Sunteți pe pagina 1din 20

Re-Learning to Count to Infinity

For the last 100 years students have been puzzled by the methods
used to count to infinity. It is not surprising that they were confused
because the method used was both illogical and inconsistent. By
applying one simple rule, the field of counting is restored to a selfconsistent and logical method that can be agreed upon by both
mathematicians and lay people.
In summary all we do is count within a finite set and then extend
that finite set to infinity. This simple change makes a world of
difference to the results achieved so that now we can quantify the
sizes of different infinite number sets with ease.

Contents
The Basic Infinite Set........................................................................................2
Limits and Number Space................................................................................3
Ratios...............................................................................................................4
Rules...........................................................................................................5
Calculation of Values....................................................................................6
Rational Numbers.............................................................................................6
Table of Results............................................................................................9
Infinite Areas..................................................................................................10
Ratio of Areas Above and Below a Sloping Line.............................................11
Ratio of Areas Above and Below a Parabola...................................................13
Further Reading..............................................................................................14
Objections to the Accepted Methods..............................................................14
Appendix 1: Computer Code to Count Rationals............................................15
Appendix 2: Proof of Uniqueness of Rationals................................................16
Version History...............................................................................................17

Leslie Green CEng MIEE

1 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

The Basic Infinite Set


The starting point for this work consists of an infinite set of
numbers.
For our purposes we will define a set as a collection of items,
separated by commas and contained within curly braces. As an
example we could define the set S to be
S { 1, 2, 3 }
We use the identically equal symbol (3 bar equals sign) to show
that the left hand side is just a symbol representing the right hand
side.
We are interested in the number of items (elements) in the list. We
will pretend this is evaluated using a computer function, count(). For
this case count(S) =3.
The count function naturally returns only counting numbers.
Counting numbers go by various other names such as positive
integers, whole numbers, and natural numbers. There is
inconsistent usage to say whether or not zero should be included,
but we will exclude it for simplicity.
The next basic symbol we will use will be the symbol for infinity, ,
which we will define as a positive counting number which has been
increased indefinitely.
We now introduce a very important new symbol, derived from the
earlier definitions.
count ( { 1, 2, 3, 4,
} )
In words, phi is the number of elements in the infinite set of
counting numbers.
Some might then get worried that we have introduced a circular
definition in the sense that we have effectively said
= count ( { 1, 2, 3, 4, } )

Leslie Green CEng MIEE

2 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

to which we reply that we have already adequately defined ; the


equality above is merely an observation, and not the definition.

Leslie Green CEng MIEE

3 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

Limits and Number Space


The notion of limits is a powerful mathematical method which we
will use throughout our discussions. Likewise the idea of a finite
number space is critical to these discussions.
In all discussions we will start from a finite number set and then
look at what changes as we increase the size of this set. Lets look
at a concrete example.
Consider the simple finite number set

S { 1, 2, 3, 4, N }

The last element in this set is N and this represents the current
boundary of what we are going to call the number space. We then
consider what happens as this number space increases without
bound.
The way we are going to use this finite number space is to include it
in the set definition like this
S(N) { 1, 2, 3, 4, (r: r N) }
Which we read as follows: The set S, limited to the finite number
space bounded by N. In other words we allow no values larger than
N in this set. The final element is written as (r: r N) which is read
as having the value r, such that r is less than or equal to the
bounded space value N.
But it would be very tiresome to have to write things like (2r+1:
2r+1 N) or worse still to work it out to (2r+1:
r (N-1)/2) ).
Therefore we are going to leave the finite number limit as a
parameter in the set definition and it is to be understood that the
last term in the series is implicitly limited by this value.

Leslie Green CEng MIEE

4 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

Ratios
Since is simply an infinitely large number, it has been said that
the number of counting numbers and the number of even counting
numbers is the same, namely infinite. We shall show that this is
unhelpful, illogical, and easily remedied.
Let S(N) { 1, 2, 3, 4, 5, 6 (2r+1) }
Let SE(N) {
2,
4,
6 (2r) }
In order to compare the sizes of these sets we are allowed to pair
off members (or sub-groups of members) in the sets, but crucially
we are not allowed to exceed the bounds of the number space.1
We can count the elements of an infinite set by comparison with our
basic infinite set. We do this for larger sets by linking groups of
numbers in the target set with one member of the basic set which
we will refer to as a one-to-many pairing. Alternatively, for a smaller
target set we can map several elements from the basic set to one
member in the target set, a many-to-one pairing.
Lets take a concrete example: We take 1 and 2 from the set S and
pair them with 2 from the set SE. Likewise 3 and 4 pair with 4, and
so forth. It is clear that for these two finite sets there are twice as
many elements in S(N) as in SE(N).
count ( S ( N ) )
2
count ( SE ( N ) )

