CHAPTER 7: REAL NUMBERS

LECTURE NOTES FOR MATH 378 (CSUSM, SPRING 2009). WAYNE AITKEN

1. Introduction
In this chapter we construct the set of real numbers R. There are sev-
eral ways to introduce the real numbers. Three popular approaches are to
introduce R with (i) new axioms, with (ii) Dedekind cuts of Q, or with
(iii) Cauchy sequences in Q. We will use the third approach and construct
real numbers as equivalence classes of Cauchy sequences of rational num-
bers. This approach is chosen since it avoids the need for additional axioms,
and gives students practice with the concept of Cauchy sequence.
The main theorem of this chapter is that R is a complete ordered field.
We begin with a discussion of the limitations of Q. Then we discuss infinite
sequences. At first the sequences we require are rational sequence. However,
much of what we do easily extends to other ordered fields. So instead of
proving theorems for Q, we will often prove the result for a general ordered
field. With this in mind, throughout this chapter we let F be an ordered
field that is either equal to Q, or contains Q.1

2. Limitations of Q
In geometry we learn the Pythagorean Theorem which allows us to com-
pute one side of a right triangle assuming we know the lengths of the other
two sides. As an application,
one easily shows that the diagonal of the unit
square has length 2. In other words, the length d of the diagonal has the
property that d2 = 2. There is a problem: there is no such d in the field Q.
This elementary observation shows that Q does not have all the numbers
required to do even basic geometry. This compels us to construct a richer
number system R called the real numbers. Any real number that is not in
Q is called an irrational real number.
Before constructing R, we prove that there is no r Q such that r2 = 2.
There are several proofs of this fact. The proof given here is designed to
build on our familiarity with modular arithmetic.
Theorem 1. There is no r Q with the property that r2 = 2.

Date: version of May 19, 2009.

1One can show that Q can be embedded into any ordered field in a unique way. So,
really, we just need to assume that F is an ordered field.
1
2 LECTURE NOTES FOR MATH 378 (CSUSM, SPRING 2009). WAYNE AITKEN

Proof. Suppose such an r exist. By a result of Chapter 6 we can write

r = a/b where a and b are relatively prime integers. This implies that a and
b cannot both be even. From the assumption r2 = 2 we get the equation
a2 = 2b2 . This, in turn, implies that
a2 2b2 mod 4
Case 1: a and b are both odd. In this case, a2 b2 1 mod 4 by the
following lemma. Substituting into a2 2b2 gives 1 2 modulo 4. This is
a contradiction.
Case 2: a is odd, b is even. In this case, a2 1 mod 4 as before, but
b2 0 mod 4 (see the following lemma). Substituting into a2 2b2 gives
1 0 modulo 4. This is a contradiction.
Case 3: a is even, b is odd. This case is similar to the previous case:
a2 0 mod 4 but b2 1 mod 4. Substituting into a2 2b2 gives us 0 2
modulo 4. This is a contradiction.
So in any case, we get a contradiction. So no such r Q exists. 
Lemma 2. If c Z is odd then c2 1 modulo 4. If c Z is even then
c2 0 modulo 4.
Proof. Suppose c Z is odd. By the Quotient-Remainder Theorem and the
definition of odd integer, c = 2q + 1. Thus
c2 = (2q + 1)2 = 4q 2 + 4q + 1,
but
4q 2 + 4q + 1 0 + 0 + 1 1 mod 4.
Now suppose c Z is even. Then c = 2d for some d Z. So c2 = 4d2 .
Since 4 | c2 the result follows. 
Exercise 1. Adapt the proof of Theorem 1 to show that if n 2 mod
4 then
there is no r Q such that r2 = n. This shows, for example, that 10 is
irrational.
Remark 1. One can generalize the above theorem to show that if n Z is
not equal to some m2 (with m N) then there is no r Q with r2 = n.
This result can, in turn, be generalized to other powers beyond 2.

3. Infinite Sequences and Limits

In this section we consider infinite sequences in an ordered field F . Recall
the definition of infinite sequences from Section 9 of Chapter 4. An infinite
sequence with values in F is a function whose domain is a set of the form {i
Z | i n0 }, and whose codomain is F . We use notation such as (ai )in0
to denote such a sequence. We simply write (ai ) when the domain is not
important to the discussion.
An important concept associated to sequences is that of a limit. What do
we mean by the limit of a sequence? Informally, a sequence (ai ) has limit b
if the terms of the sequence eventually get and stay arbitrarily close to b.
 CHAPTER 7: REAL NUMBERS 3

