Foundations Notes

Mathematical Foundations
Justin Lariviere and Tim Traynor

Contents
0 Preliminaries 5
0.1 Sentence Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
0.2 Negation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
0.3 Logical Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
0.4 Set Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
1 Structure of a Mathematical Proof 33

1.1 The Real Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.2 The Integers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2 Sets 125
2.1 Relations and Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
2.2 Real Intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
2.3 Ideals of the Integers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
2.4 Families of Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
3 Relations 187
3.1 Equivalence Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
3.2 Order Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
3.3 Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
3
4 CONTENTS
Chapter 0
Preliminaries
What to Expect
This course is an introduction to the reading and writing of mathematical proofs. Due
attention is given to the style and practices that are expected in modern written mathematics.
Grades will be a reflection of both the correctness of the mathematics as well as the style
in which it is written.
0.1 Sentence Structure
The first requirement when writing mathematics is that our sentences be phrased clearly
and precisely. We may leave no ambiguity. In particular, our ability to judge a statement
as being true or false ought never to depend on the clarity of the language.
Open Sentences
To begin our discussion of precision in language, we identify two components of a state-

ment. The first is the subject, which is the thing about which the statement is made, and the
other is the property or declaration that is being asserted about this subject. For example,
if one says ‘2 is even,’ then the subject is 2, and the property asserted about this subject
is that it is even. In general, the same property can be asserted of many different subjects,
for example ‘2 is even,’ ‘4 is even,’ and ‘6 is even’ are all statements in which the same
predicate (or asserted property) is applied to different objects. To examine the predicate
itself, without reference to a particular subject, we consider the subject as an open variable.
That is, ‘x is even’ is a predicate that would make a sensible proposition if the variable x
were replaced by any integer. For a more precise characterization of predicates we offer the
following:
An open sentence or predicate is a declarative statement containing one

or more open variables, whose truth or falsity depends on the values of
those variables.
5
6 CHAPTER 0. PRELIMINARIES
Example 0.1.1.
The following statements are examples of open sentences. The open variables in each
sentence are given in parentheses.
1. ‘The number is positive.’ (the number)
2. ‘The vector is in the null space of the matrix.’ (the vector, the matrix)
3. ‘The function is continuous at a.’ (the function, a)
4. ‘x ≤ y’ (x, y)
5. ‘The equation of the line is y = 4x + 2.’ (the line)
Z b
6. ‘ f (x) dx = 0.’ (a, b, f )
a
7. ‘2x + 1 = 4.’ (x)

8. ‘If x is even then x is divisible by 4.’ (x)
9. ‘For every value of a in the domain of f , lim f (x) = f (a).’ ( f )
x→a
Universe of Discourse
Each variable in an open sentence has a set of possible values that the variable may take.
For example, in the open sentence ‘x is even,’ the values that x can take (in order to make
this a sensible proposition) are the integers. This set of values is called the universe of
discourse for the variable. Note that different variables may have different universal sets.
For example, the open sentence ‘The vector is in the null space of the matrix’ may be
written as ‘~x is in the null space of A,’ where the universal set for ~x is say R3 , and the
universal set for A is the set of all 3 × 3 matrices.
Making an Open Sentence into a Proposition
In practice, there are three ways in which an open sentence can be made into a proposition.
We illustrate with the following example:
0.1. SENTENCE STRUCTURE 7
Example 0.1.2.
Consider the open sentence ‘the number is positive.’ This is an example of an open
sentence, because ‘the number’ is an open variable. We may use this open sentence to
make propositions in the following ways:
1. We may assign one specific value to the variable, as in ‘3 is posi-

tive’
2. We may assert that the sentence must be true for every allowable
value of the variable, as in ‘All real numbers are positive.’
3. We may assert that the sentence must be true for at least one value
of the variable, as in ‘Some real numbers are positive.’
In the proposition ‘3 is positive.’, we have simply removed the variable from the
open sentence. The variable ‘the number’ has been replaced with ‘3,’ which is not a
variable; it is one specific number. Since the sentence no longer contains a variable that
affects its truth or falsity, it now allows itself to be judged as either true or false, without
any need to clarify the language: the proposition ‘3 is positive’ is true.
In the propositions ‘All real numbers are positive’ and ‘Some real numbers are pos-
itive,’ the variable the number remains. The identity of the number is still unknown; hence
the number is still a variable. However, when judging whether the propositions ‘All real
numbers are positive’ and ‘Some real numbers are positive’ are true or false, we are no
longer concerned with the identity of the number but rather with a quantity of numbers.
Since in both cases, the quantity has been specified (as ‘all’ in one case and ‘some’ in
the other), the quantity of numbers for which the sentence must be true is not a variable.
Further, the truth or falsity of the sentence depends only on this quantity and not on the
identity of any specific number. In this way, the value of the variable no longer affects the
truth or falsity of the sentence. In fact, the proposition ‘All real numbers are positive’ is
false, and the proposition ‘Some real numbers are positive’ is true.
Quantifiers
We will focus on sentences that make a statement about a quantity of items, such as ‘all
real numbers are positive,’ or ‘some real numbers are positive.’ Formally, we introduce
two new symbols (∀ and ∃), called quantifiers, to denote the quantities of numbers whose
positiveness is being asserted (i.e. the quantity of values that make our open sentence true).
We illustrate by revisiting example 0.1.2.
Example 0.1.3.
Again, consider the open sentence ‘the number is positive.’ Since the number is an open
variable, we may write this sentence as ‘x is positive,’ or better still, ‘x > 0.’ The universe
of discourse for x is the set of all real numbers, which we denote by R.
1. We write the proposition ‘All real numbers are positive’ using the
symbolic form
‘∀x ∈ R, x > 0.’
Read: ‘For all x in R, x is positive.’
2. We write the proposition ‘Some real numbers are positive’ using

the symbolic form
‘∃x ∈ R, x > 0.’
Read: ‘There exists an x in R such that x is positive.’
The symbol ∀ is called the universal quantifier. We use ∀ to denote ‘for all.’ That is,
‘∀x ∈ R, x > 0’ is read ‘For all x in R, x is positive.’
The symbol ∃ is called the existential quantifier. We use ∃ to denote ‘for some,’
or more commonly ‘there exists.’ That is, ‘∃x ∈ R, x > 0’ is read ‘There exists an x in
R such that x is positive.’ The meaning is that there is at least one real number that is
positive. A more general description of this notation is as follows:
For an open sentence P(x), where the universe of discourse for the vari-
able x is U, the proposition
∀x ∈ U, P(x)
is true exactly when P(x) is true for every value of x in the universe of
discourse U.
The proposition
∃x ∈ U, P(x)
is true exactly when there is at least one value of x in the universe of
discourse U for which P(x) is true.
In practice, we will often quantify the variable over a subset of the universe of discourse.
That is, if A is a subset of U, one can write ∀x ∈ A, P(x), which asserts that P(x) is true for
all values of x in the set A. This same statement could be made using U as the universe of
discourse by saying ∀x ∈ U, if x ∈ A then P(x). Similarly, one may wish to write ∃x ∈ A,
P(x), asserting that there is at least one value of x in the set A for which P(x) is true. Since
this same value of x is also in the universe of discourse U, the same statement could be
made as ∃x ∈ U, x ∈ A and P(x). In general, we have the following rule:
For an open sentence P(x), where the universe of discourse for the vari-
able x is U, and for a subset A ⊆ U, we have
∀x ∈ A, P(x) ⇔ ∀x ∈ U, if x ∈ A, then P(x)
and
∃x ∈ A, P(x) ⇔ ∃x ∈ U, x ∈ A and P(x).
In the above rule, we use the symbol ⇔ to indicate that the two statements are equivalent.
That is, one is true if and only if the other is true; they assert the same fact, only in different
notation.
It is also quite common to encounter open sentences containing more than one vari-
able. For example, x ≥ y contains two open variables. In cases where the two variables
are quantified in the same way, that is either both are universally quantified, or both
are existentially quantified, we will usually alter the notation to put the two quantifiers
together. That is, rather than writing ∀x ∈ U, ∀y ∈ U, P(x, y), we will combine the two
quantifiers into one: ∀x, y ∈ U, P(x, y). This is done only for the sake of producing a more
readable notation and does not alter the meaning of the statement in any way.
Notation
For an open sentence P(x, y), where x and y take values in a universe of
discourse U, we use the following notation:
∀x, y ∈ U, P(x, y) means ∀x ∈ U, ∀y ∈ U, P(x, y)
and
∃x, y ∈ U, P(x, y) means ∃x ∈ U, ∃y ∈ U, P(x, y).
This degree of formality can seem tedious at first, but with practice we accustom ourselves
to it and begin to depend on it. Like any skill, the ability to think and speak precisely
is developed over time. One of the major goals of this course is to enable the student to
organize his or her thoughts on mathematics in a formal way and to express those thoughts
clearly and precisely. Developing the ability to identify subjects, predicates, and quantifiers
in a statement is a very important first step toward this goal.
The Benefits of Symbolic Form

Identifying the variables in a proposition, and translating the statement
into symbolic form, are useful skills for mathematicians. Not only does
this process remove ambiguity from the statement, but we will also see
that the symbolic form of a statement determines, to a large degree, the
structure of the proof of that statement. We will discuss several different
proof structures in this course, and we will frequently associate these
proof structures with the symbolic form of the statements they prove.
For now, it will be beneficial to practice our translation skills. It can be helpful to remember
that when translating a statement into symbolic form, we are writing a symbolic sentence
with the same meaning but not necessarily the same sentence structure. We do not need to
translate the sentence word for word. We need only determine the intended meaning of the
sentence and write this meaning in symbolic form.
Example 0.1.4.
Write the symbolic form of the following propositions, by identifying and quantifying each
variable and determining the universal set for each variable.
1. The square of every even integer is even.
Let E denote the set of all even integers. ∀x ∈ E, x2 ∈ E.
2. Some prime numbers have even squares.
Let P denote the set of all prime numbers. ∃x ∈ P, x2 ∈ E.
3. No odd integers have even squares.
Let O denote the set of all odd numbers. ∀x ∈ O, x2 < E.
4. Not all prime numbers have even squares.
∃x ∈ P, x2 < E.
5. Every integer has a prime divisor.
∀x ∈ Z, ∃y ∈ P, ∃a ∈ Z, x = ay.
6. There is a smallest natural number.
∃x ∈ N, ∀y ∈ N, x ≤ y.
7. There is no largest prime number.

∀x ∈ P, ∃y ∈ P, x < y.
8. Between any two distinct real numbers is another real number.
∀x, y ∈ R, if x , y then ∃z ∈ R, x < z < y or ∃z ∈ R, y < z < x.
9. If as many odd numbers as we please are added together, and their multitude is even,
then the sum is even.
∀n ∈ N, ∀x1 , x2 , . . . , xn ∈ O, if n is even, then x1 + x2 + · · · + xn is even.
10. If a cubic number multiplied by any number makes a cubic number, then the multi-
plied number is also cubic.
∀x, y ∈ N, if ∃a ∈ N, x = a3 and ∃b ∈ N, xy = b3 , then ∃c ∈ N, y = c3 .
Exercises 0.1.
Common Notation for Sets of Numbers

N: Natural numbers.
Z: Integers.
E: Even numbers.
O: Odd numbers.
Q: Rational numbers.
Qc : Irrational numbers.
R: Real numbers.
Give an example of open sentences P(x) and Q(x) and 20. ∀x ∈ R, if ∃a ∈ R, ax ≤ 0, then x ≤ 0.
a universe of discourse U for which the following hold:
21. ∀x ∈ Z, if x ≤ 0, then ∀a ∈ Z, ax ≤ 0.
1. The statement ‘∀x ∈ U, P(x) or Q(x)’ is true, but
22. ∀x ∈ R, if ∀a ∈ R, ax ≤ 0, then ∀b ∈ R, 0 ≤ bx.
the statement ‘∀x ∈ U, P(x), or ∀x ∈ U, Q(x)’ is
false. 23. ∀a ∈ R, if ∃x ∈ R, x ≤ xa, then 1 ≤ a.
2. The statement ‘∃x ∈ U, P(x), and ∃x ∈ U, Q(x)’ is 24. ∀a ∈ R, if ∀x ∈ R, ax ≤ x, then a = 1.
true, but the statement ‘∃x ∈ U, P(x) and Q(x)’ is
false. 25. ∀a ∈ R, if ∃x ∈ R, ax > 1, then ∃y ∈ R, ay < −1.
26. ∀x ∈ R, if ∃a ∈ R, ax < 0, then ∃b ∈ R, 0 < bx.
State whether the proposition is true or false.
3. ∃x ∈ R, x < 0. Write the following propositions in symbolic form by

identifying all variables and using the appropriate
4. ∀x ∈ N, 0 ≤ x. quantifiers.
5. ∀x ∈ Z, if 0 ≤ x, then x ∈ N. 27. Every natural number is positive
6. ∀x ∈ R, 0 < x2 . 28. The negative of any even integer is even.
7. ∃x ∈ R, x2 < 0. 29. The negative of any odd integer is odd.
8. ∀x ∈ R, ∃y ∈ R, x = 2y. 30. The sum of any two odd integers is even.
9. ∀x ∈ Z, ∃y ∈ Z, x = 2y. 31. Some integers have even squares.

