Вы находитесь на странице: 1из 89

Introduction to Set Theory

A Solution Manual for Hrbacek and Jech (1999)


Jianfei Shen
School of Economics, The University of New South Wales
Sydney, Australia
The Lord by wisdom founded the earth, by
understanding he established the heavens.
Proverbs 3:19
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
1 Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Introduction to Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 The Axioms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.4 Elementary Operations on Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Relations, Functions, and Orderings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1 Ordered Pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Relations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4 Equivalences And Partitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.5 Orderings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3 Natural Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.1 Introduction to Natural Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2 Properties of Natural Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.3 The Recursion Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.4 Arithmetic of Natural Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.5 Operations and Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4 Finite, Countable, and Uncountable Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.1 Cardinality of Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2 Finite Sets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.3 Countable Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4 Linear Orderings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.5 Complete Linear Orderings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.6 Uncountable Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
v
vi CONTENTS
5 Cardinal Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.1 Cardinal Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.2 The Cardinality of the Continuum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6 Ordinal Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.1 Well-Ordered Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.2 Ordinal Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Preface
Sydney, Jianfei Shen
Date
vii
Acknowledgements
ix
1
SETS
1.1 Introduction to Sets
No exercises.
1.2 Properties
No exercises.
1.3 The Axioms
IExercise 1 (1.3.1). Show that the set of all . such that . and . T exists.
Proof. Notice that

. : . and . T
_
=

. : . T
_
.
Then by the Axiom Schema of Comprehension, we know that such a set does
exist. |L
I Exercise 2 (1.3.2). Replace The Axiom of Existence by the following weaker
postulate:
Weak Axiom of Existence: Some set exists.
Prove the Axiom of Existence using the Weak Axiom of Existence and the
Comprehension Schema.
Proof. Let be a set known to exist. By the Axiom Schema of Comprehension,
there is a set X such that
X =

. : . = .
_
.
1
2 CHAPTER 1 SETS
There is no subjects . satisfying . = ., so there is no elements in X, which
proves the Axiom of Existence. |L
I Exercise 3 (1.3.3). a. Prove that a set of all sets does not exist.
b. Prove that for any set there is some . .
Proof. (a) Suppose that there exists a universe set (a set of all sets) V. Then
by the Axiom Schema of Comprehension, there is a set T = {. V : . .];
that is
. T == . V and . .. (1.1)
Now we show that T V, that is, T is not a set. Indeed, if T V, then either
T T, or T T. If T T, then, by the == direction of (1.1), T V
and T T. A contradiction; if T T, then, by the == direction of (1.1),
the assumption T V and T T yield T T. A contradiction again. This
completes the proof that T V.
(b) If there were a set such that . for all ., then is a set of all sets,
which, as we have proven, does not exist. |L
I Exercise 4 (1.3.4). Let and T be sets. Show that there exists a unique set C
such that . C if and only if either . and . T or . T and . .
Proof. Let and T be sets. The following two sets exist:
C
1
=

. : . and . T
_
=

. : . T
_
.
C
2
=

. : . and . T
_
=

. T : .
_
.
Then C = C
1
LC
2
exists by the Axiom of Union. The uniques of C follows from
the Axiom of Extensionality. |L
I Exercise 5 (1.3.5). a. Given , T, and C, there is a set 1 such that . 1 i
. = or . = T or . = C.
b. Generalize to four elements.
Proof. (a) By the Axiom of Pair, there exist two sets: {. T] and {C. C] =
{C]. By the Axiom of Union, there exists set 1 satisfying 1 = {. T] L {C] =
{. T. C].
(b) Suppose there are four sets . T. C, and D. Then the Axiom of Pair implies
that there exist {. T] and {C. D], and the Axiom of Union implies that there
exists
1 = {. T] L {C. D] = {. T. C. D] . |L
I Exercise 6 (1.3.6). Show that P(X) _ X is false for any X. In particular,
P(X) = X for any X. This proves again that a set of all sets does not exist.
SECTION 1.4 ELEMENTARY OPERATIONS ON SETS 3
Proof. Let X be an arbitrary set; then there exists a set Y = {u X: u u].
Obviously, Y _ X, so Y P(X) by the Axiom of Power Set. If Y X, then we
have Y Y if and only if Y Y [See Exercise 3(a)]. This proves that P(X) X,
and P(X) = X by the Axiom of Extensionality. |L
I Exercise 7 (1.3.7). The Axiom of Pair, the Axiom of Union, and the Axiom of
Power Set can be replaced by the following weaker versions.
Weak Axiom of Pair For any and T, there is a set C such that C and
T C.
Weak Axiom of Union For any S, there exists U such that if X and S,
then X U.
Weak Axiom of Power Set For any set S, there exists 1 such that X _ S
implies X 1.
Prove the Axiom of Pair, the Axiom of Union, and the Axiom of Power Set using
these weaker versions.
Proof. We just prove the rst axiom. By the Weak Axiom of Pair, for any
and T, there exists a set C
t
such that C
t
and T C
t
. Now by the Axiom
Schema of Comprehension, there is a set C such that C = {. C
t
: . =
or . = T]. |L
1.4 Elementary Operations on Sets
I Exercise 8 (1.4.1). Prove all the displayed formulas in this section and visu-
alize them using Venn diagrams.
Proof. Omitted. |L
I Exercise 9 (1.4.2). Prove
a. _ T if and only if T = if and only if L T = T if and only if
T = .
b. _ T C if and only if _ T and _ C.
c. T L C _ if and only if T _ and C _ .
d. T = ( L T) T = ( T).
e. T = ( T).
f. (T C) = ( T) L ( C).
g. = T if and only if ^T = .
4 CHAPTER 1 SETS
Proof. (a) We rst prove that _ T == T = . Suppose _ T. Note
that T _ is clear since a T == a and a T == a .
To prove _ T under the assumption that _ T, notice that a | .
_ T| == a | . a T| == a T. Hence, _ T == T = .
To see T = == _ T, note that = T == _ T ==
_ | . _ T| == _ T.
To see _ T == L T = T, notice rst that T _ L T holds trivially.
Hence, we need only to show L T _ T. But this is true because a L T| .
_ T| ==a a T|.a ==a T| ==a T. The direction LT =
T == _ T holds because L T = T == L T _ T == _ T.
_ T == T = holds by denition of dierence of sets: T :

. T
_
. By this denition, if _ T and a , then a T, which
contradicts the requirement a T; hence, T = when _ T. To prove
T = == _ T, we use its false antecedent. Suppose T . Then there
exists a and a T since T is a proper subset of , but which means that
T = .
(b) If _ T C, then a == a T C == a T| . a C|. The other
direction is just by denition.
(c) To see T LC _ ==T _ and C _ , let a T [a C], then a T LC _
[a T L C _ ]. To prove the inverse direction, let a T or a C; that is,
a T L C. But T _ and C _ , we have a , too.
(d) To prove T = ( L T) T, notice that a ( L T) T ==
a a T| .
_
a T
_
== a | .
_
a T
_
== a T. To prove
T = ( T), notice that
a ( T) == a | .
_
(a T)
_
== a | .
_
(a . a T)
_
== a | .
_
a a T
_
== a | .
_
a T
_
== a T.
(e) a ( T) == a |.
_
(a T)
_
== a |.
_
a a T
_
==
a | . a T| == a T.
(f) First, a (T C) i a | .
_
(a T C)
_
i a | .
_
a T a C
_
.
Then, a ( T) L ( C) ==
_
a . a T
_
a . a C| ==
a | .
_
a T a C
_
.
(g) = T == _ T| . T _ |
.a/
== T = | . T = | ==
( T) L (T ) = == ^T = . |L
I Exercise 10 (1.4.3). Omitted.
SECTION 1.4 ELEMENTARY OPERATIONS ON SETS 5
I Exercise 11 (1.4.4). Let be a set; show that a complement of does not
exist.
Proof. Suppose
c
exists. Then, by the Axiom of Union, there is a set V =
L
c
. But in this case, V is a universe. A contradiction [See Exercise 3 (a)]. |L
I Exercise 12 (1.4.5). Let S = and be sets.
a. Set T
1
= {Y P(): Y = X for some X S], and prove
_
S =
_
T
1
(generalized distributive law).
b. Set T
2
= {Y P(): Y = X for some X S], and prove
__
S
_
=
_
T
2
,

__
S
_
=
_
T
2
(generalized De Morgan laws).
Proof. (a) .
_
S i . and there is X S such that . X i there
exists X S such that . X i . T
1
.
(b) We have
.
_
_
S
_
== . | .
_

_
.
_
S
_
_
== . | .
_

_
J X S such that . X
_
_
== . | .
_
. X V X S
_
==
_
. . . X
_
V X S
== . X| V X S
== .
_
( X)
== .
_
T
2
.
and
.
_
_
S
_
== . | .
_

_
.
_
S
_
_
== . | .
_
(. X V X S)
_
== . | .
_
J X S such that . X
_
== J X S such that
_
. . . X
_
== J X S such that . X|
== .
_
( X)
== .
_
T
2
. |L
IExercise 13 (1.4.6). Prove that
_
S exists for all S = . Where is the assump-
tion S = used in the proof?
Proof. If S = , we can take a set S. Let P(.) denote . X for all X S.
Then
_
S = {. : P(.)]
6 CHAPTER 1 SETS
exists by the Axiom Schema of Comprehension.
But if S = , then
_
S is a set of all sets; that is, .
_
for all .. Suppose
not, then there must exist a set such that . , but obviously we cannot
nd such a set . |L
Remark. While
_
is not dened, we do have
_
= .
Suppose not, then there exists .
_
, that is, there exists such that
. . Now consider the antecedent
. V . (1.2)
Obviously (1.2) cannot hold since there does not exist such a set . We
thus prove that
_
= .
2
RELATIONS, FUNCTIONS, AND ORDERINGS
2.1 Ordered Pairs
I Exercise 14 (2.1.1). Prove that (a. b) P(P({a. b])) and a. b
_
(a. b). More
generally, if a and b , then (a. b) P(P()).
Proof. Notice that (a. b) =

{a]. {a. b]
_
, and P({a. b]) =

. {a]. {b] . {a. b]
_
.
Therefore, (a. b) P({a. b]) and so (a. b) P
_
P({a. b])
_
. Further,
_
(a. b) =
_
{a]. {a. b]
_
= {a. b]; hence, a. b
_
(a. b).
If a and b , then {a] _ and {a. b] _ . Then {a] P() and
{a. b] P(); that is,

{a]. {a. b]
_
_ P
_
{]
_
. Then by the Axiom of Power Set,
(a. b) =

{a]. {a. b]
_
P(P()). |L
Remark. If a and b T, then (a. b) P(P({ L T])).
Proof. We have {a] _ _ L T, {b] _ L T, and {a. b] _ L T.
Then {a]. {a. b] P( L T); that is, {{a]. {a. b]] _ P( L T). Hence, (a. b) =
{{a]. {a. b]] P(P( L T)). |L
I Exercise 15 (2.1.2). Prove that (a. b), (a. b. c), and (a. b. c. J) exist for all
a. b. c, and J.
Proof. By The Axiom of Pair, both {a] = {a. a] and {a. b] exist. Then, use
this axiom once again, we know (a. b) = {{a]. {a. b]] exists. Since (a. b. c) =
((a. b). c), it follows that the ordered triple exists. (a. b. c. J) exists because
(a. b. c. J) = ((a. b. c). J). |L
I Exercise 16 (2.1.3). Prove: If (a. b) = (b. a), then a = b.
Proof. Let (a. b) = (b. a), that is,

{a]. {a. b]
_
=

{b] . {a. b]
_
. If a = b, then
{a] = {b], which implies that a = b. A contradiction. |L
I Exercise 17 (2.1.4). Prove that (a. b. c) = (a
t
. b
t
. c
t
) implies a = a
t
, b = b
t
,
and c = c
t
. State and prove an analogous property of quadruples.
7
8 CHAPTER 2 RELATIONS, FUNCTIONS, AND ORDERINGS
Proof. Note that (a. b. c) = (a
t
. b
t
. c
t
) i ((a. b). c) = ((a
t
. b
t
). c
t
), i (a. b) =
(a
t
. b
t
) and c = c
t
. Now, (a. b) = (a
t
. b
t
) i a = a
t
and b = b
t
. The quadruples
case can be easily extended. |L
I Exercise 18 (2.1.5). Find a, b, and c such that ((a. b). c) = (a. (b. c)). Of
course, we could use the second set to dene ordered triples, with equal success.
Proof. Let a = b = c. Then
((a. a). a) = {{(a. a)]. {(a. a). a]] =

{{a]]. {{a]. a]
_
.
(a. (a. a)) =

{a]. {a. (a. a)]


_
=

{a]. {a. {a]]


_
.
Thus, ((a. a). a) = (a. (a. a)). Note that while (T) C = (TC) generally,
there is a bijection between them. |L
IExercise 19 (2.1.6). To give an alternative denition of ordered pairs, choose
two dierent sets and . (for example, = , . = {]) and dene
(a. b) =

{a. ] . {b. .]
_
.
State and prove an analogue of Theorem 1.2 [p. 18] for this notion of ordered
pairs. Dene ordered triples and quadruples.
Proof. We are going to show that
(a. b) =

a
t
. b
t

== a = a
t
and b = b
t
.
If a = a
t
and b = b
t
, then (a. b) =

{a. ]. {b. .]
_
=

{a
t
. ]. {b
t
. .]
_
=

a
t
. b
t

.
For the inverse direction, let

{a. ]. {b. .]
_
=

{a
t
. ]. {b
t
. .]
_
. There are
two cases:
v If a = b, then: (i) If a = . and b = (note that = . by assumption), then

{a. ]. {b. .]
_
=

{. .]
_
, which enforces a
t
= . and b
t
= . (ii) If a = .
or b = (or both), then {a. ] = {b. .]. We rst show that it is impossible
that {a. ] = {b
t
. .] and {b. .] = {a
t
. ]; for otherwise a = . and b = .
Hence, it must be the case that
{a. ] = {a
t
. ] and {b. .] = {b
t
. .].
i.e., a = a
t
and b = b
t
.
v If a = b, then

{a. ]. {b. .]
_
=

{a. ]. {a. .]
_
=

{a
t
. ]. {b
t
. .]
_
implies that {a. ] = {a
t
. ] and {a. .] = {b
t
. .]; that is, a = a
t
= b
t
= b.
Note that it is impossible that {a. ] = {b
t
. .] and {a. .] = {a
t
. ]; for
otherwise, a = . = . A contradiction. |L
SECTION 2.2 RELATIONS 9
2.2 Relations
I Exercise 20 (2.2.1). Let 1 be a binary relation; let =
___
1
_
. Prove that
(.. ,) 1 implies . and , . Conclude from this that D
R
and R
R
exist.
Proof. By the Axiom of Union,
:
_
_
_
1
_
== : T for some T
_
1
== : T C for some C 1.
If (.. ,) 1, then C = {{.]. {.. ,]] 1, T = {.. ,] C, and .. , T; that is,
. and , . Hence,
D
R
=

.: .1, for some ,


_
=

. : .1, for some ,


_
.
Since
___
1
_
has been proven exist by the Axiom of Union, the existence of
D
R
follows from the Axiom Schema of Comprehension. The existence of R
R
can be proved with the same logic. |L
I Exercise 21 (2.2.2). a. Show that 1
-1
and S 1 exist.
b. Show that T C exist.
Proof. (a) Since 1 _ D
R
R
R
, it follows that 1
-1
_ R
R
D
R
. Since D
R
, R
R
,
and R
R
D
R
exist, we know that 1
-1
exists.
Since S 1 = {(.. :): (.. ,) 1 and (,. :) S for some ,], we have S 1 _
D
R
R
S
. Therefore, S 1 exists.
(b) Note that T C = ( T) C. Since T exists, ( T) C exists,
too. Particularly,
T C =
_
(a. b. c) P
_
P
_
_
P(P( L T))
_
L C
_
_
: a . b T. c C
_
. |L
I Exercise 22 (2.2.3). Let 1 be a binary relation and and T sets. Prove:
a. 1 L T| = 1| L 1T|.
b. 1 T| _ 1| 1T|.
c. 1 T| _ 1| 1T|.
d. Show by an example that _ and _ in parts (b) and (c) cannot be replaced by
=.
e. Prove parts (a)(b) with 1
-1
instead of 1.
f. 1
-1
_
1|
_
_ D
R
and 1
_
1
-1
T|
_
_ TR
R
; give examples where equality
does not hold.
10 CHAPTER 2 RELATIONS, FUNCTIONS, AND ORDERINGS
Proof. (a) If , 1 L T|, then there exists . L T such that .1,; that is,
either . and .1,, or . T and .1,. Hence, , 1| L 1T|.
Now let , 1| L 1T|. Then there exists . such that .1,, or there
exists . T such that .1,. In both case, . L T, and so , 1 L T|.
(b) If , 1 T|, then there exists . T such that .1,; that is, there
exists . for which .1,, and there exists . T for which .1,. Hence,
, 1| 1T|.
(c) If , 1| 1T|, then there is . such that .1,, but there is no
.
t
T such that .
t
1,. Hence, there exists . T such that .1,; that is,
, 1 T|.
(d) Let us consider the following binary relation

1 =
_
_
(.. ,). (.. 0)
_
: (.. ,) 0. 1|
2
_
:
that is,

1 projects the .,-plane onto the .-axis, carrying the point (.. ,) into
the (.. 0); See Figure 2.1.
-
.x; y/
-
.x; 0/

R
A
B
Figure 2.1.

R
v Let = {(.. ,): . 0. 1|. , = 1], and T = {(.. ,): . 0. 1|. , = 1,2]. Then
T = , and consequently,

1 T| = . However,

1|

1T| = 0. 1|.
v Notice that

1| =

1T| = 0. 1|, so

1|

1T| = . However, T = ,
and consequently,

1 T| =

1| = 0. 1|.
(e) Just treat 1
-1
as a relation [notice that (a)(c) hold for an arbitrary binary
relation 1
t
, so we can let 1
-1
= 1
t
].
(f) If . D
R
, then . and there exists , 1| such that ,1
-1
.. Hence,
. 1
-1
_
1|
_
. To show that the equality does not hold, consider

1 in part (d).
Note that dom(

1) = ; however,

1
-1

1|| = 0. 1|
2
.
For the second claim, just notice that 1
-1
is also a binary relation with
D
R
1 = R
R
(see the next exercise). |L
SECTION 2.2 RELATIONS 11
I Exercise 23 (2.2.4). Let 1 _ X Y . Prove:
a. 1X| = R
R
and 1
-1
Y | = D
R
.
b. If a D
R
, 1{a]| = ; if b R
R
, 1
-1
{b]| = .
c. D
R
= R
R
1; R
R
= D
R
1.
d. (1
-1
)
-1
= 1.
e. 1
-1
1 _ Id
D
R
; 1 1
-1
_ Id
R
R
.
Proof. (a) , 1X| i there exists . X such that .1, i , R
R
, so 1X| =
R
R
. Similarly, . 1
-1
Y | i there exists , Y such that .1, i . D
R
.
(b) Suppose that 1{a]| = ; let b 1{a]|. But then there exists b R
R
for
which a1b; that is, a D
R
. A contradiction. Similarly, let a 1
-1
{b]|. Then
a1b; that is, b R
R
. A contradiction.
(c) . D
R
i there exists , Y such that .1,, i there exists , Y for which
,1
-1
., i . R
R
1. Similarly, , R
R
i there exists . X such that .1,, i
there exists . X such that ,1
-1
., if and only if , D
R
1.
(d) For every (.. ,) X Y , we have .(1
-1
)
-1
, i ,1
-1
. i .1,. Hence,
(1
-1
)
-1
= 1.
(e) We have (.. ,) Id
D
R
i . D
R
and . = ,. We now show that (.. .)
1
-1
1 for all . D
R
. Since . D
R
, there exists , such that (.. ,) 1, i.e.,
(,. .) 1
-1
. Hence, there exists , such that (.. ,) 1 and (,. .) 1
-1
; that is,
(.. .) 1
-1
1.
Now let (.. ,) Id
R
R
. Then . = , and , R
R
. Then there exists . such that
(.. ,) 1, i.e., (,. .) 1
-1
. Therefore, (,. ,) 1 1
-1
. |L
I Exercise 24 (2.2.5). Let X =

. {]
_
, Y = P(X). Describe
a.
Y
;
b. Id
Y
.
Proof. Y = P(X) =
_
. {] .