This informal description is a bit ambiguous and is unsuitable to use in a

proof, so we give a more precise definition.
Definition 1. Suppose F is an ordered field, and (ai ) is a sequence in F .
We say that b F is the limit of (ai ) if the following holds: for all positive
F there is a k N such that if i k then |ai b| < . We can write
this with three quantifiers as follows:
( F>0 )(k N)(i N)(i k |ai b| < ).
Here F>0 denotes the positive elements of F .
Not all sequences have limits. A sequence that has a limit is said to
converge. A sequence that does not have a limit is said to diverge.
Exercise 2. Use the rules of basic logic to negate the definition of limit.
Complete the following sentence: the sequence (ai ) does not have limit b
means that there is a positive F such that for all k N . . .
Exercise 3. Consider the sequence (i)iN in F = Q. In other words, consider
the sequence given by the identity function. Show that this sequence does
not have a limit. Hint: work with = 1, and either use the previous exercise
or give a proof by contradiction.
Notice that in the above definition we used the term the limit. It sounds
like we are treating limits as if they are unique. This is justified by the
following theorem.
Theorem 3. A convergent sequence in an ordered field has a unique limit.
Proof. Suppose otherwise that (ai ) is a sequence in F with two distinct
limits b and c. Let = |b c|/2. Since b 6= c, we have that b c 6= 0. Hence
|b c| > 0. Since 2 > 0 we have 21 > 0. Thus the product = |b c|21
is positive.
By definition of limit, there is a k1 N such that |ai b| < for all i k1 .
Likewise there is a k2 N such that |ai c| < for all i k2 . Let k be the
maximum of k1 and k2 . Then, for i k we have
|b c| = |(b ai ) + (ai c)| |b ai | + |ai c| < + .
Here we used the triangle inequality. Since 2 = |b c| we have that
|b c| < |b c|,
a contradiction. 
Informal Exercise 4. Draw a picture of a number line (representing F ).
Draw b and c in the above proof, and indicate the sets defined by |x b| <
and |x c| < where is as in the above proof. Observe that the sets do
not intersect so there can be no ai simultaneously in both. This explains
why we chose = |b c|/2. Note, we could have chosen = |b c|/4, for
instance, and gotten a contradiction. However, = 2|b c| would not work.
Why not?
4 LECTURE NOTES FOR MATH 378 (CSUSM, SPRING 2009). WAYNE AITKEN

Remark 2. In the above theorem we get to choose to be whatever we want

since we are assuming a limit. If instead you are trying to prove a limit, you
cannot choose , but must allow to be an arbitrary positive element of F .
Exercise 5. Let F = Q. Show that (1/j)j1 has limit 0. Use the following
fact from Chapter 6: for every r Q, there is a k N with k > r. The case
r = 1/ will come up in your solution.
If a sequence converges, then the terms of the sequence get and stay
arbitrarily close to each other. This is shown in the following theorem.
Such sequences are called Cauchy sequences. So an easy way to see that a
sequence does not converge is to show that the sequence is not Cauchy: the
terms do not get and stay arbitrarily close to each other.
Theorem 4. Suppose (ai ) is a convergent sequence in an ordered field F .
Then for all positive in F there is a k N such that for all i, j N
i, j k |ai aj | < .
Proof. Let > 0 be an arbitrary positive element of F . We must find a
k N that satisfies the statement of the theorem. Let 0 = /2, and let b
be the limit of (ai ) that we assume exists.
By the definition of limit (Definition 1) there is a k N with |ai b| < 0
for all i k. So, for i, j k we have
|ai aj | = |(ai b) + (b aj )| |ai b| + |b aj | < 0 + 0 .
Here we have used the triangle inequality. Since 20 = we have that
|ai aj | < . Thus k has the desired property. 
The above theorem says that all convergent sequences satisfy the following
definition:
Definition 2. Suppose (ai ) is an infinite sequence in an ordered field F .
We say that (ai ) is Cauchy if the following occurs: for all positive in F
there is a k N such that for all i, j N
i, j k |ai aj | < .
Remark 3. We can interpret Theorem 4 through its contrapositive: if a
sequence is not Cauchy, it cannot converge.
Is this the only way a sequence can fail to converge? In other words, do
all Cauchy sequences converge? The answer is no for F = Q, but will turn
out to be yes in the real numbers. The problem with Q is that it has holes.
For example, we saw above that there is no r Q with r2 = 2. Define a
sequence by the rule ai = ni /10i where ni is the largest integer such that
a2i < 2. This sequence will not be convergent in Q, but can be shown to
be Cauchy.
It will turn out that this sequence is convergent in R, and has
limit 2.2
2In fact, the equivalence class of this Cauchy sequence can be used to define 2.
 CHAPTER 7: REAL NUMBERS 5

Informal Exercise 6. Find the first five terms of (ai ) defined

in the above
remark. Assume the domain is the set of i 0. Hint: punch 2 into your
calculator.
Our approach will be to assume that all Cauchy sequences in Q should
determine a real number. Non-Cauchy sequences cannot possibly converge,
so should not determine real numbers. There is a problem: different se-
quences can determine the same real number. For example, the sequence
defined by the rule bi = ni /2i where ni is the largest integer such that b2i < 2
determines the same real number as the sequence (ai ) discussed above (in
fact, they both
determine 2: the sequence (ai ) is related tothe decimal ex-
pansion of 2 and (bi ) is related to the base 2 expansion of 2). How do we
tell if two sequences determine the same number? There is an equivalence
relation that solves the problem:
Definition 3. Suppose that (ai ) and (bi ) are two sequences in F . We write
(ai ) (bi ) if the following occurs: for all positive F there is a k N
such that, for all i N,
i k |ai bi | < .
Remark 4. Informally the above definition says that the terms of the two
sequences get and stay arbitrarily close to each other.
Theorem 5. The relation is reflexive on the set of all sequences in F .
Proof. We need to show (ai ) (ai ) for any given sequence (ai )in0 in F .
Let be an arbitrary positive element of F . We must find a k such that
i k |ai ai | < .
Let us propose k = n0 . If i k then |ai ai | < since |ai ai | = 0 and
is positive. So k has the desired property. 
Theorem 6. The relation is transitive on the set of all sequences in F .
Proof. Suppose (ai ) (bi ) and (bi ) (ci ). We need to show (ai ) (ci ). Let
be an arbitrary positive element of F . We must find a k N such that
i k |ai ci | < .
To find k we need to use the fact that (ai ) (bi ) and (bi ) (ci ). Let
0 = /2. Since (ai ) (bi ) there is a k1 N such that
i k1 |ai bi | < 0 .
Likewise, there is a k2 N such that
i k2 |bi ci | < 0 .
Let us propose for k the larger of k1 or k2 . If i k then
|ai ci | = |(ai bi ) + (bi ci )| |ai bi | + |bi ci | < 0 + 0 .
6 LECTURE NOTES FOR MATH 378 (CSUSM, SPRING 2009). WAYNE AITKEN