32. Not all integers have even squares.
10. ∀x ∈ Z, ∃y ∈ Z, y = 2x.
33. No odd integers have even squares.
11. ∀x ∈ R, ∃n ∈ Z, n < x.
34. If an integer’s square is even, then the integer itself
12. ∃x ∈ R, ∀n ∈ Z, n < x.
is even.
13. ∀x ∈ Z, if x is odd, then ∀y ∈ Z, x = 2y + 1. 35. Not all integers are even.
14. ∀x, y ∈ R, if ∃a ∈ R, x < a and a ≤ z, then x < z. 36. No odd integers are even.
15. ∀x ∈ N, ∃y ∈ N, y < x. 37. All integers are either even or odd.
16. ∀x ∈ N, ∃y ∈ N, x < y. 38. 6 is a multiple of 3.
17. ∃x ∈ N, ∀y ∈ N, y ≤ x. 39. 6 is not a multiple of 5.
18. ∃x ∈ N, ∀y ∈ N, x ≤ y. 40. No odd integer is a multiple of an even integer.
19. ∀x, y ∈ Z, if ∃a ∈ Z, y = ax, then ∃b ∈ Z, x = by. 41. Some even integers are multiples of odd integers.
42. Every real number is smaller than some natural 50. There is no real number whose square is negative.
number.
51. There is a real number whose square is not positive.
43. There is no natural number that is larger than every
real number. 52. 0 and 1 are the only real numbers that are equal to
their own squares.
44. Every element of the interval (0, 1) is smaller than
every element of the interval (1, 2). 53. 1 and 7 are the only positive divisors of 7.
45. 1 is the smallest positive integer. (NOTE: 1 is not a 54. There is no largest real number in the interval (0, 1).
variable. It is a constant.)
55. If a number multiplied by two numbers makes cer-
46. There is a smallest natural number. tain numbers, then the numbers so produced have
the same ratio as the numbers multiplied. (Here,
47. There is no largest natural number.
‘number’ should be read as ‘natural number.’)
48. Between any two distinct real numbers, there is a
56. If a number multiplied by itself makes a cubic num-
rational number.
ber, then it itself is also cubic. (Again, ‘number’
49. The equation y2 = 4x + 3 has no integer solutions. should be read as ‘natural number.’)
0.2. NEGATION 13
0.2 Negation
Our goal in the previous section was to write propositions that are free of ambiguity. By
making use of the symbolic writing style we discussed, we may ensure that our language
does not hinder our ability to judge a proposition as true or false. If we decide that a propo-
sition is true, we will attempt to prove its truth. However, if we decide that a proposition is
false, we first need to express what it means for the proposition to be false. The statement
asserting that a given proposition is false is known as the negation of that proposition.
For a proposition P, the negation (or denial) of P, denoted ¬P, is the

proposition asserting that P is false. Hence ¬P is true when P is false
and false when P is true.
Example 0.2.1.
Write the negation of each of the following propositions.

1. 2 is odd.
Negation: 2 is not odd.
2. All odd numbers are prime.
Negation: Some odd numbers are not prime.
3. Some real numbers are not smaller than their squares.
Negation: All real numbers are smaller than their squares.
Negating Statements with Quantified Variables
Consider the example ‘all odd numbers are prime.’ The negation of this statement is ‘some
odd numbers are not prime.’ ‘All odd numbers are prime’ can be written in symbolic form
as ‘∀x ∈ O, x is prime.’ The symbolic form of the negation, ‘some odd numbers are not
prime,’ is ‘∃x ∈ O, x is not prime.’
¬(∀x ∈ O, x is prime ) ⇔ ∃x ∈ B, x is not prime.
Notice that to deny the statement ‘all odd numbers are prime,’ we only need at least one
composite odd number, for example 15. This means that when we are denying a state-
ment about all odd numbers, we need only make a statement about some odd numbers. In
general, we have the following rule:
Let P(x) be an open sentence and U be the universe of discourse for the
variable x.
¬(∀x ∈ U, P(x)) ⇔ ∃x ∈ U, ¬P(x).
Next, consider the example ‘some real numbers are not smaller than their squares.’ The
negation is ‘all real numbers are smaller than their squares.’ The symbolic form of ‘some
real numbers are not smaller than their squares’ is ‘∃x ∈ R, x ≥ x2 .’ The symbolic form
of it’s negation, ‘all real numbers are smaller than their squares,’ is ‘∀x ∈ R, x < x2 .’
¬(∃x ∈ R, x ≥ x2 ) ⇔ ∀x ∈ R, x < x2 .
In this case, since ‘some real numbers are not smaller than their squares’ is true provided
there is at least one real number that is not smaller than its square, to give a denial of this
statement we must ensure that all real numbers are smaller than their squares. i.e. to deny
a statement about some real numbers, we must make a statement about all real numbers.
In general:
Let P(x) be an open sentence and U be the universal set for the variable
x.
¬(∃x ∈ U, P(x)) ⇔ ∀x ∈ U, ¬P(x).
Example 0.2.2.
Write the negations of the following propositions.

1. ∀x ∈ N, x > 1.
Negation: ∃x ∈ N, x ≤ 1.
2. ∃x, y ∈ Z, 3x + 4 = 3y + 2.
Negation: ∀x, y ∈ Z, 3x + 4 , 3y + 2.
3. ∃x ∈ R, ∀y ∈ N, x ≤ y.
Negation: ∀x ∈ R, ∃y ∈ N, x > y.
4. ∀x ∈ R, ∃y ∈ R, xy = 1.
Negation: ∃x ∈ R, ∀y ∈ R, xy , 1.
5. ∀x ∈ Z, ∃y ∈ Z, ∀z ∈ Z, xy = zy.
Negation: ∃x ∈ Z, ∀y ∈ Z, ∃z ∈ Z, xy , zy.
Compound Propositions
Some propositions may be composed of several statements joined together with the con-
nectives or, and, or the implication if, ... then. For example:
1. All integers are divisible by 2 or divisible by 3.

2. All integers are both even and positive.
3. If an integer is divisible by 2 then it is divisible by 3.

0.2. NEGATION 15
We will pay close attention to how these statements are negated and attempt general rules
for negating statements involving or, and, and if, ... then.
First, notice that all of the statements listed above are false. The reason the statement
‘All integers are divisible by 2 or divisible by 3’ is false, is that there are some integers
(for example: 5) that are not divisible by 2 and not divisible by 3. The negation of this
proposition is in fact:
1. Some integers are not divisible by 2 and not divisible by 3.
The proposition ‘All integers are both even and positive’ is false for two reasons. We may
deny this on the grounds that some integers (for example: 3) are not even, or we may deny
it on the grounds that some integers (for example: −1) are not positive. The full negation
of the statement is:
2. Some integers are not even or not positive.
Finally, the statement ‘If an integer is divisible by 2 then it is divisible by 3’ is false because
there are integers (for example: 2) that are divisible by 2, and yet are not divisible by 3.
The proper negation of this statement is:
3. Some integers are divisible by 2 and not divisible by 3.
The general rules for negating statements containing and, or, or if, then become more
evident when we examine the symbolic forms of these statements and their negations:
1. ∀x ∈ Z, x is divisible by 2 or x is divisible by 3.
Negation: ∃x ∈ Z, x is not divisible by 2 and x is not divisible by
3.
2. ∀x ∈ Z, x is even and x is positive.

Negation: ∃x ∈ Z, x is not even or x is not positive.
3. ∀x ∈ Z, If x is divisible by 2 then x is divisible by 3.
Negation: ∃x ∈ Z, x is divisible by 2 and x is not divisible by 3.
The fact that universal (∀) quantifiers are changed to existential (∃) quantifiers, under nega-
tion, was discussed in the previous section. What is of interest here is:
1. Negating a statement with an ‘or’ connective results in a statement

with an ‘and’ connective.
2. Negating a statement with an ‘and’ connective results in a state-

ment with an ‘or’ connective.
3. Negating a conditional statement with ‘if, then’ results in a state-
ment with an ‘and’ connective, with the hypothesis affirmed and
the conclusion denied.
The general rules for negating statements containing and, or, or if, then are as follows:
1. ¬(P or Q) ⇔ ¬P and ¬Q.

2. ¬(P and Q) ⇔ ¬P or ¬Q.
3. ¬(if P then Q) ⇔ P and ¬Q.
The rules concerning and and or (rules 1 and 2) are commonly called DeMorgan’s Laws.
0.2. NEGATION 17
Example 0.2.3.
Write the negations of the following propositions. Determine which is true, the proposition
or its negation.
1. ∃x, y ∈ R, x < y and y < x.
Negation: ∀x, y ∈ R, x ≥ y or y ≥ x.
The negation is true.
2. ∀x ∈ Z, x < 0 or x ≥ 1.
Negation: ∃x ∈ Z, x ≥ 0 and x < 1.
3. ∃x ∈ Z, x ≥ 0 and ∀y ∈ Z, xy ≤ 0.
Negation: ∀x ∈ Z, x < 0 or ∃y ∈ Z, xy > 0.
The original proposition is true.
4. ∀x ∈ Z, ∃a ∈ Z, x = 2a or ∃b ∈ Z, x = 2b + 1.
Negation: ∃x ∈ Z, ∀a ∈ Z, x , 2a and ∀b ∈ Z, x , 2b + 1.
5. ∀x ∈ R, ∃n ∈ N, n − 1 < x and x ≤ n.
Negation: ∃x ∈ R, ∀n ∈ N, n − 1 ≥ x or x > n.
6. ∀x ∈ R, ∃n ∈ Z, x < n and ∃m ∈ Z, m < x.
Negation: ∃x ∈ R, ∀n ∈ Z, x ≥ n or ∀m ∈ Z, m ≥ x.
7. ∀x ∈ R, if ∃y ∈ R, xy > 0 then x > 0.

Negation: ∃x ∈ R, ∃y ∈ R, xy > 0 and x ≤ 0.
8. ∀x ∈ Z, if x2 = 1 then x = 1 or x = −1.
Negation: ∃x ∈ Z, x2 = 1 and x , 1 and x , −1.
9. ∀x ∈ Z, if ∃a ∈ Z, x = 15a + 8, then ∃b ∈ Z, x = 5b + 3 and ∃c ∈ Z, x = 3c + 5.
Negation: ∃x ∈ Z, ∃a ∈ Z, x = 15a+8, and ∀b ∈ Z, x , 5b+3, or ∀c ∈ Z, x , 3c+5.
10. ∀x ∈ Z, if ∃b ∈ Z, x = 5b + 3 and ∃c ∈ Z, x = 3c + 5, then ∃a ∈ Z, x = 15a + 8.
Negation: ∃x ∈ Z, ∃b ∈ Z, x = 5b + 3, and ∃c ∈ Z, x = 3c + 5, and ∀a ∈ Z,
x , 15a + 8.
Exercises 0.2.
Write the negation of each of the following proposi- 13. ∀x, y ∈ R, if x > 0, then ∃n ∈ N, y < nx.
tions. Determine which is true, the proposition or its
negation. 14. ∀x ∈ R, if x > 0, then ∃y ∈ R, 0 < y and y < x.
1. ∀x ∈ R, x ≤ x. 15. ∀x, y ∈ Q, if x < y, then ∃z ∈ R, z < Q and

x < z < y.
2. ∃x ∈ R, x2 < 0.
16. ∀x ∈ Z, if x is prime, then ∃y ∈ Z, y is prime and
3. ∃x ∈ R, ∀y ∈ Z, x ≤ y. y > x.
4. ∀x ∈ Z, ∃y ∈ R, x ≤ y. 17. ∀x ∈ R, if ∃y ∈ R, y , 0 and xy = y, then ∀z ∈ R,
xz = z.
5. ∀x ∈ Q, ∃y ∈ Q, xy = 1.
18. ∀x ∈ R, if ∀y ∈ Z, xy ≤ 0, then ∀z ∈ Z, xz ≥ 0.
6. ∃x ∈ Q, ∀y ∈ Q, xy = 1.
19. ∀x ∈ Z, if ∃t ∈ Z, x2 = 3t, then ∃s ∈ Z, x = 3s.
7. ∀x ∈ Z, x + (x + 1) is odd and x(x + 1) is even.
20. ∀x, y ∈ Z, if ∃a ∈ Z, xy = 6a, then ∃b ∈ Z, x = 6b
8. ∀x, y ∈ Z, x divides y or y divides x.
or ∃c ∈ Z, y = 6c.
9. ∃x, y ∈ R, xy is rational, and x or y is irrational.
21. ∀n ∈ N, if n = 1 or n = 7, then ∀x ∈ N, ∃t ∈ N,
10. ∀x ∈ Z, ∃y ∈ Z, y is prime and y divides x. 7x = nt.
11. ∀x, y ∈ Z, if x − y is even, then x + y is even. 22. ∀n ∈ N, if ∀x ∈ N, ∃t ∈ N, 7x = nt then n = 1 or

n = 7.
12. ∀x, y ∈ R, if x is rational and y is irrational, then
x + y is irrational. 23. ∀x ∈ R, if ∀y ∈ R, if y > 0 then x ≤ y, then x ≤ 0.
0.3. LOGICAL OPERATORS 19
0.3 Logical Operators
Taking a closer look at the terms and, or, and if, then, we see that the function of each
of these connectives is to combine two propositions into one compound proposition. For
example, ‘16 is divisible by 4’ and ‘16 is divisible by 3’ are two propositions, the first being
true and the second false. Using the connectives and, or, and if, then, we can create the
following compound propositions:
‘16 is divisible by 4 and 16 is divisible by 3.’

