , t
n
= t

n
and also
=

. This proves (a).

3. (a): Similar to Exercise 3 in 1.1. (b) follows readily from (a). (c): If
L then by (b), = for some L. Hence = . Similarly,
, L = , some , L, hence = and = .
5. Can completely be reduced to Corollary 1.2.2 by some bijection from
X onto a set V of propositional variables.
Section 2.3
1. If / X and x / free X then /
a
x
X for each a (Theorem 2.3.1).
2. x( ), x , and Exercise 1.
3. The Theorems 3.1 and 3.5 yield / [a] /

[a] /

x
(a).
4. (a):
n

m

m
for n m, and
n

m

0
( ) for n m.
(b): Exercise 5 in 2.2, and
n

m

_
nk<m

=k
for n < m.
6 Solution hints to the Exercises
Section 2.4
1. x( ) ( )

t
x
_
=

t
x


t
x
_
.
3. W.l.o.g. y

(x, y) and z

(x, z) with disjoint tuples x, y, z.

4. Simultaneous induction on and . Clear if is prime. If the claim
holds for , then also for () and () ( ). The step
for is similar. Step for : Simply observe that .
5. x(Px yPy) xPx yPy according to (10) in 2.4.
Section 2.5
1. Proof very similar to that of Exercise 6 in 2.4
2. : S
t
x
S,
t
x
and (e) page 79. : (9) in 2.4.
3. T + T, T by the deduction theorem.
Section 2.6
1. The if part follows as Theorem 6.1 because y === =f

t
T
f

f
(

t , y)). The
only if part: y === =f

t
T
f

f
(

t , y) and T
f
x!y y === =fx. Hence also
T
f
x!y (x, y).
2. ^ x=== =0 y x ,=== = Sy. Careful calculation conrms the denition
x + y === =z x=== =y === =z === =0 z ,= 0S(x z) S(y z) = == =S(z
2
S(x y)).
Therein z
2
denotes the term z z.
3. Let xy === =xz === =e ( not written). Choose some y

with yy

=== =e. Then

yx=== =(yx)(yy

) = == =y(xy)y

=== =yey

=== =e and so ex=== =(xy)x=== =x(yx) = == =xe=== =x

for all x. In other words, e is a left and right unit element. We hence
obtain y === =ye=== =y(xz) = == =(yx)z === =ez === =z. For the additional claim derive
the axioms of T
=== =
G
from those of T
G
and conversely.
4. If < were denable then < would be invariant under automorphisms
of (Z, 0, +). This is not the case for the automorphism n n. This
approach to the problem is also called Padoas method.
Solution hints to the Exercises 7
Section 3.1
1. Let X
t
x
. Then X, x
t
x
,
t
x
. Hence X, x x.
Also X, x x (= x). Thus X x according to (2).
2. Let

:=
y
x
, u / var , u ,= y. Then x
u
y
(=
u
x
) by (1).
Hence x y

by (2), with X = x,

for , and y for x.

3. y(
y
x
) x z(
z
x
) according to Exercise 2.
4. : X X, X . : X X X, .
Section 3.2
1. First prove () T x i T

t
x
for all

t T
n
0
( L open); use
Theorem 2.3.5. Next prove
_

_
X T ( L
0
open) by
induction on , ; observe that L is === =-free. Let X x ( open)
and

t T
n
0
. Then also X :=

t
x
, hence T by
_

_
. Thus,
T x by (), and so T U.
2. K T for some T K (niteness theorem)
4. (i)(ii): (12) in 2.4. Observe also (x=== =t )
t
x

t
x
.
Section 3.3
1. Prove
PA
z(x + y) + z === =x + (y + z) by induction on z. Obvious for
z = 0. The induction step follows easily from
PA
x + Sy === =S(x + y).
Most proofs of the arithmetical laws in PA need much patience.
2. z +x=== =x z === =0 (induction on x) readily yields x y x x=== =y.
3. Informally: x < y z Sz + x=== =y z z + Sx=== =y Sx y.
The converse Sx y x < y follows from
PA
x < Sx. The induction
hypothesis of x y y x may be written as x < y y x. If x < y
then Sx y, hence Sx y y Sx (induction claim). We get the
same in the case y x, since then y Sx ( is transitive).
4. (a): Put := (y<x)
y
x
. It suces to prove (i) x( )
PA