for any N 2

Given that every large value of N gives this same ratio of 2, there is
no particular need to worry about taking N to an infinite limit. The
ratio is always 2. It would be illogical to invent a discontinuity at an
arbitrarily large value of N wherein the ratio suddenly dropped to 1,
and yet that is apparently the current thinking on the subject!
We can now compare this type of infinite set with and say that its
size is /2, or /3 if we uniformly discard 2 out of 3 elements, and
so forth. This forms a natural and intuitive counting methodology.
Suppose instead that we insert elements uniformly within the gaps
between the counting numbers. For example we could insert one
1

It is at this early point that we part company from Mathematicians.

Leslie Green CEng MIEE

5 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

number between 1 and 2 at 1.5. Another would be inserted


between 2 and 3 at 2.5. By pairing these with the lower integer in
the basic set we see that the set size has doubled. In general, if we
insert P elements uniformly between each counting number the size
of the resulting set is increased by a factor (P+1).
The definition of real numbers is all counting numbers plus all those
that have parts after the decimal point. It should be clear that the
set of positive real numbers can be created by inserting an infinite
number of elements between each adjacent pair of counting
numbers, but we would then need to include zero.
count( { real positive numbers} ) =
2

This is a very large number!

Rules

Note that any non-zero multiple of is an acceptable infinity.


However adding anything to (which is not itself a multiple of )
does not increase the result at all.
k k

multiplication by a constant is valid

addition of non-infinite values is irrelevant

a b a b

addition of the same powers of is valid

a 2 b a 2

higher powers of

overwhelm lower

powers
a a 2

higher powers of can be created

a
a

b
b

can be divided and cancelled

is not a valid operation

Leslie Green CEng MIEE

6 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

Calculation of Values
When we get to more complicated sets, where the density of
elements reduces with increasing position in the series, we
immediately run into a problem with the simple one-to-many or
many-to-one pairing schemes.
Consider the finite set of squares of counting numbers
SS(N) { 1, 2, 3, 4 . r }
count( SS(N) )= floor( N ) and so we get the obvious consequence
that
count( SS() ) = 0.5
from which we can generalise the result to give
SK(N) { 1k , 2 k , 3k , 4 k , 5 k , 6 k , r k }
count( SK() ) = k
Rational Numbers
n
Rational numbers are those of the form m
where n and m are
counting numbers. There are of course an infinite number of
rational numbers, but in the context of this work we need to define
just exactly how many that is.

Whilst the counting numbers are strictly a subset of the rationals,


we shall see that they form a very small subset. For our purposes it
is convenient to count only the rationals that do not include the
counting numbers in order to avoid double counting.
If the finite set has a limiting value of N then we will have N 2
possible ratios. Terms with a numerator of zero are all neglected.
Likewise terms with denominators of zero are all neglected. Terms
of the form r r are of course equal to 1 and therefore should not be
counted. Terms of the form

kp

kq for k>1 should not be counted


p
because they duplicate the ratio
q . It seems that the size of the

infinite set is of the form 2 a (a > 3 ) and by the -rules we can


Leslie Green CEng MIEE

7 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

simplify this to 2 . But we cannot yet be certain if some duplicated


terms have size b 2 ( b < 1 ) that also need to be subtracted from
the initial set.
Looking back to the finite set it is clear that for the limiting value N,
the greatest ratio value will be N/2 and the numerator range from 1
to N fits within this range. As the denominator is increased to 3 the
whole 1 to N range of numerators fits into the reduced ratio range
from 0 to N/3. Excluding duplicates, it is clear that the density of
rationals increases as we get closer to the zero end of the real
number range.
We can investigate further with some experimental computation.
Even with a limiting N value of 1000 we get 1E6 elements before
the duplicates are eliminated. Since the computation scales with N 2
we cannot go very many decades away from this starting value
before handing unreasonably large data sets. We cannot, for
example, realistically store and sort the data set to ensure that we
have eliminated all duplicates. The best we can do is to use
Euclidian division to establish that the numerator and denominator
of the rational are co-prime, meaning they have no common-factors
other than 1. (To be clear, this also eliminates the case where the
numerator and denominator are equal).
LIMIT Valid
100 59.870
%
1000 60.738
%
10,00 60.785
0
%
100,0 60.792
00
%

In this table are the results from a computer


program where LIMIT is the maximum value of
N allowed. The Valid column tells us what
percentage of the LIMIT2 combinations give
rational number pairs that are co-prime, and
therefore not giving duplicated values.