‘16 is divisible by 4 or 16 is divisible by 3.’
‘if 16 is divisible by 4, then 16 is divisible by 3.’
In this example, the first compound proposition is false, the second is true, and the third
is false. It is interesting to note that whether the resulting compound proposition is true
or false depends only on whether the two component propositions are true or false. It
does not depend on the meaning of the propositions or on any relationship between the
two. In fact, there does not need to be any relationship between the two propositions. For
example, though there is no relationship between the propositions ‘16 is divisible by 4’ and
‘all triangles are isosceles,’ the compound proposition ‘16 is divisible by 4 or all triangles
are isosceles’ is a true compound proposition. The compound proposition is true simply by
virtue of the fact that one of the component propositions (namely, ‘16 is divisible by 4’) is
true.
To help us phrase this observation in a more precise way, we introduce the following term:
A Boolean value is one of either False or True. The Boolean value of

a proposition is True provided the proposition is true and False provided
the proposition is false. The Boolean values {False, True} are also some-
times denoted by {F, T}, {0, 1}, {−1, 1}, or {⊥, >}.
For example, the Boolean value of the proposition ‘16 is divisible by 4’ is True; the Boolean
value of the proposition ‘16 is divisible by 3’ is False; the Boolean value of the compound
proposition ‘if 16 is divisible by 4, then 16 is divisible by 3’ is False.
Earlier, we observed that whether a compound proposition is true or false depends only
on whether the component propositions are true or false. We can rephrase this by saying
the Boolean value of a compound proposition is determined by the Boolean values of its
component propositions. Because of this, we can view the connectives and, or, and if,
then as functions that take two Boolean values as input and return a single Boolean value
as output. Functions that take a number of Boolean values as input and return a single
Boolean value as output are called logical operators. The logical operators and, or, and if,
then are given by the following rules:
The logical operator ∧ (read ‘and’) is the function accepting two

Boolean values as input and returning one Boolean value as output, ac-
cording to the following rule: for Boolean values x, y
x y x∧y
T T T
T F F
F T F
F F F
The logical operator ∨ (read ‘or’) is the function accepting two Boolean
values as input and returning one Boolean value as output, according to
the following rule: for Boolean values x, y
x y x∨y
T T T
T F T
F T T
F F F
The logical operator ⇒ (read ‘implies’ or ‘if, . . . then’) is the function

accepting two Boolean values as input and returning one Boolean value
as output, according to the following rule, for Boolean values x, y
x y x⇒y
T T T
T F F
F T T
F F T
Not all logical operators accept two Boolean values as input. For example, the logical
operator ¬ (‘not’) accepts a single Boolean value as input and returns a single Boolean
value as output, according to the rule: for a Boolean value x
x ¬x
T F
F T
To justify the definition of the logical operator ¬ above, consider for example the propo-
sition ‘2 is odd.’ The Boolean value of this proposition is False. The negation of this
proposition, ‘¬(2 is odd),’ is the proposition ‘2 is not odd,’ which has a Boolean value of
True. Likewise, the negation of a true proposition will be a false proposition. Thus, the
negation, as we understand it from section 0.2, can be viewed as a logical operator that
maps the True Boolean value to False, and the False Boolean value to True.
Since the ¬ logical operator takes only one Boolean value as input, we call it a unary
logical operator. Likewise, since the ∧, ∨, and ⇒ connectives take two Boolean values as
input, we call them binary logical operators. There are also logical operators that take three
or more Boolean values as input. For example, consider the logical operator x ⇒ (y ∨ z).
This operator is given by the rule: for Boolean values x, y, z
x y z x ⇒ (y ∨ z)
T T T T
T T F T
T F T T
T F F F
F T T T
F T F T
F F T T
F F F T
The logical operator above accepts three Boolean values as input and returns a single
Boolean value as output.
Equivalent Logical Operators
Two logical operators are called equivalent provided they are the same function. That is,
when the two operators are given the same Boolean inputs, they produce the same Boolean
output. We have already seen several examples of equivalent logical operators in the pre-
vious section. For example, consider DeMorgan’s Law: for propositions P and Q,
¬(P and Q) ⇔ ¬P or ¬Q.
One way of justifying the equivalence of these propositional forms is that the corresponding
logical operators ¬(x ∧ y) and ¬x ∨ ¬y are equivalent (in the sense that they are the same
function). Indeed, the table of inputs/outputs for these operators is as follows:
x y ¬(x ∧ y) ¬x ∨ ¬y
T T F F
T F T T
F T T T
F F T T
Since these two functions always produce the same output when given the same inputs, we
say the logical operators ¬(x ∧ y) and ¬x ∨ ¬y are equivalent. We denote this equivalence
by
¬(x ∧ y) ≡ ¬x ∨ ¬y
Likewise, the other rules for negating compound propositions
¬(P or Q) ⇔ ¬P and ¬Q
¬(if P, then Q) ⇔ P and ¬Q

are justifiable in the same way. That is, the corresponding logical operators are equivalent:
¬(x ∨ y) ≡ ¬x ∧ ¬y
¬(x ⇒ y) ≡ x ∧ ¬y.
For a slightly more elaborate example of equivalent logical operators, consider the opera-
tors x ⇒ (y ∨ z) and (x ∧ ¬y) ⇒ z. The table of values for these functions is:
x y z x ⇒ (y ∨ z) (x ∧ ¬y) ⇒ z
T T T T T
T T F T T
T F T T T
T F F F F
F T T T T
F T F T T
F F T T T
F F F T T
We see that x ⇒ (y ∨ z) and (x ∧ ¬y) ⇒ z yield the same output whenever they are given
the same inputs. Thus,
x ⇒ (y ∨ z) ≡ (x ∧ ¬y) ⇒ z.
Changing the Form of a Proposition
In practice, we often use equivalent logical operators to change the form of a proposition
into a form that is easier to work with. Consider for example the pair of equivalent logical
operators just mentioned: x ⇒ (y ∨ z) ≡ (x ∧ ¬y) ⇒ z. Since these logical operators are
equivalent, if we were to replace the Boolean variables x, y, and z with propositions P, Q,
and R, then the proposition ‘if P, then Q or R’ would necessarily have the same Boolean
value as the proposition ‘if P and not Q, then R,’ regardless of the choice of propositions P,
Q, and R. That is, any compound proposition of the form ‘if P, then Q or R’ is true if and
only if the proposition ‘if P and not Q, then R’ is true.
For example, consider the proposition ‘∀x, y ∈ Z, if xy is even, then x is even or y is even.’
For any x, y ∈ Z, if we identify P, Q and R as
P : xy is even
Q : x is even
R : y is even,
then the proposition ‘if xy is even, then x is even or y is even’ has the form
if P, then Q or R.
Since the logical operator x ⇒ (y ∨ z) is equivalent to the logical operator (x ∧ ¬y) ⇒ z,

we have that this proposition is equivalent to the proposition
if P and not Q, then R.
That is, it is equivalent to ‘if xy is even and x is not even, then y is even.’
In fact, any compound proposition of the form
if P, then Q or R
can be rewritten in the form

if P and not Q, then R,
because the corresponding logical operators x ⇒ (y ∨ z) and (x ∧ ¬y) ⇒ z are equivalent.
Example 0.3.1.
For each of the following propositions of the form P ⇒ (Q ∨ R), rewrite the proposition in
the form (P ∧ ¬Q) ⇒ R.
1. ∀x ∈ R, if 1 < x2 , then 1 < x or x < −1.
Solution: ∀x ∈ R, if 1 < x2 and x ≤ 1, then x < −1.
2. ∀x, y ∈ R, if xy is irrational, then x is irrational or y is irrational.

Solution: ∀x, y ∈ R, if xy is irrational and x is rational, then y is irrational.
3. ∀x, y ∈ R, if xy < 0, then x < 0 or y < 0.
Solution: ∀x, y ∈ R, if xy < 0 and 0 ≤ x, then y < 0.
4. ∀x, y ∈ R, if x2 = y2 , then x = y or x = −y.

Solution: ∀x, y ∈ R, if x2 = y2 and x , y, then x = −y.
Boolean Algebraic Properties of Logical Operators
A few fundamental equivalent logical operators form what we call the Boolean algebraic
properties of logical operators. We will see many of these same basic properties reoccurring
in other branches of mathematics throughout the course. All of the Boolean algebraic
properties can be easily verified by writing out a table of Boolean values for each operator.
Boolean Algebraic Properties of Logical Operators

Let T be the logical operator that returns the Boolean value T rue for every input,
and let F be the logical operator that returns the Boolean value False for every
input. Then
¬T ≡ F ¬F ≡ T
Idempotence
x∧x≡ x x∨x≡ x
Commutativity
x∧y≡y∧x x∨y≡y∨x
Associativity
x ∧ (y ∧ z) ≡ (x ∧ y) ∧ z x ∨ (y ∨ z) ≡ (x ∨ y) ∨ z
Absorption
x ∧ (x ∨ y) ≡ x x ∨ (x ∧ y) ≡ x
Distributivity
x ∧ (y ∨ z) ≡ (x ∧ y) ∨ (x ∧ z) x ∨ (y ∧ z) ≡ (x ∨ y) ∧ (x ∨ z)
Annihilator
x∧F≡F x∨T≡T
Identity
x∧T≡ x x∨F≡ x
Complementation
x ∧ ¬x ≡ F x ∨ ¬x ≡ T
Double Negation
¬(¬x) ≡ x
De Morgan’s Laws
¬(x ∨ y) ≡ ¬x ∧ ¬y ¬(x ∧ y) ≡ ¬x ∨ ¬y
The Contrapositive of an Implication
There is another pair of equivalent logical operators that is of particular importance, so we

take the time to highlight it here. The implication x ⇒ y is equivalent to the implication
¬y ⇒ ¬x. To verify that these logical operators are the same, we check the table of values:
x y x⇒y ¬y ⇒ ¬x
T T T T
T F F F
F T T T
F F T T
Since the two logical operators x ⇒ y and ¬y ⇒ ¬x always produce the same outputs,
they are the same function; thus, x ⇒ y ≡ ¬y ⇒ ¬x. This equivalence of logical operators
motivates the following definition:
For propositions P and Q, the implication
if ¬Q, then ¬P
is called the contrapositive form of the implication
if P, then Q.
Since the corresponding logical operators x ⇒ y and ¬y ⇒ ¬x are equivalent, we have that
every implication is true if and only if its contrapositive is true. In other words, we may
replace any implication with its contrapositive without changing whether the statement is
true or false.
Example 0.3.2.
Rewrite each of the following propositions by replacing the implication with its contrapos-
itive:
1. ∀x, y ∈ R, if x < y, then 1 + x < 2 + y.
Solution: ∀x, y ∈ R, if 2 + y ≤ 1 + x, then y ≤ x.
2. ∀x, y ∈ R, if x + y is irrational, then x is irrational or y is irrational.