0
x
(which is trivial) and (ii) x( )
PA

Sx
x
since by IS then
x( )
PA
x
PA
x. Now, ,
PA

PA

Sx
x
,
hence x( )
PA

Sx
x
which conrms (ii). (b): Follows from
8 Solution hints to the Exercises
(a) by contraposition. (c): For := (x<v)y z(x<v)(y<z)
holds
PA

0
v
, and
PA

Sv
v
. This yields the claim by IS.
Section 3.4
1. T v
i
,=== =v
j
[ i ,= j is satisable because each nite subset is.
2. Th/ v
n+1
< v
n
[ n N has a model with a descending -chain.
3. If / T then T has a completion T

with T

, hence / T

.
4. Consider the identical operator on the universe V and restrict it to a
given set u in AS.
5. Informally: Suppose n(a),
x
(), and ue(
x
(u)
x
(u e)).
Then holds also s(u s)(a\u ,=== = (e a\u)u e s) for the
set s := u P
a [
x
(u). Hence a s, i.e.
x
(a).
Section 3.5
2. Let T +
i
[ i N be an innite extension of T. We may assume
_
in

i

T

n+1
. Hence, T +
_
in

n+1
is consistent. Let T
n
be
a completion of T +
_
in

n+1
. Then T
n
,= T
m
. Thus, a theory
with nitely many completions cannot have an innite extension and,
in particular, no innite completion.
3. Let T
0
, . . . , T
n
be the completions of T. According to Exercise 3 in 3.4,
T i T
i
for all i n. Thus, T is decidable provided each T
i
is, and this follows from Theorem 5.2, for each T
i
is a nite extension
of T according to Exercise 2, hence is axiomatizable as well.
4. Starting with a eective enumeration (
n
)
nN
of L
0
, a Lindenbaum
completion of T as constructed in 1.4 is eectively enumerable.
5. According to Exercise 3 in 3.4, there is a bijection between the set
of consistent extensions of T (including T) and the set of nonempty
subsets of the collection T
1
, . . . , T
n
of all completions of T.
Solution hints to the Exercises 9
Section 3.6
1. x=== =y xx=== =y. The same holds for
|
, since
|
.
2. (a): Let (
n
)
nN
and (/
n
)
nN
be eective enumerations of all sentences
and of all nite T-models (up to isomorphism). In step n write down all

i
for i n with /
n

i
. (b): Let (
n
)
nN
and (
n
)
nN
be eective
enumerations of sentences provable or refutable in T, respectively. Each
L
0
occurs in one of these sequences. In the rst case is T.
3. Condition (ii) from Exercise 2 is then granted because the validity of
only nitely many axioms has to be tested in a nite structure.
Section 3.7
1. For H: Let B = h/ be a homomorphic image of /, w: Var A, and
dene hw: Var B by x
hw
:= hx
w
. Then ht
A,w
= t
B,hw
for all
terms t and there is some w: Var A with hw = w

for any given

w

: Var B. For S: (3) in 2.3 page 66. For P: Set B =

iI
/
i
.
Then t
B,w
= (t
A
i
,w
i
)
iI
with x
w
= (x
w
i
)
iI
.
Section 3.8
1. (a): Let
unc
in L
II
formalize the sentence there is a continuous order.

unc
has no countable model. In L
1
Q
one may take

Oxx=== =x for
unc
.
(b): X = i ,=== =j [ i, j I, i ,= j

Oxx=== =x has no model if I is

uncountable, although each nite subset of X has a model.
2. Dene R as a continuously ordered set with a countable dense subset.
4. Let x be a variable not in P, Q. A possible denition is provided by
x:=== = 0 ; WHILE x=== =0 DO P; x:=== = S0 END.
Section 4.1
1. Note that

t in case k = 0 is dened for ground terms only.
2. The most important case is k = 0. It deals with ground terms only.
10 Solution hints to the Exercises
Section 4.2
1. First prove (a) (iI)/
i
[w
i
] B [w] (x
w
= (x
w
i
)
iI
), prime
and B =

iI
/
i
. Then prove (b) (iI)/
i
[w
i
] B [w] by
induction over basic Horn formulas as in Theorem 2.1. (b) yields the
induction steps over , , . Observe t
B,w
= (t
A
i
,w
i
)
iI
. For a universal
Horn theory apply (ii)(i) of Theorem 2.3.2.
2. A set of positive Horn formulas has the trivial (one-element) model.
Section 4.4
1. With w
1
p
1
, p
3
, p
2
and w
2
p
2
, p
3
, p
1
we have w
1
, w
2
P. Since
w P implies w p
3
and either w p
1
or w p
2
, there is no valuation
w w
1
, w
2
such that w P.
2. For arbitrary w P, w p
m,n,m+n
follows inductively on n. Hence
w
s
w
P
, and consequently w
s
= w
P
.
3. (a): Theorem 4.2. (b): w
P
p
n,m,k
if k ,= n+m, so P, p
n,m,k