Here we use the triangle inequality and the fact that i k1 and i k2 .
Since = 20 ,
|ai ci | < .
Thus k has the desired property. 
Theorem 7. The relation is symmetric on the set of all sequences in F .
Exercise 7. Prove the above.
Corollary 8. The relation is an equivalence relation on the set of se-
quences of (ai ) in F where F is any given ordered field.
Remark 5. When we restrict just to Cauchy sequences, it is still an equiv-
alence relation. In fact, any equivalence relation on a set restricts to an
equivalence relation on any subset.
Exercise 8. Show that if (ai ) (bi ) and if (ai ) has a limit, then (bi ) converges
and has the same limit as (ai ).
Remark 6. A similar argument gives the following: if (ai ) (bi ) and if (ai )
is Cauchy, then (bi ) is Cauchy. However, in this case we choose 0 = /3
instead of 0 = /2. The key step of the proof is
|bi bj | = |(bi ai ) + (ai aj ) + (aj bj )| < 0 + 0 + 0 .
Remark 7. Suppose (ai ) and (bi ) have the property that ai = bi for suffi-
ciently large i. In other words, suppose that there is a k such that ai = bi
for all i k. Then (ai ) (bi ). This is easily proved from the definition.
As a consequence, if one is Cauchy then both are, and if one converges then
both do with the same limit.
In particular, if we take a sequence (ai ) and change a finite number of
terms, then the resulting sequence is equivalent to (ai ). Likewise, if we
change the domain of (ai )in0 by replacing n0 with a larger integer, then
the resulting sequence is equivalent.
Remark 8. Several of the definitions and proofs given in this section are
similar in nature. Be very careful to learn the subtle differences between
them.

4. Boundedness Lemmas
In this section we will consider lemmas concerning bounds on Cauchy
sequences. These lemmas will be needed in later sections.
Lemma 9. Every Cauchy sequence (ai )in0 in an ordered field F is bounded
from above and below. In other words, there is a B F such that ai B
for all i n0 , and there is an element b F such that ai b for all i n0 .
Proof. Since (ai ) is Cauchy, there is a k N such that |ai aj | < 1 for
all i, j k (choose = 1). Let A be the maximum of an0 , . . . , ak , and let
B = A + 1. We will show that B is an upper bound for (ai ).
 CHAPTER 7: REAL NUMBERS 7

First consider the case where i k. In this case

ai A < A + 1.
Since B = A + 1, we have ai B as desired.
Next consider the case where i > k. Since i, k k, we have |ai ak | < 1.
Thus 1 < ai ak < 1. So ai < ak + 1. Since ak + 1 A + 1 = B, we get
ai B as desired.
The proof of the existence of b is similar. (Subtract one from a min). 
Lemma 10. If (ai )in0 is a Cauchy sequence in an ordered field F , there is
a bound B F such that |ai | B for all i n0 .
Proof. Let b and B be as in the previous lemma. We can replace B with
anything larger, and b with anything less, and the previous lemma still holds.
If B < b replace B with b. On the other hand, if b B, then B b,
and we can replace b with B. In either case, for the new bounds, b = B.
For each i n0 , we have B ai B. Thus |ai | B. 

5. The real numbers

Our idea for constructing R is based on the notion that every Cauchy se-
quence in Q should determine a real number, and that equivalent sequences
should determine the same real  number.
 In other words, every Cauchy se-
quence in an equivalence class (ai ) should determine the same real number.
This idea leads us to the idea of defining real numbers as equivalence classes
of Cauchy sequences of rational numbers.
 
Definition 4. If (ai ) is a Cauchy sequence in Q, then let (ai ) be the
equivalence class containing (ai ) under the equivalence
 relation on the set
of all Cauchy sequences in Q. We call (ai ) a real number.
Definition 5. The set of real numbers R is defined as follows:
def    
R = (ai )  (ai ) is a Cauchy sequence in Q .
Remark 9. It might seem like a strange transition to go from every equiv-
alence class of Cauchy sequences in Q should determine a real number, to
the bolder claim that every equivalence class of Cauchy sequences in Q is
a real number. In fact, some people accepted the former, but are a bit un-
comfortable about the latter. However, the latter allows us to avoid having
to use extra axioms about the existence of new types of objects.
In order to make R into a field we need to define an addition and multi-
plication operation on R.
   
Definition 6. Let (ai ) and (bi ) be real numbers. Then
    def  
(ai )in0 + (bi )im0 = (ai + bi )il0
and     def  
(ai )in0 (bi )im0 = (ai bi )il0 .
8 LECTURE NOTES FOR MATH 378 (CSUSM, SPRING 2009). WAYNE AITKEN