Solution: ∀x ∈ R, if x is rational and y is rational, then x + y is rational.
3. ∀x, y ∈ Z, if xy is odd, then x is odd and y is odd.
Solution: ∀x ∈ Z, if x is even or y is even, then xy is even.
4. ∀a ∈ R, if ∃x ∈ R, x , 1 and ax = a, then a = 0.
Solution: ∀a ∈ R, if a , 0, then ∀x ∈ R, x = 1 or ax , a.
Alternate Solution: ∀a ∈ R, if a , 0, then ∀x ∈ R, if x , 1, then ax , a.
Alternate Solution: ∀a ∈ R, if a , 0, then ∀x ∈ R, if ax = a, then x = 1.
1
5. ∀x, y ∈ R, if ∀n ∈ N, |x − y| < , then x = y.
n
1
Solution: ∀x, y ∈ R, if x , y, then ∃n ∈ N, ≤ |x − y|.
n
The Converse of an Implication
For the implication x ⇒ y, another related, though not equivalent, logical operator is the
implication that reverses the antecedent and the consequent. That is, the implication y ⇒ x.
We call the operator y ⇒ x the converse of the operator x ⇒ y and we use the same word
to describe propositions of the corresponding forms:
For propositions P and Q, the implication
if Q, then P
is called the converse of the implication
if P, then Q.
It is important to stress that the converse of an implication is not equivalent to the original
implication. That is,
x ⇒ y . y ⇒ x.
This can easily be seen by comparing the tables of Boolean values for these two operators:
x y x⇒y y⇒x
T T T T
T F F T
F T T F
F F T T
Since these logical operators are not equivalent, it is possible to find an implication that
does not have the same Boolean value as its converse. This is possible even when the
implication is part of a proposition with quantified variables. For example, the proposition
∀x, y ∈ Z, if x is even and y is even, then x + y is even
is true. However, if we replace the implication in this proposition with its converse, we get
the proposition
∀x, y ∈ Z, if x + y is even, then x is even and y is even
which is false. The converse is false because it is possible to find integers, say x = 1 and
y = 3 for which x + y is even, but in this case neither x nor y are even.
Exercises 0.3.
“.. you should say what you mean,” the March Hare went on.
“I do,” Alice hastily replied; “at least... at least I mean what I say... that’s the same
thing, you know.”
“Not the same thing a bit!” said the Hatter. “You might just as well say that ‘I see what
I eat’ is the same thing as ‘I eat what I see’!”
“You might just as well say,” added the March Hare, “that ‘I like what I get’ is the same
thing as ‘I get what I like’!”
“You might just as well say,” added the Dormouse, who seemed to be talking in his
sleep, “that ‘I breathe when I sleep’ is the same thing as ‘I sleep when I breathe’!”
(Alice’s Adventures in Wonderland) [1]
The following exercises are related to logical operators (d) ∀x ∈ Z, if x is even or x is odd, then x2 −3x+1
and equivalent propositional forms. is odd.
(e) ∀x, y ∈ Z, if xy is odd, then x is odd and y is
1. Let x, y ∈ Z. Let P be the Boolean value of the
odd.
proposition ‘x is odd,’ and let Q be the Boolean
value of the proposition ‘y is even.’ Using ∧, ∨, and
¬, write each of the following as a logical operator The following exercises are related to the converse and
applied to P and Q: contrapositive of an implication.
(a) x is odd and y is even. 3. State the converse and contrapositive of each of the
following implications:
(b) x is even and y is odd.
(c) Both x and y are even. (a) ∀x ∈ R, if 0 < x, then −x < 0.
(d) Neither x nor y is even. (b) ∀x, y, a ∈ R, if ax = ay and a , 0, then x = y.
(c) ∀a ∈ R, if ∀x ∈ R, ax = 0, then a = 0.
(e) At least one of x or y is even.
(d) ∀a ∈ R, if ∃x ∈ R, x , 0 and ax = 0, then
(f) At least one of x or y is odd. a = 0.
(g) At most one of x or y is even. (e) ∀a, b, x ∈ R, if a < x and x < b, then
(h) At most one of x or y is odd. ∃t ∈ (0, 1), x = (1 − t)a + tb.
(i) Exactly one of x or y is even. 4. In the passage above, taken from ”Alice’s Adven-
(j) Exactly one of x or y is odd. tures in Wonderland” by Lewis Carroll, Alice, the
March Hare, the Dormouse, and the Mad Hatter
2. Given the following pairs of equivalent logical op- have a conversation about whether or not converse
erators: statements are equivalent to one another.
(a) x ⇒ (y ⇒ z) ≡ (x ∧ y) ⇒ z (a) Rewrite each of the implications in the con-

versation in the form ‘if P then Q’ and ‘if Q
(b) x ⇒ (y ∨ z) ≡ (x ∧ ¬y) ⇒ z
then P’ to see that each pair are in fact con-
(c) (x ∧ y) ⇒ z ≡ (x ∧ ¬z) ⇒ ¬y verse statements.
(d) (x ∨ y) ⇒ z ≡ (x ⇒ z) ∧ (y ⇒ z) (b) Write the negation of each statement.
(e) x ⇒ (y ∧ z) ≡ (x ⇒ y) ∧ (x ⇒ z) (c) Is Alice right when she says “that’s the same
thing, you know” or is the Hatter right when
rewrite the following propositions using the corre- he says they are “Not the same thing a bit!”?
sponding equivalences above.
5. Give, if possible, an example of a true implication
(a) ∀a, x ∈ R, if a , 0, then if ax = a, then x = 1. statement for which:
(b) ∀x, y, z ∈ R, if xy = 0, then y = 0 or x = 0. (a) the converse is true.
(c) ∀x, y, z ∈ Z, if xz = yz and z , 0 then x = y. (b) the converse is false.
(c) the contrapositive is true. 7. (a) Show, using a truth table, that for any Boolean
(d) the contrapositive is false. values x and y, at least one of x ⇒ y or y ⇒ x
must be true.
6. Give, if possible, an example of a false implication
statement for which: (b) Give reasons why the following propositions
(a) the converse is true. are both false:
(b) the converse is false.
i. ∀x ∈ R, if x < 0, then x2 < 1.
(c) the contrapositive is true.
(d) the contrapositive is false. ii. ∀x ∈ R, if x2 < 1, then x < 0.
(c) Explain why part 7b does not contradiction

part 7a.
0.4. SET NOTATION 29
0.4 Set Notation
We use the word set to describe a collection of objects. The objects in the set are called
elements of the set. The notation that we use to describe this relationship between sets and
elements is
x∈A
which denotes that the object x is an element of the set A. To describe a particular set,
we commonly refer to the collection of objects satisfying a certain condition. To do this,
we need a universe of discourse from which the objects are taken and an open sentence
that gives the condition that these objects must satisfy to be considered elements in the set.
Consider for example
{x ∈ N | x < 6}
Here, the universe of discourse is N. This tells us that the type of objects considered when
defining the set are natural numbers. The open sentence that must be satisfied for x to be
an element is x < 6. This tells us that those and only those natural numbers satisfying the
condition x < 6 are considered elements of the set. Hence for any given natural number
a ∈ N, if a < 6 we write
a ∈ {x ∈ N | x < 6}
whereas if a ≥ 6 we write
a < {x ∈ N | x < 6}.
In general, given a universe of discourse U and an open sentence P(x),

we can define the set
{x ∈ U | P(x)}.
Some common examples of sets defined in this way, which will make frequent appearances
throughout the course, are the real intervals:
Definition 0.4.1. For a, b ∈ R,
(a, b) = {x ∈ R | a < x and x < b}

(a, b] = {x ∈ R | a < x and x ≤ b}
[a, b) = {x ∈ R | a ≤ x and x < b}
[a, b] = {x ∈ R | a ≤ x and x ≤ b}
(a, ∞) = {x ∈ R | a < x}
[a, ∞) = {x ∈ R | a ≤ x}
(−∞, b) = {x ∈ R | x < b}
(−∞, b] = {x ∈ R | x ≤ b}
(−∞, ∞) = R.
Another example that may be less familiar but will also be used very often, is the ideals of
the integers:
Definition 0.4.2. For a ∈ Z, the ideal of Z generated by a is the set
hai = {x ∈ Z | ∃t ∈ Z, x = at}.
Another set that at first glance may not seem very important but we will see is useful in
many arguments is the empty set ∅. The empty set is the set with no elements. Hence,
given a universe of discourse U, the statement
∀x ∈ U, x < ∅ is true.
Equivalently,
∃x ∈ U, x ∈ ∅ is false.
Further, given a set A whose elements are in the universe of discourse U,
A , ∅ means ∃x ∈ U, x ∈ A.
Equivalently,
A = ∅ means ∀x ∈ U, x < A.
For example, the real interval (1, 0) = ∅. This is because (1, 0) is a set whose elements are
in the universe of discourse R, and the statement ∀x ∈ R, x < (1, 0) is true. To see that this
is the case, notice that for x ∈ R, to say x ∈ (1, 0) means 1 < x and x < 0. Hence, to say
x < (1, 0) means x ≤ 1 or 0 ≤ x. The statement ‘x ≤ 1 or 0 ≤ x’ is true for all values of x in
the real numbers.
Subsets and Set Equality
Given two sets A and B, whose elements belong to a common universe of discourse U, to
say that A is a subset of B means that B contains all of the elements of A and possibly, but
not necessarily, more. Since mathematics requires the use of clear and precise language,
we will give this definition formally in symbolic form:
0.4. SET NOTATION 31
Definition 0.4.3. For sets A and B whose elements belong to the universe
of discourse U, A is a subset of B, denoted A ⊆ B, means
∀x ∈ U, if x ∈ A, then x ∈ B.
For example, the number systems discussed earlier have the following inclusions
N⊆Z⊆Q⊆R⊆C
Although it may seem unnecessary, we provide an explicit interpretation of what it means

for two sets to be equal.
For sets A and B whose elements belong to the universe of discourse U,

A = B means
A ⊆ B and B ⊆ A.
Notice that if we write out this definition in terms of the definitions of A ⊆ B and B ⊆ A,
we have that A = B means
∀x ∈ U, if x ∈ A then x ∈ B, and ∀x ∈ U, if x ∈ B then x ∈ A.
Equivalently, this can be written as
∀x ∈ U, x ∈ A if and only if x ∈ B.
Hence two sets are equal means that the sets have exactly the same elements.
The reason it is necessary to specify this, is that in general two objects are considered
equal provided they are one and the same object. In the case of sets, however, the notion of
equality makes reference only to the elements of the sets and not to anything pertaining to
containment of those elements. This can hopefully provide a better intuitive view of what
the word set is meant to signify. The set is determined only by its elements. Hence, we do
not view a set as a container with several objects inside. In this flawed view, the container
is seen as a substance in itself that has an identity beyond the objects that it contains. In
such a view, two containers could be interpreted as distinct even though the objects they
contain are the same; the additional identity of the container provides the distinction. This
is not the case with sets. As set is only a collective reference to its elements and has no
other distinguishing properties or characteristics. Hence the method by which the elements
are collected, or any other property related to the containment of the elements does not
contribute to the identity of the set. Only the collective identities of the elements determine
the identity of the set.
To make this point more clear, consider the example

aπ
A = x ∈ R ∃a ∈ Z, x = sin
2
B = {x ∈ R | x ∈ Z and x2 < 2}.