{]
_
.

. {]
_
_
. Then

Y
=

(a. b): a Y , b Y , and a b


_
=

(. {]). (. {. {]]). ({]. {{]]). ({]. {. {]])


_
.
and
Id
Y
=

(a. b)

a Y , b Y , and a = b
_
=

(. ). ({]. {]). ({{]]. {{]]). ({. {]]. {. {]])


_
. |L
I Exercise 25 (2.2.6). Prove that for any three binary relations 1, S, and T
T (S 1) = (T S) 1.
12 CHAPTER 2 RELATIONS, FUNCTIONS, AND ORDERINGS
Proof. Let 1, S, and T be binary relations. Then
(n. :) T (S 1) == there exists , for which n(S 1),. ,T :
== there exists , and . for which n1.. .S,. ,T :
== there exists . for which .(T S):. n1.
== (n. :) (T S) 1. |L
I Exercise 26 (2.2.7). Give examples of sets X, Y , and 7 such that
a. X Y = Y X.
b. X (Y 7) = (X Y ) 7.
c. X
3
= X X
2
[i.e., (X X) X = X (X X)].
Proof. (a) Let X = {1] and Y = {2. 3]. Then X Y = {(1. 2) . (1. 3)], but Y X =
{(2. 1) . (3. 1)].
(b) Let X = {1], Y = {2], and 7 = {3]. Then X (Y 7) = {(1. (2. 3))], and
(XY )7 = {((1. 2). 3)]. But (1. (2. 3)) = ((1. 2). 3) since 1 = (1. 2) and (2. 3) = 3.
(c) Let X = {a]. Then X
3
= {((a. a). a)] = {({{a]]. a)], but XX
2
= {(a. (a. a))] =
{(a. {{a]])]. It is clear that X
3
= XX
2
since a = {{a]]. [Remember that a = (a)
is an one-tuple, but {{a]] = (a. a) is an ordered pair.] |L
I Exercise 27 (2.2.8). Prove:
a. T = if and only if = or T = .
b. (
1
L
2
) T = (
1
T) L(
2
T), and (T
1
L T
2
) = ( T
1
) L( T
2
).
c. Same as part (b), with L replaced by , , and ^.
Proof. (a) T = i
_
J a and b T
_
i a | b T| i =
or T = .
(b) We have
(a. b) (
1
L
2
) T == a
1
L
2
and b T
==
_
a
1
and b T
_
or
_
a
2
and b T
_
==
_
(a. b)
1
T
_
or
_
(a. b)
2
T
_
== (a. b) (
1
T) L (
2
T) .
and
(a. b) (T
1
L T
2
) == a and b T
1
or b T
2
|
==
_
a and b T
1
_
or
_
a and b T
2
_
==
_
(a. b) T
1
_
or
_
(a. b) T
2
_
== (a. b) ( T
1
) L ( T
2
).
SECTION 2.3 FUNCTIONS 13
(c) We just prove the rst part.
(a. b) (
1

2
) T == a
1
. a
2
| . b T|
== a
1
. b T| . a
2
. b T|
== (a. b) (
1
T) (
2
T).
(a. b) (
1

2
) T ==
_
a
1
. a
2
_
. b T|
== a
1
. b T| .
_
a
2
_
==
_
(a. b)
1
T
_
.
_
(a. b)
2
T
_
== (a. b) (
1
T) (
2
T) .
and
(
1
^
2
) T =
_
(
1

2
) L (
2

1
)
_
T
=
_
(
1

2
) T
_
L
_
(
2

2
) T
_
=
_
(
1
T) (
2
T)
_
L
_
(
2
T) (
1
T)
_
= (
1
T)^(
2
T). |L
2.3 Functions
I Exercise 28 (2.3.1). Prove: If R
f
_ D
g
, then D
gf
= D
f
.
Proof. It is clear that D
gf
= D
f

-1
D
g
| _ D
f
. For the other inclusion
direction, we have
D
gf
= D
f

-1
D
g
| _ D
f

-1
_
R
f
_
= D
f
.
where we use the fact that
-1
R
f
| = D
f
:
.
-1
_
R
f
_
== J , R
f
such that (,. .)
-1
== J , R
f
such that (.. ,)
== . D
f
. |L
I Exercise 29 (2.3.2). The functions
i
, i = 1. 2. 3 are dened as follows:

1
= (2. 1 : . R) .

2
=
_
_
. : . > 0
_
.

3
=

1,. : . R. . = 0

.
Describe each of the following functions, and determine their domains and
ranges:
2

1
,
1

2
,
3

1
, and
1

3
.
14 CHAPTER 2 RELATIONS, FUNCTIONS, AND ORDERINGS
Proof. The domain of
2

1
is determined as
1
D
f
2
f
1
= D
f
1

-1
1
_
D
f
2
_
= R
-1
1
R

|
= {. R: . > 1,2] .

2

1
=

(.. :) : . > 1,2 and, for some ,. 2. 1 = , and


_
, = :
_
=
_
_
2. 1: . > 1,2
_
.
0 . 3 2 1 1 2 3

3
2

2
Figure 2.2.
Further, D
f
1
f
2
= D
f
2

-1
2
_
D
f
1
_
= R


-1
2
R| = R

, and
1

2
=
_
2
_
. 1: . > 0
_
. |L
I Exercise 30 (2.3.3). Prove that the function
1
,
2
,
3
from Exercise 29 are
one-to-one, and nd the inverse functions. In each case, verify that D
f
i
= R
f
1
i
,
R
f
i
= D
f
1
i
.
Proof. As an example, we consider
2
.
0 . 1 2 3
1
2
3
2

-
1
2
Figure 2.3. f
2
and f
1
2
.
1
Throughout this book, R
CC
:

x R[ x > 0
_
, and R
C
:

x R[ x > 0
_
.
SECTION 2.3 FUNCTIONS 15
We have (.. ,)
-1
2
i (,. .)
2
i . =
_
, and , > 0 i , = .
2
and
. > 0. |L
I Exercise 31 (2.3.4). Prove:
a. If is invertible,
-1
= Id
D
f
,
-1
= Id
R
f
.
b. Let be a function. If there exists a function g such that g = Id
D
f
then
is invertible and
-1
= g R
f
. If there exists a function h such that
h = Id
R
f
then may fail to be invertible.
Proof. (a) We have proven in Exercise 23 (e) that [since is a relation]

-1
_ Id
D
f
and
-1
_ Id
R
f
; hence, we need only to show the inverse
directions. To see
-1
_ Id
D
f
, let . D
f
. Then
(.. ,)
-1
==J : such that (.. :) and (:. ,)
-1
==J : such that (.. :) and (,. :)
==. = , since is invertible
==(.. ,) Id
D
f
.
To see
-1
_ Id
R
f
, let , R
f
. Then
(,. .)
-1
==J : such that (,. :)
-1
and (:. .)
==J : such that (:. ,) and (:. .)
==, = .
==(,. .) Id
R
f
.
(b) Suppose that there exists a function g such that g = Id
D
f
. Let .. .
t
D
f
with . = .
t
. Then (.. .) g and (.
t
. .
t
) g . Thus
J , such that (.. ,) and (,. .) g. (2.1)
J ,
t
such that (.
t
. ,
t
) and (,
t
. .
t
) g. (2.2)
It follows that , = ,
t
; for otherwise, by (2.1) and (2.2), we would have (,. .)
g and (,. .
t
) g, which contradicts the fact that g is a function.
To see that
-1
= g R
f
, rst notice that g = Id
D
f
implies that
D
f

-1
D
g
| = D
f
, which implies that D
f
_
-1
D
g
|, which implies that
D
f
| = R
f
_
-1
D
g
| = D
g
since is invertible. Hence,
D
gR
f
= D
g
R
f
= R
f
= D
f
1.
Further, for every , D
f
1, there exists . such that . =
-1
(,), i.e., , = (.).
Then g R
f
(,) = (g R
f
)(.) = .. Hence, g R
f
=
-1
.
Finally, as in Figure 2.4, let : {.
1
. .
2
] { ,] dened by (.
1
) = (.
2
) = ,.
Let h: { ,] {.
1
] dened by h( ,) = .
1
. Then h: { ,] { ,] is given by
( h)( ,) = ,; that is, h = Id
R
f
. However, is not invertible since it is not
injective. |L
16 CHAPTER 2 RELATIONS, FUNCTIONS, AND ORDERINGS
,
.
1
.
2
,
h

Figure 2.4. f is not invertible
I Exercise 32 (2.3.5). Prove: If and g are one-to-one functions, g is also
a one-to-one function, and (g )
-1
=
-1
g
-1
.
Proof. Let .. , D
gf
and (g )(.) = (g )(,). Then (.) = (,) since g
is injective; then . = , since is injective. Thus, g is injective.
X Y
7
g

g
Figure 2.5.
To see that (g )
-1
=
-1
g
-1
, notice that
(:. .) (g )
-1
== (.. :) g
== J , such that (.. ,) and (,. :) g
== J , such that (,. .)
-1
and (:. ,) g
-1
== (:. .)
-1
g
-1
. |L
IExercise 33 (2.3.6). The images and inverse images of sets by functions have
the properties exhibited in Exercise 22, but some of the inequalities can now be
replaced by equalities. Prove
a. If is a function,
-1
T| =
-1
|
-1
T|.
b. If is a function,
-1
T| =
-1
|
-1
T|.
Proof. (a) If .
-1
T|, then (.) T, so that (.) and (.) T.
But then .
-1
| and .
-1
T|, i.e., .
-1
|
-1
T|. Conversely, if
.
-1
|
-1
T|, then .
-1
| and .
-1
T|. Therefore, (.) and
(.) T, i.e., (.) T. But then .
-1
T|.
(b) If .
-1
T|, then (.) T, so that (.) and (.) T.
But then .
-1
| and .
-1
T|, i.e., .
-1
|
-1
T|. Conversely, if
SECTION 2.3 FUNCTIONS 17
.
-1
|
-1
T|, then .
-1
| and .
-1
T|. Therefore, (.) and
(.) T, i.e., (.) T. But then .
-1
T|. |L
I Exercise 34 (2.3.7). Give an example of a function and a set such that

2
= .
Proof. Let (.
t
) = ,
t
, where .
t
and ,
t
. Then (.
t
. ,
t
) , but
(.
t
. ,
t
)
2
. |L
I Exercise 35 (2.3.8). Show that every system of sets can be indexed by a
function.
Proof. For every system of sets , consider Id: . Then = {Id(i ): i
]. |L
I Exercise 36 (2.3.9). a. Show that the set T
A
exists.
b. Let (S
i
: i 1) be an indexed system of sets; show that

iI
S
i
exists.
Proof. (a) _ T for all T
A
, and so P(T). Then T
A
_ P(T).
Therefore, T
A
= { P( T): : T] exists by the Axiom Schema of
Comprehension.
(b) By denition,

iI
S
i
= { : is a function on 1 and
i
S
i
for all i 1].
Hence, for all

iI
S
i
, if (i. s
i
) , then (i. s
i
) 1 S
i
_ 1
_
iI
S
i
; that
is, _ 1
_
iI
S
i
. Hence, P(1
_
iI
S
i
) for all

iI
S
i
, and hence

iI
S
i
_ P(1
_
iI
S
i
). Therefore, the existence of

iI
S
i
follows the Axiom
Schema of Comprehension. |L
IExercise 37 (2.3.10). Show that unions and intersections satisfy the following
general form of the associative law:
_
a
_
S
J
a
=
_
CS
_
_
_
aC
J
a
_
_
.
_
a
_
S
J
a
=
_
CS
_
_
_
aC
J
a
_
_
.
if S is a nonempty system of nonempty sets.
Proof. We have
.
_
a
_
S
J
a
== J a
_
S such that . J
a
== J a C S. such that . J
a
== .
_
CS
_
_
_
aC
J
a
_
_
.
and
18 CHAPTER 2 RELATIONS, FUNCTIONS, AND ORDERINGS
.
_
a
_
S
J
a
== . J
a
. V a
_
S
== . J
a
. V C S. V a C
== .
_
CS
_
_
_
aC
J
a
_
_
. |L
I Exercise 38 (2.3.11). Other properties of unions and intersections can be
generalized similarly.
De Morgan Laws
T
_
_
_
aA
J
a
_
_
=
_
aA
(T J
a
) . T
_
_
_
aA
J
a
_
_
=
_
aA
(T J
a
) .
Distributive Laws
_
_
_
aA
J
a
_
_

_
_
_
bB
G
b
_
_
=
_
.a;b/AB
(J
a
G
b
) .
_
_
_
aA
J
a
_
_
L
_
_
_
bB
G
b
_
_
=
_
.a;b/AB
(J
a
L G
b
) .
Proof. We have
. T
_
_
_
aA
J
a
_
_
== . T| .
_
_
_
_
_
.
_
aA
J
a
_
_
_

_
== . T| .
_

_
J a such that . J
a
_
_
== . T| .
_
V a . J
a
_
== V a
_
. T . . J
a
_
== .
_
aA
(T J
a
) .
and
SECTION 2.3 FUNCTIONS 19
. T
_
_
_
aA
J
a
_
_
== . T| .
_
_
_
_
_
.
_
aA
J
a
_
_
_

_
== . T| .
_
(V a . . J
a
)
_
== . T| .
_
J a such that . J
a
_
== J a such that
_
. T . . J
a
_
== J a such that . T J
a
|
== .
_
aA
(T J
a
) .
and
.
_
_
_
aA
J
a
_
_

_
_
_
bB
G
b
_
_
==
_
_
.
_
aA
J
a
_
_
.
_
_
.
_
bB
G
b
_
_
== J a such that . J
a
and
J b T such that . G
b
== J (a. b) T such that . J
a
G
b
== .
_
.a;b/AB
(J
a
G
b
) .
Finally,
.
_
_
_
aA
J
a
_
_
L
_
_
_
bB
G
b
_
_
==
_
_
.
_
aA
J
a
_
_

_
_
.
_
bB
G
b
_
_
== V a . . J
a
| V b T. . G
b
|
== V (a. b) T . J
a
. G
b
|
== .
_
.a;b/AB
(J
a
L G
b
) . |L
I Exercise 39 (2.3.12). Let be a function. Then

_
_
_
aA
J
a
_
_
=
_
aA
J
a
| .
-1
_
_
_
aA
J
a
_
_
=
_
aA

-1
J
a
| .

_
_
_
aA
J
a
_
_
_
_
aA
J
a
| .
-1
_
_
_
aA
J
a
_
_
=
_
aA

-1
J
a
| .
If is one-to-one, then _ in the third formula can be replaced by =.
Proof. Let be a function. Then
20 CHAPTER 2 RELATIONS, FUNCTIONS, AND ORDERINGS
,
_
_
_
aA
J
a
_
_
== J .
_
aA
J
a
such that (.. ,)
== J a . J . J
a
. such that (.. ,)
== J a such that , J
a
|
== ,
_
aA
J
a
| .
(2.3)
and
.
-1
_
_
_
aA
J
a
_
_
== (.)
_
aA
J
a
== J a such that (.) J
a
== J a such that .
-1
J
a
|
== .
_
aA

-1
J
a
| .
(2.4)
and
,
_
_
_
aA
J
a
_
_
==J .
_
aA
J
a
such that (.. ,)
==V a . . J
a
such that (.. ,) (+)
==V a . , J
a
| (++)
==,
_
aA
J
a
| :
(2.5)
hence,
__
aA
J
a
_
_
_
aA
J
a
|. But if is not one-to-one, then (++) does
not imply (+) in (2.5). For example, let , J
1
| J
2
|, but it is possible that
(.
1
) = (.
2
) = ,, where .
1
J
1
, .
2
J
2
, and .
1
= .
2
. However, if is one-
to-one, then it must be that .
1
= .
2
. More explicitly, to derive (+) from (++) in
(2.5), notice that
V a . , J
a
| ==J .
_
aA
J
a
such that (.. ,)
==V a . . J
a
such that (.. ,) .
Finally,
.
-1
_
_
_
aA
J
a
_
_
== (.)
_
aA
J
a
== V a . (.) J
a
== V a . .
-1
J
a
|
== .
_
aA

-1
J
a
| . |L
SECTION 2.3 FUNCTIONS 21
I Exercise 40 (2.3.13). Prove the following form of the distributive law:
_
aA
_
_
_
bB
J
a;b
_
_
=
_
f B
A
_
_
_
aA
J
a;f.a/
_
_
.
assuming that J
a;b
1
J
a;b
2
= for all a and b
1
. b
2
T, b
1
= b
2
.
Proof. First note that
J
a;f.a/
_
_
bB
J
a;b
(2.6)
for any T
A
since T
A
[there exists b T such that b = (a)]. Hence
_
aA
J
a;f.a/
_
_
aA
_
_
_
bB
J
a;b
_
_
(2.7)
follows (2.6), and which proves that
_
f B
A
_
_
_
aA
J
a;f.a/
_
_
_
_
aA
_
_
_
bB
J
a;b
_
_
. (2.8)
To prove the inverse direction, pick any .
_
aA
__
bB
J
a;b
_
. Put (a. b)
if and only if . J
a;b
. We now need to show that is a function on into T.
Because .
_
aA
__
bB
J
a;b
_
, for any a ,
.
_
bB
J
a;b
== J b T such that . J
a;b
:
hence, for any a , there exists b T such that . J
a;b
, that is, for any
a , there exists b T such that (a. b) , which is just the denition of a
function. Since we have proven that T
A
, we obtain
.
_
aA
_
_
_
bB
J
a;b
_
_
==V a . .
_
bB
J
a;b
==V a . J b T such that . J
a;b
==V a . J T
A
such that (a) = b and . J
a;f.a/
==.
_
aA
J
a;f.a/
==.
_
f B
A
_
_
_
aA
J
a;f.a/
_
_
.
(2.9)
Therefore, (2.8) and (2.9) imply the claim. |L
22 CHAPTER 2 RELATIONS, FUNCTIONS, AND ORDERINGS
2.4 Equivalences And Partitions
I Exercise 41 (2.4.1). For each of the following relations, determine whether
they are reexive, symmetric, or transitive:
a. Integer . is greater than integer ,.
b. Integer n divides integer m.
c. . = , in the set of all natural numbers.
d. _ and in P().
e. in .
f. in a nonempty set .
Solution. (a) is transitive; (b) is reexive and transitive; (c) is symmetric; (d):
_ is an equivalence relation, but is not reexive; (e) and (f) are equivalence
relations. |L
I Exercise 42 (2.4.2). Let be a function on onto T. Dene a relation 1 in
by: a1b if and only if (a) = (b).
a. Show that 1 is an equivalence relation on .
b. Dene a function on ,1 onto T by (a|
E
) = (a) (verify that (a|
E
) =
(a
t
|
E
) if a|
E
= a
t
|
E
).
c. Let be the function on onto ,1 given by (a) = a|
E
. Show that = .
Proof. (a) 1 is an equivalence relation on since (i) a1a as (a) = (a);
(ii) a1b i (a) = (b) i (b) = (a) i b1a; (iii) Let a1b and b1c; that is,
(a) = (b) and (b) = (c). Then (a) = (c) and so a1c.
(b) Let (a|
E
) = (a) for any a|
E
,1. If a|
E
= a
t
|
E
, then a
t
1a. Therefore,
(a) = (a
t
) by the denition of 1. Thus, (a|
E
) = (a) = (a
t
) = (a
t
|
E
).
(c) First, D
'j
= D
f
= since D
'j
= D
j