HR
.
Section 4.5
2. : x
i
var t
j
x

j
= t
j
,= t

j
= x

2
j
, hence ,=
2
. : t

i
= t
i
since
necessarily x

= x for all x var t

i
.
3. Let be a unier of K
0
K
1
. Then K

0
= K

1
is a singleton. Put
x

= x

for x var K

0
and x

= x

else. Then K

0
= K

0
= K

0
since
2
= , and K

1
= K

1
. Thus, K

0
K
1
is unied by

. The
converse need not hold. Let r
2
be a binary relation symbol, f a unary
operation symbol, and 0 a constant. K
0
= r
2
fvfx and K
1
= r
2
f0v
are not uniable, but K

0
and K
1
are, with =
_
v
u
_
. Indeed, for
=
0
u
fx
v
we get K

0
= r
2
fufx

= r
2
f0fx = K

1
.
Section 4.6
1. Join T
g
and T
h
and add to the resulting program the rules
r
f
(x, 0, u) :r
g
(x, u) and r
f
(x, Sy, u) :r
f
(x, y, v), r
h
(x, y, v, u).
2. Add to the programs the rule r
f
xu :r
g
1
xy
1
, . . . , r
gm
xy
m
, r
h
yu.
Solution hints to the Exercises 11
Section 5.1
1. Let = (x), a A
n
, and / (a). Then ( (a) as well, and since
B (, also B (a).
3. Prove rst the following simple lemma: Let 0 < b < c < 1. Then there
is a strictly monotonic bijection f : [0, 1] [0, 1] (an automorphism of
the closed interval [0, 1]) such that fb = c. W.l.o.g. a
1
< < a
n
,
n 2, and b [a
1
, a
n
] irrational. Let a
k
< b < a
k+1
. W.l.o.g. we may
assume a
k
= 0 and a
k+1
= 1. Choose some c Q with b < c < 1 and
an automorphism f : [0, 1] [0, 1] with fb = c according to the above
lemma. f can be extended in a trivial way to an automorphism of the
whole of (R, <) by setting fx = x outside [0, 1].
4. W.l.o.g. AB = . It suces to show that D
el
/D
el
B is consistent.
Assume the contrary. Then there is some conjunction (

b) of members
of D
el
B and some

b B
n
such that D
el
/, (

b) . Thus, D
el
/
(

b). Since A B = , the b

1
, . . . , b
n
do not occur in A, hence
constant quantication yields D
el
/ x and so / x. But
clearly B x. a contradiction to / B.
5. (a): t
A
[ t T
G
is closed with respect to all f
A
. It is the smallest
such set and hence exhausts A. (b): By (a), we may choose to each
a A
\
G some t
a
T
G
such that D/ a=== =t
a
. Thus, T + D/ can be
regarded as a denitorial and hence conservative extension of T +D
G
/,
so that D/
T
D/
E

T
for all sentences LG.
Section 5.2
2. T
suc
IS because (N, 0, S) IS and T
suc
is complete. To prove the
no circle scheme which is equivalent to () xS
n
x ,=== =x (n 1), we
start from (#) S
n+1
x = S
n
(Sx) for every n. (#) is easily veried by
metainduction on n, while the induction schema IS is needed in order
to prove () by induction on x. Clearly, S
n
0,=== =0 by the axiom x0,=== =Sx.
From the induction hypothesis S
n
x ,=== = x we get the induction claim
S
n
(Sx) = == =S(S
n
x),=== =Sx by applying (#) and the second axiom of T
suc
.
3. Let a G T and
a
n
the element with n
a
n
= a, and
m
n
: a m
a
n
for
m
n
Q. Then G becomes the vector group of a Q-vector space. This
group is easily shown to be
1
-categorical.
12 Solution hints to the Exercises
4. Each consistent T

T is the intersection of its completions in L.

