You are on page 1of 15

Section 5.

1 452 Construction of the Reals

Section 5.1 Construction of the Real Numbers:


Numbers:  →  →  → 

Purpose of
of Section Starting with the non negative integers, we construct, in
order, the integers, rational numbers and the real numbers, using equivalence
relations.

Introduction

At last we have reached the Holy Grail of mathematics, which is the


understanding of the “continuum”. People have tried to understand space,
time, motion, and the concept of the continuum for thousands of years. This
quest led the Pythagoreans to the discovery of irrational numbers, Zeno’s
paradoxes, infinitesimal calculus, Cantor’s set theory, and other intriguing
ideas.

The idea of a continuous range of values has always been accepted as


the model of the three basic physical measurements of length, mass, and time.
However, quantum physicists tell us that in the tiny world of quantum physics,
we cannot admit the possibility of continuous observations; namely that all
observations must take place at discrete, isolated instances. In mathematics,
we can observe continuously, at least in our heads. For instance, take a
continuum, say from 0 to 1, and calling a variable x , we can imagine
continuous quantities like x 2 or sin x . However, a physicist might argue, if a
mathematician were to peer closer and closer into the continuum, strange
things may be observed, just as it does in the physical world. It is the purpose
of this and the next section to ask ourselves, just what are real numbers, and
find out what happens when we look at them ... up close.

The path from the natural numbers 1,2,3,… to the real numbers was a
journey that took several thousands years. The natural (or counting) numbers
probably arose from counting; counting goats, sheep, or whatever possessions
early humans possessed. Fractions, or rational numbers, are simply a
refinement of counting finer units, like 1/2 bushel of wheat, 1/3 of a mile, and
were well-known numbers to Greek geometers. Although Greek
mathematicians routinely used rational1 numbers, they did not accept zero2, or
negative numbers as legitimate numbers. To them, numbers represented
something and 0 and negative numbers didn’t represent something. Believe it
or not, Columbus discovered America before mathematicians discovered
negative numbers, or maybe we should say accepted them as numbers to

1
The word rational is the adjective form of ratio.
22
The first occurrence of a symbol used to represent 0 goes back to Hindu writings in India in the
9th century.
Section 5.1 453 Construction of the Reals

include in an arithmetic system. Even such prominent Renaissance


mathematicians as Cardano in Italy, and Viete in France called negative
numbers “absurd” and “fictitious.” But gradually, mathematicians realized the
need to enlarge their thinking and address paradoxes like 2 − 5 , or the solution
of equations like x + 3 = 1 so negative numbers finally were brought into the
family of numbers. But the last step in evolution of real numbers from natural
numbers took considerably more thought. One of the great mathematical
achievements of the 19th century, was the understanding of what we call the
“real numbers”: those numbers which can be expressed in decimal notation,
whether the decimal digits stop, go on forever in a repeating pattern, or go on
in a pattern that never repeats. So now we arrive at this foundation of real
analysis, the real numbers. So what are they?

There are two basic ways to define the real numbers. First, we can play
God and bring the “laws” down from the mountaintop, so to speak, where we
lay down the rules of the game and say, Here they are, these are the real
numbers.” This approach would be called the synthetic approach, whereby we
list a series of axioms, which we feel are the embodiment of what we think a
“continuum” should be. On the other hand, we can “construct” the real
numbers, much like a carpenter does in building a house. In this approach, we
begin with a foundation of the simplest numbers, like the natural numbers
1, 2,3,... , then doing some “mathematical construction” one builds the real
numbers step by step, passing through the integers and rational numbers on
the way to the reals. It is this “construction” approach we carry out in this
section. The synthetic axiomatic approach will be carried out in the next
section.

The Building of the Real Numbers

The construction of the real numbers begins with the simplest of


mathematical objects, the non negative integers 1,2,3,….
Section 5.1 454 Construction of the Reals

 = {1, 2,3,...}

then by a series of steps, we construct the integers

 = {.... − 3, −2, −1, 0,1, 2,3,...}

followed by the rational numbers

 = { p / q : p, q ∈ , q ≠ 0}

and finally the real numbers  .