-1
D
'
| =
-1
,1| = . Next,
( )(.) = (.|
E
) = (.) for all . . |L
I Exercise 43 (2.4.3). Let 1 = {(r. ;) R R: r > 0], where R is the set of all
real numbers. View elements of 1 as polar coordinates of points in the plane,
and dene a relation on 1 by
(r. ;) - (r
t
. ;
t
) if and only if r = r
t
and ; ;
t
is an integer multiple of 2.
Show that - is an equivalence relation on 1. Show that each equivalence class
contains a unique pair (r. ;) with 0 6 ; 6 2. The set of all such pairs is there-
fore a set of representatives for -.
SECTION 2.5 ORDERINGS 23
Proof. (r. ;) - (r. ;) is obvious: r = r and ; ; = 0 2. To see - is symmetric,
let (r. ;) - (r
t
. ;
t
); then r = r
t
and ; ;
t
= n 2, where n Z. Therefore, r
t
= r
and ;
t
; = (n) 2; that is, (r
t
. ;
t
) - (r. ;). Finally, to see - is transitive, let
(r. ;) - (r
t
. ;
t
), and (r
t
. ;
t
) - (r
tt
. ;
tt
). In this case, r = r
t
= r
tt
, so r = r
tt
, and
; ;
t
= m 2. ;
t
;
tt
= n 2.
where m. n Z. But then ; ;
tt
= (; ;
t
) (;
t
;
tt
) = (m n) 2. Hence,
(r. ;) = (r
tt
. ;
tt
). The above steps show that - is an equivalence relation on 1.
Consider an arbitrary element of 1, -, say, (r
t
. ;
t
)|
~
. Since ;
t
R, there
must exist ; such that ;
t
; = n 2, where n Z. Then, there exists ; R such
that ; = ;
t
n 2. Hence, we can nd a n Z satisfying ;
t
,2 1 6 n 6 ;
t
,2,
and let (r. ;) = (r
t
. ;
t
n 2). |L
2.5 Orderings
I Exercise 44 (2.5.1). a. Let 1 be an ordering of , S be the corresponding
strict ordering of , and 1
+
be the ordering corresponding to S. Show that
1
+
= 1.
b. Let S be a strict ordering of , 1 be the corresponding ordering, and S
+
be
the strict ordering corresponding to 1. Then S
+
= S.
Proof. (a) Let (a. b) 1, where a. b . If a = b, then (a. b) 1
+
because
orderings are reexive; if a = b, then (a. b) S. But then (a. b) 1
+
. Hence,
1 1
+
. To see the inverse direction, let (a. b) 1
+
. Firstly, a = b implies that
(a. b) 1 since 1 is reexive. So we suppose a = b. In this case, (a. b) S.
Because S is 1s corresponding strict ordering of , we know (a. b) S if and
only if (a. b) 1 and a = b. Hence, 1
+
1. This proves that 1
+
= 1.
(b) Let (a. b) S, then a = b. Since 1 is Ss corresponding ordering, we have
(a. b) 1. Since (a. b) 1 and a = b, we have (a. b) S
+
. The revers direction
can be proven with the same logic. |L
IExercise 45 (2.5.2). State the denitions of incomparable elements, maximal,
minimal, greatest, and least elements and suprema and inma in terms of strict
orderings.
Solution. If (1. <) is a partially ordered set, X is a nonempty subset of 1,
and a 1, then:
v a and b are incomparable in < if a = b and neither a < b nor b < a holds;
v a is a maximal element of X if a X and (V . X) a .;
v a is a minimal element of X if a X and (V . X) . a;
24 CHAPTER 2 RELATIONS, FUNCTIONS, AND ORDERINGS
v a is the greatest element of X if a X and (V . X) . 6 a;
v a is the least element of X if a X and (V . X) a 6 .;
v a is an upper bound of X if (. X) . 6 a;
v a is a lower bound of X if (V . X) a 6 .;
v a is the supremum of X if a is the least upper bound of X;
v a is the inmum of X if a is the greatest lower bound of X. |L
I Exercise 46 (2.5.3). Let 1 be an ordering of . Prove that 1
-1
is also an
ordering of , and for T _ ,
a. a is the least element of T in 1
-1
if and only if a is the greatest element of T
in 1;
b. Similarly for (minimal and maximal) and (supremum and inmum).
Proof. (a) (i) a1
-1
a since a1a. (ii) Suppose (a. b) 1
-1
and (b. a) 1
-1
. Then
(b. a) 1 and (a. b) 1, and so a = b since 1 is antisymmetric. (iii) Let a1
-1
b
and b1
-1
c. Then b1a and c1b. Hence, c1a since 1 is transitive. But which
means that a1
-1
c, i.e., 1
-1
is transitive.
(b) If a is the least element of T in 1
-1
, then a T and a1
-1
. for all . T.
But then .1a for all . T, i.e., a is the greatest element of T in 1; if a be the
greatest element of T in 1, that is, a T and .1a for all . T, then a1
-1
. for
all . T, and so a is the least element of T in 1
-1
. With the same logic as (a)
we can get (b). |L
I Exercise 47 (2.5.4). Let 1 be an ordering of and let T _ . Show that
1 T
2
is an ordering of T.
Proof. (i) For every b T we have (b. b) T
2
and (b. b) 1; hence, (b. b)
1T
2
; that is, 1T
2
is reexive. (ii) Let (a. b) 1T
2
and (b. a) 1T
2
. Then
(a. b) 1 and (b. a) 1 imply that a = b. Therefore, 1 T
2
is antisymmetric.
(iii) Let (a. b) 1T
2
and (b. c) 1T
2
. Then (a. b) 1 and (b. c) 1 implies
that (a. c) 1. Furthermore, since both a T and c T, we have (a. c) T
2
.
Hence, (a. c) 1 T
2
; that is, 1 T
2
is transitive. |L
I Exercise 48 (2.5.5). Give examples of a nite ordered set (. 6) and a subset
T of so that
a. T has no greatest element.
b. T has no least element.
c. T has no greatest element, but T has a supremum.
d. T has no supremum.
SECTION 2.5 ORDERINGS 25
Proof. (a) Let = {a. b. c. J], T = {a. b. c], and
6= {(a. a). (b. b). (c. c). (J. J). (a. J). (b. J). (c. J)] .
In this example, a is not the greatest element of T because (a. b), (a. c) are
incomparable; similarly, b and c are not the greatest elements of T.
(b) As the example in (a), there is no least element.
(c) As the example in (a), there is no greatest element, but J is an upper bound
of T, and it is the least upper bound of T, so J is the supremum of T.
(d) Let = {a. b. c. c], T = {a. b. c], and 6= {(a. a). (b. b). (c. c). (J. J)]. Then
there is no upper bound of T, and consequently, T has no supremum. |L
I Exercise 49 (2.5.6). a. Let (. <) be a strictly ordered set and b . Dene
a relation - in T = L {b] as follows:
. - , if and only if (.. , and . < ,) or (. and , = b).
Show that - is a strict ordering of T and -
2
=<.
b. Generalize part (a): Let (
1
. <
1
) and (
2
. <
2
) be strict orderings,
1

2
= .
Dene a relation - on T =
1
L
2
as follows:
. - , if and only if .. ,
1
and . <
1
,
or .. ,
2
and . <
2
,
or .
1
and ,
2
.
Show that - is a strict ordering of T and -
2
1
=<
1
, -
2
2
=<
2
.
Proof. (a) Let . - ,. Then either .. , and . < , or . and , = b. In the
rst case, , . because , .; in the later case, , . be denition. Therefore,
- is asymmetric.
Let . - , and , - :. Then , = b; otherwise, , - : cannot hold. With the
same logic, . = b, too. If : = b, then . - : = b by denition; if : , then
. < , and , < : implies . < : and so . - :.
To prove (-
2
) =<, let (.. ,) (-
2
). Then .. , and (.. ,) -,
which means that (.. ,) <. Now let (.. ,) <. Then .. , == (.. ,)
2
and (.. ,) - by denition of -; hence, (.. ,) (-
2
).
(b) Let . - ,. If .. ,
1
, then . <
1
, and so , .; if .. ,
2
, then . <
2
,
and so , .; if .
1
and ,
2
, then , . by denition.
Let . - , and , - :. There are four cases:
v .. ,. :
1
. In this case, . <
1
, <
1
: ==. <
1
: ==. - :.
v .. ,. :
2
. In this case, . <
2
, <
2
: ==. <
2
: ==. - :.
v .. ,
1
and :
2
. In this case, . - : by denition.
26 CHAPTER 2 RELATIONS, FUNCTIONS, AND ORDERINGS
v .
1
and ,. :
2
. In this case, . - : by denition.
To prove
_
-
2
1
_
=<
1
, suppose (.. ,)
_
-
2
1
_
rstly. Then (.. ,) - and
.. ,
1
; hence . - , == . <
1
,. Now suppose (.. ,) <
1
. Then . - , and
.. ,
1
; that is, (.. ,)
_
-
2
1
_
.
The result that
_
-
2
2
_
=<
2
can be proved with the same logic. |L
I Exercise 50 (2.5.7). Let 1 be a reexive and transitive relation in (1 is
called a preordering of ). Dene 1 in by
a1b if and only if a1b and b1a.
Show that 1 is an equivalence relation on . Dene the relation 1,1 in ,1 by
a|
E
(1,1)b|
E
if and only if a1b.
Show that the denition does not depend on the choice of representatives for
a|
E
and b|
E
. Prove that 1,1 is an ordering of ,1.
Proof. We rst show that 1 is an equivalence relation on . (i) 1 is reexive
since 1 is. (ii) 1 is symmetric: if a1b, then a1b and b1a, i.e., b1a and a1b;
therefore, b1a by the denition of 1. (iii) 1 is transitive: if a1b and b1c, then
a1b and b1a, and b1c and c1b. Hence, a1c and c1a by the transitivity of 1.
We thus have a1c.
Let a|
E
(1,1)b|
E
if and only if a1b. We show that if c a|
E
and J b|
E
,
then a|
E
(1,1)b|
E
if and only if c1J. We rt focus on the IF part. Since
c a|
E
, we have c1a, i.e., a1c and c1a; similarly, J1b and b1J. Let c1J. We
rst have a1J since a1c; we also have J1b; hence a1b, i.e., c1J implies that
a|
E
(1,1)b|
E
. To prove the ONLY IF part, let a|
E
(1,1)b|
E
. Then a1b. Since
c1a and b1J, we have c1J.
1,1 is an ordering of ,1 since (i) 1,1 is reexive: for any a|
E
,1,
we have a a|
E
and a1a, so a|
E
(1,1)a|
E
; (ii) 1,1 is antisymmetric: if
a|
E
(1,1)b|
E
and b|
E
(1,1)a|
E
, then a1b and b1a, i.e., a1b. Hence, a|
E
=
b|
E
; (iii) 1,1 is transitive: if a|
E
(1,1)b|
E
and b|
E
(1,1)c|
E
, then a1b and
b1c and so a1c, that is, a|
E
(1,1)c|
E
. |L
I Exercise 51 (2.5.8). Let = P(X), X = . Prove:
a. Any S _ has a supremum in the ordering _
A
; supS =
_
S.
b. Any S _ has an inmum in _
A
; inf S =
_
S if S = ; inf = X.
Proof. (a) Let U =

u [ s _
A
u. V s S
_
, i.e., U is the set of all the upper
bounds of S according to _
A
. Note that U = since X U. Now we show that
the least element of U exists, and which is
_
S. Since s _
A
s _
A
_
S for any
s S, we have
_
S U; to see that
_
S is the least element of U, take any
u U. Then s _
A
u for all s S and so
_
S _
A
u; therefore, supS =
_
S.
SECTION 2.5 ORDERINGS 27
(b) Let 1 =

[ _
A
s. V s S
_
, i.e., 1 is the set of all the lower bounds
of S according to _
A
, and 1 = since 1. We rst consider the case that
S = , and show that sup1 =
_
S. Firstly, it is clear that
_
S 1; secondly,
if 1, then _
A
s for all s S, so _
A
_
S. Therefore, inf S =
_
S if S = .
Finally, let S = . Then inf = X because for all T _ X, T _
A
C, V C =
S. Suppose it were not the case. Then there exists C
t
such that T
A
C
t
.
However, there does not exist such a C
t
since there is no element in .
Therefore, all subsets of X, including X itself, is a lower bound of according
to _
A
. Then the greatest element according to _
A
is X. |L
I Exercise 52 (2.5.9). Let Fn(X. Y ) be the set of all functions mapping a subset
of X into Y [i.e., Fn(X. Y ) =
_
Z_X
Y
Z
]. Dene a relation 6 in Fn(X. Y ) by
6 g if and only if _ g.
a. Prove that 6 is an ordering of Fn(X. Y ).
b. Let J _ Fn(X. Y ). Show that supJ exists if and only if J is a compatible
system of functions; then supJ =
_
J.
Proof. (a) The relation 6 is reexive since _ for any Fn(X. Y ). If
6 g and g 6 , then _ g and g _ . By the Axiom of Extentionality, we
have = g; hence, 6 is antisymmetric. Finally, let 6 g, and g 6 h, where
. g. h Fn(X. Y ). Then _ g and g _ h implies that 6 g; that is, 6 is
transitive. Therefore, 6 is an ordering of Fn(X. Y ).
(b) Let J _ Fn(X. Y ). If supJ exists, there is a function supJ Fn(X. Y ) such
that for any . g Fn
_
X. ,
_
, _ supJ and g _ supJ. Suppose (.. ,) ,
and (.. :) g. Then (.. ,) supJ, and (.. :) supJ. Hence, it must be the
case that , = :; otherwise, supJ would be not a function. This proves J is a
compatible system of functions.
Now suppose J is a compatible system of functions. Then,
_
J is a function
with D
F
=
_
D
f
[ J
_
_ X; therefore,
_
J Fn(X. Y ). It is easy to see
that
_
J is an upper bound of J since _
_
J == 6
_
J for any
J. Finally, let G be any upper bound of J, then _ G for any J;
consequently,
_
_
_
J =
_
f F
_ G
_
_
==
_
J 6 G.
for any upper bound of J. This proves that supJ =
_
J. |L
I Exercise 53 (2.5.10). Let = ; let Pt () be the set of all partitions of .
Dene a relation 4 in Pt () by
S
1
4 S
2
if and only if for every C S
1
there is D S
2
such that C _ D.
(We say that the partition S
1
is a renement of the partition S
2
if S
1
4 S
2
holds.)
28 CHAPTER 2 RELATIONS, FUNCTIONS, AND ORDERINGS
a. Show that 4 is an ordering.
b. Let S
1
. S
2
Pt (). Show that {S
1
. S
2
] has an inmum. How is the equivalence
relation 1
S
related to the equivalence 1
S
1
and 1
S
2
?
c. Let T _ Pt (). Show that inf T exists.
d. Let T _ Pt (). Show that supT exists.
Proof. (a) It is clear that 4 is reexive. To see 4 is antisymmetric, let S
1
4 S
2
and S
2
4 S
1
, where S
1
. S
2
Pt (). Since S
1
4 S
2
, for every C
1
S
1
there
is D
2
S
2
such that C
1
_ D
2
. Suppose that C
1
D
2
. Since S
2
4 S
1
, there
is D
1
S
1
such that D
2
_ D
1
. Then C
1
D
1
. But then C
1
D
1
= A
contradiction. Hence, S
1
_ S
2
. Similarly, S
2
_ S
1
.
To verify that 4 is transitive, let S
1
4 S
2
, and S
2
4 S
3
, where S
1
. S
2
. S
3

Pt (). Then for every C S
1
, there is D S
2
and 1 S
3
such that C _ D _ 1;
that is, C _ 1. Hence, S
1
4 S
3
.
(b) Let S
1
. S
2
Pt (). Let L = {S Pt () : S 4 S
1
and S 4 S
2
]. Note that
L = because {{a]: a ] L. We now show
M = {C D : C S
1
and D S
2
]
is the greatest element of L. If m M, then there exist C S
1
and D S
2
such
that m = C D. Then m _ C and m _ D; that is, M 4 S
1
and M 4 S
2
; that is,
M L.
Pick an arbitrary N L. Then for every n N, there exists C S
1
such that
n _ C, and there exists D S
2
such that n _ D; that is, n _ C D M. Hence,
N 4 M and so M = inf {S
1
. S
2
].
(c) The same as (b).
(d) Let T _ Pt (). Dene U = {S Pt () : t 4 S V t T ]. Notice that U =
because U. Now we show that
supT =
_
_
_
_
C
i
t
i
C
i
: t
i
T
_
_
_
= 1.
This can be proved as follows:
v 1 U. For any C
i
t
i
T , C
i
_ C
i
L
_
C
j
t
j
C
j
1, where = i ; hence
t
i
4 1, V t
i
T .
v 1 is the least element of U. Suppose Q U. Then t
i
4 Q, V t
i
T ; then,
for any C
i
t
i
, there exists q Q such that C
i
_ q, for all t
i
T . But which
means that
_
C
i
t
i
C
i
_ q. V t
i
T . Hence, 1 4 Q, V Q U. |L
I Exercise 54 (2.5.11). Show that if (1. <) and (Q. -) are isomorphic strictly
ordered sets and < is a linear ordering, then - is a linear ordering.
SECTION 2.5 ORDERINGS 29
Proof. Let h: 1 Q be the isomorphism. Pick any q
1
. q
2
Q with q
1
= q
2
.
There exist
1
.
2
1 with
1
=
2
such that q
1
= h(
1
) and q
2
= h(
2
).
Since < is a linear ordering,
1
and
2
are comparable, say,
1
<
2
. Then
h(
1
) = q
1
- q
2
= h(
2
). |L
I Exercise 55 (2.5.12). The identity function on 1 is an isomorphism between
(1. <) and (1. <).
Proof. The function Id
P
: 1 1 is bijective, and
1
<
2
i Id
P
(
1
) <
Id
P
(
2
). |L
I Exercise 56 (2.5.13). If h is an isomorphism between (1. <) and (Q. -), then
h
-1
is an isomorphism between (Q. -) and (1. <).
Proof. Since D
h
1 = R
h
= Q, and R
h
1 = D
h
= 1, the function h
-1
: Q 1
is bijective. For all q
1
. q
2
Q, there exists unique
1
.
2
1 such that q
1
=
h(
1
) and q
2
= h(
2
); then
q
1
- q
2
== h(
1
) - h(
2
) ==
1
<
2
== h
-1
(q
1
) < h
-1
(q
2
). |L
I Exercise 57 (2.5.14). If is an isomorphism between (1
1
. <
1
) and (1
2
. <
2
),
and if g is an isomorphism between (1
2
. <
2
) and (1
3
. <
3
), then g is an iso-
morphism between (1
1
. <
1
) and (1
3
. <
3
).
Proof. First, D
gf
= D
f

-1
D
g
| = 1
1

-1
1
2
| = 1
1
. Next, for every

3
1
3
, there exists
2
1
2
such that
3
= g(
2
), and for every
2
1
2
, there
exists
1
1
1
such that
2
= (
1
). Therefore, for every
3
1
3
, there exists

1
1
1
such that
3
= g(
2
) = g((
1
)) = (g )(
1
). Hence, g : 1
1
1
3
is surjective.
To see that g is injective, let
1
=
t
1
. Then (
1
) = (
t
1
), and so
g((
1
)) = g((
t
1
)).
Finally, to see g is order-preserving, notice that

