Anda di halaman 1dari 8

A SMALL DIGRESSION

RELATIONS AND COUNTABILITY


Let us start by recalling a few facts about functions and sets.
Recall that a function f : A B is said to be
injective or one-to-one if whenever a
1
, a
2
A, f(a
1
) = f(a
2
) a
1
= a
2
.
surjective or onto if for every b B there exists a A such that f(a) = b.
bijective if it is one-to-one and onto.
Given two sets A and B, we dene
their union A B := c : either c A or c B.
their intersection A B := c : c A and c B.
their dierence A B := c : c A and c / B.
One can extend the denitions of union and intersection to an arbitrary family of sets,
A
i
: i I, as follows:


iI
A
i
:= c : c A
i
for some i I.


iI
A
i
:= c : c A
i
for every i I.
Recall also that, given sets A
1
, A
2
, . . . , A
n
, their cartesian product, A
1
A
2
A
n
,
is the collection of all ordered n-tuples (a
1
, a
2
, . . . , a
n
) such that a
i
A
i
(1 i n). We
will write A
n
for AA A
. .
n times
.
Throughout, we let N, Z, Q and R stand for the sets of natural, integer, rational
and real numbers, respectively.
1. Relations.
Denition 1.0.1 (Relation). A relation on a set A is simply a subset of the cartesian
product AA.
It is customary to write ab (to be read a is related to b) instead of (a, b) . There
are dierent kinds of relations. Here we shall look only at order relations and equivalence
relations.
Order relation: A relation on a set A is said to be an order relation if it satises the
following:
i) reexivity: aa for every a A.
ii) anti-symmetry: ab and ba a = b.
iii) transitivity: ab and bc ac.
Example: On N, the relation dened by
m n m is less than or equal to n,
is an order relation, for
i) n n (n N).
1
2 RELATIONS AND CARDINALS
ii) n m and m n m = n.
iii) l m and m n l n.
Example: Let A be a set and let T(A) be the power set of A, i.e., the set formed by all
subsets of A. On T(A) dene a relation by
a b a is a subset of b.
Then is an order relation. Indeed,
i) a a (a T(A)).
ii) a b and b a a = b.
iii) a b and b c a c.
Equivalence relation: A relation on a set A is said to be an equivalence relation if it
has the following properties:
i) reexivity: aa for every a A.
ii) symmetry: ab ba.
iii) transitivity: ab and bc ac.
Example: On N, let < be dened by m < n m is smaller than n. Then < is not
an equivalence relation because it is not reexive. For instance, 1 1.
Example: On N, let

be dened by m

n m divides n. Then

cannot be an
equivalence relation because it is not symmetric. For instance, 3

6 but 6 3.
Example: Let A be a nonempty set and let = be the relation on A dened by
a = b a is equal to b.
Then = is an equivalence relation, for
i) a = a (a A).
ii) a = b b = a.
iii) a = b and b = c a = c.
Example: Fix p N and dene a relation
p
on Z by
m
p
n p

(mn),
or equivalently,
m
p
n mn kp : k Z.
Then
i) m
p
m because p

(mm).
ii) m
p
n p

(mn) p

(n m) n
p
m.
iii) m
p
n and n
p
k p

(mn) and p

(n k) p

(mn + n k) m
p
k.
So
p
is an equivalence relation. It is called congruence module p. When m
p
n one
says that m is congruent with n modulo p. It is customary to write this last also as
m n(modp).
Example: On Z (Z 0) dene a relation by
(a, b) (c, d) ad = bc.
Then
i) (a, b) (a, b) because ab = ba (a Z, b Z 0).
RELATIONS AND CARDINALS 3
ii) (a, b) (c, d) ad = bc cb = da (c, d) (a, b).
iii) (a, b) (c, d) and (c, d) (e, f) ad = bc and cf = de adf = bcf and
bcf = bde adf = bde af = be (a, b) (e, f).
So is an equivalence relation.
An equivalence relation splits a set into disjoint pieces. To make this more precise we
need a couple of concepts.
Denition 1.0.2. Given a set A, a collection of pairwise disjoint subsets of A whose
union is the whole of A is called a partition of A.
Denition 1.0.3. Given an equivalence relation on a set A, we dene for every a A
its equivalence class by to be the set [a] := b A : ba. The collection of all
equivalence classes [a] : a A will be denoted A/.
Example: The equivalence class by
p
of an element n Z is [n] = i Z : i
p
n =
i Z : p

(i n) = i Z : i = kp + n for some k Z = kp + n : k Z.
Example: Consider in R
2
the equivalence relation dened by (x
0
, y
0
)(x
1
, y
1
)
x
1
x
0
= y
1
y
0
. Then
_
(x
0
, y
0
)