The only elements in A are −1, 0, and 1. The same is true for B. Because the two sets
contain exactly the same elements, we can say that A ⊆ B and B ⊆ A; hence A = B. Thus A
and B are considered to be identical, regardless of the fact that their elements were collected
in different ways.
In the case where A ⊆ B but A , B, we say that A is a proper subset of B. This is often
denoted as A ( B. It is worth while to take a minute to dissect the statement ‘A ⊆ B but
A , B.’ Since A = B means A ⊆ B and B ⊆ A, the statement A , B can only mean A * B
or B * A. Hence, in the case where A ( B, we must have A ⊆ B and B * A. That is,
∀x ∈ U, if x ∈ A then x ∈ B, and ∃y ∈ U, y ∈ B and y < A.
Complements
It is often useful to describe a set by the objects that are not elements. For example, we
define the irrational numbers to be all of the real numbers that are not rational. To formalize
this idea, we introduce the complement of a set.
Definition 0.4.4. Let A be a set whose elements are in the universe of

discourse U. We define the complement of A, denoted Ac , to be
Ac = {x ∈ U | x < A}.
Further, given any two sets A and B, we define the complement of A

relative to B, denoted B \ A to be
B \ A = {x ∈ U | x ∈ B and x < A}.
For a simple example, we may observe that in the universe of discourse Z, the complement
of the even integers is the odd integers, or that
{1, 2, 3, 4} \ {1, 4} = {2, 3}.
For now, the intention is not to examine these concepts in any great detail. A more rigorous
treatment will be given in chapter 2. Presently, it will suffice to give the formal definitions
of some basic set notation, for reference purposes in the coming chapters, and to develop a
brief intuitive view of that which these notations represent.
Chapter 1
Structure of a Mathematical Proof
The linguistic rigor introduced in the previous section will be maintained throughout the
course. Clear, precise language is the first requirement for writing mathematics. The sec-
ond requirement is good definitions. Our definitions are the foundations upon which we
build all of our theory.
When a mathematician believes he or she has discovered a mathematical truth, he or she

is required to demonstrate this truth using only the following:
What we can use in a mathematical proof
1. Definitions
2. Rules of Logic
3. Prior theorems
In particular, we may not appeal to our intuition about what is true and what is false when
writing a mathematical proof. Developing a good mathematical intuition is important,
because it is often what leads us to our discoveries. However, intuition varies from one
mathematician to another and can often deceive us. It is therefore not considered a valid
justification of truth in a mathematical demonstration.
Since we have not proven any results as of yet, in the early stages of our theoretical devel-
opment we will only have definitions and logic available to us. We begin by attempting a
clear, precise definition of a real number. There are several different systems of numbers
in mathematics, including natural numbers, integers, rational numbers, and real numbers.
We will define each of these systems individually, however the observant reader will notice
many similarities between the definitions.
33
34 CHAPTER 1. STRUCTURE OF A MATHEMATICAL PROOF
1.1 The Real Numbers
Intuitively, the real numbers may be viewed as the set of points on a continuous unbroken
line extending infinitely in both directions. We denote the set of real numbers with the
symbol R.
The definition we will give for the real numbers is what is known as an axiomatic defini-
tion. Axioms are defining properties of an object. They are propositions that are true of
the object, by definition of the object. We give an axiomatic definition of the real numbers
by listing several axioms of the real numbers and defining the real numbers as any system
satisfying these axioms.
Definition 1.1.1. The system of real numbers, denoted R is defined to be a set, containing
constants 0 and 1 with 0 , 1, binary operations given by (x, y) 7→ x + y and (x, y) 7→ xy,
unary operations given by x 7→ −x and for x , 0, x 7→ x−1 , and a relation <, satisfying the
following axioms:
A1 For all x, y ∈ R, x + y = y + x. (Addition is commutative)
A2 For all x, y, z ∈ R, (x + y) + z = x + (y + z). (Addition is associative)
A3 For every x ∈ R, x + 0 = x and 0 + x = x. (0 is an additive identity)

A4 For every x ∈ R, x + (−x) = 0 and (−x) + x = 0. (−x is an additive inverse of x)
M1 For all x, y ∈ R, xy = yx. (Multiplication is commutative)
M2 For all x, y ∈ R, (xy)z = x(yz). (Multiplication is associative)
M3 For every x ∈ R, x1 = x and 1x = x. (1 is a multiplicative identity)

M4 For all x ∈ R with x , 0, xx−1 = 1 and x−1 x = 1. (x−1 is a multiplicative inverse of
non-zero x)
DL For all x, y, z ∈ R, x(y + z) = (xy) + (xz) and (y + z)x = (yx) + (zx). (Multiplication
distributes over addition)
O1 For all x, y ∈ R, exactly one of x < y, x = y, or y < x holds. (Trichotomy)
O2 For all x, y, z ∈ R, if x < y and y < z then x < z. (Transitivity)
O3 For all x, y, z ∈ R, if x < y then x + z < y + z. (Addition preserves order)
O4 For all x, y, z ∈ R, if x < y and 0 < z, then xz < yz. (Multiplication by a positive
preserves order)
C If A and B are non-empty subsets of R such that for all a ∈ A and all b ∈ B, a < b,
then there is an element x ∈ R such that for all a ∈ A, a ≤ x, and for all b ∈ B, x ≤ b.
(Completeness)
1.1. THE REAL NUMBERS 35
To say that the operation given by (x, y) 7→ x + y is a binary operation means that +
combines a pair of real numbers (x, y) into a single real number x + y. This means
that for all x, y ∈ R, x + y ∈ R and that for x, y, a, b ∈ R, if x = a and y = b then
x + y = a + b. Similarly, to say that there is a binary operation given by (x, y) 7→ xy
means that for all x, y ∈ R, xy ∈ R and that for x, y, a, b ∈ R, if x = a and y = b then xy = ab.
That there is a unary operation given by x 7→ −x means for every x ∈ R, −x ∈ R,

and for all x, a ∈ R, if x = a then −x = −a. Similarly, for x ∈ R with x , 0 we have
x−1 ∈ R, and for all x, a ∈ R with x , 0 and a , 0, if x = a then x−1 = a−1 .
The relation < is intended to convey the usual intuitive meaning of relative size. However,
formally, we require only that it is a relation, meaning for x, y ∈ R, x < y is a statement that
is either true or false, and this relation obeys properties O1 through O4. We will also add
to this the following commonly used notation:
Notation
For x, y ∈ R, the notation ‘x ≤ y’ means ‘x < y or x = y.’ Similarly, ‘x ≥ y’ means ‘x > y
or x = y.’
The final axiom (completeness) is the most difficult to decipher. It is worthwhile to pause
a minute to comment on this axiom and its intended meaning.
A remark about the completeness axiom

The completeness axiom (axiom C) describes in what sense the real numbers are
continuous and unbroken. If we imagine the real numbers as a line extending infinitely in
both directions and we cut the line into two pieces A and B, then the point at which the
cut is made must be a real number. That is, there is a real number x lying between all the
points in piece A and the points in piece B.
Note that this condition is not satisfied by the rational numbers, since we may divide
the rational numbers into two pieces A = {x ∈ Q | x3 ≤ 2} and B = {x ∈ Q | x3 ≥ 2}. In
this case, the cut is made at a gap in the rational numbers. That is, there√ is no rational
3
number lying between the points in piece A and the points in piece B, since 2 is irrational.
All of the properties listed as axioms are indeed properties that a set we call the real num-
bers should have. There are, however, several more properties of the real numbers that we
know are true, which are not listed. Any other properties of the real numbers will need
to be proven using our definition. To illustrate how this is done, we will prove some very
trivial properties about the real numbers.
To ensure that we are only using the axioms, logic, and previously proven propositions,
we will cite the source of each step in the proof. In general, this much citation is not
needed in a proof. As propositions and their proofs become more involved, we will drop
this practice. When writing your own proofs, you may freely use any part of a definition,
logic, and previously proven proposition. Strict citation of everything used in the proof
will become less necessary as the course progresses. However, references to propositions
and explanations of logical steps are always valuable if they add clarity to your argument.
A good rule of thumb to ensure we are liberal enough with our citations and expla-
nations is to write as if the reader is a fellow student in the class. That is, we should
assume that the reader is only familiar with the concepts that have been introduced as of
the particular stage in the course in which we are writing. Further, we should assume
that the reader is only as comfortable with the material as a student in the class would be
at that point in time. For example, since we have only just introduced the axioms of the
real numbers, a student would not be expected to have committed them to memory at this
point in time. Hence, in our proofs at this stage, we should include references to these
axioms wherever we use them. Later, as we grow more familiar with these axioms and the
references to them consequently become tedious and unnecessary, we will begin to drop
the more obvious of these references.
Remember
A proof is written for the purpose of demonstrating the truth of your statement to a reader.
Helpful explanations of how you are making your steps can make your proofs easier to
read.
Proving Statements with Universal Quantifiers
The first proposition we will prove is that any number multiplied by zero will give zero as a
result. In symbolic form, this can be written as ∀x ∈ R, 0x = 0 and x0 = 0. This may seem
intuitively obvious, and it is. In fact, any set claiming to be the set of real numbers would
need to abandon such a claim if ∀x ∈ R 0x = 0 and x0 = 0 were found to be false. Hence,
for a proposition this intuitive, the purpose of the proof is not so much to determine its truth,
but rather to ensure that its truth can be derived from our definition. In addition, it provides
a good example of the process by which one derives further properties from the defining
axioms. This process, and the various techniques involved, will occupy our attention for
the remainder of the chapter. In particular, in the case of the proposition ∀x ∈ R, 0x = 0 and
x0 = 0, since this is a general statement about all real numbers, we are forced to consider
what must be done in order to prove such statements containing universal quantifiers.
We add notes, pointing out the structure of the proof, in the left margin. These notes are not
part of the proof itself and are only included to indicate the proof structures and techniques
in use.
Proposition 1.1.1.
∀x ∈ R, x0 = 0 and 0x = 0.
Proof.
Arbitrary element I Let x ∈ R

 The constants 0 and 1 are real numbers.
 Since 1 + 0 = 1, by axiom A3, we have x(1 + 0) = x1.
(x1) + (x0) = x1, by axiom DL.


x + (x0) = x, by axiom M3.


 Applying the additive inverse unary operation, −x ∈ R; hence −x + (x + (x0)) = −x + x.
Demonstration  
 (−x + x) + (x0) = −x + x, by axiom A2.
 0 + (x0) = 0, by axiom A4.
x0 = 0, by axiom A3.


In addition to this, 0x = 0, by axiom M1.


Therefore, x0 = 0 and 0x = 0.
Conclusion I Therefore, ∀x ∈ R, x0 = 0 and 0x = 0.
Notice, in the above proof, that we begin by letting x represent an arbitrary real number.
We do this because the proposition we are proving,
‘∀x ∈ R, x0 = 0 and 0x = 0,0
is a statement about all real numbers. It is not enough to prove the statement for one
particular real number. i.e. we do not merely prove 1(0) = 0, or 2(0) = 0. We must
prove every real number multiplied by zero gives zero; hence we must prove x0 = 0 for a
nonspecific, arbitrary, real number x.
In general, when we are proving any statement about all real numbers, we will prove the
statement for an arbitrary real number. More generally, proofs of propositions with the
form ‘∀x ∈ U, P(x),’ where P(x) is an open sentence and U is the universe of discourse for
the variable x, will have the following form:
Proof of ∀x ∈ U, P(x):
Proof.
Arbitrary element I Let x ∈ U.
 ..

 .
Demonstration  Demonstrate P(x).
 ..
.
Conclusion I Therefore P(x).
Therefore, ∀x ∈ U, P(x).
The next proposition we consider is that for all real numbers x and y,
(−x)y = −(xy).
Notice that this proposition is again a statement about all real numbers x and y; hence our
proof will involve arbitrary real numbers x and y. Indeed, the first line of the proof will be
‘Let x, y ∈ R.’
Proposition 1.1.2.
∀x, y ∈ R, (−x)y = −(xy).
Proof.
Arbitrary elements I Let x, y ∈ R.
(−x)y + (xy + (−(xy))) = ((−x)y + xy) + (−(xy)), by axiom A2.


 (−x)y + 0 = ((−x) + x)y + (−(xy)), by axioms A4 and DL.
Demonstration 
 (−x)y = (0)y + (−(xy)), by axioms A3 and A4.
(−x)y = 0 + (−(xy)), by proposition 1.1.1.
Conclusion I Therefore, (−x)y = −(xy), by axiom A4.
Therefore, ∀x, y ∈ R, (−x)y = −(xy).
For practice in identifying the axioms that allow us to perform algebraic manipulations, the
reader is invited to fill in the blanks in the following proof:
Proposition 1.1.3.
∀x, y ∈ R, (−x)(−y) = xy.
Proof.
(−x)(−y) + (−(xy) + xy) = ((−x)(−y) + (−(xy))) + xy, by axiom .


 (−x)(−y) + 0 = ((−x)(−y) + (−x)y) + xy, by axiom and prop. .
(−x)(−y) = (−x)((−y) + y) + xy, by axioms

Demonstration  and .

 (−x)(−y) = (−x)(0) + xy, by axiom .
(−x)(−y) = 0 + xy, by proposition .
Conclusion I Therefore, (−x)(−y) = xy, by axiom .
Therefore, ∀x, y ∈ R, (−x)(−y) = xy.
Given the axioms of the real numbers, we see that the only real numbers whose existence
is explicitly given in the definition are 0 and 1. Of course we know that there are more
real numbers than only these two. Since the existence of other real numbers is not given in
the definition, it will need to be proven. To prove that there are real numbers other than 0
and 1, we need to use the order axioms. That the non-order axioms are not sufficient on
their own to guarantee the existence of more numbers can be seen by noticing that the set
{0, 1} satisfies all of the non-order axioms if we define addition and multiplication by the
following tables:
+ 0 1 · 0 1
0 0 1 0 0 0
1 1 0 1 0 1
The reader may wish to verify that this system does in fact satisfy axioms A1, A2,
A3, A4, M1, M2, M3, M4, and DL.
Proof By Contradiction
We introduce a proof technique known as proof by contradiction, or reductio ad absurdum.