1
<
1

t
1
== (
1
) <
2
(
t
1
) == (g )(
1
) <
3
(g )(
t
1
). |L
3
NATURAL NUMBERS
3.1 Introduction to Natural Numbers
I Exercise 58 (3.1.1). . _ S(.) and there is no : such that . : S(.).
Proof. It is clear that . _ . L{.] = S(.). Given ., suppose there exists a set :
such that . :. Then there must exist some set a = such that : = . L a. If
a = {.], then : = S(.); if a = {.], then there must exist J a such that J = ..
Therefore, we have a {.]. Consequently, : = . L a . L {.] = S(.)(J). |L
3.2 Properties of Natural Numbers
I Exercise 59 (3.2.1). Let n N. Prove that there is no k N such that n < k <
n 1.
Proof. (Method 1) Let n N. Suppose there exists k N such that n < k. Then
n k; that is, n k [See Exercise 65]. If k < n 1, then k n 1 = S(n). That
is impossible by Exercise 58.
(Method 2) Suppose there exists k such that k < n1. By Lemma 2.1, k < n1
if and only if k < n or k = n. Therefore, it cannot be the case that n < k. |L
I Exercise 60 (3.2.2). Use Exercise 59 to prove for all m. n N: if m < n, then
m 1 6 n. Conclude that m < n implies m 1 < n 1 and that therefore the
successor S(n) = n 1 denes a one-to-one function in N.
Proof. m < m 1 for all m N. It follows from Exercise 59 that there is no
n N satisfying m < n < m1. Since < is linear on N, it must be the case that
m 1 6 n. Then m 1 6 n < n 1 implies that m 1 < n 1. To see S(n) is
one-to-one, let m < n. Then S(m) = m1, S(n) = n1, and so m1 < n1. |L
I Exercise 61 (3.2.3). Prove that there is a one-to-one mapping of N onto a
proper subset of N.
31
32 CHAPTER 3 NATURAL NUMBERS
Proof. Just consider S : n n 1. By Exercise 60, S is injective. By denition,
S is dened on N, i.e., D
S
= N, and by the following Exercise 62, R
S
= N{0].
Therefore, S : N N {0], as desired. |L
I Exercise 62 (3.2.4). For every n N, n = 0, there is a unique k N such that
n = k 1.
Proof. We use the induction principle in Exercise 69 to prove this claim. Let
P(.) be there is a unique k N such that . = k 1. It is clear that P(1) holds
since 1 = 0 1. The uniqueness of 0 = is from Lemma 3.1 in Chapter 1. Now
suppose that P(n) holds and consider P(n 1). We have n 1 = (k 1) 1 by
the induction assumption P(n). Note that k 1 = S(k) N. Let k 1 = k
t
. The
uniqueness of k implies that k
t
is unique. We thus complete the proof. |L
I Exercise 63 (3.2.5). For every n N, n = 0. 1, there is a unique k N such
that n = (k 1) 1.
Proof. We know from Exercise 62 that for every nonzero n N there is a
unique k
t
N, such that n = k
t
1. Now consider k
t
N. If n = 1, then k
t
= 0.
Therefore, we can impose the result of Exercise 62 on k
t
; that is, there is a
unique k N such that k
t
= k 1. Combining these above two steps, we know
for all n N, n = 0. 1, there is a unique k N such that n = (k 1) 1. |L
I Exercise 64 (3.2.6). Prove that each natural number is the set of all smaller
natural numbers; i.e., n = {m N: m < n].
Proof. Let P(.) denote . = {m N: m < .]. It is evident that P(0) holds
trivially. Assume that P(n) holds and let us consider P(n 1). We have
n 1 = n L {n] = {m N: m < n] L {n] = {m N: m < n 1] . |L
I Exercise 65 (3.2.7). For all m. n N, m < n if and only if m n.
Proof. Let P(.) be the property m < . if and only if m .. It is clear that
P(0) holds trivially. Assume that P(n) holds. Let us consider P(n 1). First
let m < n 1; then m < n or m = n. If m < n, then m n (n 1) by the
induction assumption P(n); if m = n, then m = n (n 1), too. Now assume
that m (n 1). Then either m = n or m n. We get m < n 1 in either
case. |L
I Exercise 66 (3.2.8). Prove that there is no function : N N such that for
all n N, (n) > (n 1). (There is no innite decreasing sequence of natural
numbers.)
Proof. Suppose there were such a function . Then = {(n) N: n N] _
N. Because (N. <) is well-ordered, the set {(n) N: n N] has a least element
; that is, there is m N such that (m) = . But (m1) < (m) = , which
contradicts the assumption that is the least element. |L
SECTION 3.2 PROPERTIES OF NATURAL NUMBERS 33
I Exercise 67 (3.2.9). If X _ N, then

X. < X
2

is well-ordered.
Proof. Let Y _ X be nonempty. Y has a least element , when Y is embedded
in N. But clearly , is still a least element of Y when Y is embedded in X _
N. |L
I Exercise 68 (3.2.10). In Exercise 49, let = N, b = N. Prove that - as
dened there is a well-ordering of T = NL {N]. Notice that . - , if and only if
. , holds for all .. , T.
Proof. The relation - in T = NL {N] is dened as
. - , == (.. , N and . < ,) or (. N and , = N) == . ,.
Let X _ T = NL {N] be nonempty. There are two cases:
v If N X, then X _ N, and so X has a least element since (N. <) is well-
ordered.
v If N X, then X = Y L {N], where Y _ N. Hence, Y has a least element .
But - N since N; that is, is the least element of X. |L
I Exercise 69 (3.2.11). Let P(.) be a property. Assume that k N and
a. P(k) holds.
b. For all n > k, if P(n) then P(n 1).
Then P(n) holds for all n > k.
Proof. If k = 0, then this is the original Induction Principle. So assume that
k > 0 and P(k) holds. Then, by Exercise 62, there is a unique k
t
N such that
k
t
1 = k. Dene
T = {n N: n 6 k
t
]. and C =

n N: n > k and P(n)


_
.
Notice that T C = .
We now show that = T L C is inductive. Obviously, 0 . If n T,
then either n < k
t
and so n 1 T, or n = k
t
and so n 1 = k C. If
n C, then n 1 C by assumption. Hence, N = (since _ N), and so
{n N: n > k] = N T = C. |L
I Exercise 70 (3.2.12, Finite Induction Principle). Let P(.) be a property. As-
sume that k N and
a. P(0).
b. For all n < k, P(n) implies P(n 1).
Then P(n) holds for all n 6 k.
34 CHAPTER 3 NATURAL NUMBERS
Proof. Suppose there were n < k such that P(n). Then it must be the case
that P (m), where m 1 = n. Thus, X =

a N: a < k and P(a)
_
= , and
so X has a least element, . Also = 0 since P(0) holds by assumption.
However, if P(), then P(), where 1 = , is also true. But < ,
which contradicts the assumption that is the least element of X. Therefore,
P(n) holds for all n < k.
To see P(k) holds, too, notice that there exists m N and m < k such
that m 1 = k (by Exercise 62). Because we have shown that P(m) holds,
P(m1) = P(k) also holds. |L
I Exercise 71 (3.2.13, Double Induction). Let P(.. ,) be a property. Assume
If P(k. ) holds for all k. N such that k < m or (k = m and < n).
then P(m. n) holds.
(++)
Conclude that P (m. n) holds for all m. n N.
Proof. We proceed by induction on m. Fix n N. Then P(m. n) is true for all
m N by the second version of Induction Principle. Now for every m N,
(m. n) is true for all n by the second version of Induction Principle. Hence,
P(m. n) holds for all m. n N. |L
3.3 The Recursion Theorem
I Exercise 72 (3.3.1). Let be an innite sequence of elements of , where
is ordered by -. Assume that
n
-
n1
for all n N. Prove that n < m implies

n
-
m
for all n. m N.
Proof. We proceed by induction on m in the form of Exercise 69. For an arbi-
trary n N, let P(.) denote
n
-
x
if n < .. Let k = n 1. then P (k) holds
since
n
-
n1
=
k
by assumption.
Suppose that P(m) holds, where m > k, and consider P(m 1). Since
m
-

m1
by the assumption of the exercise, and
n
-
m
by induction hypothesis
of P(m), we have
n
-
m1
.
Using the Induction Principle in the form of Exercise 69, we conclude that
P(m) holds for all m > k = n 1 > n. |L
I Exercise 73 (3.3.2). Let (. -) be a linearly ordered set and . q . We say
that q is a successor or if - q and there is no r such that - r - q.
Note that each can have at most one successor. Assume that (. -) is
nonempty and has the following properties:
a. Every has a successor.
b. Every nonempty subset of has a --least element.
SECTION 3.3 THE RECURSION THEOREM 35
c. If is not the --least element of , then is a successor of some q .
Prove that (. -) is isomorphic to (N. <). Show that the conclusion need not hold
if one of the conditions (a)(c) is omitted.
Proof. We rst show that each can have at most one successor. If q
1
and q
2
are both the successors of , and q
1
= q
2
, say, q
1
- q
2
, then - q
1
- q
2
,
in contradiction to the assumption that q
2
is a successor of .
Let a be the least element of (by (b)) and let g(.. n) be the successor of .
(for all n). Then a and g: N is well dened by (a). The Recursion
Theorem guarantees the existence of a function : N such that
v
0
= a = the least element of ;
v
n1
= g(
n
. n) = the successor of
n
.
By denition,
n
-
n1
for all n N; by Exercise 72
n
-
m
whenever
n < m. Consequently, is injective. It remains to show that is surjective.
If not, R
f
= ; let be the least element of R
f
. Then = a, the
least element of . It follows from (c) that there exists q such that is the
successor of q. There exists m N such that
m
= q; for otherwise q R
f
and q - . Hence,
m1
= by the recursive condition. Consequently, R
f
,
a contradiction. |L
I Exercise 74 (3.3.3). Give a direct proof of Theorem 3.5 in a way analogous
to the proof of the Recursion Theorem.
Proof. We rst show that there exists a unique innite sequence of nite se-
quences (J
n
Seq(S): n N) = J satisfying
J
0
= ( ) . (A)
J
n1
= G
_
J
n
. n
_
. (B)
where
G
_
J
n
. n
_
=

J
n
L
_

n. g(J
n
0
. . . . . J
n
n-1
)

_
if J
n
is a sequence of length n
( ) otherwise.
It is easy to see that G: Seq(S) N Seq(S).
Let T : (m1) Seq(S) be an m-step computation based on J
0
= ( ) and G.
Then
T
0
= ( ) . and T
k1
= G(T
k
. k) for 0 6 k < m.
Notice that T P(N Seq(S)). Let
F =

T P(N Seq(S)): T is an m-step computation for some m N


_
.
Let J =
_
F . Then
36 CHAPTER 3 NATURAL NUMBERS
v J is a function. We need only to prove the system of functions F is com-
patible. Let T. U F , D
T
= m N, D
U
= n N. Assume, e.g., m 6 n; then
m _ n ==m n = m, and it suces to show that
_
T
k
0
. . . . . T
k
k-1
_
= T
k
= U
k
=
_
U
k
0
. . . . . U
k
k-1
_
for all k < m. This can be done by induction [Exercise 70]. Surely, T
0
= ( ) =
U
0
. Next let k be such that k 1 < m, and assume T
k
= U
k
. Then
T
k1
= T
k
L
_
_
k. g(T
k
)
_
_
= U
k
L
_
_
k. g(U
k
)
_
_
= U
k1
.
Thus, T
k
= U
k
for all k < m.
v D
F
= N; R
F
_ Seq(S). We know that D
F
=
_
D
T
[ T F
_
_ N, and
R
F
_ N. To show that D
F
= N, it suces to prove that for each n N
there is an n-step computation T . We use the Induction Principle. Clearly,
T = {

0. ( )

] is a 0-step computation.
Assume that T is an n-step computation. Then the following function T

on
(n 1) 1 is an (n 1)-step computation:

T
k

= T
k
. if k 6 n
T
n1

= T
n
L

(n. g(T
n
))
_
.
We conclude that each n N is in the domain of some computation T F ,
so N _
TF
D
T
= D
F
.
v J satises condition (A) and (B). Clearly, J
0
= ( ) since T
0
= ( ) for all T
F . To show that J
n1
= G (J
n
. n) for any n N, let T be an (n 1)-step
computation; then T
k
= J
k
for all k D
T
, so J
n1
= T
n1
= G (T
n
. n) =
G (J
n
. n).
Let H : N Seq(S) be such that
H
0
= ( ) . (A
t
)
and
H
n1
= G (H
n
. n) V n N. (B
t
)
We show that J
n
= H
n
, V n N, again using induction. Certainly J
0
= H
0
.
If J
n
= H
n
, then J
n1
= G (J
n
. n) = G (H
n
. n) = H
n1
; therefore, J = H, as
claimed.
Now we can dene a function by
=
_
nN
J
n
. |L
SECTION 3.3 THE RECURSION THEOREM 37
I Exercise 75 (3.3.4). Derive the parametric version of the Recursion Theo-
rem: Let a: 1 and g: 1 N be functions. There exists a unique
function : 1 N such that
a. (. 0) = a() for all 1;
b. (. n 1) = g
_
. (. n). n
_
for all n N and 1.
Proof. Dene G:
P
N
P
by
G(.. n)() = g(. .(). n)
for .
P
and n N. Dene J : N
P
by recursion:
J
0
= a
P
. J
n1
= G(J
n
. n). (3.1)
Then, by the Recursion Theorem, there exists a unique J : N
P
satisfying
(3.1). Now let (. n) = J
n
(). Then
v (. 0) = J
0
() = a(), and
v (. n 1) = J
n1
() = G(J
n
. n)() = g(. J
n
(). n) = g(. (. n). n). |L
I Exercise 76 (3.3.5). Prove the following version of the Recursion Theorem:
Let g be a function on a subset of N into , a . Then there is a unique
sequence of elements of such that
a.
0
= a;
b.
n1
= g(
n
. n) for all n N such that (n 1) D
f
;
c. is either an innite sequence or is a nite sequence of length k 1 and
g(
k
. k) is undened.
Proof. Let

= L { a] where a . Dene g:

N

as follows:
g (.. n) =

g(.. n) if dened
a otherwise.
(3.2)
Then, by the Recursion Theorem, there exists a unique innite sequence

: N

such that

0
= a.

n1
= g(

n
. n).
If

`
= a for some N, consider

for the least such . |L
I Exercise 77 (3.3.6). Prove: If X _ N, then there is a one-to-one (nite or
innite) sequence such that R
f
= X.
Proof. Dene g: X N X by
g(.. n) = min{, X: , > .].
38 CHAPTER 3 NATURAL NUMBERS
Let a = minX. Then, by Exercise 76, there exists a unique function satisfying

0
= a and
n1
= g(
n
. n).
For every m N, we have
m1
>
m
1 >
m
; hence, is injective. It
follows from the previous exercise that is surjective. |L
3.4 Arithmetic of Natural Numbers
I Exercise 78 (3.4.1). Prove the associative low of addition: (k m) n =
k (mn) for all k. m. n N.
Proof. We use induction on n. So x k. m N. If n = 0, then
(k m) 0 = k m.
and
k (m0) = k m.
Assume that (k m) n = k (mn) and consider n 1:
(k m) (n 1) = (k m) n| 1
= k (mn)| 1
= k (mn) 1|
= k m(n 1)|. |L
I Exercise 79 (3.4.2). If m. n. k N, then m < n if and only if mk < n k.
Proof. We rst need to prove the following proposition: for any m. n N,
m < n == m1 < n 1. (3.3)
The == half has been proved in Exercise 60, so we need only to show the
== part. Assume that m1 < n 1. Then m < m1 6 n. Hence, m < n.
For the == half we use induction on k. Consider xed m. n Nwith m < n.
Clearly, m < n == m0 < n0. Assume that m < n ==mk < nk. Then
by (3.3), (mk) 1 < (n k) 1, i.e., m(k 1) < n (k 1).
For the == half we use the trichotomy law and the == half. If m
k < n k, then we cannot have m = n (lest n k < n k) nor n < m (lest
n k < mk < n k). The only alternative is m < n. |L
I Exercise 80 (3.4.3). If m. n N then m 6 n if and only if there exists k N
such that n = mk. This k is unique, so we can denote it n m, the dierence
of n and m.
Proof. For the == half we use induction on n. If n = 0, the proposition
trivially hods since there is no natural number m < 0. Assume that m < n
implies that there exists a unique k
m;n
N such that m k
m;n
= n. Now
SECTION 3.4 ARITHMETIC OF NATURAL NUMBERS 39
consider n 1. If m < n 1, then m = n or m < n. If m = n, let k
m;n1
= 1 and
so mk
m;n1
= n 1; if m < n, then by the induction hypothesis, there exists
a unique k
m;n
N such that mk
m;n
= n. Let k
m;n1
= k
m;n
1. Then
mk
m;n1
= m(k
m;n
1) = (mk
m;n
) 1 = n 1.
For the == half we use induction on k. If k = 0, it is obvious that m = n.
Now assume that m k = n implies that m 6 n. Let us suppose that for all
m. n N there exists a unique k 1 such that m (k 1) = n. Then by
Exercise 79 we have
0 < k 1 ==m0 < m(k 1)
==m < n. |L
I Exercise 81 (3.4.4). There is a unique function = ( multiplication) from N
N N such that
m= 0 = 0 for all m N:
m= (n 1) = m= n m for all m. n N.
Proof. We use the parametric version of the Recursion Theorem. Let a: N
Nbe dened as a() = 0, and g: NNN Nbe dened as g(. .. n) = ..
Then, there exists a unique function =: N N such that
m= 0 = a(m) = 0.
and
m= (n 1) = g(m. m= n. n) = m= n m. |L
I Exercise 82 (3.4.5). Prove that multiplication is commutative, associative,
and distributive over addition.
Proof. ( is commutative) We rst show that 0 commutes by showing 0 m = 0
(since m 0 = 0) for all m N. Clearly, 0 0 = 0, and if 0 m = 0, then
0 (m1) = 0 m0 = 0.
Let us now assume that n commutes, and let us show that n 1 commutes.
We prove, by induction on m, that
m (n 1) = (n 1) m for all m N. (3.4)
If m = 0, then (3.4) holds, as we have already shown. Thus let us assume that
(3.4) holds for m, and let us prove that
(m1) (n 1) = (n 1) (m1). (3.5)
We derive (3.5) as follows:
40 CHAPTER 3 NATURAL NUMBERS
(m1) (n 1) = (m1) n| (m1) = n (m1)| (m1)
= (n mn) (m1)
= (n mm) (n 1)
= (m n m) (n 1)
= m (n 1) (n 1)
= (n 1) m(n 1)
= (n 1) (m1).
( is distributive over addition) We show that for all m. n. N,
m (n ) = m n m . (3.6)
Fix m. n N. We use induction on . It is clear that m (n0) = m n = m n0 =
m n m 0. Now assume that (3.6) holds for , and let us consider 1:
m n ( 1)| = m (n ) 1|
= m (n ) m
= m n m m
= m n (m m)
= m n m ( 1).
( is associative) Fix m. n N. We use induction on . Clearly, m (n 0) =
m 0 = 0, and (m n) 0 = 0 as well. Now suppose that
m (n ) = (m n) .
Then
m n ( 1)| = m (n n)
= m (n ) m n
= (m n) m n
= (m n) ( 1). |L
IExercise 83 (3.4.6). If m. n Nand k > 0, then m < n if and only if mk < nk.
Proof. For the == half we x m. n N and use induction on k. Clearly,
m 1 < n 1 since
m 1 = m (0 1) = m 0 m = m.
and similarly for n 1. Let us assume that m < n implies m k < n k with k > 0,
and let us consider k 1:
SECTION 3.5 OPERATIONS AND STRUCTURES 41
m (k 1) = m k m
< n k m
< n k n
= n (k 1).
where the inequalities follow from Exercise 79.
The other half then follows exactly as in Exercise 79. |L
I Exercise 84 (3.4.7). Dene exponentiation of nature numbers as follows:
m
0
= 1 for all m N (in particular, 0
0
= 1):
m
n1
= m
n
m for all m. n N (in particular, 0
n
= 0 for n > 0).
Prove the usual laws of exponents.
Proof. We show that m
np
= m
n
m
p
for all m. n. N using induction on .
It is evident that
m
n0
= m
n
= m
n
1 = m
n
m
0
.
so let us assume m
np
= m
n
m
p
and consider 1:
m
n.p1/
= m
.np/1
= m
np
m
= (m
n
m
p
) m
= m
n
(m
p
m)
= m
n
m
p1
. |L
3.5 Operations and Structures
I Exercise 85 (3.5.1). Which of the following sets are closed under operations
of addition, subtraction, multiplication, and division of real number?
a. The set of all positive integers.
b. The set of all integers.
c. The set of all rational numbers.
d. The set of all negative rational numbers.
e. The empty set.
Solution. See the following table:
42 CHAPTER 3 NATURAL NUMBERS
division of real numbers
(a) Yes No Yes No
(b) Yes Yes Yes No
(c) Yes Yes Yes No
(d) Yes No Yes No
(e) Yes Yes Yes Yes
|L
I Exercise 86 (3.5.4). Let = , T = P(). Show that (T. L
B
.
B
) and
(T.
B
. L
B
) are isomorphic structures.
Proof. Dene a function h: T T as h(.) = T .. It is evident that h is
injective. To see h is surjective, notice that if , T, then , _ and so , T;
hence h( ,) = ,.
Since T = P(), both L
B
and
B
are well dened. For all .. , T,
h(. L
B
,) = T (. L
B
,) = (T .)
B
(T ,) = h(.)
B
h(,).
and similarly, h(.
B
,) = h(.) L
B
h(,). |L
I Exercise 87 (3.5.5). Refer to Example 5.7 for notation.
a. There is a real number a such that a a = a (namely, a = 0). Prove from
this that there is a
t

t
such that a
t
a
t
= a
t
. Find this a
t
.
b. For every a there is b such that a b = 0. Show that for every a
t

t
there is b
t

t
such that a
t
b
t
= 1. Find this b
t
.
Proof. It is from Example 5.7 that (. 6
A
. ) (
t
. 6
A
0 . ), and the isomor-
phism h:
t
is h(.) = e
x
.
(a) If a a = a, then
h(a a) = e
aa
= e
a
e
a
= e
a
.
Hence, there exists a
t
= e
a
= e
0
= 1 such that a
t
a
t
= a
t
.
(b) For every a
t

t
, there exists a unique a such that h(a) = a
t
. Let
b such that a b = 0. Then
h(a b) = h(a) h(b) = a
t
h(b) = e
0
= 1.
Hence, for every a
t

t
, there exists b
t
= h(b) such that a
t
b
t
= 1. |L
I Exercise 88 (3.5.6). Let Z

and Z
-
be, respectively, the sets of all positive
and negative integers. Show that (Z