Here l0 is the maximum of n0 and m0 . Our definitions give two binary

operations R R R.
In order to check that these definitions are well-defined we need to verify
facts that are not totally obvious: (i) (ai + bi ) and (ai bi ) are Cauchy, (ii) if
(a0i ) (ai ) then we can replace ai with a0i in the definition and the result
will be the same real number, and (iii) a similar invariance when (bi ) (b0i ).
The remainder of this section will be devoted to verifying all the non-
obvious facts needed to confirm that the definition is well-defined.
Lemma 11. Suppose that (ai )in0 and (bi )im0 are Cauchy sequences in F .
Then (ai + bi )il0 and (ai bi )il0 are also Cauchy. Here l0 is the maximum
of n0 and m0 .
Proof. For the first claim, suppose is a positive element of F . We must
find a suitable k for the sequence (ai + bi ).
Let 0 = /2. By assumption, there is an integer k1 such that i, j k1
implies |ai aj | < 0 . Similarly, there is an integer k2 such that i, j k2
implies |bi bj | < 0 . Let k be the larger of k1 and k2 . If i, j k, then
|(ai + bi ) (aj + bj )| = |(ai aj ) + (bi bj )| |ai aj | + |bi bj |.
Since |ai aj |+|bi bj | < 0 +0 , we get |(ai + bi ) (aj + bj )| as desired.
For the second claim, suppose is positive. We must find a suitable k
for (ai bi ). By Lemma 10, there is a bound A such that |ai | A for all terms
ai of the first sequence. Likewise, there is a bound B such that |bi | B for
all terms bi of the second sequence. Clearly we can assume that A and B
are chosen to be positive.
Let 1 = /(2B). By assumption (ai ) is Cauchy. So there is an integer k1
such that i, j k1 implies |ai aj | < 1 . Similarly, if 2 = /(2A), there is
an integer k2 such that i, j k2 implies |bi bj | < 2 . Let k be the larger
of k1 and k2 . If i, j k, then
|ai bi aj bj | = |ai bi ai bj + ai bj aj bj |
|ai bi ai bj | + |ai bj aj bj | .
Observe that
|ai bi ai bj | = |ai ||bi bj | A|bi bj | < A2 = /2.
Similarly,
|ai bj aj bj | = |ai aj ||bj | |ai aj |B < 1 B = /2.
Thus |ai bi aj bj | < /2 + /2 = as desired. 
Lemma 12. Given sequences with values in F , if (ai ) (a0i ) then
(ai + bi ) (a0i + bi ),
and if (bi ) (b0i ) then
(ai + bi ) (ai + b0i ).
 CHAPTER 7: REAL NUMBERS 9

Exercise 9. Prove the above. Hint: you may wish to look at Lemma 13 for
guidance. However, Lemma 13 is more difficult since it requires the use of
Lemma 10.
Lemma 13. Given Cauchy sequences with values in F , if (ai ) (a0i ) then
(ai bi ) (a0i bi ),
and if (bi ) (b0i ) then
(ai bi ) (ai b0i ).
This is valid for any ordered field F.
Proof. We prove the first statement; the second statement is similar. So
suppose that (ai ) (a0j ). We must prove (ai bi ) (ai b0i ). In other words, for
each positive  in F we must find a k such that
i k |ai bi a0i bi | < .
By Lemma 10, there is a B F such that |bi | B for all i in the domain
of (bi ). Clearly we can choose B to be positive. Thus B 1 is also positive.
Let 0 = B 1 . Since (ai ) (a0j ), there is a k N such that
i k |ai a0i | < 0 .
Thus, if i k then
|ai bi a0i bi | = |ai a0i | |bi | |ai a0i |B < 0 B.
Since 0 B =  we have
i k |ai bi a0i bi | < 
as desired. 

6. The Real Numbers R as a Commutative Ring

It is fairly easy to show that R is a commutative ring. It is a bit harder
to show it is a field. We start with showing it is a commutative ring.
Theorem 14. Addition and multiplication on R are commutative and as-
sociative.
Exercise 10. Prove the above theorem.
Definition 7. Let F be an ordered field. If c F then the constant sequence
determined by c is the sequence (ai )iN defined by the rule i 7 c. In other
words, ai = c for all i N. We sometimes write (c)iN or just (c) to denote
this constant sequence.
Theorem 15. If (c)iN is a constant sequence in F , then it has limit c. In
particular, such sequences are Cauchy.
Exercise 11. Prove the above theorem. Hint: for each > 0, let k = 0.
Theorem 16. An additive identity for R exists and is [(0)]. A multiplicative
identity for R exists and is [(1)].
10 LECTURE NOTES FOR MATH 378 (CSUSM, SPRING 2009). WAYNE AITKEN

Remark 10. Identities, if they exist, are unique.3 Thus we can say the
additive identity and the multiplicative identity for R.
 
Proof. Let x = (ai ) be an arbitrary real number. By definition of + in R,
       
x + [(0)] = (ai ) + (0) = (ai + 0) = (ai ) = x
where the next to last equality is due to the fact that 0 is the additive
identity of Q (Chapter 5). By the commutative law (Theorem 14) we get
[(0)] + x = x + [(0)] = x. Thus [(0)] is the additive identity.
We leave the proof that [(1)] is the multiplicative identity to the reader.


We now consider inverses.

Lemma 17. If (ai ) is a Cauchy sequence in an ordered field F , then (ai )
is also Cauchy.
Proof. Suppose (ai ) is Cauchy. Let  be an arbitrary positive element of F .
In order to show that (ai ) is Cauchy, we must find a k N such that
(ai ) (aj ) <  for all i, j k. Since (ai ) is Cauchy, there is a k N
such that |ai aj | <  for all i, j k. Using this k, if i, j k then
   
(ai ) (aj ) = (1)(ai aj ) = | 1||ai aj | = |ai aj |.
 
But |ai aj | < , so (ai )(aj ) <  as desired. Thus (ai ) is Cauchy. 
Theorem 18. Every element of R has an additive inverse.
 
Proof. Let x = (ai ) be a real number. Here (ai ) is a Cauchy
 sequence.
By Lemma 17 the sequence (ai ) is also Cauchy, so y = (ai ) is a real
number. We leave it to the reader to show that y is the additive inverse
of x. 
Exercise 12. Complete the proof of the above theorem. Hint: show x + y =
y + x = 0.
As we we will see, multiplicative inverses are trickier. Fortunately we do
not need multiplicative inverses for the following.
Theorem 19. The real numbers R form a commutative ring.
Exercise 13. Prove the above. Hint: what about the distributive law? What
about the other laws?
Now that we know that R is a commutative ring, we can use all the
familiar algebraic manipulations and laws valid in rings.