=
_
(x, y) R
2
: (x, y)(x
0
, y
0
)
_
=
_
(x, y) R
2
:
x x
0
= y y
0
_
=
_
(x, y) R
2
: y = x + (y
0
x
0
)
_
. In geometric terms the class
of (x
0
, y
0
) is simply the straight line in R
2
passing through (x
0
, y
0
) and parallel to the
straight line y = x.
Now the fundamental theorem stating the connection between equivalence relations on
a set and partitions of it reads as follows.
Theorem 1.0.1. (I) Let be an equivalence relation on a set A. Then A/ form a
partition of A.
(II) Let A
i
: i I be a partition of a set A. Then the relation on A dened by ab
a belongs to the same subset of the partition as b is an equivalence relation and
A/ = A
i
: i I.
Proof. (I) Since is reexive, a [a] (a A), so A =

aA
[a]. Let us show that
[a] [b] ,= [a] = [b]. Let c [a] [b] and let [a] arbitrary. Then ca, cb and a
a, ac and cb b [b]. This shows [a] [b]. Similarly, one shows
that [b] [a].
(II) It is easy to verify that , as dened in (II), is an equivalence relation. Let
[a] A/ and suppose a A
i
. Then [a] = b : ba = b : b belongs to the same sub-
set of the partition as a = b : b belongs to A
i
= A
i
. This shows A/ A
i
: i I.
Since A =

aA
[a] the two sets must be equal.
One immediate consequence of Theorem 1.0.1 is the following.
Corollary 1.0.1. There is a bijective correspondence between equivalence relations on a
set and partitions of it.
2. Countable and uncountable sets.
We shall call a set nite if it has a nite number of elements and innite otherwise.
4 RELATIONS AND CARDINALS
Denition 2.0.4. A set A is said to be countably innite if there exists a bijective map
f : A N. A set that is either nite or countably innite is called countable. A set that
is not countable is called uncountable.
Example: The function f : Z N, dened by
f(z) :=
_
2z, if z > 0
1 2z, if z 0
is a bijection, hence, Z is countably innite.
Example: The function f : N N N, dened by
f(m, n) :=
1
2
(m + n 1)(m + n 2) + n ((m, n) N N),
is a bijection. The following diagram illustrates the correspondence established by f.
6
-
1
1
1
3
6
10
2 4 7
5
9
8
Hence, also N N is countably innite.
As suggested by our last example, showing that a given set is countable by explicitly
exhibiting a bijective map from the set onto N can be dicult. In this respect, our next
result will prove useful. It will reduce the task of showing that a given set is countable to
nding either a one-to-one map from the set into N or an onto map from N to the set.
Proposition 2.1. Any subset of a countable set is countable.
Proof. (Sketch) Let A be a countable set, so A can be written as a sequence a
1
, a
2
, . . . ,
a
n
, . . ., and let B A. If B is nite then we are done. Otherwise, there is an increasing
sequence of positive integers, (n
k
) say, such that B = a
n
1
, a
n
2
, . . . , a
n
k
, . . .. The map
f : N B dened by f(k) := a
n
k
(k N) is a bijection.
Corollary 2.1.1. Let A be a set. If there is either a one-to-one function f : A N or
an onto function g : N A then A is countable.
Proof. First suppose there exists a one-to-one map f : A N. The corestriction f

f(A)
:
A f(A), a f(a), is a bijection and, by Proposition 2.1, there is : f(A) N
bijective. Then f

f(A)
: A N is a bijection, and so, A is countable.
Next, suppose there is g : N A onto. For each a A choose (a) i N : g(i) = a.
Then : A N is one-to-one, and so, by the previous part, A is countable.
Example: Let Q
+
(resp. Q

) stand for the set of all positive (resp. negative) rational


numbers. Let f : Q
+
N N, p/q (p, q), where we assume p and q to be relatively
prime. It is easy to see that f is one-to-one. Let : N N N be a bijection. Then
f : Q
+
N is one-to-one, and we can apply Corollary 2.1.1.
RELATIONS AND CARDINALS 5
A similar argument shows that Q

is countable.
Example: Let k N be arbitrary, let p
1
, p
2
, . . . , p
k
be distinct prime numbers, and let
f : N N N
. .
k times
N, (n
1
, n
2
, . . . , n
k
) p
n
1
1
p
n
2
2
. . . p
n
k
k
.
By the Fundamental Theorem of Arithmetic, f is one-to-one. Hence, by Corollary 2.1.1,
N
k
must be countable.
The last example can be generalized as follows.
Proposition 2.2. The cartesian product of any nite number of countable sets is count-
able.
Proof. Let X
1
, . . . , X
k
be countable sets. We can write each X
i
as x
i,1
, x
i,2
, x
i,3
, . . ..
Dene f : X
1
X
k
N
k
by f(x
1,i
1
, x
2,i
2
, . . . , x
k,i
k
) := (i
1
, i
2
, . . . , i
k
). It is easy to
see that f is one-to-one. As N
k
is countable there is a bijective map : N
k
N. The
composition map f is then a one-to-one map from X
1
X
k
to N. Thus, by
Corollary 2.1.1, X
1
X
k
must be countable.
Proposition 2.3. A countable union of countable sets is countable.
Proof. Let X
i
: i I N be a collection of countable sets. Then each X
i
can be written
as a sequence x
i,1
, x
i,2
, x
i,3
, . . .. Dene a map f : N N