. <. ) is isomorphic to (Z
-
. >. ) (where
< is the usual ordering of integers).
Proof. Dene h: Z

Z
-
by letting h(:) = :. Then h is bijective. Let
:
1
. :
2
Z

. Then :
1
< :
2
i :
1
> :
2
i h(:
1
) > h(:
2
). It is evident that
SECTION 3.5 OPERATIONS AND STRUCTURES 43
both operations on Z

and Z
-
are well dened, and h(:
1
:
2
) = (:
1
:
2
) =
(:
1
) (:
2
) = h(:
1
) h(:
2
). Thus, (Z

. <. ) (Z
-
. >. ). |L
I Exercise 89 (3.5.14). Construct the sets C
0
, C
1
, C
2
, and C
3
in Theorem 5.10
for
a. A = (R. S) and C = {0].
b. A = (R. . ) and C = {0. 1].
Proof. (a) C
0
= C = {0], C
1
= C
0
LSC
0
| = {0] L{1] = {0. 1], C
2
= C
1
LSC
1
| =
{0. 1] L {1. 2] = {0. 1. 2], and C
3
= C
2
L SC
2
| = {0. 1. 2] L {1. 2. 3] = {0. 1. 2. 3].
(b) C
0
= C = {0. 1], C
1
= C
0
L C
2
0
| L C
2
0
| = {0. 1] L {0. 1. 2] L {1. 0. 1] =
{1. 0. 1. 2], C
2
= C
1
L C
2
1
| L C
2
1
| = {1. 0. 1. 2] L {2. 1. 0. 1. 2. 3. 4] L
{3. 2. 1. 0. 1. 2. 3] = {3. 2. 1. 0. 1. 2. 3. 4], and C
3
= C
2
L C
2
2
| L C
2
2
| =
{7. 6. . 7. 8]. |L
4
FINITE, COUNTABLE, AND UNCOUNTABLE SETS
4.1 Cardinality of Sets
I Exercise 90 (4.1.1). Prove Lemma 1.5.
a. If [[ 6[T[ and[[ =[C[, then[C[ 6[T[.
b. If [[ 6[T[ and[T[ =[C[, then[[ 6[C[.
c. [[ 6[[.
d. If [[ 6[T[ and[T[ 6[C[, then[[ 6[C[.
Proof. (a) If [[ =[C[, then [C[ =[[, and so there is a bijection : C .
Since[[ 6[T[, there is an injection g: T. Then g : C T is an injection
and so[C[ 6[T[.
(b) Since[[ 6[T[, there is a bijection g: R
g
, where R
g
_ T is the image
of under g. Since[T[ =[C[, there is a bijection : T C. Let h : R
g
be
the restriction of on R
g
. Let D
t
:R
h
_ C. Then h: R
g
D
t
is a bijection.
To prove[[ 6[C[, consider h g: D
t
. This is a one-to-one correspondence
from to D
t
_ C.
(c) This claim follows two facts: (i) Id
A
is a one-to-one mapping of onto ,
and (ii) _ .
(d) Since [[ 6[T[, there is a bijection : R
f
, where R
f
_ T. Since [T[ 6
[C[, there is a bijection g: T R
g
, where R
g
_ C. Let h : g R
f
. Then
h : C is a injection and so[[ 6[C[. |L
I Exercise 91 (4.1.2). Prove
a. If [[ <[T[ and[T[ 6[C[, then[[ <[C[.
b. If [[ 6[T[ and[T[ <[C[, then[[ <[C[.
Proof. (a) [[ < [T[ means [[ 6 [T[ and [[ =[T[. We thus have [[ 6 [C[ by
Exercise 90 (d). If [[ =[C[, then[T[ 6[[ by Exercise 90 (b). But then[[ =[T[
by the Cantor-Bernstein Theorem. A contradiction.
45
46 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
(b) [T[ <[C[ means[T[ 6[C[ and[T[ =[C[. We thus have[[ 6[C[ by Exercise 90
(d). If [[ = [C[, then [C[ 6 [T[ by Exercise 90 (a). But then [T[ = [C[ by the
Cantor-Bernstein Theorem. A contradiction. |L
I Exercise 92 (4.1.3). If _ T, then[[ 6[T[.
Proof. Just consider Id
A
. This is an embedding on T, and so[[ 6[T[. |L
I Exercise 93 (4.1.4). Prove:
a. [ T[ =[T [.
b. [( T) C[ =[ (T C)[.
c. [[ 6[ T[ if T = .
Proof. (a) Let : (a. b) (b. a) for all (a. b) T. It is easy to see is a
function. To see is injective, let (a
1
. b
1
) = (a
2
. b
2
). Then (a
1
. b
1
) = (b
1
. a
1
) =
(b
2
. a
2
) = (a
2
. b
2
). To see is surjective, let (b. a) T . There must exist
(a. b) T such that (a. b) = (b. a). We thus proved that : T T
is bijective; consequently, [ T[ =[T [.
(b) Remember that ( T) C = (T C) [see Exercise 26 (b)], but as we
are ready to prove, these two sets are equipotent. Let
:
_
(a. b). c
_

_
a. (b. c)
_
. V
_
(a. b). c
_
( T) C.
With the same logic as in (a), we see that is bijective and so[( T) C[ =
[ (T C)[.
(c) If T = , we can choose some b T. Let : a (a. b) for all a . Then
: b _ T is bijective, and so[[ 6[ T[ if T = . |L
I Exercise 94 (4.1.5). Show that [S[ 6[P (S)[.
Proof. If a S, then {a] _ S; hence, {a] P (S) for each a S. Dene
A =

{a]: a S
_
.
It is clear that A _ P(S). Consider the embedding : a {a] for all a S.
Then : S A is bijective, and so[S[ 6[P(S)[.
In fact, [S[ <[P(S)[. To prove this, we need the following claim.
Claim. There is a one-to-one mapping from = to T i there is a mapping
from T onto .
Proof. If : T is one-to-one, and R
f
= T
+
_ T, then let
g(.) =

-1
(.) if . T
+
a
0
if . T T
+
, where a
0
.
SECTION 4.1 CARDINALITY OF SETS 47
Then this g is a mapping from T onto .
Conversely, let g: T be a mapping of T onto . The relation . - ,
if g(.) = g(,) is an equivalence relation on T [See Exercise 42, p. 22]. Let
h be a choice function on the set of equivalence classes, i.e., if .|
~
is an
equivalence class, then h
_
.|
~
_
is an element of .|
~
. It is clear that the map
(.) =
_
h g
-1
_
(.) is a one-to-one mapping of into T.
To verify[S[ <[P(S)[, we want to show that there is no mapping from S onto
P(S) [note that P(S) = since P(S) at least; hence here P(S) takes the
role of in the above claim]. Let : S P(S) be any mapping. We have to
show that is not onto P(S). Let
:

a S : a (a)
_
P(S).
[Notice that by the Axiom Schema of Comprehension, is a subset of S, and
so is an element of P(S) by the Axiom of Power Set.] We claim that does not
have a preimage under . In fact, suppose that is not the case, and (a
0
) =
with some a
0
S. Then, because _ S, there are two possibilities:
v a
0
, i.e., a
0
(a
0
) which is not possible for then a
0
cannot be in by
the denition of .
v a
0
, which is gain not possible, for then a
0
(a
0
), so a
0
should belong
to .
Thus, in either case we have arrived at a contradiction, which means that a
0
with the property (a
0
) = does not exist. |L
I Exercise 95 (4.1.6). Show that [[ 6

for any and any S = .


Proof. For every a , we construct a constant function
a
: S by letting

a
(s) = a for all s S. Now J : {
a
: a ] _
S
. Let g: a
a
. It is easy
to see that g is surjective. To see g is injective, let a. a
t
and a = a
t
; then
g(a) =
a
=
a
0 = g
_
a
t
_
. This proves that [[ =[J[; that is, [[ 6

, where
S = . |L
I Exercise 96 (4.1.7). If S _ T , then

; in particular, [
n
[ 6[
m
[ if
n 6 m.
Proof. For any
S
, we dene a corresponding function g
f

T
as follows
g
f
(.) =

(.) if . S
a
0
if . T S, where a
0
.
Then T :
_
g
f

T
:
S
_
_
T
. Hence, we have a bijection
S
T. If
n 6 m, then either n = m or n m. Therefore, [
n
[ 6[
m
[ if n < m. |L
I Exercise 97 (4.1.8). [T [ 6

S
T

if [S[ > 2.
48 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
Proof. Since [S[ > 2, we can pick u. S with u = . For any t T , dene a
function
t
S
T
as follows

t
(.) =

u if . = t
if . = t.
Notice that :
_

t
S
T
: t T
_
_ S
T
. Then we can dene a function g: T
as g(t ) =
t
. It is clear g is a one-to-one mapping from T onto T; therefore,
[T [ 6

S
T

. |L
I Exercise 98 (4.1.9). If [[ 6 [T[ and if is nonempty then there exists a
mapping of T onto .
Proof. [[ 6 [T[ implies that there is a one-to-one correspondence from
= onto | _ T. Dene g: T as follows:
g(.) =

-1
(.) if . |
a
0
if . T |.
where a
0
. See also the claim in Exercise 94. |L
(For Exercise 99Exercise 101) Let F be a function on P.A/ into P.A/. A
set X A is called a xed point of F if F.X/ D X. The function F is called
monotone if X Y A implies F.X/ F.Y/.
I Exercise 99 (4.1.10). Let J : P() P() be monotone. Then J has a xed
point.
Proof. Let T = {X _ : J(X) _ X]. Note that T = since, e.g., T . Now
let

X =
_
T and so

X _ X for any X T . Since J is monotone, we have
J(

X) _ J(X) _ X for every X T . Then
J
_

X
_
_

X. (4.1)
Hence,

X T .
On the other hand, (4.1) and the monotonicity of J implies that
J
_
J
_

X
_
_
_ J
_

X
_
. (4.2)
But (4.2) implies that J
_

X
_
T , too. Then, by the denition of

X, we have

X _ J
_

X
_
. (4.3)
Therefore, (4.1) and (4.3) imply that J
_

X
_
=

X, i.e.,

X is a xed point of J. |L
SECTION 4.1 CARDINALITY OF SETS 49
I Exercise 100 (4.1.11). Use Exercise 99 to give an alternative proof of the
Cantor-Bernstein Theorem.

1
Figure 4.1. Cantor-Bernstein Theorem
Proof. We use Exercise 99 to prove Lemma 4.1.7: If
1
_ T _ and[
1
[ =[[,
then[T[ =[[. Let J : P() P() be dened by
J(X) = ( T) L X|.
where :
1
is a bijection from onto
1
. Then J is monotone since
is, and so there exists a xed point C _ of J such that
C = ( T) L C|.
Let D = C. Dene a function g: T as
g(.) =

(.). if . C
.. if . D.
We now show that g is bijective.
g is surjective We have
R
g
= C| L D = C| L ( C) = C| L
_

_
( T) L C|
_
_
= C| L
_
( T)
c
C|
_
= C| L
_
T
c
C|
_
= C| L T
= T.
where the last equality holds since C| _
1
_ T [remember that : -

1
]. Thus, g is surjective indeed.
50 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
g is injective Both g C and g D are injective functions, so we need only
to show C| D = . This holds because
C| D = C|
_
T
c
C|
_
= .
Therefore, g: T is bijective, and so[T[ =[[. |L
I Exercise 101 (4.1.12). Prove that

X in Exercise 99 is the least xed point of
J, i.e., if J(X) = X for some X _ , then

X _ X.
Proof. Notice that if J(X) = X, then J(X) _ X, and so X T . Then we obtain
the conclusion just because

X =
_
T . |L
(For Exercise 102 and Exercise 103) A function F W P.A/ !P.A/ is continu-
ous if
F
_
[
i 2N
X
i
_
D
[
i 2N
F .X
i
/
holds for any nondecreasing sequence of subsets of A. [hX
i
W i 2 Ni is non-
decreasing if X
i
X
j
holds whenever i 6 j.]
I Exercise 102 (4.1.13). Prove that J used in Exercise 100 is continuous.
Proof. Let (X
i
: i N) _ P() be a nondecreasing sequence of . Then
J
_
_
_
iN
X
i
_
_
= ( T) L
_
_
_
iN
X
i
_
_
= ( T) L
_
_
_
iN
X
i
|
_
_
=
_
iN
_
( T) L X
i
|
_
=
_
iN
J (X
i
) . |L
I Exercise 103 (4.1.14). Prove that if

X is the least xed point of a monotone
continuous function J : P() P(), then

X =
_
iN
X
i
, where we dene
recursively X
0
= , X
i1
= J(X
i
).
Proof. We prove this statement with several steps.
(1) We rst show that the innite sequence (X
i
: i N) dened by X
0
= ,
X
i1
= J (X
i
) is nondecreasing [(X
i
: i N) exists by the Recursion Theorem].
We use the Induction Principle to prove this property. Let P(.) denote X
x
_
X
x1
. Then
v P(0) holds because X
0
= .
v Assume that P(n) holds, i.e., X
n
_ X
n1
. We need to show P(n 1). Notice
that
X
.n1/1
= J (X
n1
)
(1)
_ J (X
n
) = X
n1
.
SECTION 4.2 FINITE SETS 51
where (1) holds because X
n
_ X
n1
by P(n) and since J is monotone. We
thus prove P(n 1)
Therefore, by the Induction Principle, X
n
_ X
n1
, for any n N. Then by
Exercise 72, X
i
_ X
j
holds whenever i 6 , i.e., (X
i
: i N) is a nondecreasing
innite sequence.
(2) We now show
_
iN
X
i
is a xed point of J. Since J is continuous and
(X
i
: i N) is nondecreasing, we have
J
_
_
_
iN
X
i
_
_
=
_
iN
J (X
i
) = J (X
0
) L J (X
1
) L = L J (X
0
) L J (X
1
)
= X
0
L X
1
L X
2
L
=
_
iN
X
i
:
therefore,

X :
_
iN
X
i
is a xed point of J.
(3) To see

X is the least xed point of J, let X be any xed point of J, that is,
J(X) = X. Then, since _ X, we have J() _ J(X) = X as J is monotone
and X is a xed point of J. Furthermore, X
1
: J() _ X means that X
2
=
J (X
1
) _ J (X) = X. With this process, we have X
i1
= J (X
i
) _ X. Therefore,

X =
_
iN
X
i
_ X for any xed point X of J; that is,

X is the least xed point
of J.
(4) Till now, we have just proved that

X =
_
iN
X
i
is a least xed point of J,
but the exercise asks us to prove the inverse direction. However, that direction
must hold because there is only one least element in the set of all xed points
of J. |L
4.2 Finite Sets
I Exercise 104 (4.2.1). If S = {X
0
. . . . . X
n-1
] and the elements of S are mutu-
ally disjoint, then

_
S

n-1
i=0
[X
i
[.
Proof. We use the Induction Principle to prove this claim. The statement
is true if [S[ = 0. Assume that it is true for all S with [S[ = n, and let
S = {X
0
. . . . . X
n-1
. X
n
] be a set with n 1 elements, where each X
i
S is
nite, and the elements of S are mutually disjoint. By the induction hypothe-
sis, [
_
n-1
i=1
X
i
[ =

n-1
i=0
[X
i
[, and we have
[S[ =

_
_
n-1
_
i=1
X
i
_
_
L X
n

(1)
=

n-1
_
i=1
X
i

[X
i
[
(2)
=
n-1

i=1
[X
i
[ [X
n
[ =
n

i=1
[X
i
[ .
where (1) is from Theorem 4.2.7, and (2) is from the induction hypothesis. |L
52 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
I Exercise 105 (4.2.2). If X and Y are nite, then X Y is nite, and[X Y [ =
[X[ [Y [.
Proof. Let X = {.
0
. . . . . .
m-1
], and let Y = {,
0
. . . . . ,
n-1
], where (.
0
. . . . . .
m-1
)
and (,
0
. . . . . ,
n-1
) are injective nite sequences. Then
X Y =

(.. ,): . X and , Y


_
=
_
x
0
X

(.
t
. ,): , Y
_
.
Note that

(.
t
. ,): , Y
_
is nite for a xed .
t
X since Y is nite. Precisely,
since [Y [ = m, there is a bijective function : m Y , so we can construct
a bijective function g: m

(.
t
. ,): , Y
_
as g
i
= (.
t
.
i
) for all i 6 m 1.
Therefore,

{(.
t
. ,): , Y ]

= m for all .
t
X. Thus, by Theorem 4.2.7, a nite
union of nite sets is nite, we conclude that X Y is nite, and
[X Y [ =

_
x
0
X

(.
t
. ,): , Y
_

x
0
X

(.
t
. ,): , Y
_

x
0
X
[Y [ =[X[ [Y [ .
where the second equality comes from Exercise 104 because

(.
t
. ,): , Y
_

(.
tt
. ,): , Y
_
= whenever .
t
. .
tt
X and .
t
= .
tt
. |L
I Exercise 106 (4.2.3). If X is nite, then[P(X)[ = 2
[X[
.
Proof. We proceed by induction on the number of elements of X. The state-
ment is true if[X[ = 0: in this case, P() = {], and so[P()[ = 1 = 2
0
. Assume
that it is true for all X with [X[ = n. Let Y be a set with n 1 elements, i.e.,
Y = {,
0
. . . . . ,
n-1
. ,
n
]. Let X = {,
0
. . . . . ,
n-1
] and U =

U : U _ Y and ,
n
U
_
.
Then P (Y ) = P(X) L U. Since P(X) U = , and [P (X)[ =[U[, we have by
Exercise 104
[P (Y )[ =[P(X)[ [U[ =[P(X)[ [P(X)[ = 2
n
2
n
= 2
n1
= 2
[Y [
. |L
I Exercise 107 (4.2.4). If X and Y are nite, then X
Y
has[X[
[Y [
elements.
Proof. Let X = {.
0
. . . . . .
m-1
] and Y = {,
0
. . . . . ,
n
], where (.
0
. . . . . .
m-1
) and
(,
0
. . . . . ,
n
) are injective nite sequences. We use the Induction Principle on Y
to prove this claim. If [Y [ = 0, then X
Y
= X

=

( )
_
= {], and so

X
Y

=
1 =[X[
0
=[X[
[Y [
. Assume that for any nite X,

X
Y

=[X[
[Y [
if [Y [ = n N.
Now consider a nite set Y with [Y [ = n 1. Let Y = {,
0
. . . . . ,
n
]. Let Y
t
=
{,
0
. . . . . ,
n-1
]; that is,

Y
t

= n. By the induction hypothesis,

X
Y
0

=[X[
[Y
0
[
=
m
n
, i.e., there are m
n
functions in X
Y
0
. For any X
Y
0
, we can construct a set
J( ) as follows:
J( ) :

g
i
X
Y
: g
i
(,) =

(,) if , Y
t
.
i
if , = ,
n
, and i 6 m 1

.
SECTION 4.2 FINITE SETS 53
It is easy to see that X
Y
=
_
f X
Y
0 J( ), and[J( )[ =[X[ = m. Since