3For any binary operation on a set S, one can show that if there is an identity, it must
be unique. For example, if 0 and 00 are additive identities, 0 = 0 + 00 = 00 .
 CHAPTER 7: REAL NUMBERS 11

7. The Canonical Embedding

Now that we have constructed R we wish to regard Q as a subset of R. To
do so we need to embed Q into R. This will require an injective map Q R.
What we will do is send any r Q to the constant sequence (r)iN .
Theorem 20. Let (ai ) be a sequence in an ordered field F , and let b F .
Then (ai ) has limit b if and only if (ai ) (b) where (b) is the constant
sequence with value b.
Proof. Write the constant sequence (b) as (bi )iN where bi is defined to be b.
We will show (ai ) (bi ) if and only if (ai ) has limit b.
First suppose (ai ) (bi ). Let  be an arbitrary positive element of F . In
order to show that (ai ) has limit b we must find a k N such that |ai b| < 
for all i k. We do know that (ai ) (bi ), so, by definition, there is a k N
such that |ai bi | <  for all i k. However, bi = b, so, using the k
mentioned above, if i k then
|ai b| = |ai bi | < .
Thus b is the limit of (ai ) as desired.
Now suppose (ai ) has limit b. Let  be an arbitrary positive element of F .
In order to show that (ai ) (bi ), we must find a k N such that |ai bi | < 
for all i k. We do know that (ai ) has limit b, so, by definition, there is a
k N such that |ai b| <  for all i k. However, bi = b, so, using the k
mentioned above, if i k then
|ai bi | = |ai b| < .
Thus (ai ) (bi ) as desired. 
Theorem 21. Let b, c F where F is an ordered field. Suppose b 6= c.
Then (b)iN and (c)iN are non-equivalent.
Exercise 14. Prove the above theorem. Hint: start by showing the two
sequences have distinct limits.
Corollary 22. Let b, c Q be distinct. Then [(b)] 6= [(c)] in R.
Proof. This follows from basic properties of equivalence classes: two non-
equivalent elements give non-equal (and disjoint) equivalence classes. 
Definition 8 (Canonical embedding).
 We define the canonical embedding
Q R by the rule c 7 (c)iN . In other words, the map sends c to the
constant sequence with value c. By Theorem 15, the image is an equivalence
class of Cauchy sequences, so the image is, indeed, in R.
Theorem 23. The canonical embedding Q R is injective.
Exercise 15. Prove the above theorem using Corollary 22.
Once we have a canonical embedding Q R we can use this to identify
elements of Q with certain elements of R. Thus we can think of Q as being
a subset of R.
12 LECTURE NOTES FOR MATH 378 (CSUSM, SPRING 2009). WAYNE AITKEN

When we think of Q as a subset of R, we need to check that the addition

and multiplication of R extends the addition and multiplication of Q defined
in Chapter 5. In other words, when we are working with addition and
multiplication on Q we want to be assured that we get the same result
whether we use the addition of Q (Chapter 6) or the addition of R (this
chapter). This is demonstrated in the following lemma.
Lemma 24. The definitions of addition and multiplication on R extend the
definitions of addition and multiplication on Q.
Proof. We give the proof for addition; the proof for multiplication is similar.
Let a, b Q be given and let +Q be the addition defined in Chapter 5. Let
+R be the addition defined in the current chapter. We must show that
a +Q b is identified with the same real number as a +R b (via the canonical
embedding).
This is actually easy once we figure out what is involved. The canonical
embedding identifies the rational number a+Q b with the equivalence class of
the constant sequence (a +Q b)iN . Since a is identified with the equivalence
class of the constant sequence (a)iN and b is identified with
 the equivalence
 
class of (b)iN , the sum a +R b is really equal to the sum (a)iN +R (b)iN .
By the definition of +R
     
(a)iN +R (b)iN = (a + Q b)iN .
The result follows. 
Remark 11. Since 0 in Q is identified with the equivalence class [(0)] of the
constant sequence (0), and since [(0)] is the additive identity of R, we usually
write 0 for the additive identity of R. This is consistent with the practice of
writing 0 for the additive identity of any ring.
Similarly, we write 1 for the multiplicative identity of R.
Remark 12. Subtraction in R extends the subtraction in Q. This follows
from the definition of r s (in any ring) as r + (s). As mentioned above,
addition in R extends addition in Q, and additive inverse in R extends
additive inverse in Q since [(r)] = [(r)].

8. Positive Real Numbers

An important part of showing that R is an ordered field is to define the
set of positive real numbers P , and to show that this set has the required
properties: closure and trichotomy. We will do this even before proving that
R is a field.
Since a real number can be thought of as [(ai )] where (ai ) is Cauchy, we
might be tempted to say that x = [(ai )] is positive if each ai is positive.
Warning: this does not work. For example, the sequence (1/i2 ) converges
to 0, and so is equivalent to the constant sequence (0). Thus [(1/i2 )] is zero
even though all its terms are strictly positive.
 CHAPTER 7: REAL NUMBERS 13