In this technique, we begin by assuming the negation of the proposition we wish to prove.
From this assumption, we then derive a logical contradiction, demonstrating that the as-
sumption must be false. We will illustrate this technique by using it to prove the existence
of more real numbers. We begin by investigating the order relation between our two known
real numbers, 0 and 1. If fact, we will demonstrate that 0 < 1. We will then prove the
existence of the number 2, which we define as 1 + 1.
Proposition 1.1.4.
0<1
Proof.
Assumption I Suppose that 0 ≮ 1
By the trichotomy axiom, we have either 1 < 0 or 1 = 0.



 By axiom M3, we have that 1 , 0, hence it must be the case that 1 < 0.
 Then 1 + (−1) < 0 + (−1), by axiom O3.
This reduces to 0 < −1, by axioms A3 and A4.


Demonstration 
 By axiom O4 we then have (0)(−1) < (−1)(−1).

 Therefore, 0 < (−1)(−1), by proposition 1.1.1,
 and hence 0 < (1)(1), by proposition 1.1.3.
This is then 0 < 1, due to axiom M3.

Contradiction I We now have 0 < 1 and 0 ≮ 1, which is a logical contradiction.

Therefore, our initial assumption that 0 ≮ 1 must be false.
Conclusion I Therefore 0 < 1.
Notice that, in this proof, we assume the negation of the statement we are proving and
derive a logical contradiction from that assumption. In doing so, we demonstrate that the
negation of our proposition is false, because it implies a contradiction. Since the negation
of our proposition is shown to be false, we may then conclude that our proposition is true.
In general, to prove a proposition P by contradiction, we use the following structure:
Proof by contradiction of a proposition P:

Proof.
Assumption I Suppose ¬P.
..

.


Demonstration  Demonstrate any contradiction.
..

.
Contradiction I Therefore R and ¬R, which is a contradiction.
Conclusion I Therefore P.
The next two propositions demonstrate that the real number 1 + 1 is neither 1 nor 0. It is
therefore a new real number, which proves the existence of real numbers beyond 0 and 1.
In both of the following propositions, we will use the proof by contradiction technique.
Proposition 1.1.5.
1+1,1
Proof.
Assumption I Suppose 1 + 1 = 1.
 Then (1 + 1) + (−1) = 1 + (−1).
By associativity, we have 1 + (1 + (−1)) = 1 + (−1).


Therefore 1 + 0 = 0,

Demonstration 

 and hence 1 = 0.
However, by axiom M3, we have 1 , 0.
Contradiction I Therefore, 1 = 0 and 1 , 0, which is a contradiction.
Conclusion I Therefore, 1 + 1 , 1
Proposition 1.1.6.
1+1,0
Proof.
Assumption I Suppose 1 + 1 = 0.

 0 < 1 from proposition 1.1.4.
 Therefore, 0 + 1 < 1 + 1.
Demonstration  Hence 1 < 1 + 1, by axiom A4.

Our assumption reduces this to 1 < 0.
Contradiction I Therefore 0 < 1 and 1 < 0, which contradicts axiom O1.
Conclusion I Therefore, 1 + 1 , 0
From proposition 1.1.5, proposition 1.1.6, we see that 1 + 1 is a new real number. The
common notation for this real number is 2. We now have the existence of the real number
2 and the fact that 1 + 1 = 2.
Proving Implications with a ‘Direct Proof’
A direct proof is one method we can use to prove a statement of the form ‘If P then Q,’
where P and Q are propositions. In an implication of the form ‘If P then Q,’ we call the
proposition P the antecedent and the proposition Q the consequent. In a direct proof, we
assume that the antecedent P is true and use our assumption to derive the consequent Q,
thereby proving that if P is true, then Q must also be true. A direct proof follows the form:
Proof of if P then Q.
Proof.
Assumption I Assume P.
 ..

 .
Demonstration  
 Demonstrate Q.
 ..
.
Conclusion I Therefore Q.
Therefore, if P then Q.
We will continue our development of the basic properties of real numbers with several
examples of proofs using the direct proof method.
Proposition 1.1.7.
∀x ∈ R, if x < 0 then 0 < −x.
Proof.
Arbitrary element I Let x ∈ R.
Assumption I Assume x < 0
Demonstration [ Then −x + x < −x + 0.
Conclusion I Therefore, 0 < −x.
Conclusion I Therefore, if x < 0 then 0 < −x.
Therefore, ∀x ∈ R, if x < 0 then 0 < −x.
Proposition 1.1.8.
∀x, y, a ∈ R, if a + x = a + y then x = y.
Proof.
Arbitrary elements I Let x, y, a ∈ R.
Assumption I Assume a + x = a + y
Then −a + (a + x) = −a + (a + y).


Demonstration  associativity, (−a + a) + x = (−a + a) + y.

By
Hence, 0 + x = 0 + y.

Conclusion I Therefore, x = y.
Conclusion I Therefore, if a + x = a + y then x = y.
Therefore, ∀x, y, a ∈ R, if a + x = a + y then x = y.
Proposition 1.1.9.
∀x, y, z ∈ R, if x < y and z < 0 then yz < xz.
Proof.
Arbitrary elements I Let x, y, z ∈ R.
Assumption I Assume x < y and z < 0.
Then 0 < −z by proposition 1.1.7.



 We now have x < y and 0 < −z.
 By axiom O4, we then have x(−z) < y(−z).
Therefore, −(xz) < −(yz) by proposition 1.1.2.


 Adding xz to both sides gives us xz + (−(xz)) < xz + (−(yz)).
 Hence 0 < xz + (−(yz)).
yz to both sides gives us 0 + yz < xz + (−(yz) + yz).

 Adding
Hence, 0 + yz < xz + 0.

Conclusion I Therefore, yz < xz.
Conclusion I Therefore, if x < y and z < 0 then yz < xz.
Therefore, ∀x, y, z ∈ R, if x < y and z < 0 then yz < xz.
Proving Statements with Existential Quantifiers
To prove a statement of the form ∃x ∈ U, P(x), we must demonstrate the existence of at

least one element x ∈ U that satisfies the open sentence P(x). One technique for proving
existence is construction. In this method, we assign a specific value to the constant x. The
specific value of x that we choose must be an element of the universe of discourse U, and
it must satisfy the desired property P(x). We then demonstrate that our chosen value of x
does indeed have the property P(x). We illustrate this technique with a very basic example,
followed by some more involved examples.
Proposition 1.1.10.
∃x ∈ R, x < 0.
Proof.
Set a value I Put x = −1.
Demonstration [ Since 0 < 1, we have 0 + (−1) < 1 + (−1); hence −1 < 0.
Conclusion I Therefore x < 0.
Therefore, ∃x ∈ R, x < 0.
In the above example, we proved ∃x ∈ R, x < 0 by identifying an example of a real number

x with that property. In general, to prove a statement of the form ∃x ∈ U, P(x), we may
exhibit one example of a value of x that makes P(x) true. We then demonstrate that P(x) is
indeed true for the x we found.
Proof of ∃x ∈ U, P(x):
Proof.
Set a value  I Put x = .
..
.


Demonstration  Demonstrate P(x).


 .
..
Conclusion I Therefore, P(x).
Therefore, ∃x ∈ U, P(x).
Continuing with our investigation of the basic properties of real numbers, we can now show
that one may construct more and more real numbers without end. In fact, the real numbers
are unbounded. We show this by proving
∀x ∈ R, ∃y ∈ R, x < y.
Notice that this proposition contains a universal quantifier followed by an existential quan-
tifier. To prove this statement, we will first introduce an arbitrary real number x. For this
x, we then need to demonstrate the statement ‘∃y ∈ R, x < y.’ To do so, we must prove the
existence of the real number y, hence we must exhibit a real number y that has the desired
property of being larger than x. We do this as follows:
Proposition 1.1.11.
∀x ∈ R, ∃y ∈ R, x < y.
Proof.
" I
Set a value Put y = x + 1.
Demonstration By proposition 1.1.4, 0 < 1.
We then have x + 0 < x + 1.
Conclusion I Therefore, x < y.
Conclusion I Therefore, ∃y ∈ R, x < y.
Therefore, ∀x ∈ R, ∃y ∈ R, x < y.
Notation
For x, y, z ∈ R, the notation x < y < z means x < y and y < z. Also, for x, y ∈ R with y , 0,
x
means xy−1 . Note that by commutativity, this is the same as y−1 x.
y
Proposition 1.1.12.
∀x, y ∈ R, if x < y then ∃z ∈ R, x < z < y.
Proof.
Assumption I Assume x < y.
x+y
Set a value I Put z = .
 2
Since x < y, we have x + x < x + y.


 Therefore, 2x < x + y.
 x+y
 By axiom O4, we then have x < .
 2
Therefore, x < z.


Demonstration  Further, since x < y, we have x + y < y + y.
Therefore, x + y < 2y.

x+y

< y.

 We then have
 2
 Therefore, z < y.
We now have x < z and z < y.

Conclusion I Therefore, x < z < y.

Conclusion I Therefore, ∃z ∈ R, x < z < y.
Conclusion I Therefore, if x < y then ∃z ∈ R, x < z < y.
Therefore, ∀x, y ∈ R, if x < y then ∃z ∈ R, x < z < y.
Proving Implications by ‘Contraposition’
Recall from section 0.3 that any implication ‘if P, then Q’ can be rewritten in its contra-
positive form: ‘if ¬Q, then ¬P.’ The resulting proposition is necessarily equivalent to the
original implication because the corresponding logical operators x ⇒ y and ¬y ⇒ ¬x are
equivalent. Thus, to prove a statement of the form ‘if P, then Q,’ we can prove the equiv-
alent statement ‘if ¬Q, then ¬P.” We do this by applying the direct proof method to the
contrapositive form ‘if ¬Q, then ¬P.” That is, to prove a statement of the form ‘if P, then
Q,’ we can assume ¬Q and demonstrate ¬P.
In general, a proof by contraposition has the form:
Proof of if P then Q:
Proof.
Assumption I Assume ¬Q.
 ..
.


Demonstration  Demonstrate ¬P.
 ..
.
Conclusion I Therefore ¬P.
Contrapositive I Therefore, if ¬Q then ¬P.
Therefore, if P then Q.
We illustrate this technique by proving the statement
∀x, y ∈ R, if xy ≤ 0 then x ≤ 0 or y ≤ 0 .
| {z } | {z }
P Q
Rewriting the implication in this statement in its contrapositive form gives us the equivalent
proposition
∀x, y ∈ R, if x > 0 and y > 0 then xy > 0.
| {z } | {z }
¬Q ¬P
We will then assume x > 0 and y > 0 and demonstrate xy > 0.
Proposition 1.1.13.
∀x, y ∈ R, if xy ≤ 0 then x ≤ 0 or y ≤ 0.
Proof.
Assumption I Assume x > 0 and y > 0.
Demonstration [ Then by axiom O4, xy > 0y.
Conclusion I Therefore xy > 0, by proposition 1.1.1.
Contrapositive I Therefore, if x > 0 and y > 0 then xy > 0.
Conclusion I Therefore, if xy ≤ 0 then x ≤ 0 or y ≤ 0.
Therefore, ∀x, y ∈ R, if xy ≤ 0 then x ≤ 0 or y ≤ 0.
For another example, consider the statement
∀a, b ∈ R, if ∀x ∈ R, a < x ⇒ b ≤ x, then b≤a

| {z } |{z} .
P Q
Rewriting the implication in its contrapositive form gives us
a < b then ∃x ∈ R, a < x and x < b

∀a, b ∈ R, if |{z}
| {z } .
¬Q ¬P
We will then assume a < b and demonstrate ∃x ∈ R, a < x and x < b.

Proposition 1.1.14.
∀a, b ∈ R, if ∀x ∈ R, a < x ⇒ b ≤ x, then b ≤ a.
Proof.
Arbitrary elements I Let a, b ∈ R.
Assumption I Assume a < b.
a+b
Set a value I Put x = .
 2

 Since a < b, we have a + a < a + b.
 Therefore, 2a < a + b.
 a+b
 By axiom O4, we then have a < .
 2
Therefore, a < x.