X
Y
0

=
m
n
by induction hypothesis, and for each there is a corresponding set J( )
with m elements; furthermore, J( ) J(
t
) = whenever =
t
. It then
follows from Exercise 104 that

X
Y

f X
Y
0
[J( )[ = m
n
m = m
n1
=[X[
[Y [
. |L
I Exercise 108 (4.2.5). If [X[ = n > k = [Y [, then the number of one-to-one
functions : Y X is n (n 1) (n k 1).
Proof. Let X = {.
0
. . . . . .
n-1
] and Y = {,
0
. . . . . ,
k-1
], where (.
0
. . . . . .
n-1
) and
(,
0
. . . . . ,
k-1
) are injective nite sequences. To construct a injective function
: Y X, we just pick k dierent elements from X. Because there are n (n
1) (n k 1) dierent ways to pick n elements from k > n elements, there
are n (n 1) (n k 1) injective functions : Y X. |L
I Exercise 109 (4.2.6). X is nite i every nonempty system of subsets of X
has a _-maximal elements.
Proof. To see the == half, let X = {.
0
. . . . . .
n-1
]. If = U _ P(X), let
m : max

[Y [ : Y U
_
. Such a set m exists since X is nite, so Y _ X is nite
[see Theorem 4.2.4], and P (X) is nite, too [see Theorem 4.2.8]. Let

Y U
satisfying

= m. Now we show

Y is a _-maximal element in U. Suppose
not; then there exists Y
t
P(X) such that

Y Y
t
, but then

<

Y
t

. A
contradiction.
For the == half, assume that X is innite, and every nonempty system of
X has a _-maximal element. Let
V :

Y _ X: Y is nite
_
.
However, there are no maximal elements in V. To see this, suppose Y V is a
_-maximal element, then consider Y
t
= Y L {,], where , Y [such a , exists
since X is innite]; then Y Y
t
and Y
t
is nite. A contradiction. |L
I Exercise 110 (4.2.7). Use Lemma 2.6 and Exercise 105 and Exercise 107 to
give easy proofs of commutativity and associativity for addition and multiplica-
tion of natural numbers, distributivity of multiplication over addition, and the
usual arithmetic properties of exponentiation.
Proof. As an example, we only prove the commutativity of addition of natural
numbers. Let [X[ = m and[Y [ = n, where X Y = and m. n N. It follows
from Lemma 2.6 that
[X L Y [ =[X[ [Y [ = mn.
Similarly, we have[Y L X[ =[Y [ [X[ = nm. Since[X L Y [ =[Y L X[, we know
that mn = n m. |L
54 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
I Exercise 111 (4.2.8). If , T are nite and X _ T, then[X[ =

aA
k
a
,
where k
a
=

X ({a] T)

.
Proof. Let 1
a
= X ({a] T) for all a . We rst show
_
aA
1
a
= X. Since
1
a
_ X for all a , we have
_
aA
1
a
_ X. Let (a. b) X. Then a and
b T, so there exists 1
a
such that (a. b) {a] T; therefore, (a. b) X 1
a
.
Consequently, X _
_
aA
1
a
. We then show that 1
a
1
a
0 = if a = a
t
, but
this is straightforward because (a. b) = (a
t
. b
t
) for any b. b
t
T when a = a
t
.
Now, follows Exercise 104, we have
[X[ =

_
aA
1
a

aA
[1
a
[ =

aA
k
a
. |L
4.3 Countable Sets
Remark. We verify that the mapping : N N N dened by
(.. ,) =
(. ,)(. , 1)
2
.
is bijective (see Figure 4.2).
(0. 0)
0
(1. 0)
2
(2. 0)
5
(3. 0)
9
(0. 1)
1
(1. 1)
4
(2. 1)
8
(0. 2)
3
(1. 2)
7
(0. 3)
6
Figure 4.2. .x; y/ |- .x y/.x y 1/=2 x.
Look at the diagonal where . , = 3 (positions 6, 7, 8, 9 in the diagram).
(.,)(.,1),2 = 6 is the sum of the rst ., = 3 integers, which accounts
for all previous diagonals (. , = 0. 1. 2). Then . locates the position within
the diagonal; e.g., . = 0 yields position 6, . = 1 position 7, . = 2 position 8,
. = 3 position 9.
To go backwards, say we are given the integer 11. Since 1 2 3 4 = 10 <
11 < 12345, we are on the diagonal with ., = 4; . = 0 gives position
10, . = 1 gives 11. Therefore . = 1, , = 4 1 = 3.
I Exercise 112 (4.3.1). Let [
1
[ =[T
1
[, [
2
[ =[T
2
[. Prove
a. If
1

2
= , T
1
T
2
= , then[
1
L
2
[ =[T
1
L T
2
[.
SECTION 4.3 COUNTABLE SETS 55
b. [
1

2
[ =[T
1
T
2
[.
c. [Seq(
1
)[ =[Seq(T
1
)[.
Remark. See the original exercise. I am afraid that there are some mistakes in
the original one.
Proof. (a) Let :
1

2
, and g: T
1
T
2
be bijections. Dene a function
h : (
1
L
2
) (T
1
L T
2
) as follows:
h(a) =

(a) if a
1
g(a) if a
2
.
It can be see that
h = L g: (
1
L
2
) (T
1
L T
2
)
is bijective since
1

2
= T
1
T
2
= .
(b) Let and g be dened as in part (a). We dene a function h :
1

2

T
1
T
2
as follows:
h(a
1
. a
2
) =
_
(a
1
). g(a
2
)
_
.
Then h is bijective.
(c) We know that

n
1

T
n
1

, V n N [see Lemma 5.1.6]. Notice that Seq(


1
) =
_
nN

n
, and Seq(T
1
) =
_
nN
T
n
1
, and

m
1

n
1
= . T
m
1
T
n
1
.
for any m = n, m. n N [because, say,
m
1
and
n
1
have dierent domains].
Therefore,
[Seq(
1
)[ =

_
nN

n
1

nN

n
1

nN

T
n
1

_
nN
T
n
1

=[Seq(T
1
)[ . |L
I Exercise 113 (4.3.2). The union of a nite set and a countable set is count-
able.
Proof. Let [[ = m, [T[ =
0
, and
t
= T . Then C = L T =
t
L T.
Since is nite,

= n 6 m. Then there exists two bijections : n


t
and
g: N T. Dene a function h : N
t
L T as follows
h(i ) =

(i ) if i < n
g(i n) if i > n.
It is easy to see that h is a bijection; thus[ L T[ =

t
L T

=
0
. |L
I Exercise 114 (4.3.3). If = is nite and T is countable, then T is
countable.
56 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
Proof. Write as = {a
0
. . . . . a
n-1
], where (a
0
. . . . . a
n-1
) is a one-to-one nite
sequence. Since T is countable, there is a bijection : N T. Pick a
i
and
consider the set

i
=

(a
i
. (n)): (n) T and n N
_
.
Then
i
is countable because there is a bijection g: n (a
i
. (n)).
Since T =
_
in

i
, that is, T is the union of a nite system of
countable sets, and so it is countable by Corollary 4.3.6. |L
I Exercise 115 (4.3.4). If = is nite, then Seq() is countable.
Proof. It suces to prove for = n N. We rst show that [Seq(n)[ >
0
. Be-
cause n = 0, we can pick an i n. Consider the following set of nite sequences
on n:
S =

s
0
= ( ) . s
1
= (i ) . s
2
= (i. i ) . s
3
= (i. i. i ) . . . .
_
.
Dene : N S by letting (n) = s
n
; then is bijective. Because S _ Seq(n),
we have
0
=[S[ 6[Seq(n)[.
We then show that Seq(n) 6
0
. This is simply because Seq(n) _ Seq(N) and
Seq(N) =
0
.
Now, by Cantor-Bernstein Theorem, [Seq(n)[ =
0
. |L
I Exercise 116 (4.3.5). Let be countable. The set |
n
=

S _ : [S[ = n
_
is
countable for all n N, n = 0.
Proof. It is enough to prove the statement for = N. We use the Induction
Principle in Exercise 69. |
1
is countable since |
1
=

{a] : a
_
, and we can
dene a bijection : |
1
by letting (a) = {a] for all a . Therefore,

|
1

= [[ =
0
. Assume that |
n
is countable; particularly, we write |
n
as |
n
= {S
1
. S
2
. . . .]. We need to prove that |
n1
is countable, too. For any
S
i
|
n
, we construct a set

i
=

S
i
L { ] : N S
i
_
.
Notice that J
i
= N S
i
is countable; in particular, there exists a bijection
g: N J
i
. Dene a bijection h: J
i

i
by letting h( ) = S
i
L { ], and we see
that [
i
[ =
0
.
Since |
n1
=
_
iN

i
, the set |
n1
is a countable union of countable sets.
Now for each i N, let a
i
= (a
i
(n): n N), where
a
i
(n) = S
i
L {g(n)] .
Then
i
= {a
i
(n): n N]. It follows from Theorem 4.3.9, |
n1
is countable.
|L
SECTION 4.3 COUNTABLE SETS 57
I Exercise 117 (4.3.6). A sequence (s
n
)
o
n=0
of natural numbers is eventually
constant if there is n
0
N, s N such that s
n
= s for all n > n
0
. Show that the
set of eventually constant sequences of natural numbers is countable.
Proof. Let C be the set of eventually constant sequences of natural numbers.
A generic element of C is

b
0
. . . . . b
n
0
-1
. s. s. . . .

, where

b
0
. . . . . b
n
0
-1

N
n
0
, and
s N.
Let Seq(N) be the set of all nite sequences of elements of N. Dene

n
0
: C Seq(N) as follows:

b
0
. . . . . b
n
0
-1
. s. s. . . .

_
=

b
0
. . . . . b
n
0
-1
. s

.
Then is bijective, and so[C[ =
0
. |L
I Exercise 118 (4.3.7). A sequence (s
n
)
o
n=0
of natural numbers is (eventually)
periodic if there are n
0
. N, > 1, such that for all n > n
0
, s
np
= s
n
. Show
that the set of all periodic sequences of natural numbers is countable.
Proof. Let P be the set of all eventually periodic sequences of natural num-
bers. A generic element of P is

b
0
. . . . . b
n
0
-1
. a
n
0
. a
n
0
1
. . . . . a
n
0
p-1
. a
n
0
. a
n
0
1
. . . . . a
n
0
p-1
. a
n
0
. . . .

.
Dene : P Seq(N) by letting

b
0
. . . . . b
n
0
-1
. a
n
0
. a
n
0
1
. . . . . a
n
0
p-1
. a
n
0
. a
n
0
1
. . . . . a
n
0
p-1
. a
n
0
. . . .

_
=

b
0
. . . . . b
n
0
-1
. a
n
0
. a
n
0
1
. . . . . a
n
0
p-1

.
is bijective, and so[P[ =
0
. |L
I Exercise 119 (4.3.8). A sequence (s
n
)
o
n=0
of natural numbers is called an
arithmetic progression if there is J N such that s
n1
= s
n
J for all n N.
Prove that the set of all arithmetic progressions is countable.
Proof. Let A be the set of all arithmetic progressions. A generic element of A
is
(a. a J. a 2J. a 3J. . . .) .
Now dene a function : A N N by letting

_
(a. a J. a 2J. . . .)
_
= (a. J) .
is bijection and so[A[ =
0
. |L
I Exercise 120 (4.3.9). For every s = (s
0
. . . . . s
n-1
) Seq(N {0]), let (s) =

s
0
0

s
n1
n-1
, where
i
is the i -th prime number. Show that is one-to-one and
use this fact to give another proof of [Seq(N)[ =
0
.
58 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
Proof. (i) We use the Induction Principle on n to show that (s) = (s
t
),
whereever s. s
t
Seq(N{0]) and s = s
t
. It is clear that
s
0
0
=
s
0
0
0
if s
0
= s
t
0
, i.e.,
this claim holds for [s[ = 1. Assume which holds for [s[ = n. We need to show
it holds for[s[ = n 1.
Suppose[s[ =

s
t

= n 1 and s = s
t
, but (s) = (s
t
); that is,

s
0
0

s
n1
n-1

s
n
n
=
s
0
0
0

s
0
n1
n-1

s
0
n
n
. (4.4)
There are two cases make (4.4) hold:
v s
n
= s
t
n
. In this case, (s
0
. . . . . s
n-1
) =

s
t
0
. . . . . s
t
n-1

, and by the inductive hy-


pothesis,
s
0
0

s
n1
n-1
=
s
0
0
0

s
0
n1
n-1
. Therefore, (4.4) implies that

s
n
n
=
s
0
n
n
. (4.5)
but which means that s
n
= s
t
n
. A contradiction.
v s
n
= s
t
n
. In this case, (4.5) must hold. Under this case, there are two cases
further:
(s
0
. . . . . s
n-1
) =

s
t
0
. . . . . s
t
n-1

. Then,

s
0
0

s
n1
n-1
=
s
0
0
0

s
0
n1
n-1
. (4.6)
However, (4.6) and (4.5) imply that (4.4) fails to hold.
(s
0
. . . . . s
n-1
) =

s
t
0
. . . . . s
t
n-1

. In this case, we know by the inductive hy-


pothesis that

s
0
0

s
n1
n-1
=
s
0
0
0

s
0
n1
n-1
. (4.7)
Without loss of generality, we assume that s
n
< s
t
n
. Then (4.4) implies that

s
0
0

s
n1
n-1
=
s
0
0
0

s
0
n1
n-1

s
0
n
-s
n
n
. (4.8)
But we know from the Unique Factorization Theorem [see, for example,
Apostol 1974] that every natural number n > 1 can be represented as a
product of prime factors in only one way, apart form the order of the
factors. Therefore, (4.8) cannot hold since
n
=
i
, V i 6 n 1, and
s
t
n
s
n
> 0.
(ii) We now show is indeed onto N {0. 1]. This is follows the Unique Fac-
torization Theorem again; hence, [Seq(N 0)[ =
0
. To prove[Seq(N)[ =
0
, we
consider the following function g: Seq(N) N
g(s
0
) =
_
s
0
1
_
. V s
0
Seq(N).
where 1 is the nite sequence (1. 1. . . .) which has the same length as s
0
. Then
g is one-to-one and onto N {0. 1], which mean that
[Seq(N)[ =
0
. |L
SECTION 4.4 LINEAR ORDERINGS 59
IExercise 121 (4.3.10). Let (S. <) be a linearly ordered set and let (
n
: n N)
be an innite sequence of nite subsets of S. Then
_
o
n=0

n
is at most countable.
Proof. Because (S. <) is a linearly ordered set, and
n
_ S is nite for all
n N, we can write
n
as

n
=

s
0
. s
1
. . . . . s
[A
n
[-1
_
.
and rank the elements of
n
as
s
0
< s
1
< . . . < s
[A
n
[-1
.
Then we can construct

a
n
(k): k <[
n
[ 1

, a unique enumeration of
n
, by
letting a
n
(k) = s
k
. Therefore,
_
o
n=0

n
is at most countable. |L
I Exercise 122 (4.3.11). Any partition of an at most countable set has a set of
representatives.
Proof. Let P be a partition of . Then there exists an equivalence relation -
on induced by P. Since is at most countable, the set of equivalence classes,
, -= {a|
~
: a ], is at most countable. Hence,
, -= (a
1
|
~
. a
2
|
~
. . . .) .
and so there is a set of representatives: {a
1
. a
2
. . . .]. |L
4.4 Linear Orderings
I Exercise 123 (4.4.1). Assume that (
1
. <
1
) is similar to (T
1
. -
1
) and (
2
. <
2
)
is similar to (T
2
. -
2
).
a. The sum of (
1
. <
1
) and (
2
. <
2
) is similar to the sum of (T
1
. -
1
) and (T
2
. -
2
),
assuming that
1

2
= = T
1
T
2
.
b. The lexicographic product of (
1
. <
1
) and (
2
. <
2
) is similar to the lexico-
graphic product of (T
1
. -
1
) and (T
2
. -
2
).
Proof. We use (. <) (T. -) to denote that (. <) is similar to (T. -).
(a) Let (. <) be the sum of (
1
. <
1
) and (
2
. <
2
), and let (T. -) be the sum of
(T
1
. -
1
) and (T
2
. -
2
). Then both (. <) and (T. -) are linearly ordered sets (by
Lemma 4.4.5 and Exercise 49). Because (
1
. <
1
) (T
1
. -
1
), there is an isomor-
phism
1
: (
1
. <
1
) (T
1
. -
1
); similarly, there is an isomorphism
2
:
2
T
2
since (
2
. <
2
) (T
2
. -
2
). Dene a bijection g: T by g =
1
L
2
.
To see a
1
< a
2
i g(a
1
) - g(a
2
), notice that (i) If a
1
. a
2

1
, then g(a
1
) =

1
(a
1
) and g(a
2
) =
2
(a
2
); hence, a
1
<
1
a
2
i a
1
< a
2
i
1
(a
1
) -
1

1
(a
2
) i
g(a
1
) - g(a
2
). (ii) If a
1
. a
2

2
we get the similarly result. (iii) If a
1

1
and
a
2

2
, then a
1
< a
2
by the denition of <. Moreover, by the denition of g,
60 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
g(a
1
) T
1
and g(a
2
) T
2
; then by the denition of -, we have g(a
1
) - g(a
2
).
For the inverse direction, suppose g(a
1
) - g(a
2
). However, since a
1

1
and
a
2

2
, we know immediately that a
1
< a
2
by denition of <. We thus proved
(. <) (T. -).
(b) Let =
1

2
and T = T
1
T
2
. We need to show that (. <) (T. -),
where < and - are the lexicographic orderings of and T. First notice that
both (. <) and (T. -) are linearly ordered sets by Lemma 4.4.6. For any
(a
1
. a
2
) , let : T be dened as
(a
1
. a
2
) =
_

1
(a
1
).
2
(a
2
)
_
.
where
1
:
1
T
1
and
2
:
2
T
2
are isomorphisms. It is easy to see that
is bijective.
Now let (a
1
. a
2
). (a
t
1
. a
t
2
) . Suppose (a
1
. a
2
) < (a
t
1
. a
t
2
); then either a
1
<
1
a
t
1
, or a
1
= a
t
1
and a
2
<
2
a
t
2
. In the rst case,
1
(a
1
) -
1

1
(a
t
1
), and so
(
1
(a
1
).
2
(a
2
)) - (
1
(a
t
1
).
2
(a
t
2
)); in the second case,
1
(a
1
) =
1
(a
t
1
) and

2
(a
2
) -
2

2
(a
t
2
) and so (
1
(a
1
).
2
(a
2
)) - (
1
(a
t
1
).
2
(a
t
2
)).
To see the inverse direction, let (
1
(a
1
).
2
(a
2
)) - (
1
(a
t
1
).
2
(a
t
2
)). Then ei-
ther
1
(a
1
) -
1

1
(a
t
1
) or
1
(a
1
) =
1
(a
t
1
) and
2
(a
2
) -
2

2
(a
t
2
). In the rst case,

1
(a
1
) -
1

1
(a
t
1
) and so a
1
<
1
a
t
1
and so (a
1
. a
2
) < (a
t
1
. a
t
2
); in the second case,
a
1
= a
t
1
and a
2
<
2
a
t
2
and so (a
1
. a
2
) < (a
t
1
. a
t
2
). |L
I Exercise 124 (4.4.2). Give an example of linear orderings (
1
. <
1
) and
(
2
. <
2
) such that the sum of (
1
. <
1
) and (
2
. <
2
) does not have the same
order type as the sum of (
2
. <
2
) and (
1
. <
1
) (addition of order types is not
commutative). Do the same thing for lexicographic product.
Proof. (i) Let (
1
. <
1
) =
_
N {0] . <
-1
_
, and (
2
. <
2
) = (N. <), where < de-
notes the usual ordering of numbers by size. Then the sum of
_
N {0] . <
-1
_
and (N. <) is just (Z. <). Particularly, there is no greatest element in (Z. <).
However, there is a greatest element in the sum of (N. <) and
_
N {0] . <
-1
_
,
namely, 1.
(ii) This is just the case of lexicographic ordering and antilexicographic or-
dering. |L
I Exercise 125 (4.4.3). Prove that the sum and the lexicographic product of
two well-orderings are well-orderings.
Proof. Let (
1
. <
1
) and (
2
. <
2
) be two well-ordered sets.
(i) Let
1

2
= and (. <) be the sum of (
1
. <
1
) and (
2
. <
2
). Let T _
be nonempty. Write T = (T
1
) L(T
2
). Let T
1
= T
1
and T
2
= T
2
.
Then T
1
_
1
, T
2
_
2
, and T
1
T
2
= . There are three cases:
v If T
1
= and T
2
= , then T
1
has a least element b
1
, and T
2
has a least
element b
2
. By denition, b
1
< b
2
and so b
1
is the least element of T.
v If T
1
= and T
2
= , then Ts least element is just b
1
.
SECTION 4.4 LINEAR ORDERINGS 61
v If T
1
= and T
2
= , then Ts least element is just b
2
.
(ii) Let < be the lexicographic ordering on =
1

2
. Take an arbitrary
nonempty subset C _ . Let C
1
be the projection of C on
1
. Then C
1
= and
so has a least element c
1
. Now take the set {c
2

2
: ( c
1
. c
2
) C]. This set is
nonempty hence has a least element c
2
. We now show that ( c
1
. c
2
) is the least
element of C: for every (c
1
. c
2
) C, either c
1
< c
1
, or c
1
= c
1
and c
2
< c
2
. In
both case, ( c
1
. c
2
) < (c
1
. c
2
). Thus, (. <) is well-ordered. |L
I Exercise 126 (4.4.4). If (
i
: i N) is an innite sequence of linearly ordered
sets of natural numbers and[
i
[ > 2 for all i N, then the lexicographic order-
ing of
iN

i
is not a well-ordering.
Proof. Because [
i
[ > 2 for all i N, we can pick a
1
i

i
, a
2
i

i
, and a
1
i
<
a
2
i
, where < is the usual linear ordering on N. Consider the innite sequence
(a
0
. a
1
. . . .), where
a
0
=
_
a
2
0
. a
2
1
. a
2
2
. a
2
3
. a
2
4
. . . .
_
.
a
1
=
_
a
1
0
. a
2
1
. a
2
2
. a
2
3
. a
2
4
. . . .
_
.
a
2
=
_
a
1
0
. a
1
1
. a
2
2
. a
2
3
. a
2
4
. . . .
_
.