Construction of the Integers:  → 

So how does one “construct” the integers 0, ±1, ±2,... from the natural
numbers 1, 2,3,... ? The idea is define integers as pairs of nonnegative
integers, where we “think” of each pair ( m, n ) as representing the difference
m − n , and thus a pair like ( 2,5 ) would represent −3 . If we then define
addition, subtraction, and multiplication of the pairs ( m, n ) in a way that is
consistent with the arithmetic of the nonnegative integers, we have
successfully defined the integers in terms of the nonnegative integers. To
carry out this program, we start with the set

 ×  = {( m, n ) : m, n = 1, 2,...}

of pairs of nonnegative integers, called grid points, which are illustrated as


dots in Figure 1
Section 5.1 455 Construction of the Reals

Partitioning  ×  into Equivalence Classes


Figure 1

We now say ( m, n ) and ( m′, n′ ) are equivalent3 if and only if

( m, n ) ≡ ( m′, n′) if and only if m + n′ = m′ + n


It is an easy matter to show this relation is an equivalence relation on  ×  ,
and thus partitions  ×  into equivalence classes, each equivalence class
being the grid points on a straight line of the form n = m + k , k = 0, ±1, ±2,... .
See Figure 1. A few equivalences classes are listed in Table 1, along with
their designated representative ( ). Each equivalence class corresponds to
an integer. For example, the equivalence class ( 0, 0 ) denotes zero, and (1, 2 )
would be −1 and so on.

Equivalence Class Integer Correspondence


(1,3) = {(1,3) , ( 2, 4 ) , ( 3,5)…} −2
(1, 2 ) = {(1, 2 ) , ( 2,3) , ( 3, 4 )…} −1
(1,1) = {(1,1) , ( 2, 2) , ( 3, 3)…} 0

( 2,1) = {( 2,1) , ( 3, 2) , ( 4, 3)…} 1

3
The idea behind the equivalence relation is that ( m, n ) ≡ ( m′, n′) if and only if m − n = m′ − n ′
except we are not allowed to use negative numbers as of now. Hence, we get by this and say the
equivalent statement ( m, n ) ≡ ( m′, n′) if and only if m + n ′ = m′ + n
Section 5.1 456 Construction of the Reals

( 3,1) = {( 3,1) , ( 4, 2) , ( 5,3)…} 2

Five Equivalence Classes in S


Table 1

We now define the integers  as the collection of equivalences classes of S :

{
 = ... (1, 4 ), (1, 3), (1, 2 ) , (1,1), ( 2,1), ( 3,1), ( 4,1) ,... }
= {... − 3, −2, −1, 0,1, 2,3,...}

or in general
( k ,1) = k (positive integers)
(1,1) = 0 (zero)
(1, k ) = −k (negative integers)

We now define addition, subtraction, and multiplication of these newly found


integers in a manner consistent with the arithmetic of the natural numbers.
We define for p, q, r, s ∈  :

Addition: ( p, r ) ⊕ ( q, s ) = ( p + q, r + s )
Subtraction: ( p, q ) ž ( r , s ) = ( p + s, q + r )

Multiplication ( p, q ) ⊗ ( r , s ) = ( pr + qs, ps + qr )
Multiplication:

For example

Addition: ( 3,5) ⊕ (1, 4 ) = ( 4,9 ) or − 5


Subtraction: ( 3, 6 ) ž ( 2, 7 ) = (10,8 ) or +2
Multiplication (1,3) ⊗ ( 7, 2 ) = (13, 23)
Multiplication: or − 10

______________________________________

Construction of the Rationals:  →  The next step is to construct the


rational numbers  , which we do by defining them in terms of pairs ( m, n ) of
integers  . For example, we associate the fraction 2/3 with any pair ( m, n ) of
integers that satisfy 2n = 3m . Typical values for 2/3 would be
( 2, 3) , ( 4, 6) , ( −6, −9) , ( 20,30 ) ,... which we would likely choose 2 / 3 = ( 2, 3) as the
Section 5.1 457 Construction of the Reals

representative. These pairs belong to the same equivalence class on


 × (  − {0} ) defined by

( m, n ) ≡ ( m′, n′) ⇔ mn′ = m′n


For example

(1, 2 ) ≡ ( 2, 4 ) ≡ ( −3, −6 ) ≡ (13, 26) ≡ ( −5, −10)


which we would associate with the fraction 1/2. The equivalence classes can
be illustrated graphically as grid points on straight lines passing through the
origin as illustrated in Figure 2.

Equivalence Classes Defining Rational Numbers


Figure 2

A few equivalence classes with designated representative are


Section 5.1 458 Construction of the Reals

Equivalence Class Rational Correspondence


1
(1, 2 ) = {(1, 2 ) , ( 2, 4 ) , ( 3, 6 )…}
2
(1, −1) = {(1, −1) , ( 2, −2) , ( 3, −3)…} −1
3
( 3, −5) = {( 3, −5) , ( −3, 5) , ( 6, −10 )…} −
5
( 0,1) = {( 0,1)( 0, 2 ) , ( 0,3)…} 0
Five Equivalence Classes in  × (  − {0} )
Table 1

We now define a rational number p / q as

p
= ( p, q ) , p , q ∈  , q ≠ 0 .
q

Of course, just defining fractions is not enough. It is necessary we know how


to carry out arithmetic on them, like adding and subtracting.