The computer code has clearly established that


the duplicated ratios do in fact form a
significant proportion of the available ratios. However, we cannot
yet be sure that we have eliminated all possible duplicates.
Specifically, if we define p, q, r & s as counting numbers, if p r
then can there be duplicates of the form

Leslie Green CEng MIEE

8 of 20

p
r

q
s

v1.10: Dec 2015

Re-Learning to Count to Infinity

The answer is no, but the proof requires some knowledge of


analytical number theory so we have pushed the proof into
Appendix 2 for the interested reader.
Using the definition that
{ irrationals }

reals

rationals

we can immediately write down the count of the irrationals as well.

Leslie Green CEng MIEE

9 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

Table of Results
We are now in a position to write down a table of values to clarify
the results. Notice that the table is sorted into size-order, with the
smallest sets listed first. (In some cases, however, the exact
ordering is dependant on the parameters chosen.)
set
count()
the cubes of all positive counting numbers
1 / 3
the cubes of all integers
2 1 / 3
the squares of all positive counting numbers
1 / 2
the squares of all integers
2 1 / 2
2 P
the squares of all real numbers with finite precision
/ ln
all prime numbers (from the Prime Number Theorem)
/N
all positive counting numbers that are evenly divisible by
N
/3
all positive counting numbers that are evenly divisible by
3
/2
all even positive counting numbers
/2
all odd positive counting numbers

all positive counting numbers

all even integers

all odd integers

the squares of all positive real numbers


2
all integers
P
all positive real numbers with finite precision
10 D
all positive real numbers with D digits after the decimal
point
2P
all real numbers with finite precision
all complex numbers with positive counting numbers for
the real parts and squares of the counting numbers for
1.5
the imaginary parts
0.39 2
all positive irrational numbers
0.61 2
all positive rational numbers
all positive real numbers
2
all real numbers
2 2
all complex numbers with integer real and imaginary parts
4 2
all complex numbers with real values for both the real &
4 4
imaginary parts
All of these results are infinite, but the sizes of the infinities have a
huge range.
Leslie Green CEng MIEE

10 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

This then completes the tools we need to classify the size of any
infinite set as being between limits defined by some multiple of
some power of .

Leslie Green CEng MIEE

11 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

Infinite Areas
If we can compare infinite counts (a 1-dimensional number space),
we should be able to compare both infinite areas and infinite
volumes by an extension of the same method. We will, however,

y
N
A

N
C

just stick to infinite areas for simplicity.


We start from a trivial case where we are asked to consider the ratio
of the areas above and below the X-axis. We start off with a limited
Cartesian space whose x extent is from N to +N and whose y
extent is also from N to +N.
The required ratio of areas is

area(A) area(B) N 2 N 2

1
area(C) area(D) N 2 N 2

Increasing N to infinity has no effect as it divides out. We have a


unique and well defined answer to the question.
We could have been asked to calculate the ratio of the areas of the
infinite quadrant B to the other three quadrants and we would have
similarly achieved a result of 1/3.

Leslie Green CEng MIEE

12 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

Ratio of Areas Above and Below a Sloping Line


If we want to deal with a sloping line through the origin we only
have to do the drawing to recognise that the triangular area missing
from one side of the Y-axis is reclaimed on the other side. The ratio
of areas remains at 1, regardless of the slope of the line up to a 45
angle. After a 45 angle the question becomes one of the ratio
between the left and right sides of the line, but the ratio still holds
as 1.

y
N

If the line does not pass through the origin we can readily see that if
the scales are increased (N increased by 10 for example) the line
will eventually look as if it does in fact pass through the origin.
Therefore the areas above and below in the limit become 1 again.
More formally we have shown that slope of the line is not relevant
to the areas above and below so we could just consider a line
parallel to the X-Axis with an offset of A. The ratio is then
ratio

Leslie Green CEng MIEE

N N A A

1
N N A A lim

13 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

We now ask: what is the ratio of areas above and below the red
lines as the number space tends to infinity?

y
N

a
N

ratio

aN 2
a

2
2
4a
4 N aN

for 0 a 1

Again there is no requirement for convergence when N becomes


infinite because there is no dependence on N.