In this sequence, a
n1
- a
n
by the lexicographic ordering -. More explicitly,
di(a
n1
. a
n
) = n, and a
n1
(n) = a
1
n
< a
2
n
= a
n
(n). Then the set {a
0
. a
1
. . . .]
does not have a least element, that is, the lexicographic ordering of
iN

i
is not well-ordering. |L
I Exercise 127 (4.4.5). Let ((
i
. <
i
) : i 1) be an indexed system of mutually
disjoint linearly ordered sets, 1 _ N. The relation - on
_
iI

i
dened by:
a - b i either a. b
i
and a <
i
b for some i 1 or a
i
, b
j
and
i < (in the usual ordering of natural numbers) is a linear ordering. If all <
i
are well-orderings, so is -.
Proof. We rst show that - is a linear ordering (compare with Exercise 49).
(Transitivity) Let a. b. c
_
iI

i
with a - b and b - c. If a. b. c
i
for some
i 1, then a <
i
b and b <
i
c imply that a - c; if a. b
i
, c
j
, and
i < , then a - c; if a
i
, b. c
j
, and i < , then a - c. (Asymmetry) Let
a. b
_
iI

i
and a - b. If a. b
i
, then a <
i
b, which implies that a ,
i
b,
which implies that a b; if a
i
, b
j
, and i < , then, by denition,
a b. (Linearity) Given a. b
_
iI

i
, one of the following cases has to occur:
If a. b
i
for some i 1, then a. b is comparable since <
i
is; if a
i
, b
j
,
and i < , then a - b; if a
i
, b
j
, and i > , then b - a.
Now suppose that all <
i
are well-orderings. Pick an arbitrary nonempty sub-
set _
_
iI

i
. For each a , there exists a unique i
a
1 such that a
i
a
.
Let
62 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
1
A
=

i 1 : a
i
for some a
_
.
Notice that 1
A
= . Then 1
A
has a least element i
t
. Since
i
0 is also
nonempty,
i
0 has a least element a
i
0 . Hence, a
i
0 is the least element of . |L
I Exercise 128 (4.4.6). Let (Z. <) be the set of all integers with the usual linear
ordering. Let - be the lexicographic ordering of Z
N
as dened in Theorem 4.4.7.
Finally, let FS _ Z
N
be the set of all eventually constant elements of Z
N
; i.e.,
(a
i
: i N) FS i there exists n
0
N, a Z such that a
i
= a for all i > n
0
(compare with Exercise 117). Prove that FS is countable and (FS. - FS
2
) is a
dense linear ordered set without endpoints.
Proof. The countability of FS is obtained by a similar proof as in Exercise 117.
It is also easy to see that (FS. - FS
2
) is a linear ordered set without endpoints.
So we just show that it is dense.
Take two arbitrary elements a = (a
i
: i N) and b = (b
i
: i N) in FS, and
assume that a - b. Then there exists n
0
N such that a
n
0
< b
n
0
, where n
0
is
the least element of di(a. b). Dene c = (c
i
: i N) by letting
c
i
=

a
i
if i 6 n
0
max {a
i
. b
i
] if i > n
0
.
This innite sequence c is well-dened since both a and b are eventually con-
stant. Then a - c - b. |L
I Exercise 129 (4.4.7). Let - be the lexicographic ordering of N
N
(where N
is assumed to be ordered in the usual way) and let 1 _ N
N
be the set of all
eventually periodic, but not eventually constant, sequences of natural numbers
(see Exercises 117 and 118 for denitions of these concepts). Show that (1. -
1
2
) is a countable dense linearly ordered set without endpoints.
Proof. It is evident that (1. - 1
2
) is a countable linearly ordered set, so we
focus on density. Take two arbitrary elements a. b N
N
with a - b. Then
there exists n
0
N such that a
n
0
< b
n
0
, where n
0
is dened as in the previous
exercise. Dene c 1 as in the previous exercise, we have a - c - b. |L
I Exercise 130 (4.4.8). Let (. <) be linearly ordered. Dene - on Seq() by:
(a
0
. . . . . a
m-1
) - (b
0
. . . . . b
n-1
) i there is k < n such that a
i
= b
i
for all i < k
and either a
k
< b
k
or a
k
is undened (i.e., k = m < n). Prove that - is a linear
ordering. If (. <) is well-ordered, (Seq(). -) is also well-ordered.
Proof. Transitivity: Let (a
0
. . . . . a
m-1
) - (b
0
. . . . . b
n-1
) - (c
0
. . . . . c
`-1
). Then
there exists k
1
< n such that a
i
= b
i
for all i < k
1
and either a
k
1
< b
k
1
or a
k
1
is undened. Similarly, there exists k
2
< such that b
i
< c
i
for all i < k
2
and
either b
k
2
< c
k
2
or b
k
2
is undened. Assume that k
1
< k
2
.
v If a
i
= b
i
for all i < k
1
, a
k
1
< b
k
1
, b
i
= c
i
for all i < k
2
, and b
k
2
< c
k
2
, then
a
i
= c
i
for all i < k
1
, and a
k
1
< c
k
1
, i.e., (a
0
. . . . . a
m-1
) - (c
0
. . . . . c
`-1
).
SECTION 4.4 LINEAR ORDERINGS 63
v If a
i
= b
i
, k
1
= m < n, b
i
= c
i
for all i < k
2
, and b
k
2
< c
k
2
, then a
i
= b
i
= c
i
for all i < k
1
, and a
k
1
is undened, i.e., (a
0
. . . . . a
m-1
) - (c
0
. . . . . c
`-1
).
v If a
i
= b
i
for all i < k
1
, a
k
1
< b
k
1
, b
i
= c
i
for all i < k
2
, and k
2
= n < ,
then a
i
= b
i
= c
i
for all i < k
1
, and a
k
1
< b
k
1
= c
k
1
, i.e., (a
0
. . . . . a
m-1
) -
(c
0
. . . . . c
`-1
).
We can see that (a
0
. . . . . a
m-1
) - (c
0
. . . . . c
`-1
) also holds for k
1
> k
2
.
Asymmetry: Follows from denition immediately.
Linearity: Given (a
0
. . . . . a
m-1
) . (b
0
. . . . . b
n-1
) Seq(). If m < n, then either
there exists k < m such that a
i
= b
i
for all i < k and a
k
< b
k
or a
k
> b
k
, which
implies that (a
0
. . . . . a
m-1
) - (b
0
. . . . . b
n-1
) or (a
0
. . . . . a
m-1
) > (b
0
. . . . . b
n-1
); or
a
i
= b
i
for all i < m, which implies that (a
0
. . . . . a
m-1
) - (b
0
. . . . . b
n-1
). All other
cases can be analyzed similarly.
Well-ordering: Let X _ Seq() be nonempty, and (. <) be well-ordered. Let
T
i
= {a
i
: (a
0
. . . . . a
i
. . . . . a
n-1
) X].
Then T
i
_ is nonempty and so has a least element b
i
. The sequence
(b
0
. . . . . b
`-1
) is the least element of X and so (Seq(). -) is well-ordered. |L
I Exercise 131 (4.4.10). Let (. <) be a linearly ordered set without endpoints,
= . A closed interval a. b| is dened for a. b by a. b| = {. : a 6
. 6 b]. Assume that each closed interval a. b|, a. b , has a nite number of
elements. Then (. <) is similar to the set Z of all integers in the usual ordering.
Proof. Take arbitrary a. b with a 6 b. Denote a. b| as {a
i
0
. a
i
1
. . . . . a
i
k
]
(since it is nite), where a
i
0
= a and a
i
k
= k, with a
i
0
< < a
i
k
. Let
h
a;b
=

(a
i
0
. 0). (a
i
1
. 1). . . . . (a
i
k
. k)
_
.
Clearly, h is a partial isomorphism. Now for any c , either c < a or c > b.
For example, assume that c < a. Let c. a| = {c
j
`
. . . . . c
j
0
], where c
j
`
= c and
c
j
0
= a, with c
j
`
< < c
j
0
. Let
h
c;a
=

(c
j
`
. ). . . . . (c
j
1
. 1). (c
j
0
. 0)
_
.
Let h =
_
a;bA
h
a;b
. Then h is an isomorphism and so (. <) (Q. <). |L
I Exercise 132 (4.4.11). Let (. <) be a dense linearly ordered set. Show that
for all a. b , a < b, the closed interval a. b|, as dened in Exercise 131, has
innitely many elements.
Proof. If a. b| has nitely element, then (. <) (Z. <). However, (Z. <) is
not dense. |L
I Exercise 133 (4.4.12). Show that all countable dense linearly ordered sets
with both endpoints are similar.
64 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
Proof. Let (1. -) and (Q. <) be such two sets. Let (
n
: n N) be an injective
sequence such that 1 = {
n
: n N], and let (q
n
: n N) be an injective se-
quence such that Q = {q
n
: n N]. We also assume that
0
-
1
- - and
q
0
< q
1
< < q, where is the greatest element of 1 and q is the greatest
element of Q.
Let h
0
:
0
q
0
. Having dened h
n
: {
0
. . . . .
n
] {q
0
. . . . . q
n
], we let
h
n1
: h
n
L {(
n1
. q
n1
)]. Now let h =
_
o
=0
h
i
. Then h is an isomorphism and
so (1. -) (Q. <). |L
I Exercise 134 (4.4.13). Let (Q. <) be the set of all rational numbers in the
usual ordering. Find subsets of Q similar to
a. the sum of two copies of (N. <);
b. the sum of (N. <) and (N. <
-1
);
c. the lexicographic product of (N. <) and (N. <).
Proof. For (a) and (b), we take the subset as Z. For (c), let = {m 1,(n
1): m. n N, irreducible]. We show that N N. Let h: N N be
dened as h(m 1,(n 1)) = (m. n). It is clear that h is bijective. First assume
that m
1
1,(n
1
1) < m
2
1,(n
2
1). Then it is impossible that m
1
> m
2
; for
otherwise,
1
n
1
1

1
n
2
1
> m
2
m
1
> 1.
which is impossible. If m
1
< m
2
, there is nothing to prove. So assume that
m
1
= m
2
, but then n
1
< n
2
and hence (m
1
. n
1
) < (m
2
. n
2
). The other hand can
be proved similarly. |L
4.5 Complete Linear Orderings
Remark (p. 87). Let (1. <) be a dense linearly ordered set. (1. <) is complete
i it does not have any gaps.
Proof. We rst show that if (1. <) does not have any gaps, then it is complete.
Suppose (1. <) is not complete, that is, there is a nonempty set S _ 1 bounded
from above, and S does not have a supremum. Let
=

. 1 : . 6 s for some s S
_
.
T =

. 1 : . > s for every s S


_
.
Then (. T) is a gap: = since S _ , and T = since S is bounded from
above. Next, for every 1, if > s for all s S then T; if 6 s for some
s S then , i.e., L T = 1. Finally, T = , and if a and b T
then there exists s S such that a 6 s < b, i.e., a < b.
SECTION 4.5 COMPLETE LINEAR ORDERINGS 65
If has a greatest element, or T has a least element, then has a supre-
mum, but which means that S has a supremum, too. To see this, let sup = ;.
Then ; > a for all a , and if ;
t
< ;, there exists a such that ;
t
< a 6 ;.
Since S _ , we get s 6 ; for all s S. So we need only to prove that there
exists s S such that ;
t
< s 6 ;. By denition, there exists s S such that
s > a; therefore, ;
t
< a 6 s 6 ; implies that ;
t
< s 6 ; since < is transitive.
For the other direction, assume that (1. <) has a gap (. T). Then = _
1, is bounded from above (since any element of T is an upper bound of ).
But does not have a supremum; hence (1. <) is not complete. |L
I Exercise 135 (4.5.1). Prove that there is no . Q for which .
2
= 2.
Proof. (See Rudin, 1976, for this exercise and Exercise 136.) If there were such
a . Q, we could write . = m,n, where m and n are integers that are not both
even. Let us assume this is done. Then .
2
= 2 implies
m
2
= 2n
2
. (4.9)
This shows that m
2
is even. Hence m is even (if m were odd, then m = 2k 1,
k Z, then m
2
= 2
_
2k
2
2k
_
1 is odd), and so m
2
is divisible by 4. It follows
that the right side of (4.9) is divisible by 4, so that n
2
is even, which implies
that n is even.
The assumption that .
2
= 2 holds thus leads to the conclusion that both m
and n are even, contrary to our choice of m and n. Thus, .
2
= 2 for all . Q. |L
I Exercise 136 (4.5.2). Show that (. T), where
=
_
. Q: . 6 0 or (. > 0 and .
2
< 2)
_
. T =
_
. Q: . > 0 and .
2
> 2
_
.
is a gap in (Q. <).
Proof. To show that (. T) is a gap in (Q. <), we need to show (a)(c) of the
denition hold. Since (a) and (b) are clear [note that
_
2 Q by Exercise 135],
we need only to verify (c); that is, does not have a greatest element, and T
does not have a least element.
More explicitly, for every we can nd a rational q such that < q,
and for every T such that q < . To to this, we associate with each rational
> 0 the number
q =

2
2
2
=
2 2
2
. (4.10)
Then
q
2
2 =
2(
2
2)
( 2)
2
. (4.11)
v If then
2
2 < 0, (4.10) shows that q > , and (4.11) shows that
q
2
< 2. Thus q .
66 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
v If T then
2
2 > 0, (4.10) shows that 0 < q < , and (4.11) shows that
q
2
> 2. Thus q T. |L
I Exercise 137 (4.5.3). Let 0.a
1
a
2
a
3
be an innite, but not periodic, decimal
expansion. Let
=

. Q: . 6 0.a
1
a
2
a
k
for some k N {0]
_
.
T =

. Q: . > 0.a
1
a
2
a
k
for all k N {0]
_
.
Show that (. T) is a gap in (Q. <).
Proof. It is easy to see that and T are nonempty, disjoint, and L T = Q.
Further, if a and b T, then there exists k N {0] such that a 6
0.a
1
a
2
a
k
< b.
If has a greatest element , then = 0.a
1
a
2
a
k
for some k N{0]. But
< 0.a
1
a
2
a
k
1 . Similarly, T does not have a least element. |L
I Exercise 138 (4.5.4). Show that a dense linearly ordered set (1. <) is com-
plete i every nonempty S _ 1 bounded from below has an inmum.
Proof. We rst suppose (1. <) is complete. Then by denition, every nonempty
S
t
_ 1 bounded from above has a supremum. Now suppose = S _ S is
bounded from below. Let S
t
be the set of all lower bounds of S. Since S is
bounded from below, S
t
= , and since S
t
consists of exactly those s
t
1
which satisfy the inequality s
t
6 s for every s S, we see that every s S is an
upper bound of S
t
. Thus S
t
is bounded above and
= supS
t
exists in 1 by denition of completion. We show that indeed = inf S.
v If ; < then ; is not an upper bound of S
t
, hence ; S. It follows that
6 s for every s S since s is an upper bound of S
t
. Thus is an lower
bound of S, i.e., S
t
.
v If < then S
t
, since is an upper bound of S
t
.
We have shown that S
t
but S
t
if > . In other words, is a lower
bound of S, but is not if > . This means that = inf S.
With the same logic, we can prove the inverse direction. Suppose every
nonempty S _ 1 bounded from below has an inmum. Let = S
t
_ 1 is
an arbitrary set bounded from above. We want to show that S
t
has a supre-
mum. Let S be the set of all upper bounds of S
t
. Since S
t
is bounded above,
S = , and since S consists of exactly those s 1 which satisfy the inequal-
ity s > s
t
for every s
t
S
t
, we see that every s
t
S
t
is an lower bound of S.
Therefore, S is bounded from below and
= inf S
SECTION 4.5 COMPLETE LINEAR ORDERINGS 67
exists in 1. We show that = supS
t
, too.
v As before, we rst show S. If ; > , then ; is not an lower bound of S,
hence ; S
t
. It follows that s
t
6 for every s
t
S
t
; that is, is an upper
bound of S
t
, so S.
v If < then S, since is an upper bound of S
t
.
We have shown that S but S if < . Therefore, = supS
t
. |L
I Exercise 139 (4.5.5). Let D be dense in (1. <), and let 1 be dense in (D. <).
Show that 1 is dense in (1. <).
Proof. It seems that the denition of denseness in the Theorem 4.5.3(c) is
wrong. We use the denition from Jech (2006):
Definition 4.1. a. A linear ordering (1. <) is dense if for all a < b there exists
a c such that a < c < b.
b. A set D _ 1 is a dense subset if for all a < b in 1 there exists a J D such
that a < J < b.
Let
1
.
2
1 and
1
<
2
. Since D is dense in (1. <), there exists J
1
D
such that

1
< J
1
<
2
. (4.12)
Because J
1
D _ 1, we know there exists a J
2
D such that
J
1
< J
2
<
2
. (4.13)
Because 1 is dense in (D. <), there exists e 1 such that
J
1
< e < J
2
. (4.14)
Now combine (4.12)(4.14) and adopt the fact that < is linear, we conclude
that for any
1
<
2
, there exists e 1 such that
1
< e <
2
; that is, 1 is
dense in (1. <). |L
I Exercise 140 (4.5.8). Prove that the set R Q of all irrational numbers is
dense in R.
Proof. We want to show that for any a. b R and a < b, there is an . R Q
such that a < . < b. We can chose such an . as follows:
. =

(a b),2 if . R Q
(a b),
_
2 otherwise.
|L
68 CHAPTER 4 FINITE, COUNTABLE, AND UNCOUNTABLE SETS
4.6 Uncountable Sets
IExercise 141 (4.6.1). Use the diagonal argument to show that N
N
is uncount-
able.
Proof. Consider any innite sequence
_
a
n
N
N
: n N
_
, we prove that there
is some d N
N
, and d = a
n
for all n N. This can be done by dening
d(n) = a
n
(n) 1.
Note that a
n
(n) 1 N, and d = a
n
for all n N. |L
I Exercise 142 (4.6.2). Show that