Addition: ( p, q ) ⊕ ( r , s ) = ( ps + qr , qs )
Subtraction: ( p, q ) ž ( r , s ) = ( ps − qr , qs ) ps ≥ rq
Multiplication: ( p, q ) ⊗ ( r , s ) = ( pr , qs )

For example

14 7
Addition: (1, 2 ) ⊕ ( 2,10 ) = (14, 20 ) =or
20 10
6 1
Subtraction: ( 3, 6 ) ž (1, 4 ) = ( 6, 24 ) or =
24 4
3 1
Multiplication ( 3, 6 ) ⊗ (1, 4 ) = ( 3, 24 ) or
Multiplication: =
24 8

_______________________________________________________

Construction of Reals:  → 
There are different ways to define the real numbers and each has its
advantages and disadvantages. It is well known (See Problem 1) that for
decimal representations of rational numbers, there will always be repeating
blocks of digits. For example 1/3 = 0.333333… repeats in blocks of 1,
Section 5.1 459 Construction of the Reals

starting at the first digit, whereas the rational number 7 / 22 = 0.3181818...


(often written 0.318 ) has repeating blocks of 2, starting at the second digit. In
fact, repeating decimal digits defines the rational numbers4. If a decimal
5
expansion does not repeat, the number is not rational , which we call
irrational, like 2 = 1.414213... The net result is that the real numbers can be
defined as all positive and negative decimal expansions, with repeating and
nonrepeating digits, one group called rational numbers, the other irrational.

The disadvantage of the above definition of real numbers is that decimal


expansions doesn’t relate to points on a continuum. To relate real numbers to
a continuum, there are two basic approaches, one due to Cantor and the other
by his good friend and supporter, Richard Dedekind. Cantor’s approach
defines the real numbers as “limits” of sequences of rational numbers, like

3, 3.1, 3.14, 3.141, 3.1415, 3.14159, 3.141592,... → π

But this approach, while having an intuitive appeal, leads us into the study of
sequences, convergence, null sequences, and other ideas from real analysis
we have not introduced, hence we follow the approach of Dedekind.

Historical Note: The German mathematician Richard Dedekind (1831=1916)


was one of the greatest mathematicians of the 19th century, as well as one of
the greatest contributors to number theory and abstract algebra. His invention
of ideals in ring theory and his contributions to algebraic numbers, fields,
modules, lattice, etc were crucial in the development of modern algebra. . In
188 his book ‘Was sind and was sollen die Zahlen?’ (What are numbers and
what should they be?) laid the foundation for the real number system and was a
milestone in the history of mathematics. His definition of the real numbers by
Dedekind cuts, and his formulation of the Dedekind-Peano axioms were
important for the early development of set theory.

Although the rational numbers have many desirable properties, including


the ability to add, subtract, multiply and divide them, even totally order them,
they have the undesirable property that they contain gaps, say at 2 and π
and other points (in fact an uncountable infinite number of points). The idea is
to “fill in” those gaps, getting a new system of numbers (real numbers) which
can be placed in a one-to-one correspondence with points on a continuous
line.

4
In order that decimal expansions be unique, one agrees to represent all non-terminating blocks of
9s, like 0.2399999… by 0.2400000… .
5
The word rational is the adjective form of the word “ratio.”
Section 5.1 460 Construction of the Reals

Dedekind’s idea appeals to our intuitive grasp of the rational numbers all
aligned on a line. Dedekind’s basic idea was to partition the rational numbers
into two (disjoint) sets L, U where every rational number x in the lower set
L is less than every rational number U of the upper set U . That is

L ∪ U = , L ∩ U = ∅,  x ∈ L ∧ y ∈ U  ⇒ x < y

Such a pair of sets ( L,U ) of rational numbers is called a Dedekind cut.


cut But
even though there are an infinite number of ways to make this “cut,” Dedekind
made the seminal observation there are only three distinct types:

1. (max, no min) The lower set L has a largest member x * ∈ L but the upper
set U has no smallest member. An example of such a cut would be when
L consists of all rational numbers less than or equal to 1, and U all
rational numbers strictly greater than 1.

2. (min, no max) The upper set U has a smallest member y * ∈U but the
lower set L has no largest member. An example of such a cut would be
when L consists of all rational numbers strictly less than 1 , and U all
rational numbers greater than or equal to 1.