Leslie Green CEng MIEE

14 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

Ratio of Areas Above and Below a Parabola


Next we ask for the ratio of areas above and below the parabola
y x 2 . If you feel that the cyan shaded areas are not strictly below
the parabola then realise that we have two and only two areas,
bounded by the curve. Anything that is not above is necessarily the
other side of the boundary and therefore below.

Since we restrict the number space to N in each axis, the parabola


is constrained between the two vertical blue lines. We neglect the
actual area under the curve that is between the blue lines for
simplicity.
ratio

2N N
1

2
4N 2N N
2 N 1

Without much effort we have shown that the ratio of areas above
and below the parabola tends to zero as N tends to infinity, even
though the area above the parabola is actually infinite! If the
parabola is not exactly at the origin it is evident from our previous
discussions that the relative offset will become negligible as we
zoom out from the plot and the ratio of areas will remain at zero.
Leslie Green CEng MIEE

15 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

Further Reading
Since the methods presented here are generally in disagreement
with the prior literature, the best you can do with further reading is
to see the Mathematicians viewpoint. There is a vast range of
mathematical literature available which can be accessed using the
search terms below.
Georg Cantor (1845-1918)
injection / surjection / bijection
Aleph-null, Aleph-one
countably infinite, uncountable sets
continuum hypothesis, axiom of choice
cardinality, cardinality of the continuum
Dedekind-infinite set
There is also a term called density for an infinite series, but again,
whilst this has similarities to the methods presented here, it
diverges rapidly giving illogical results such as that the density of
squares and primes is zero. The values we have calculated for these
values are non-zero and very different to each other.
Objections to the Accepted Methods
We hid the primary objections here at the end so as to not
confuse the reader. In the accepted method, two sets are found to
be of the same size by using different number spaces. The sets S
and SE are considered to be the same size because they can be
paired as shown below, pairing vertically adjacent elements.
Let S { 1, 2, 3, 4, r }
Let SE { 2, 4, 6, 8, 2r }
It is this fundamental step which we stated as being logically
inconsistent and intuitively wrong. Everyone knows that there are
twice as many counting numbers as there are even counting
numbers. It takes years to convince mathematics students that
their intuition is wrong, coupled with the fact that they will fail their
exams unless they agree!
It is important to note that all we have done here is to add one rule
to the counting step, namely that we must count within a

Leslie Green CEng MIEE

16 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

common number space, and all the sets mentioned readily


become countable.
Actually there is one other thing we have done. We also state that
2 which no doubt will make a lot of people happy; the old
counting method being counter-intuitive.

Leslie Green CEng MIEE

17 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

Appendix 1: Computer Code to Count Rationals

Leslie Green CEng MIEE

18 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

Appendix 2: Proof of Uniqueness of Rationals


The computer code eliminated integer and some duplicated ratios
by insisting that the numerator and denominator were co-prime, by
which we mean the numerator and denominator have no common
divisors excepting 1.
We now wish to prove that this rule in fact eliminated all possible
duplicates.
If we define p, q, r & s as counting numbers, the elimination of nonco-prime pairs removes a significant fraction of the N 2 possible
ratios. However, we cannot yet be sure that if p r then
Suppose that there are duplicates such that
Then

p
r

q
s

p
r

q
s

for p r .

ps qr t

Whilst t is clearly composite, t must have a unique prime


factorisation (provided the primes are listed in ascending order)
according to the Fundamental Theorem of Arithmetic. Since p, q, r
& s are not defined as prime, we can factorise them into unique
prime products
p p1 p2
r r1r2

q q1q 2
s s1s2

We have defined p and q as being co-prime, meaning that none of


the ps can be equal to any of the qs. Likewise none of the ss can
be equal to any of the rs. But since t has a unique prime
factorisation we would have to conclude that p r and s q , a
situation which we explicitly excluded at the beginning. The
contradiction assures us that there can be no duplicate rationals
created after our elimination of non-co-prime pairs.

Leslie Green CEng MIEE

19 of 20

v1.10: Dec 2015

Re-Learning to Count to Infinity

Version History
v1.00:
22 Dec 2015 First publication on scribd.
V1.10:
31 Dec 2015 Changed definition of and other sets to
exclude 0.

Leslie Green CEng MIEE

20 of 20

v1.10: Dec 2015

S-ar putea să vă placă și