N
N

= 2
R
0
.
Proof. We rst show that N
N
_ P(N N). A generic element of N
N
can be
written as {(1. a
1
). (2. a
2
). (3. a
3
). . . .]. Since (n. a
n
) NN for all n N, we have
{(1. a
1
). (2. a
2
). . . .] _ N N; that is, {(1. a
1
). (2. a
2
). . . .] P(N N). Therefore,
2
N
_ N
N
_ P(N N).
Because [N N[ = [N[, we have [P(N N)[ = [P(N)[ (by Exercise 143); fur-
thermore,

2
N

=[P(N)[, so

2
N

=[P(N N)[. It follows from Cantor-Bernstein


Theorem that

N
N

= 2
R
0
= c. |L
I Exercise 143 (4.6.3). Show that [[ =[T[ implies[P()[ =[P(T)[.
Proof. Let : T be a bijection. For every subset a _ , we dene a
function g: P() P(T) as follows:
g(a) = a|.
where a| is the image of a under . Then it is easy to see that g is bijective.
Hence, [P()[ =[P(T)[. |L
5
CARDINAL NUMBERS
5.1 Cardinal Arithmetic
I Exercise 144 (5.1.1). Prove properties (a)(n) of cardinal arithmetic stated in
the text of this section.
a. k z = z k.
b. k (z j) = (k z) j.
c. k 6 k z.
d. If k
1
6 k
2
and z
1
6 z
2
, then k
1
z
1
6 k
2
z
2
.
e. k z = z k.
f. k
_
z j
_
= (k z) j.
g. k (z j) = k z k j.
h. k 6 k z if z > 0.
i. If k
1
6 k
2
and z
1
6 z
2
, then k
1
z
1
6 k
2
z
2
.
j. k k = 2 k.
k. k k 6 k k, whenever k > 2.
l. k 6 k

if z > 0.
m. z 6 k

if k > 1.
n. If k
1
6 k
2
and z
1
6 z
2
, then k

1
1
6 k

2
2
.
Proof. We let [[ = k, [T[ = z, and[C[ = j throughout this exercise.
(a & b) L T = T L , and L (T L C) = ( L T) L C.
(c) Let T = . Then kz =[ L T[. Considering the embedding Id
A
:
L T. Then[[ 6[ L T[, i.e., k 6 k z.
69
70 CHAPTER 5 CARDINAL NUMBERS
(d) Let [
1
[ = k
1
, [
2
[ = k
2
, [T
1
[ = z
1
, [T
2
[ = z
2
,
1
T
1
= =
2
T
2
,
[
1
[ 6[
2
[, and[T
1
[ 6[T
2
[. Let :
1

2
and g: T
1
T
2
be two injections.
Dene h:
1
L T
1

2
L T
2
by letting
h(.) =

(.) if .
1
g(.) if . T
1
.
Then h is an injection, and so k
1
z
1
6 k
2
z
2
.
(e) Let : T T with ((a. b)) = (b. a) for all (a. b) T. Then
is bijective, and so[ T[ =[T [, i.e., k z = z k.
(f) By letting : (a. (b. c)) ((a. b). c) for all (a. (b. c)) (T C), we see
that [ (T C)[ =[( T) C[; hence, k (z j) = (k z) j.
(g) (T L C) = ( T) L ( C).
(h) Pick b T (since z > 0). Dene : {b] by letting for all a :
(a) = (a. b).
Then is bijective. Since {b] _ T, we have (h).
(i) Let [
1
[ = k
1
, [
2
[ = k
2
, [T
1
[ = z
1
, [T
2
[ = z
2
, k
1
6 k
2
, and z
1
6 z
2
. Let
:
1

2
and g: T
1
T
2
be two injections. By dening h:
1
T
1

2
T
2
with
h(a. b) = ((a). g(b)).
we see that h is injective. Therefore, k
1
z
1
6 k
2
z
2
.
(j) In the book.
(k) k k 6 2 k 6 k k if k > 2, by part (j) and (i).
(l) For every a , let
a

B
be dened as
a
(b) a for all b T. Then
we dene a function J :
B
by letting J(a) =
a
. Then J is injective and
so k 6 k

if z > 0.
(m) Take a
1
. a
2
(since k > 1). For every b T, we dene a function

b
: T by letting

b
(.) =

a
1
if . = b
a
2
if . = b.
Then dene a function J : T
B
as J(b) =
b
. This function J is injective,
and so[T[ 6

.
(n) Let [
1
[ = k
1
, [
2
[ = k
2
, [T
1
[ = z
1
, [T
2
[ = z
2
, k
1
6 k
2
, and z
1
6 z
2
. Let
:
1

2
and g: T
1
T
2
be two injections. For any k
B
1
1
, we can pick a
h
k

B
2
2
such that
h
k
(.) =

( k g
-1
)(.) if . gT
1
|

b
2
if .
2
gT
1
|.
where

b
2
T
2
. Then the function J :
B
1
1

B
2
2
dened by (k) = h
k
is
injective, and so k

1
1
6 k

2
2
. |L
SECTION 5.1 CARDINAL ARITHMETIC 71
I Exercise 145 (5.1.2). Show that k
0
= 1 and k
1
= k for all k.
Proof. k
0
= 1 because

= ( ) for all .
Let [[ = k and T = {b]. Then
b
= {b] ; that is,

=[[. |L
I Exercise 146 (5.1.3). Show that 1

= 1 for all k and 0

= 0 for all k > 0.


Proof. Let = {a] and [T[ = k. In this case, {a]
B
= { : T {a]: (b) =
a for all b T]; that is,

{a]
B

= 1 =

{a]

.
Since
B
= for all T, we have

= 0 =[[. |L
I Exercise 147 (5.1.4). Prove that k

6 2

.
Proof. Let [[ = k. We look for an injection J :
A
{0. 1]
AA
. For every
element
A
, let J( ): {0. 1] be dened as
J( )(a. b) =

0 if b = (a)
1 if b = (a).
To verify J is injective, take arbitrary .
t

A
with =
t
. Then there exists
a such that (a) =
t
(a). For the pair (a. (a)) ,
J( )(a. (a)) = 1 = 0 = J(
t
)(a. (a)).
Hence, J( ) = J(
t
) whenever =
t
. Thus, k

6 2

. |L
I Exercise 148 (5.1.5). If [[ 6[T[ and if = , then there is a mapping of T
onto .
Proof. Let : T be an injection, and let a . Dene g: T as
g(b) =

-1
(b) if b |
a if b T |.
It is evident that g is surjective. |L
I Exercise 149 (5.1.6). If there is a mapping of T onto , then 2
[A[
6 2
[B[
.
Proof. Let g: T be surjective. Dene : P() P(T) as (X) = g
-1
X|.
Then is injective and so 2
[A[
=[P()[ 6[P[ (T) = 2
[B[
. |L
I Exercise 150 (5.1.7). Use Cantors Theorem to show that the set of all sets
does not exist.
Proof. Suppose U is the set of all sets. Then Y = P(
_
U) _
_
U, and so
[Y [ 6

_
U

. But Cantors Theorem says that [Y [ >

_
U

. A contradiction. |L
72 CHAPTER 5 CARDINAL NUMBERS
I Exercise 151 (5.1.8). Let X be a set and let be a one-to-one mapping of X
into itself such that X| X. Then X is innite.
Proof. : X X| is bijective, and so[X[ =[ X|[. If X is nite, it contra-
dicts Lemma 4.2.2. |L
I Exercise 152 (5.1.9). Every countable set is Dedekind innite.
Proof. It suces to consider N. Let : N N{0] be dened as (n) = n1.
Thus, N is Dedekind innite. |L
I Exercise 153 (5.1.10). If X contains a countable subset, then X is Dedekind
innite.
Proof. Let _ X be countable. Then there is an bijection : N . Dene a
function g: X X by
g((n)) = (n 1) for n N
g(.) = . for . X
(see Figure 5.1). By this construction, g: X X {g(0)] is bijective.
0 1 2 3

X
g
Figure 5.1. f.0/ is not in R
g
.
lL
I Exercise 154 (5.1.11). If X is Dedekind innite, then it contains a countable
subset.
Proof. Let X be Dedekind innite. Then there exists a bijection : X Y ,
where Y X. Pick . X Y . Let
.
0
= .. .
1
= (.
0
). . . . . .
n1
= (.
n
). . . . .
Then the set {.
n
: n N] is countable. |L
IExercise 155 (5.1.12). If and T are Dedekind nite, then LT is Dedekind
nite.
Proof. If and T are Dedekind nite, then and T does not contain a count-
able subset; hence, LT does not contain a countable subset, and so LT is
Dedekind nite. |L
SECTION 5.2 THE CARDINALITY OF THE CONTINUUM 73
IExercise 156 (5.1.13). If and T are Dedekind nite, then T is Dedekind
nite.
Proof. If and T are Dedekind nite, then and T does not contain a count-
able subset; hence, T does not contain a countable subset, and so T is
Dedekind nite. |L
I Exercise 157 (5.1.14). If is innite, then P(P()) is Dedekind innite.
Proof. For each n N, let
S
n
=

X : [X[ = n
_
.
The set {S
n
: n N] is a countable subset of P(P()), and hence P(P()) is
Dedekind innite. |L
5.2 The Cardinality of the Continuum
I Exercise 158 (5.2.1). Prove that the set of all nite sets of reals has cardinal-
ity c.
Proof. Every nite set of reals can be written as a nite union of open inter-
vals with rational endpoints. For example, we can write {a. b. c] as (a. b) L(b. c).
Thus, the cardinality of the set of all nite sets of reals is c. |L
I Exercise 159 (5.2.2). A real number . is algebraic if it is a solution of some
equation
a
n
.
n
a
n-1
.
n-1
a
1
. a
0
= 0. (+)
where a
0
. . . . . a
n
are integers. If . is not algebraic, it is called transcendental.
Show that the set of all algebraic numbers is countable and hence the set of all
transcendental numbers has cardinality c.
Proof. Let A
n
denote the set of algebraic numbers that satisfy polynomials of
the form a
k
.
k
a
1
. a
0
where k < n and max{[a
j
[] < n. Note that there
are at most n
n
polynomials of this form, and each one has at most n roots.
Hence, A
n
is a nite set having at most n
n1
<
0
elements. Let A denote the
set of all algebraic numbers. Then [A[ =

_
nN
A
n

6
0

0
=
0
.
On the other hand, consider the following set of algebraic numbers:
A
t
= {. R: a
0
. = 0. a
0
Z] .
Obviously, [A
t
[ = [Z[ and so [A[ > [Z[ =
0
. It follows from Cantor-Benstein
Theorem that [A[ =
0
. |L
I Exercise 160 (5.2.4). The set of all closed subsets of reals has cardinality c.
74 CHAPTER 5 CARDINAL NUMBERS
Proof. Let C be the set of closed sets in R, and O the set of open sets in R. A
set 1 C i R 1 O; that is, there exists a bijection : C O dened by
(1) = R 1. Thus, [C[ = [O[ = c by Theorem 5.2.6(b). |L
I Exercise 161 (5.2.5). Show that, for n > 0, n 2
c
=
0
2
c
= c 2
c
= 2
c
2
c
=
(2
c
)
n
= (2
c
)
R
0
= (2
c
)
c
= 2
c
.
Proof. We have
2
c
6 n 2
c
6
0
2
c
6 c 2
c
6 2
c
2
c
= 2
cc
= 2
c
.
2
c
6 (2
c
)
n
6 (2
c
)
R
0
6 (2
c
)
c
6= 2
c
2
= 2
c
.
and
2
c
6 n
c
6
c
0
6 (2
R
0
)
c
= 2
R
0
c
= 2
c
.
Thus, by the Cantor-Bernstein Theorem, we get the result. |L
I Exercise 162 (5.2.6). The cardinality of the set of all discontinuous functions
is 2
c
.
Proof. Let C denote the set of all continuous functions, and D the set of all
discontinuous functions. Suppose that [D[ = k < 2
c
. Then by Cantors Theo-
rem,

R
R

=[D[ [C[ = k c < 2


c
6 2
2
c
c
.
Since
2
c
c 6 2
c
2
c
= 2 2
c
= 2
c
by Exercise 161, we have

R
R

< 2
c
=

R
R

.
A contradiction. |L
I Exercise 163 (5.2.7). Construct a one-to-one mapping of R R onto R.
Proof. Using the hints. |L
6
ORDINAL NUMBERS
6.1 Well-Ordered Sets
I Exercise 164 (6.1.1). Give an example of a linearly ordered set (1. <) and
an initial segment S of 1 which is not of the form {. : . < a], for any a 1.
Proof. We know from Lemma 6.1.2 that if 1 is a well-ordered set, then every
initial segment is of the form 1a| for some a 1. Hence, we have to nd a
linear ordered set which is not well-ordered. We also know from Lemma 4.4.2
that every linear ordering on a nite set is a well-ordering. Therefore, our st
task is to nd an innite linear ordered (1. <) which is not well-ordered.
As an example, let 1 = R and S = (o. 0|. Then (R. <) is a linear ordered
set, and S is an initial segment of 1, but S = Ra| for any a R. |L
I Exercise 165 (6.1.2). o 1 is not isomorphic to o (in the well-ordering by ).
Proof. We rst show that o = N is an initial segment of o 1. By denition,
o 1 = o L {o], so o o 1. Choose any o, and let . Both and
are natural numbers, and so o. Then, by Corollary 6.1.5 (a), o 1 is not
isomorphic to o since o 1 is a well-ordered sets. |L
I Exercise 166 (6.1.3). There exist 2
R
0
well-orderings of the set of all natural
numbers.
Proof. There are
R
0
0
= c well-orderings on N. |L
I Exercise 167 (6.1.4). For every innite subset of N, (. <) is isomorphic to
(N. <).
Proof. Let _ N be innite. Notice that (. <) is a well-ordered set, and is
not an initial segment of N; for otherwise, = Nn| for some n N and so
is nite.
cannot be isomorphic to Nn| for all n N since Nn| is nite; similarly,
n| cannot be isomorphic to N. Hence, by Theorem 6.1.3, is isomorphic to
N. |L
75
76 CHAPTER 6 ORDINAL NUMBERS
I Exercise 168 (6.1.5). Let (W
1
. <
1
) and (W
2
. <
2
) be disjoint well-ordered sets,
each isomorphic to (N. <). Show that the sum of the two linearly ordered
sets is a well-ordering, and is isomorphic to the ordinal number o o =
{0. 1. 2. . . . . o. o 1. o 2. . . .].
Proof. Let (W. -) be the sum of (W
1
. <
1
) and (W
2
. <
2
). We have known that
(W. -) is a linearly ordered set. To see (W. -) is well-ordered, take an arbitrary
nonempty set X W. Then X = (W
1
X)(W
2
X), and (W
1
X)(W
2
X) = .
For i = 1. 2, if W
i
X = , then it has a least element
i
. Let = min{
1
.
2
].
Then is the least element of X.
Let
i
: W
i
N, i = 1. 2, be two isomorphisms. To see (W. -) (o o. <),
let : W
1
L W
2
o o be dened as
(n) =

1
(n) if n W
1
o
2
(n) if n W
2
.
It is clear that is an isomorphism and so (W. <) (o o. <). |L
I Exercise 169 (6.1.6). Show that the lexicographic product (N N. <) is iso-
morphic to o o.
Proof. Dene a function : NN o o as follows: for an arbitrary (m. n)
N N,
(m. n) = o mn.
Clearly, is bijective. To see is an isomorphism, let (m. n) < (. q). Then
either m < or m = and n < q. For every case, o mn < o q. |L
I Exercise 170 (6.1.7). Let (W. <) be a well-ordered set, and let a W. Extend
< to W
t
= W L {a] by making a greater than all . W. Then W has smaller
order type than W
t
.
Proof. We have W
t
a| = W. Dene a bijection : W W
t
a| as (.) = . for
all . W. Then is an isomorphism. |L
I Exercise 171 (6.1.8). The sets W = N {0. 1] and W
t
= {0. 1] N, ordered
lexicographically, are nonisomorphic well-ordered sets.
Proof. See Figures 6.1 and 6.2. The rst ordering is isomorphic to (o. <), but
the second ordering is isomorphic to (o o. <). Since o o is not isomorphic
to o (by Exercise 165, we get the result.
6.2 Ordinal Numbers
Remark. Let be a nonempty set of ordinals. Take , and consider the
set .
SECTION 6.2 ORDINAL NUMBERS 77
(0. 0) (1. 0) (2. 0) (3. 0)
(0. 1) (1. 1) (2. 1) (3. 1)
Figure 6.1. The lexicographic ordering on N0; 1.
(0. 0) (0. 1) (0. 2) (0. 3)
(1. 0) (1. 1) (1. 2) (1. 3)
Figure 6.2. The lexicographic ordering on 0; 1 N.
lL
a. If = , then is the least element of .
b. If = , then ;, where ; is the least element of , is the least element
of .
Proof. (a) If = , then for every . It follows from Theorem
6.2.6(c) that 6 for all . Hence, is the least element of .
(b) For every , if , then 6 ; if , then < . If = ,
it has a least element ; in the ordering

; that is ; 6 for any .


Further, since ; _ , we have ; < and ; . In sum,

; < 6 if
; 6 if .
Hence, ; is the least element of . |L
I Exercise 172 (6.2.1). A set X is transitive if and only if X _ P(X).
Proof. Take an arbitrary . X. If X is transitive, then . _ X, and so . P(X),
i.e., X _ P(X). On the other hand, if X _ P(X), then . X implies that
. P(X), which is equivalent to . _ X; hence X is transitive. |L
I Exercise 173 (6.2.2). A set X is transitive if and only if
_
X _ X.
Proof. Take any .
_
X, then there exists .
i
X such that . .
i
, that is,
. .
i
X; therefore, . X if X is transitive and so
_
X _ X. To see the
converse direction, let
_
X _ X. Take any .
_
X. There exists .
i
X such
that . .
i
; but . X since
_
X _ X, so X is transitive. |L
I Exercise 174 (6.2.3). Are the following sets transitive?
a.

. {]. {{]]
_
,
b.

. {]. {{]]. {. {]]
_
,
78 CHAPTER 6 ORDINAL NUMBERS
c.

. {{]]
_
.
Proof. (a) and (b) are transitive. However, (c) is not since {] {{]], but
{] {. {{]]]. |L
I Exercise 175 (6.2.4). Which of the following statements are true?
a. If X and Y are transitive, the X L Y is transitive.
b. If X and Y are transitive, the X Y is transitive.
c. If X Y and Y is transitive, then X is transitive.
d. If X _ Y and Y is transitive, then X is transitive.
e. If Y is transitive and S _ P(Y ), then Y L S is transitive.
Proof. (a), (b), and (e) are correct. |L
I Exercise 176 (6.2.5). If every X S is transitive, then
_
S is transitive.
Proof. Let u
_
S. Then there exists X S such that u X and so
u X since X is transitive. Therefore, u
_
S, i.e.,
_
S is transitive. |L
I Exercise 177 (6.2.7). If a set of ordinals X does not have a greatest element,
then supX is a limit ordinal.
Proof. If X does not have a greatest element, then supX > for all X,
and supX is the least such ordinal. If there were such that supX = 1,
then would be the greatest element of X. A contradiction. |L
I Exercise 178 (6.2.8). If X is a nonempty set of ordinals, then
_
X is an
ordinal. Moreover,
_
X is the least element of X.
Proof. If u
_
X, then u for all X, and so u for all
X, i.e., u
_
X. Hence,
_
X is transitive. It is evident to see that
_
X
is well-ordered. Thus,
_
X is an ordinal. For every X, we have
_
X _ ;
hence,
_
X 6 for all X.
We nally show that
_
X X. If not, then
_
X < ;, where ; is the least
element of X. It is impossible. |L
References
[1] Apostol, Tom M. (1974) Mathematical Analysis: Pearson Education, 2nd
edition. [58]
[2] Hrbacek, Karel and Thomas Jech (1999) Introduction to Set Theory,
220 of Pure and Applied Mathematics: A Series of Monographs and Text-
books, New York: Taylor & Francis Group, LLC, 3rd edition. [i]
[3] Jech, Thomas (2006) Set Theory, Springer Monographs in Mathematics,
Berlin: Springer-Verlag, the third millennium edition. [67]
[4] Rudin, Walter (1976) Principles of Mathematical Analysis, New York:
McGraw-Hill Companies, Inc. 3rd edition. [65]
79

Вам также может понравиться