5. Each / T has a countable elementary substructure (Theorem 1.5).
Section 5.3
1. For SO
00
: In the rst round player II may play arbitrarily, then ac-
cording to the winning strategies for models of SO
01
or SO
10
in the
decomposed segments.
2. If player I starts with a A and to the right and the left of a remain at
least 2
k1
elements, player II should choose correspondingly. Otherwise
he should answer with the elements of the same distance from the left
or right edge element, respectively.
3. SO
11
FO is obvious. FO SO
11
: If / SO
11
then for each k > 0
there is some nite B SO
11
such that /
k
B.
4. Prove rst that SO
11

i
[ i > 0 is complete. Then apply Theorem 2.3.
Section 5.4
1. Let h: / B be a homomorphism, / = (/, w), /

= (B, w

) with
x
w

= hx
w
. Verify / [a] /

[ha] by induction on .
2. Let / = (A, <) be ordered. Replacing each a A by a copy of (Z, <)
or of (Q, <) results in a discrete or a dense order B /, respectively.
3. Let /
0
T
0
. Choose /
1
with /
0
/
1
T
1
, /
2
with /
1
/
2
T
0
etc. This results in a chain /
0
/
1
/
2
such that /
2i
T
0
and /
2i+1
T
1
. Then /

:=

iN
/
2i
=

iN
/
2i+1
T
0
, T
1
and
hence /

T := T
0
+ T
1
. This shows that T is consistent and model
compatible with T
0
(hence likewise with T
1
). Clearly, T is an -theory
and therefore also inductive.
4. The union S of a chain of inductive theories model compatible with T
has again these properties. By Zorns lemma there exists a maximal,
hence in view of Exercise 3 a largest theory of this kind.
Solution hints to the Exercises 13
Section 5.5
1. Let (i, j) ,= (0, 0). Then DO
ij
has models / B with / , B. To show
that DO
00
is the model completion of DO note rst that T := DO
00
+T/
is model complete for each / DO. Moreover, T is complete since T
has a prime model: For instance, let / DO
10
. Then the ordered sum
Q+/ (i.e., (xQ)(yA)x < y) is a prime model of T.
2. (a) Lindstrms criterion. T is
1
-categorical because a T-model can be
understood as a Q-vector space. (b) Each T
0
-model G is embeddable
in a T-model H. One gains such H by dening a suitable equivalence
relation on the set of all pairs
a
n
with a G and n Z
\
0.
3. Uniqueness follows similarly to uniqueness of the model completion. If
/ T

and / B T then B ( T

for some (, hence / ( in

view of / (, and therefore /
ec
B according to Lemma 4.8.
4. The algebraic closure T
p
of the prime eld T
p
is equal to

n1
T
p
n,
where T
p
n is the nite eld of p
n
elements. Thus, an -sentence valid
in all nite elds is valid in all a.c. elds of prime characteristics and
hence in all a.c. elds (proof indirectly with (1) in 3.3).
Section 5.6
1. Let /, B ZG, / B. Then also /

for the ZGE-expansions /

and B

of / and B, respectively, because m has in ZG both an - and

an -Denition. Thus /

and hence / B.
2. Similiar to quantier elimination in ZGE but somewhat more simple.
3. Inductively over quantier-free = (x) follows: either
A
or ()
A
is nite for each / RCF

. This is not the case for (x).

4. CS holds in the real closed eld R, hence in each / RCF. The proofs
from CS of (x0)y x=== =y y, and that each polynomial of odd degree
has a zero must be carried out without a theory of continuous functions,
which is very instructive.
14 Solution hints to the Exercises
Section 5.7
1. If F is trivial then there is some i
0
I with i
0
J for each J F
by Exercise 3 in 1.5. Then a
F
b i
0
I
a=b
a
i
0
= b
i
0
, for all
a, b

iI
A
i
. This implies

F
iI
/
i
/
i
0
.
2. x x
I
/F (x A) is an embedding (to be checked in detail) and
moreover an elementary embedding.
3. Let X
K
and I, J

F dened as in the proof of Theorem 7.3 and

assume that for each i I there is some /
i
K and w
i
: Var A
i
such
that w
i
D
A
i
for all i but w
i
/ D
A
i
. Put ( :=