3. (no max, no min) The lower set L has no largest member and the upper
set U has no smallest member. An example of such a cut would be when
L consists of all rational numbers less than or equal to 0 and positive
rational numbers r satisfying r 2 < 2 while the upper set U consists of all
positive rational numbers r satisfying 2 < r 2 . (The reader can think
r < 2, 2 < r but we are not allowed to talk about irrational numbers 2
at this stage).

These three types of Dedekind cuts are illustrated in Figure 3.

Three Types of Dedekind Cuts


Figure 3
Section 5.1 461 Construction of the Reals

At this point we make the observation that it is impossible for the lower set to
have a maximum x * and the upper set to have a minimum y * since then the
average (x *
)
+ y * / 2 , which lies between x * and y * would be larger than the
largest member of L and smaller than y * the smallest member of U , which of
course violates how the Dedekind cut is made.

We are now ready to define the real numbers.

Case 1 (Rational Numbers):


Numbers): In the two cases when the lower set has a
maximum element r or the upper set has a smallest element r , we assign such
cuts to the number r = ( Lr ,U r ) . Dedekind cuts of this type occur when r is a
rational number and in this way we have a correspondence between the rational
numbers and Dedekind cuts of this type.

Case 2 (Irrational Numbers): In case neither the lower set has a maximum or
the upper set has a minimum, Dedekind simply calls such cuts an irrational
number α = ( Lα ,U α ) .

numbers denoted by  , is
Dedekind’s Definition of the Real Numbers: The real numbers,
the collection of all Dedekind cuts ( L, U ) on the rational numbers, with each
real number being associated with a specific Dedekind cut. If the lower set L
has a largest rational number r , or if the upper set U has a smallest rational
number r , we associate such cuts with the rational number r = ( Lr ,U r ) . If the
upper set U does not have a minimum rational number and the upper has has
no minimum rational number, we associate such cuts with an irrational number,
say α = ( Lα ,U ∂ ) .

Arithmetic and Ordering the Real Numbers

Our task is not yet complete. We must define the many properties we
desire of the real numbers, like how to add, subtract, multiply and divide as
well as well as how they are ordered such as a < b, a ≤ b, a > b, a ≥ b .
Associating real numbers a, b to their Dedekind cuts

a = ( La , U a ) , b = ( Lb , U b )

where La , Lb , U a , U b are the corresponding lower and upper intervals we define


the following arithmetic operations and order relation. The idea is to
construct the lower Dedekind set from two lower Dedekind cut sets (we could
Section 5.1 462 Construction of the Reals

just as well construct the upper Dedekind set) and then “associate” with the
arithmetic operation the (real) number where the cut occurs.

Operation6 Example
a + b ∼ La +b = { x + y : x ∈ La , y ∈ Lb } 2 + 3 = { x + y : x ∈ L2 , y ∈ L3 } = {rational numbers < 5} or 5
a − b ∼ La −b = { x − y : x ∈ La , y ∈ U b } 3 − 2 = { x − y : x ∈ L3 , y ∈U 2 } = {rational numbers < 1} or 1
a, b > 0 ⇒ ab ∼ Lab = { xy : x ∈ La , y ∈ Lb } 2 ⋅ 3 = { xy : x ∈ L2 , y ∈ L3 } = {rational numbers < 6} or 6
a ≤ b ⇔ La ⊆ Lb 2 ≤ 3 ⇔ L2 ⊆ L3

The Algebrization of Geometry:


Geometry: Dedekind cuts (i.e. real numbers), are based
on the fundamental property of the Euclidean line that “if all points on a line fall
into one of two classes, such that every point in the first class lies to the left of
every point in the second class, then there is one and only one point that
produces this division. It was this obvious geometric property of a line that
inspired Dedekind’s arithmetic formulation of continuity. It might be said that
Dedekind separated arithmetic from geometry in the process by creating a
purely arithmetic description of the Euclidean line.

6
The values of x, y in the following definitions are taken as rational numbers.
Section 5.1 463 Construction of the Reals

Problems:
Problems: Section 5.1, Construction
Construction of the Real Numbers

3. (Equivalence Relation I) Show that the relation ≡ defined by

( m, n ) ≡ ( m′, n′) if and only if m + n′ = m′ + n

between pairs of of natural numbers ( m, n ) and ( m′, n′ ) is an equivalence relation.