 Further, since a < b, we have a + b < b + b.
Therefore, a + b < 2b.


 a+b
 We then have < b.
 2
Therefore, x < b.
Conclusion I We now have a < x and x < b.
Conclusion I Therefore, ∃x ∈ R, a < x and x < b.
Contrapositive I Therefore, if a < b then ∃x ∈ R, a < x and x < b.
Conclusion I Therefore, if ∀x ∈ R, if a < x then b ≤ x, then b ≤ a.
Therefore, ∀a, b ∈ R, if ∀x ∈ R, a < x ⇒ b ≤ x, then b ≤ a.
Proving Implications by Contradiction
Yet another method for proving a statement of the form ‘if P then Q’ is to use the proof by
contradiction technique discussed earlier. Recall that in a proof by contradiction, we as-
sume the negation of the statement we are trying to prove and derive a logical contradiction
from that assumption. If the statement we wish to prove has the form ‘if P then Q,’ then its
negation will be ‘P and ¬Q.’ Hence, to prove an implication using a proof by contradiction,
we use the following form:
Proof of if P then Q:
Proof.
Assumption I Suppose P and ¬Q.
..

.


Demonstration  Demonstrate any contradiction.
..

.
Contradiction I Therefore R and ¬R, which is a contradiction.
Conclusion I Therefore if P then Q.
The above technique is not different from the proof by contradiction technique discussed
earlier. In fact it is simply a special case of the earlier proof by contradiction method, in
the case where the statement we are proving is an implication. However, since so many
mathematical statements contain implications, it can be useful to practice using the proof
by contradiction technique in this specific setting. To get accustomed to the application of
the proof by contradiction method to implications, we give three examples:
Proposition 1.1.15.
∀x ∈ R, if x , 0, then x−1 , 0.
Proof.
Arbitrary elements I Let x ∈ R.
Assumption I Suppose x , 0 and x−1 = 0.
 Then xx−1 = x(0).
By axiom M4 and proposition 1.1.1, this becomes 1 = 0.

 However, by axiom M3, we have 1 , 0.
Contradiction I We now have the contradiction: 1 = 0 and 1 , 0.
Conclusion I Therefore, if x , 0 then x−1 , 0.
Conclusion I Therefore, ∀x ∈ R, if x , 0, then x−1 , 0.
Proposition 1.1.16.
∀x ∈ R, if x > 0, then x−1 > 0.
Proof.
Arbitrary elements I Let x ∈ R.
Assumption I Suppose x > 0 and x−1 ≤ 0.
We then have x−1 < 0 or x−1 = 0.


Since x , 0, proposition 1.1.15 gives us x−1 , 0.


 It must then be the case that x−1 < 0.
 Since 0 < x and x−1 < 0, by proposition 1.1.9 we have xx−1 < (0)x−1 .
By axiom M4 and proposition 1.1.1, this becomes 1 < 0.


However, by proposition 1.1.4, we have 0 < 1.
Contradiction I By trichotomy, 1 < 0 and 0 < 1 is a contradiction.
Conclusion I Therefore, if x > 0, then x−1 > 0.
Conclusion I Therefore, ∀x ∈ R, if x > 0, then x−1 > 0.
Proposition 1.1.17.
∀x, y ∈ R, if xy = 0, then x = 0 or y = 0.
Proof.
Assumption" I Suppose xy = 0 and x , 0 and y , 0.
By axiom M4, there is a real number x−1 for which x−1 x = 1.
Demonstration
Therefore, y = (1)y = (x−1 x)y = x−1 (xy) = x−1 (0) = 0.
Contradiction I We now have y = 0 and y , 0, which is a contradiction.
Conclusion I Therefore, if xy = 0, then x = 0 or y = 0.
Conclusion I Therefore, ∀x, y ∈ R, if xy = 0, then x = 0 or y = 0.
Three Methods for Proving Implications
We have now discussed three methods for proving statements of the form ‘if P then Q.’
These methods are direct proof, contraposition, and proof by contradiction. Their struc-
tures are as follows:
Direct Proof Contraposition Contradiction

Proof of if P then Q: Proof of if P then Q: Proof of if P then Q:
Proof. Proof. Proof.
Assume P. Assume ¬Q. Suppose P and ¬Q.
.. .. ..
. . .
Demonstrate Q. Demonstrate ¬P. Find a
.. .. contradiction.
. . ..
∴ Q. ∴ ¬P. .
∴ if P then Q. ∴ if ¬Q then ¬P. ∴ R and ¬R.
∴ if P then Q. ∴ if P then Q.
Notice that the assumption in a proof by contradiction is the combined assumption of both
the direct proof and the proof by contraposition. However, this superior assumption comes
at a price: The direction of the proof is no longer decided. When doing a direct proof
or a proof by contraposition, the direction of the proof is certain, and it is often useful to
have knowledge of the conclusion one is working towards. On the other hand, in a proof
by contradiction, we have only our assumption. It often takes some creativity to steer our
assumption into a contradiction. There is no general rule to tell us which method we will
find easiest. In practice, we try to use the method that results in the most natural and elegant
proof.
Proof by Exhaustion (Cases)
One technique that has grown in popularity since computers have started to be used in
mathematical proofs is called proof by exhaustion. We can use this method if there are
only finitely many elements in our universe of discourse, or if the universe of discourse
can be divided into finitely many categories (such as the integers into even and odd, or the
real numbers into positive, negative, and zero). We can prove a statement with a universal
quantifier by examining all of the possible values, or categories, the variable can take and
proving the statement for each of these individually. For example:
Proposition 1.1.18.
∀x ∈ R, x2 ≥ 0.
Proof.
By the trichotomy axiom, we have either x > 0 or x < 0, or x = 0.
We will consider each of these three cases.
Case 1: x > 0.
We have x(x) > 0(x), by axiom O4.


Demonstration  Hence, x2 > 0, by proposition 1.1.1.

Then the statement x2 > 0 or x = 0 is true.

Conclusion I Therefore, x2 ≥ 0.
Case 2: x < 0.
We have −x + x < −x + 0, hence 0 < −x by axioms A3 and A4.


By axiom O4, we then have 0(−x) < (−x)(−x).


Demonstration  By proposition 1.1.1, this becomes 0 < (−x)(−x).
By proposition 1.1.3, we then have 0 < (x)(x).


Therefore, x2 > 0.
" Case 3: x = 0.
Demonstration Then (x)(x) = (0)(0).
Therefore, x2 = 0, by proposition 1.1.1.
In all cases we have x2 ≥ 0.
Therefore, ∀x ∈ R, x2 ≥ 0.
Note that every case ends with the same conclusion, that being the statement that is to
be proven. This is an essential feature of a proof by exhaustion; we must end every case
by concluding that the statement we are proving is true in that case. Further, notice that
we have used the trichotomy axiom to prove that the cases we are considering cover the
entire universe of discourse. In general, we must ensure that the cases chosen exhaust all
possibilities. That is, we must show that every element in the universe of discourse, that is
open for consideration under our hypotheses, falls into at least one of the categories being
considered as cases.
The next three examples prove that axioms O2, O3, and O4 hold for the ≤ relation
just as they do with the < relation.
Proposition 1.1.19.
∀x, y, z ∈ R, if x ≤ y and y ≤ z, then x ≤ z.
Proof.
Arbitrary element I Let x, y, z ∈ R.
Assume x ≤ y and y ≤ z.
Then (x < y or x = y) and (y < z or y = z).
Case 1: x < y and y < z
Demonstration [ In this case, we have x < z by transitivity.
Conclusion I Therefore, x ≤ z.
Case 2: x < y and y = z
Demonstration [ Then x < z.
Case 3: x = y and y < z
Demonstration [ Then x < z.
Case 4: x = y and y = z.
Demonstration [ Then x = z.
In all cases, x ≤ z.
Therefore, if x ≤ y and y ≤ z, then x ≤ z.
Therefore, ∀x, y, z ∈ R, if x ≤ y and y ≤ z, then x ≤ z.
Proposition 1.1.20.
∀x, y, z ∈ R, if x ≤ y, then x + z ≤ y + z.
Proof.
Arbitrary element I Let x, y, z ∈ R.
Assume x ≤ y.
Then x < y or x = y.
Case 1: x < y
Demonstration [ Hence, x + z < y + z by axiom O3.
Conclusion I Therefore, x + z ≤ y + z.
Case 2: x = y
Demonstration [ Then x + z = y + z.
Conclusion I Therefore, x + z ≤ y + z.
In both cases, x + z ≤ y + z.
Therefore, if x ≤ y, then x + z ≤ y + z.
Therefore, ∀x, y, z ∈ R, if x ≤ y, then x + z ≤ y + z.
Proposition 1.1.21.
∀x, y, z ∈ R, if x ≤ y and 0 ≤ z, then xz ≤ yz.
Proof.
Let x, y, z ∈ R.
Assume x ≤ y and 0 ≤ z.
Then x < y or x = y.
Case 1: x < y
Since 0 ≤ z, we have 0 < z or 0 = z.
We consider these two subcases.
Case 1.1: 0 < z.
Since x < y and 0 < z, we have xz < yz by axiom O4.
Therefore, xz ≤ yz.
Case 1.2: 0 = z.
In this subcase, we have xz = 0 and yz = 0; hence xz = yz.
In both subcases, xz ≤ yz.
Case 2: x = y
Then xz = yz.
In all cases, xz ≤ yz.
Therefore, if x ≤ y and 0 ≤ z, then xz ≤ yz.
Therefore, ∀x, y, z ∈ R, if x ≤ y and 0 ≤ z, then xz ≤ yz.
The max, min, and absolute value functions
We take some time now to introduce the max, min, and absolute value functions. These are
functions of real numbers that will be used throughout the course and provide an opportu-
nity to practice some of the proof techniques discussed above.
Definition 1.1.2. For x, y ∈ R, we define the maximum of x and y to be

(
x if x ≥ y
max(x, y) = .
y if x < y
Similarly, we define the minimum of x and y to be

(
x if x ≤ y
min(x, y) = .
y if y < x
Proposition 1.1.22.
∀x, y ∈ R, max(x, y) ≥ x and max(x, y) ≥ y.
Proof.
Let x, y ∈ R.
We have either x ≥ y or x < y, which we consider in cases.
Case 1: x ≥ y.
In this case, we have max(x, y) = x.
Hence, max(x, y) ≥ x.
Since max(x, y) = x and x ≥ y, we have max(x, y) ≥ y.
Therefore, max(x, y) ≥ x and max(x, y) ≥ y.
Case 2: x < y.
In this case, we have max(x, y) = y.
Hence, max(x, y) ≥ y.
Since max(x, y) = y and x < y, we have x < max(x, y).
Therefore, max(x, y) ≥ x.
We then have max(x, y) ≥ x and max(x, y) ≥ y.
In both cases, we have max(x, y) ≥ x and max(x, y) ≥ y.
Therefore, ∀x, y ∈ R, max(x, y) ≥ x and max(x, y) ≥ y.
In the next example, we prove a statement of the form ‘P if and only if Q.’ The meaning
of this is that both the implication ‘if P then Q,’ and its converse ‘if Q then P’ are true.
Hence, to prove a statement of the form ‘P if and only if Q,’ we are required to prove both
‘if P then Q’ and ‘if Q then P.’ These two proofs are generally done consecutively and
are independent of one another. That is, we write a full proof of the statement ‘if P then
Q’ including whichever assumptions and conclusions need to be made, and when this is
finished we proceed to write a full proof of ‘if Q then P.’ When both proofs are completed,
we may conclude ‘P if and only if Q.’ The two proofs need not be done using the same
technique. That is, we could prove ‘if P then Q’ using a direct proof and ‘if Q then P’ using
a proof by contraposition or any other combination of proof techniques.
Proposition 1.1.23.
∀x, y ∈ R, max(x, y) = min(x, y) if and only if x = y.
Proof.
Let x, y ∈ R.
Assume x = y.
Then x ≥ y; hence max(x, y) = x.
Also, x ≤ y; hence min(x, y) = x.
Therefore, max(x, y) = min(x, y).
Therefore, if x = y then max(x, y) = min(x, y).
Conversely, assume x , y.
Then by trichotomy we have either x < y or y < x.
Case 1: x < y.
In this case, we have max(x, y) = y.
Also, since x < y, we have x ≤ y; hence min(x, y) = x.
Since min(x, y) = x, max(x, y) = y, and x < y,
we have min(x, y) < max(x, y).
Therefore, min(x, y) , max(x, y).
Case 2: y < x.
In this case, we have min(x, y) = y.
Also, since y < x, we have x ≥ y; hence max(x, y) = x.
Since max(x, y) = x, min(x, y) = y, and y < x,
we have min(x, y) < max(x, y).
Therefore, min(x, y) , max(x, y).
In both cases, we have min(x, y) , max(x, y).
Therefore, if x , y then min(x, y) , max(x, y).
Therefore, if min(x, y) = max(x, y) then x = y.
We have now shown if x = y then max(x, y) = min(x, y),
and if min(x, y) = max(x, y) then x = y.
Therefore, max(x, y) = min(x, y) if and only if x = y.
Therefore, ∀x, y ∈ R, max(x, y) = min(x, y) if and only if x = y.