Furthermore, if a sequence has a finite number of zero or negative terms,

and the rest are positive, then the sequence could represent a positive num-
ber. Thus there are two ways in which the naive definition of positive is
defective. The following definition corrects both deficiencies.
Definition 9 (Positive). A positive-type Cauchy sequence in an ordered
field F is a Cauchy sequence (ai ) with the following property: there is a
positive d F and a k N such that ai d for all i k.
A positive real number is a real number of the form [(ai )] where (ai ) is a
positive-type Cauchy sequence with values in Q.
Exercise 16. Suppose (ai ) and (bi ) are Cauchy sequences where (ai ) (bi ).
Show that (ai ) is positive-type Cauchy if and only if (bi ) is.
Hint: suppose there is a d > 0 and a k N such that ai d for all i k.
We must find d0 and k 0 that work for (bi ). Let = d/2 and choose a k0 so
that |ai bi | < for all i k0 . Why does such a k0 exist? Choose k 0 as the
maximum of k and k0 . What do you think d0 should be? Prove that your
choice of k 0 and d0 work for (bi ).
Remark 13. Suppose r Q. It is easy to check that r is a positive element
of Q if and only if the real number [(r)] is a positive element of R. Thus the
present definition of positive is compatible with the definition of Chapter 6
Theorem 25 (Closure). If x, y R are positive then so is x + y and xy.
Proof. Let x = [(ai )] and y = [(bi )] where (ai ) and (bi ) are positive-type
Cauchy sequences of rational numbers. By definition there is a positive
d1 Q and a k1 N such that ai d1 for all i k1 . Likewise, there
is a positive d2 Q and a k2 N such that bi d2 for all i k2 . Let
d = d1 + d2 . We know that d > 0 (Chapter 6). Let k be the maximum of k1
and k2 . If i k, then
ai + bi d1 + bi d1 + d2 = d
by properties of rational numbers (Chapter 6). Thus x + y = [(ai + bi )] is
positive.
The proof for xy is similar. 
Exercise 17. Prove the above for the case of multiplication.
We now want to prove a trichotomy law: for all x R exactly one of the
following occurs (i) x is positive, (ii) x = 0, or (iii) x is positive. In the
third case we also say that x is negative.
We divide the proof of this law into lemmas:
Lemma 26. Let x R. If x is positive, then x 6= 0.
Proof. Suppose otherwise that x is positive and x = 0 at the same time.
Since 0 is the equivalence class [(0)] we have x = [(0)]. Since x is positive,
there is a positive-type Cauchy sequence (ai ) such that x = [(ai )]. By
properties of equivalence classes, (0) (ai ). By Exercise 16, (0) is a positive-
type Cauchy sequence. This is clearly a contradiction. 
14 LECTURE NOTES FOR MATH 378 (CSUSM, SPRING 2009). WAYNE AITKEN

Lemma 27. Let x R. It is not possible for both x and x to be positive.

Exercise 18. Prove the above. Hint: suppose not. Write x = [(ai )]. Then
x = [(ai )]. Define a d1 and k1 for (ai ) and d2 and k2 for (ai ). Let i be
the maximum of k1 and k2 , and show that ai is both positive and negative
in Q. This contradicts a certain law in Chapter 6.
Remark 14. Notice how we use a trichotomy law for Q (Chapter 6) to help
prove a trichotomy law for R.
Now we come to a lemma that is easy to state, but a bit tricky to prove.
Part of its trickiness comes from the need to manipulate quantifiers.
Lemma 28. If x 6= 0 is a real number, then either x or x is positive.
 
Proof. Write x =  (ai ) where (ai ) is a Cauchy sequence in Q. This implies
that x = (ai ) .    
By assumption x 6= 0. In other words, (ai ) 6= (0) . Thus we have that
(ai ) and the constant sequence (0) are non-equivalent. When we negate the
definition of equivalence (see Exercise 2 for a similar negation) we find that
there exists a positive Q such that for all k N there is an integer i k
with |ai 0| . Fix such an for what follows.
Let 0 = /2. Since (ai ) is Cauchy, there is a k0 N such that
i, j k0 |ai aj | < 0 .
Above we determined that for all k N there is an integer i k with
|ai 0| . In particular, there is an integer i0 k0 with |ai0 | . By a
result of Chapter 6, this implies that ai0 or ai0 .
We first consider the case where ai0 . Suppose j k0 . Since i0 , j k0
we have |aj ai0 | < 0 by the Cauchy condition. By a result of Chapter 6,
0 < aj ai0 . This gives us 0 + ai0 < aj . Therefore,
aj > ai0 0 0 = /2 = /2 = 0 .
In summary, we have 0 > 0 and k0 such that 0
 j k0 implies aj . By
definition of positive, we conclude that x = (ai ) is positive.
Next we consider the case where ai0 . Suppose j k0 . Since
i0 , j k0 we have |aj ai0 | < 0 by the Cauchy condition. By a result of
Chapter 6, aj ai0 < 0 . This gives us aj < 0 + ai0 . Thus
aj < 0 + ai0 0 = /2 = /2 = 0 .
Hence, aj 0 . In summary, we have 0 > 0 and k0 such that j k0
implies aj 0 . By definition of positive, we conclude that x = (ai )
is positive.
So either x or x is positive, assuming x 6= 0. 
Theorem 29 (Trichotomy: version 1). For every x R exactly one of the
following occurs: (i) x is positive, (ii) x = 0, or (iii) x is positive.
Exercise 19. Prove the above. Hint: if you make generous use of the previous
lemmas, the proof should be fairly short.
 CHAPTER 7: REAL NUMBERS 15