i
X
i
by f(i, j) := x
i,j
. The
map f is clearly onto. Let : N N N be a bijection. Then f : N

i
X
i
is onto,
and so, by Corollary 2.1.1,

i
X
i
is countable.
Example: Q is countable. Indeed, Q
+
and Q

are countable, and Q = Q

0 Q
+
, so
one can apply Proposition 2.3.
Example: By Proposition 2.2 and our previous example, Q
n
is countable (n N).
At this stage, one may be wondering whether there are uncountable sets at all.
Example (Cantor): The set of points in the interval [0, 1] is uncountable. To see this,
suppose towards a contradiction that there is a bijective function f : N [0, 1]. Let
f(1) = 0.a
11
a
12
a
13
. . . a
1n
. . .
f(2) = 0.a
21
a
22
a
23
. . . a
2n
. . .
f(3) = 0.a
31
a
32
a
33
. . . a
3n
. . .
.
.
.
.
.
.
.
.
.
f(n) = 0.a
n1
a
n2
a
n3
. . . a
nn
. . .
.
.
.
.
.
.
.
.
.
where the expressions on the right hand side stand for the corresponding decimal repre-
sentations. For each n N, choose b
n
1, . . . , 8 a
nn
. Then 0.b
1
b
2
b
3
. . . b
n
. . . is a
number in [0, 1] which is not in the list, contradicting the fact that f is onto. Thus [0, 1]
cannot be countable.
Note: The reason for not allowing b
n
to take the values 0 and 9 is that numbers with a
nite decimal representation admit also a representation involving innitely many 9s,
e.g., 0.5 = 0.4999... (innitely many 9s!). Thus, in this way, we avoid ending with a
number which is already in the list.
6 RELATIONS AND CARDINALS
It follows readily from the last example and Proposition 2.1 that R is uncountable too.
RELATIONS AND CARDINALS 7
If we agree to say that two sets have the same size if there is a bijective correspondence
between them then the fact that it is impossible to match each real number with a positive
integer (i.e., the fact that we cannot count the real numbers) shows that not all innite
sets have the same size. Let us make this a bit more precise. In what follows, we shall
refer to the size of a set A as its cardinal and denote it by [A[.
Denition 2.3.1. Given two sets A and B, we shall say that the cardinal of A is equal
to the cardinal of B if there exists a bijective function f : A B. We will write this
as [A[ = [B[.
We will say that the cardinal of A is smaller than or equal to the cardinal of
B (or that the cardinal of B is greater than or equal to the cardinal of A) if there
is a one-to-one function f : A B. We will write this as [A[ [B[ (or as [B[ [A[).
Moreover, we will say that the cardinal of A is smaller than the cardinal of B (or
that the cardinal of B is greater than the cardinal of A) if [A[ [B[ and [A[ , = [B[.
We will write this as [A[ < [B[.
Example: By our previous results, [Z[ = [Q[ = [N[. It is customary to denote the cardinal
of N, and hence, the cardinal of any countably innite set, by the symbol
0
(to be read
Aleph zero).
Example: Also by our previous results, [N[ < [[0, 1][. The cardinal of [0, 1] is commonly
denoted by c, so the last inequality can be stated as
0
< c.
Note that, in the case of nite sets, two sets have the same cardinal if and only if they
have the same number of elements, so we can still write n for the cardinal of any set of
n elements. In this way, the notion of a cardinal becomes an extension of the notion of a
natural number.
The relation dened on the class of all sets by A B [A[ = [B[ is an equivalence
relation (Tutorial), while the relation dened by
A _ B [A[ [B[,
is an order relation. Reexivity and transitivity of the latter are easy to verify. Verifying
its anti-symmetry is far more dicult and it is the content of the following celebrated
result.
Theorem 2.3.1 (Schroeder-Bernstein theorem (SB-theorem, in short)). Let A and B be
sets such that there are injective functions f : A B and g : B A. Then there exists
a bijective function h : A B. In other words, if [A[ [B[ and [B[ [A[ then [A[ = [B[.
Proof. Omitted.
Thus, in order to show that two sets A and B have the same cardinal it suces to
exhibit injective maps f : A B and g : B A.
Example: [(0, 1)[ = [(0, 1][ = [R[ = c. To see this, simply note that the maps f : (0, 1)
(0, 1], t t; g : (0, 1] R, t t; and h : R (0, 1), t
1
2
+
t
2(|t|+1)
, are all one-to-one,
so [(0, 1)[ [(0, 1][ [R[ [(0, 1)[. Then apply Theorem 2.3.1.
Are there other cardinal numbers apart from
0
and c?
Theorem 2.3.2 (Cantors Theorem). For every set S, [S[ < [2
S
[, where we have written
2
S
for the power set of S, i.e., the set T(S) formed by all subsets of S.
8 RELATIONS AND CARDINALS
Proof. Clearly, [S[ [2
S
[. Suppose [S[ = [2
S
[. Then there is a bijective function f : S
2
S
. Let F := s S : s / f(s). As F 2
S
, F = f( s) for some s in S. But s F
s / f( s) = F, which is clearly a nonsense.

Anda mungkin juga menyukai