Since we are particularly interested in the cardinality of the Reals, it turns
out to be notationally simpler to work with the "open" unit interval
0 1,
, written
0,1
.
9.5 Lemma
0,1
Proof:
This amounts to producing the appropriate 1 to 1
onto map of Dedekind cuts.
0,1
.
For clarity, this can be done as a composition,
-1,1
where
x
-1,1
0,1
where
x
We will want to calculate the cardinality of the Continuum of Real Numbers. We
begin by identifying the set of Dedekind cuts in
0,1
with
a particular set of infinite decimal fractions. Explicitly,
Let
0,1,2,3,4,5,6,7,8,9
Note
that applying 9.4,
gives
Let
be all maps not excluded by the following two bullets.
The constant map
x
9
is not in
x
is not in
if there exists an n such that
x
for all x
n.
Notationally, given
we write
i
d
We are identifying
with the infinite
decimal fractions,
.dd
d
......d
.....
Where not all the
d9
and the sequence does not end in an infinite string of 0's.
Definition
We define a map
(0,1)
as follows:
Given
,let
d(i)=.d
d
d
......d
, the ith. (finite) partial fraction.
Let
where,
d(i)
for
some i
9.6 Lemma:
(0,1)
is one to one and onto, hence
(0,1)
Proof:
As before, but note that in the definition of
,
had we not excluded .99999999...,
would have produced
In the text, the notation
"c" is used
for .
In these notes we will stick to the second notation.
The proof of the following theorem turns out to be deceptively straight forward, given the material that we will discuss on Page 14.
9.7 Theorem:
Proof:
Using 9.5 and 9.6, it suffices
to show,
, which in turn ,applying 9.1, requires producing 1 to
1 maps
and
.
One direction is trivial since inclusion gives a one to one into map
Next let
1,2,3,4,5,6,7,8
Apply 9.5 to conclude that
,
hence there is a 1 to 1 onto map
Finally,
composing this with the inclusion
gives a 1 to 1 map
From the point of view of set maps we have natural strict inclusions.
We have to make a choice for
We have shown
It is not difficult to find other sets between
and
of size
The natural question to ask is are other cardinalities between
and
?
In particular, can we find a set
such that
and
?
Cantor conjectured that this was not possible. We shall see that this conjecture has has resulted in some very interesting mathematics.
Evidence - Measure
Working Definition - Let
We say that has measure 0, if for any
there exists a, possibly countable, set of closed intervals
of length greater than 0 such that
and
1 . If
then
has measure 0.
choose some 1 to 1, onto map
.
For each
choose
such that
and
note that
and
For example if
, and
is the
i
rational
in some ordering of the rationals by
then set
2. The Cantor Middle Third Set -
Take the closed unit interval [0,1] and delete the open middle third (1,3)
Call the resulting set
. note that the "length" of
, equal the sum of the lengths of the intervals in
,
is
Suppose we have defined
as a union of closed intervals in [0,1], and that the length of
is
.
Define
to be the result of deleting the open middle third out of each of the
intervals in
.
Note that the length of
is
Finally, define
Almost by definition,
has measure 0. On the other hand
Check that
is more or less,
0,2
,using base three arithmetic.