9. The Real Numbers R as an Ordered Field

We will now see that every non-zero element of R has a multiplicative
inverse. We first focus on positive elements. Recall that a sequence (ai ) is
positive-type Cauchy if there is a k and and d > 0 such that i k implies
ai d. In particular, such sequences are allowed to have a finite number of
zero (or even negative) terms. This is why we have to change from n0 to k0
in the following lemma.
Lemma 30. Suppose (ai )in0 is a positive-type Cauchy sequence in an or-
1


dered field F . Then there is an integer k0 such that ai ik0 is a Cauchy
sequence.
Proof. By definition of positive-type Cauchy sequence, there is a k0 N and
positive d F such that ai d for all i k0 . If i k0 then ai 6= 0, so ai has
a multiplicative inverse in F . By properties of ordered fields (Chapter 6)
a1 1
i d , and ai
1
is positive.
We wish to show that a1


i ik0
is Cauchy. In other words, let F be
 
 1 1 
positive. We want a k such that if i, j k then ai aj  < .
Let 0 = d2 . Since d and are positive, so is 0 . Since (ai ) is Cauchy,
there is a k 0 such that |ai aj | < 0 for all i, j k 0 . Let k be the maximum
of k 0 and k0 . If i, j k then
 
|a1 a 1
| = (a a )a1 1 
a

i j  j i i j 
  1 

= |aj ai | a1

 a 
i j

= |aj ai | a1 1
i aj
|aj ai | d1 d1
< 0 d2 =
So |a1 1 1


i aj | < as desired. So ai ik0
is Cauchy. 

Lemma 31. Let x R. If x is positive, then x has a multiplicative inverse.

Proof. Write x = [(ai )] where (ai ) is a positive-type Cauchy sequence

of
rational numbers. By the previous lemma, there is a k0 such that a1
i ik0
is Cauchy. Thus
y = (a1
 
i )ik0
is a real number. By definition of multiplication in R,
xy = [(ai )] (a1 1
   
i ) = (a i a i ) = [(1)].
Thus xy = 1. By the commutative law for multiplication, yx = xy = 1. We
have shown that x has a multiplicative inverse. 
Theorem 32. Let x R. If x 6= 0, then x has a multiplicative inverse.
16 LECTURE NOTES FOR MATH 378 (CSUSM, SPRING 2009). WAYNE AITKEN

Exercise 20. Prove the above theorem. Hint: if x is positive, use the lemma.
If x is positive, try y where y is the multiplicative inverse of x. Use
the fact that x(y) = (x)y (true in any ring).
We now come to one of the main theorems of this Chapter.
Theorem 33. The real numbers R are an ordered field.
Proof. We know that R is a commutative ring by Theorem 19. We know
that 0 6= 1 by Corollary 22. Multiplicative inverses exist by Theorem 32.
We conclude that R is a field.
To show that R is an ordered field we need to check that (i) the positive
elements are closed under addition and multiplication, and (ii) the positive
elements satisfy the trichotomy law. Both these were done in the previous
section (Theorem 25 and Theorem 29). 
Remark 15. Now that we know that R is an ordered field, we can use all
the definitions and results about ordered fields F from the end of Chap-
ter 6 including facts about < and absolute values. You should review these
definitions and results from Chapter 6.
Remark 16. As mentioned above, positivity defined for R is compatible with
the earlier concept of positivity defined for Q. Since x < y means y x is
positive, if follows that inequality in R is compatible with inequality in Q.

10. Relationship between R and Q

In this section we will consider a few useful results relating R and Q.
For example, we will see that Cauchy sequences of rational numbers always
converge to real numbers, and that all real numbers are limits of rational
sequences. We will also see that Q is dense in R; in other words, between
any two distinct real numbers we can always find a rational number.
We begin with a lemma that can be used to establish x y where x, y R.
Lemma 34. Suppose (ai ) and (bi ) are two Cauchy sequences of rational
numbers, and let x = [(ai )] and y = [(bi )] be the corresponding real numbers.
If there is a k N such that ai bi for all i k, then x y.
Remark 17. If we have ai < bi instead, we cannot necessarily conclude that
x < y. Without extra information, we can only conclude that x y.
Proof. Suppose otherwise that x > y. Then x y is positive. Now x y
is given by [(ai bi )]. So there is a k 0 N and a positive d Q such that
ai bi d for all i k 0 . Since d > 0 this means that ai > bi for all i k 0 .
Let i be the maximum of k and k 0 , where k is as in the statement of the
lemma. Then ai bi by hypothesis and ai > bi by the above argument.
This contradicts trichotomy of Chapter 6. 
Corollary 35. Suppose x R is given by x = [(ai )]. Suppose that b is a
rational number. If there is a k N such that ai b for all i k, then
 CHAPTER 7: REAL NUMBERS 17

x b (where here we are thinking of b as a real number). If, instead, there

is a k N such that ai b for all i k, then x b.
Proof. For the first statement, apply Lemma 34 to the sequences (ai ) and (b).
For the second statement, switch the order and apply Lemma 34 again. 
The following lemma gives a special case of the density of Q in R. The
general case will come later.
Lemma 36. If is a positive real number, then we can find a rational
number r Q such that 0 < r < .
Proof. Write as [(ai )] where (ai ) is a positive-type Cauchy sequence of
rational numbers. By definition, there is a k0 N and a positive d Q such
that ai d for all i k0 . Identify d with the constant sequence (di ) where
di = d for all i N. Observe that di ai for all i k0 since di = d. By
Lemma 34, we have that d .
From Chapter 6 we know that between the rational number 0 and d there
is a rational number r. Thus 0 < r and r < d . 
The following says that if a Cauchy sequence of rational numbers repre-
sents a certain real number, then the Cauchy sequence converges to the real
number.
Theorem 37. Let (ai ) be a Cauchy sequence of rational numbers. Then
(ai ), considered as a sequence of real numbers, converges to the real number x
where x = [(ai )].
Proof. Let be an arbitrary positive real number. We must find a k N
such that |ai x| < for all i k. It seems like we should be able to use the
definition of Cauchy sequence to find such a k. There is a slight problem:
(ai ) is a Cauchy sequence in Q, but R. We first need to replace with
a rational number. Lemma 36 gives us a positive 0 Q such that 0 < .
By definiton of a Cauchy sequence in Q, we can use this 0 and find a k N
such that |ai aj | < 0 for all i, j k. We must show that this k has the
desired properties.
So let k be as above, and let i k. We must show that |ai x| < . We
will actually show that |ai x| < 0 . Since we are fixing i, write ai as c. If
j k then (in Q)
|c aj | = |ai aj | 0
since (aj ) is Cauchy. Thus 0 c aj 0 for all j k (Chapter 6:
properties of absolute values). By Corollary 35 (in R)
0 [(c aj )] 0 .
Now [(c aj )] = [(c)] [(aj )], so [(c aj )] is the real number c x. Recall
c = ai . Thus
0 ai x 0 .
This implies that |ai x| 0 (Ch. 6). Since 0 < we get |ai x| < . 
18 LECTURE NOTES FOR MATH 378 (CSUSM, SPRING 2009). WAYNE AITKEN