F
iI
/
i
( K)
and w = (w
i
)
iI
. Then wX D
C
and w / D
C
, hence X
C
, a
contradiction to X
K
.
4. W.l.o.g. / = 2 and 2 B 2
I
for some set I by Stones representation
theorem. 2 2
I
B according to Theorem 7.5.
Section 6.1
1. b ran f (ab)fa = b (this predicate is p.r. i f is p.r.).
2. Injectivity: Let (a, b) = (c, d). In order to prove a = c and b = d
assume rst that a + b < c + d. This leads to a contradiction since
(a, b) < (a, b) + b + 1 = t
a+b
+ a + b + 1 = t
a+b+1
t
c+d
(c, d).
Thus a + b = c + d. But then a = (a, b) t
a+b
= (c, d) t
c+d
= c,
hence also b = d. Surjectivity: Since (0, 0) = 0 ran it suces
to prove (a, b) + 1 ran , for all a, b. Clear for b = 0 because
(a, 0) + 1 = t
a
+ a + 1 = t
a+1
= (0, a + 1). In case b ,= 0 is
(a, b) + 1 = t
a+b
+ a + 1 = t
a+1+b1
+ a + 1 = (a + 1, b 1). This
proof also conrms the correctness of the diagram for , that is, the
arrows truly reect the successively growing values of .
3.
1
n = (k n)[(m n)(k, m) = n].
4. lcmf[ n = k

n
f[k ,= 0 & ( n)f k].
5. : Let R be recursive, M = a N [ bRab, and c M xed. Put
fn = k in case (mn) n = (m, k) & Rkm, and fn = c otherwise.
Solution hints to the Exercises 15
Section 6.2
1. Let
0
,
1
, . . . be a recursive enumeration of X,
n
=
n
. . .
n
. .
n
. By
Exercise 1 in 6.1,
n
[ n N is recursive and axiomatizes T as well.
2. Follow the proof of the unique term reconstruction property.
3. Similar to Exercise 2 with the unique formula reconstruction.
4. (a): A proof = (
0
, . . . ,
n
) of =
n
from an axiom system
X in T + can easily and in a p.r. manner be converted into a
somewhat longer proof

of in T, following the case distinc-

tion in Lemma 1.6.3:
i
should in case
i
= be replaced
by a proof of in T, and in case
i
X by a proof of

i

i
in T followed by
i
and
i
. If
k
results from

i
and
j
=
i

k
by applying MP, then the axiom
(
i

k
) (
i
)
k
, followed by (
i
)
k
and
k
should replace
k
. One may also proceed inductively on
the length of in constructing

.
Section 6.3
1. xy
N
z(xz)(yz)(z === =(x, y) ) where z / var . Simi-
larly for xy. Note also that xy
N
z(xz)(yz). In all
these equivalences,
N
can be replaced by
PA
.
2. (z<y)x
PA
u(z<y)(x<u). Contraposition and renaming of
readily yields (z<y)x
PA
u(z<y)(x<u).
3. Prove R
=
by case distinction.
4. Prove by induction on that both and satisfy the claim.
Section 6.4
1.
(a): p, a ap xy xa + 1=== =yp (Euclids lemma)
xy b === =ypb xab p b.
(b): Let m := lcma

[ n, so that m = a

for suitable c

. Assume
that (n)p, a

. Then (n)p c

by (a). Thus m = pm

and
c

= pc

for suitable m

, c

. This leads to contradiction to the denition

of m. (c) easily follows from (b).
16 Solution hints to the Exercises
2. u[beta u02 (v<x)(w, w

y)(beta uvw beta uSvw

w < w

primw primw

(z<w

)(primz zw) beta uxy)].

3. (a): Prove this rst for x instead of x. (b): It suces to show that
sb
x
( , x) = for x / free . (sb
x
((x)

, x) = (x)

for closed ).
Section 6.5
2. (ii)(i): If T is complete and T

+T is consistent then T

T provided
T and T

belong to the same language.