Ans: reflexive:
reflexive ( m, n ) ≡ ( m, n ) since m + n = m + n . Hence ≡ is reflexive.
symmetric If ( m, n ) ≡ ( m′, n′ ) ⇔ m + n′ = m′ + n ⇔ ( m′, n′ ) ≡ ( m, n )
symmetric:
transitive: Let ( m, n ) ≡ ( m′, n′) and ( m′, n′ ) ≡ ( m′′, n′′ ) . Hence
m + n′ = m′ + n, m′ + n′′ = m′′ + n′ . Adding these two equations and subtracting the
common factors we get m + n′′ = m′′ + n which says ( m, n ) ≡ ( m′′, n′′ ) . Hence, the
relation is transitive.

4. (Equivalence Relation II) Show that the relation ≡ defined by

( m, n ) ≡ ( m′, n′) ⇔ mn′ = m′n


between pair
s of integers ( m, n ) and ( m′, n′ ) is an equivalence relation.

Ans: reflexive:
reflexive ( m, n ) ≡ ( m, n ) since mn = mn . Hence ≡ is reflexive.
symmetric If ( m, n ) ≡ ( m′, n′ ) ⇔ mn′ = m′n ⇔ ( m′, n′ ) ≡ ( m, n )
symmetric:
transitive: Let ( m, n ) ≡ ( m′, n′) and ( m′, n′ ) ≡ ( m′′, n′′ ) . Hence
mn′ = m′n, m′n′′ = m′′n′ . Multiplying these equations and canceling the common
factor m′n′ we get mn′m′n′′ = m′nm′′n′ ⇒ mn′′ = m′′n . Hence ( m, n ) ≡ ( m′′, n′′ ) and so
the relation is transitive.

5. (Arithmetic in  ) We have created the integers  from equivalence classes


of natural numbers pair ( m, n ) . When we see ( 3, 5 ) we think “-2”, when we see
( 6, 3) we think “3” and so on. Perform the following arithmetic steps on pairs of
natural numbers.

a) (1,5) ⊕ ( 3, 2 ) Ans: (1,5) ⊕ ( 3, 2 ) = ( 4, 7 ) or − 3


b) (1, 5) ž ( 3, 2) Ans: (1, 5) ž ( 3, 2) = ( 3,8) or − 5
Section 5.1 464 Construction of the Reals

c) (1, 5) ⊗ ( 3, 2 ) Ans: (1,5) ⊗ ( 3, 2 ) = (13,17 ) or − 4

6. (Arithmetic in  )  ) We have created the rational numbers  from


equivalence classes of pairs of integers ( m, n ) . When we see ( 3, 5 ) we think
“3/5”, when we see ( 6, 3 ) we think “6/3 = 2” and so on. Perform the following
arithmetic steps on pairs of integers.

17
a) (1,5) ⊕ ( 3, 2 ) Ans: (1, 5) ⊕ ( 3, 2 ) = (17,10 ) or
10
1
b) (1, 2 ) ž ( 3, 2) Ans: (1, 2 ) ž ( 4, 9 ) = (1,18) or 18
3
c) (1, 5) ⊗ ( 3, 2 ) Ans: (1, 5) ⊗ ( 3, 2 ) = ( 3,10) or 10

7. (Decimal to Fractions) Find the fraction for each of the following numbers
in decimal form.

a) 0.9999…. ( 0.9 )
Ans:

0.9999... = 0.9 + 0.09 + .009 + 


9 9 9
= + + +
10 100 1000
9 1 1 
= 1 + + + 
10  10 100 
9 1 
=  
10  1 − (1 / 10 ) 
9  10 
=  
10  9 
=1

b) 0.23232323…. ( 0.23)
Ans:
Section 5.1 465 Construction of the Reals

0.2323...... = 0.23 + 0.0023 + .000023 + 


23 23
= + +
100 10000
23  1 1 
=  1+ + + 
100  100 10000 
23  1 
=  
100  1 − (1 / 100 ) 
23  100 
=
100  99 
23
=
99

c) 0.0123123123… ( 0.0123)
Ans:
125 ∞

k
0.0125125 =
1000 k =0
( 0.001)

125  1 
=
1000  1 − 0.001 

125
=
9990

d) 0.001111… ( 0.001)
Ans:
1  1 1 
0.0011111 =  1+ + + 
1000  10 100 
1  1 
=  
1000  1 − (1 / 10 ) 
1  10 
=
1000  9 
1
=
900

In general the fraction resulting from a rational number in decimal form is


Section 5.1 466 Construction of the Reals

d1d 2 ...d k
0. 000000


dd ...d k d1d 2 ...d k  =


1 2

999...999 ...0
n zeros k repeats k repeats 
000

k nines n zeros