Corollary 38. Every Cauchy sequence of rational numbers converges to

some real number.
Proof. Let (ai ) be a Cauchy sequence of rational numbers. Let x = [(ai )].
By Theorem 37, (ai ) has limit x. 
Remark 18. Let (ai ) be a sequence of rational number. There is some am-
biguity of what Cauchy means for (ai ) when we embed Q into R. We can
mean the Cauchy condition holds for all positive in F = Q, or we can
mean that the Cauchy condition holds for all positive in F = R. In the
above theorem and corollary we are thinking of the former. However, since
such sequences converge to a real number (in the sense of F = R), they are
also Cauchy for F = R (Theorem 4).
Conversely, if (ai ) is Cauchy in the sense of F = R, then it is automatically
Cauchy in the sense of F = Q since any rational is automatically real.
We conclude that if (ai ) is a sequence of rational numbers, there is no
difference between being Cauchy for F = Q or for F = R.
Corollary 39. Every real number is the limit of a sequence of rational
numbers
Proof. Let x = [(ai )] be a real number. By Theorem 37, (ai ) has limit x. 
Corollary 40. If x R and if R is positive, then there is a rational
number r Q with |x r| < .
Proof. Since x is the limit of a sequence (ai ) of rational numbers, there is a
k N such that |ai x| < for all i k. Let r = ak . 
Theorem 41. The field Q is dense in R. In other words, if x, y R are
such that x < y, there is a r Q with x < r < y.
Proof. Let = (y x)/2. By Corollary 40 there are rational numbers
a, b Q such that |x a| < and |y b| < . Let r = (a + b)/2. Observe
that r Q since a, b, 2 Q.
Observe that < x a < (Chapter 6). Since x a < , we get
x < a. Likewise, y < b. Adding inequalities gives x + y 2 < a + b.
However, 2 = y x. Thus 2x = x + y (y x) < a + b. So x < (a + b)/2.
This shows x < r.
A similar argument shows that r < y. 
Exercise 21. Complete the above theorem: show that r < y.

11. Complete Ordered Fields

As we have discussed before, Q has holes. For example, Q is missing a
square root for 2. In an ordered field without holes, we expect every Cauchy
sequence to converge. In contrast, Q has Cauchy sequences that do not
converge. For example, consider (ni /10i ) where ni is the largest integer such

that (ni /10i )2 2. This sequence is related to the decimal expansion of 2.
 CHAPTER 7: REAL NUMBERS 19

In the next chapter, we will see that this sequence is Cauchy. This Cauchy

sequence does not converge in Q, but does converge in R with limit 2.
We will now show that the real numbers R do not have such Cauchy
sequences that fail to converge. This means that R is complete: it does
not have holes. In order to make this discussion formal, we need the
following:
Definition 10. An ordered field F is said to be complete if every Cauchy
sequence in F converges to a limit in F .
We already know, from the previous section, that every Cauchy sequence
of rational numbers converges in R. However, we need to show that every
Cauchy sequence of real numbers converges in R in order to be able to assert
that R is complete. We begin with a lemma.
Lemma 42. If (ai ) is a sequence of real numbers, then there is a se-
quence (bi ) of rational numbers such that (ai ) (bi ). (Equivalence is taken
with F = R).
Proof. For each ai , we know by Corollary 40 that there is a rational num-
ber bi such that |ai bi | < 1/i. Consider the sequence (bi ) formed from such
rational numbers.4 We must show that (ai ) (bi ).
Let R be an arbitrary positive real number. We must find a k N
such that |ai bi | < for all i k. By Lemma 36 we can find a positive
rational number m/n < . Here m, n are chosen to be positive integers.
Then 1/n m/n. So 1/n < . Let k = n. If i k then
|ai bi | < 1/i (choice of bi )
1/n (i k and k = n)
< (choice of n)

Thus (ai ) (bi ) as desired. 

Now for the main theorem.
Theorem 43. The field R is a complete ordered field.
Proof. We must show that if (ai ) is a Cauchy sequence of real numbers, then
(ai ) converges. By Lemma 42 there is a sequence (bi ) of rational numbers
such that (bi ) (ai ). By Remark 6, the sequence (bi ) must also be Cauchy.
By Corollary 38 (bi ) has a limit b. By Exercise 8 and the fact that (ai ) (bi )
we conclude that (ai ) has limit b. Thus (ai ) converges. 

4In order to avoid using the axiom of choice, we can select b to have the smallest
i
possible denominator, and among fractions with the smallest possible denominator we
choose the smallest possible numerator.