4. Trivial if T + is inconsistent. Otherwise let be the conjunction
of all sentences x!y(x, y), running through all dening formulas
for operations from . If T is decidable than so is T + . Moreover

T+

T+

rd
.
5. Set fa = (((

)))
last
if there is a proof in Q with a =

, and fa = 0
otherwise. ran f = 0 [
Q
is not recursive, since otherwise

Q
would be recursive which is not the case.
Section 6.6
1. Let T T
1
be consistent. S = L
0
[
P
T

+ CA is a
theory, see the proof of Theorem 6.2. S extends T
0
consistently, hence
is undecidable. The same then holds for T

+CA, hence for T

(since
CA is nite), and therefore also for T.
2. Identify P with and dene for arbitrary n, m, k
n +m=== =k ab(a nb ma b === =k a b).
For an explicit denition of multiplication on the cross product has
to be used. These denitions reect the naive set-theoretic standard
denitions of addition and multiplication in N.
Section 6.7
2.
0
is r.e. but not
1
(Remark 2 in 6.4).

Q is
1
but not
1
.
3. T is -inconsistent i (L
1
)(nbwb
T
(n) & bwb
T
x).
Solution hints to the Exercises 17
Section 7.1
1. Prove
PA
r
rem
(a, b, r) for b ,= 0 by induction on a.
2. (a): Follow the proof of Euclids lemma in 6.4. (b): Use <-induction.
(c): Let p ab. p, a x,y xa+1 = yp x,y xab+b = ybp p b.
3. Similar to part (c) of Exercise 1 in 6.4.
4. Existence: <-induction. Uniqueness: Prove rst p, q
k
(p, q prime) by
induction on k, applying Exercise 2(c).
5. (a):
T+

T

T
( ) formalizes part (b) of Exercise 4 in 6.2.
Section 7.3
1.
T

T

T
Con
T
, since Con
T

T
by (5).
Thus, T

is inconsistent by (1), hence

T
.
3. Clear if n = 0. Let T
n
= T +
n
and Con
T
n
T

n+1
(the
induction hypothesis). Now,
n

T

n+1
by D3. Hence, we obtain
T
n+1
= (T+
n
)+
n+1
= T+
n+1
. Further, by (5) page 281,
Con
T
n+1
T
(
n+1
)
T

n+2
.
4. For arithmetical sentences the statement If is provable in PA then
is true in ^ is provable in ZFC. Formalized:
ZFC

PA
.
Section 7.4
1. p p is responsible for transitivity, Lbs formula for irreexivity.
2.
G
p p p
G
(p p p)
G
p (p p).
Section 7.5
1. Prove rst ()
Gn
H
G

n
H for all H F

. The direction
in () follows by induction on
Gn
H. Then continue as follows:

Gn
H
G

n
H (by())

PA
(
n
H)

for all (Theorem 5.2)

PA

n
H

for all (property of )

PAn
H

for all (PA

n
= PA +
n
).
18 Solution hints to the Exercises
2. The rst claim follows immediately from Exercise 3 in 7.3. For deter-
mining the provability logic of PA
n

, use (6) in 7.3 and Theorem 5.3.

4. Prove that
GS
[(p q) (p q) (q p) (q p)]
and observe Theorem 5.4.
Section 7.7
1. We show there is some : g n with P < Q P < Q for n := lhg
(the length of a longest path in g). Trivial for lhg = 0, with P = 0 for
all P g. Let lhg = n+1 and g

:= g
\
max g where max g denotes the
set of all maximal points in g. Then lhg

= n and g

has property (p) as

well as is readily checked. Hence g

is a preference order with a mapping

: g

n by the induction hypothesis. Extend

to : g n + 1 by
putting P = n for all P max g. Obviously, P < Q P < Q.
For proving the converse let P < Q with Q max g. Then certainly
P

max g for some P

> P. Hence, by (p), either P < Q or Q < P

.
The latter is impossible since Q max g. Thus P < Q.
2. If (i) is falsied in g (that is, if 3(p3q) 3(q p) is satisable
in some point O g) then g contains the diagram from page 296 as a
subdiagram, with no arrow from P to Q and from Q to P

. It easily
follows that the nite poset g cannot be a preference order.
3. It is a matter of routine to check that (pp q) (q p) is
satised in an ordered G-frame. For the converse assume that g is initial
but not (totally) ordered. Then g contains the fork from page 298 as
a subframe, in which the Gj-axiom can easily be refuted.
4. Soundness of the G-axioms and rules is shown as the soundness part
of Theorem 7.3 which was given in the text. Soundness of the Gj-
axiom follows by contraposition. Assume that there are cardinals ,
such that V

, and V

. Then each of the

assumptions < , > , or = yields a contradiction.