Definition 1.1.3. For x, ∈ R, we define the absolute value of x to be

(
x if x ≥ 0
|x| = .
−x if x < 0
Proposition 1.1.24.
∀x ∈ R, |x| ≥ 0.
Proof.
Let x ∈ R.
By trichotomy we have 0 < x, x < 0 or x = 0, which we consider in cases.
Case 1: 0 < x.
In this case, we have x ≥ 0; hence |x| = x.
Since x ≥ 0 and |x| = x, we have |x| ≥ 0.
Case 2: x < 0.
In this case, we have |x| = −x.
Since x < 0, we have −x + x < −x + 0; hence 0 < −x.
Therefore, −x ≥ 0; hence |x| ≥ 0.
Case 3: x = 0.
In this case, we have x ≥ 0; hence |x| = x = 0.
Therefore, |x| ≥ 0.
In all cases we have |x| ≥ 0.
Therefore, ∀x ∈ R, |x| ≥ 0.
Proposition 1.1.25.
∀x ∈ R, |x| = | − x|.
Proof.
Let x ∈ R.
By trichotomy we have 0 < x, x < 0 or x = 0, which we consider in cases.
Case 1: 0 < x.
In this case, we have x ≥ 0; hence |x| = x.
Since 0 < x, we have −x + 0 < −x + x; hence −x < 0.
Therefore, | − x| = −(−x) = x.
Therefore, |x| = | − x|.
Case 2: x < 0.
Since x < 0, we have −x + x < −x + 0; hence 0 < −x.
Therefore, −x ≥ 0; hence | − x| = −x.
Case 3: x = 0.
In this case, we have x ≥ 0; hence |x| = x = 0.
Further, since −x = 0, we have −x ≥ 0; hence | − x| = −x = 0.
In all cases we have |x| = | − x|.
Therefore, ∀x ∈ R, |x| = | − x|.
Proposition 1.1.26.
∀x ∈ R, x ≤ |x|.
Proof.
Let x ∈ R.
We consider the two cases x ≥ 0 and x < 0.
Case 1: x ≥ 0.
In this case, we have |x| = x; hence x ≤ |x|.
Case 2: x < 0.
Since x < 0, we have 0 < −x by proposition 1.1.7.
Now, since x < 0 and 0 < −x, we have x < −x by transitivity.
Therefore, x < |x|; hence x ≤ |x|.
In both cases we have x ≤ |x|.
Therefore, ∀x ∈ R, x ≤ |x|.
Proposition 1.1.27.
∀x, a ∈ R, |x| < a if and only if −a < x < a.
Proof.
Let x, a ∈ R.
Assume |x| < a.
Then x ≤ |x| and |x| < a, hence x < a.
Further, since −x ≤ | − x| and | − x| = |x|, we have −x ≤ |x|.
We now have −x ≤ |x| and |x| < a, hence −x < a.
Adding (x − a) to both sides gives us −x + (x − a) < (x − a) + a.
Hence −a < x.
Therefore, −a < x and x < a, which means −a < x < a.
Therefore, if |x| < a then −a < x < a.
Conversely, assume −a < x < a.
Since either x ≥ 0 or x < 0, we consider two cases:
Case 1: x ≥ 0.
In this case, |x| = x and x < a, hence |x| < a.
Case 2: x < 0.
In this case, |x| = −x.
Adding a − x to both sides of −a < x gives us −a + (a − x) < (a − x) + x.
Therefore, −x < a; hence |x| < a.
In both cases we have |x| < a.
Therefore, if −a < x < a then |x| < a.
We have now shown if |x| < a then −a < x < a, and if −a < x < a then |x| < a.
Therefore, |x| < a if and only if −a < x < a.
Therefore, ∀x, a ∈ R, |x| < a if and only if −a < x < a.
Theorem 1.1.28 (The Triangle Inequality).
∀x, y ∈ R, |x + y| ≤ |x| + |y|.
Proof.
Let x, y ∈ R.
Since |x| ≤ |x|, we have −|x| ≤ x ≤ |x|.
Similarly, since |y| ≤ |y|, we have −|y| ≤ y ≤ |y|.
In particular, since −|x| ≤ x and −|y| ≤ y, we have −|x| − |y| ≤ x + y.
Similarly, since x ≤ |x|, and y ≤ |y|, we have x + y ≤ |x| + |y|.
We now have −|x| − |y| ≤ x + y ≤ |x| + |y|.
By proposition 1.1.27, we then have |x + y| ≤ |x| + |y|.
Therefore, ∀x, y ∈ R, |x + y| ≤ |x| + |y|.
Proposition 1.1.29.
∀x, y ∈ R, ||x| − |y|| ≤ |x − y|.
Proof.
Let x, y ∈ R.
By the triangle inequality, we have |(x − y) + y| ≤ |x − y| + |y|.
Therefore, |x| ≤ |x − y| + |y|; hence |x| − |y| ≤ |x − y|.
Similarly, |(x − y) + (−x)| ≤ |x − y| + | − x|.
Therefore, | − y| ≤ |x − y| + | − x|; hence |y| ≤ |x − y| + |x|.
This gives us −|x − y| ≤ |x| − |y|.
We now have −|x − y| ≤ |x| − |y| ≤ |x − y|.
By proposition 1.1.27, this gives us ||x| − |y|| ≤ |x − y|.
Therefore, ∀x, y ∈ R, ||x| − |y|| ≤ |x − y|.
Proposition 1.1.30.
∀x, y ∈ R, if 0 ≤ x and 0 ≤ y, then |x − y| ≤ max(x, y).
Proof.
Let x, y ∈ R.
Assume 0 ≤ x and 0 ≤ y.
We consider two cases: x ≥ y and x < y.
Case 1: x ≥ y.
In this case, we have max(x, y) = x.
Further, adding −y to both sides of x ≥ y gives us x − y ≥ 0.
We then have |x − y| = x − y.
Now, since 0 ≤ y, we have −y ≤ 0; hence x − y ≤ x. Therefore, |x − y| ≤
max(x, y).
Case 2: x < y.
In this case, max(x, y) = y.
Further, adding −y to both sides of x < y gives x − y < 0.
Therefore, |x − y| = −(x − y) = y − x.
Now, since 0 ≤ x, we have −x ≤ 0; hence y − x ≤ y. Therefore again,
|x − y| ≤ max(x, y).
In both cases we have |x − y| ≤ max(x, y).
Therefore, ∀x, y ∈ R, if 0 ≤ x and 0 ≤ y, then |x − y| ≤ max(x, y).
The techniques discussed above will often be used in combination with one another. Noting
that some of the techniques allow us to make assumptions, we should take a moment to
discuss the nature and limits of these assumptions. The first aspect of making assumptions
that must be emphasized is that one cannot make any extra assumptions in addition to those
determined by the structure of the sentence being proven. If an additional assumption is
made, then it will change the conclusion that can be drawn from it accordingly; it will thus
alter the proposition that has been proven. For example, if one is to prove the statement ‘if
x ∈ Z, then x2 + 3x + 1 is odd,’ he or she is able to assume that x ∈ Z. If the additional
assumption that ‘x is odd’ is added into the proof, then even though the conclusion ‘x2 +
3x + 1’ may be reached, the statement ‘if x ∈ Z, then x2 + 3x + 1 is odd’ will not have been
proven. Rather, the statement ‘if x ∈ Z and x is odd, then x2 + 3x + 1 is odd’ will have
been proven. Although this point may seem obvious, it suggests a general rule that must
be followed strictly: The only assumptions that can be made are those determined by the
structure of the sentence and the structure of the proof.
Assumptions Containing Quantifiers
Another consideration about assumptions is that the assumption itself may have a complex
sentence structure involving quantifiers, connectives, or implications. We will examine
some examples of these. The first situation to consider is when our assumption contains
an existential quantifier. That is, we assume ∃x ∈ U, P(x) for some predicate P(x) and
universe of discourse U. In this case, the assumption is that there is at least one value of x
for which the statement P(x) is true. This assumption tells us nothing of the identity of this
value of x; it only tells us that such a value of x exists. For example, suppose a biologist
were to inform us that there is at least one fish in the world that is able to breathe outside
of the water. In saying this, the biologist does not provide us with the identity of such a
fish; we only have knowledge of its existence. It is easy, in this context, to see that it would
be a mistake to identify a fish of our choosing with the one that we are told exists. Indeed,
if I confused my own goldfish with the unknown fish whose existence has been assured,
then the consequences could be fatal for my goldfish. Thus, if a statement of the form
∃x ∈ U, P(x) has been assumed, it is a mistake to assign a specific value of our choosing
to x. Under such an assumption, we know only that there is at least one value in U that
satisfies the predicate P(x), but we do not know what that value is or anything about it
except that P(x) is true. However, since we know one exists, we may choose a constant to
represent such a value in our proof. That is, we may assign a constant to denote the thing
that is known to exist. For example, if in a proof we are able to assume ∃x ∈ R, x2 < x,
we may choose a ∈ R with a2 < a and use this nonspecific constant a in our proof. By
‘nonspecific’, we mean that we do not know the value of a; in fact, we know nothing about
the constant a except that it is a real number for which a2 < a. The following proposition
provides an example:
Proposition 1.1.31.
∀x, y ∈ R, if ∃a ∈ R, x < a and a < y, then x < y.
Proof.
" I
Assumption Assume ∃a ∈ R, x < a and a < y.
Choose t ∈ R with x < t and t < y.
Demonstration
We then have x < t and t < y.
Conclusion I Therefore, x < y by transitivity.
Conclusion I Therefore, if ∃a ∈ R, x < a and a < y, then x < y.
Therefore, ∀x, y ∈ R, if ∃a ∈ R, x < a and a < y, then x < y.
The next situation to consider is when our assumption contains a universal quantifier. That
is, we assume ∀x ∈ U, P(x), for some predicate P(x) and universe of discourse U. In
this case, our assumption tells us that P(x) is true for every value of x in the universe of
discourse U. It does not, however, tell us that such an x exists. After all, the universe of
discourse could be empty. For example, the assumption that every perfect leader must be
benevolent does not in itself make the claim that there is such a thing as a perfect leader.
The assumption would be true even if there were no perfect leaders, but in this case it would
not be very informative. In fact, the assumption ∀x ∈ U, P(x) is only useful to us if we
independently know of some element or elements of the set U. That is, if, independent of
this assumption, we establish that some value a is an element of U, then the assumption
∀x ∈ U, P(x) informs us that P(a) is true. For example, suppose that at some point in
a proof we assume ∀x ∈ R, x2 < x. This assumption alone does not inform us of the
existence of any real numbers for which x2 < x. However, since we know, independent
of this assumption, that 1 ∈ R, the assumption ∀x ∈ R, x2 < x tells us that 12 < 1 (If
this particular example were part of a proof by contradiction, we would then have our
contradiction). We give an example of the application of this principle:
Proposition 1.1.32.
∀x, y ∈ R, if ∀a ∈ R, if x ≤ a then y ≤ a, then y ≤ x.
Proof.
Assumption I Assume ∀a ∈ R, if x ≤ a then y ≤ a.
Demonstration [ Since x ∈ R and x ≤ x, we then have y ≤ x.
Conclusion I Therefore, y ≤ x.
Conclusion I Therefore, if ∀a ∈ R, if x ≤ a then y ≤ a, then y ≤ x.
Therefore, ∀x, y ∈ R, if ∀a ∈ R, if x ≤ a then y ≤ a, then y ≤ x.
Notice also that in the above proof, the assumption contains an implication. It is important
to note that when we assume a statement of the form ‘if P then Q,’ this does not make
claims about the truth or falsity of the propositions P and Q. In particular, it does not
tell us that P and Q are true. It only tells us that if P were true, then we would be able
to conclude that Q would also be true. Such an assumption only becomes usable if we
can establish, independently, that P is true. That is, having assumed ‘if P then Q,’ if we
are able at any point in our proof to demonstrate that P is true, then our assumption will
inform us that Q is also true. If we are not able to establish the truth of P independently,
then our assumption tells us nothing of the truth of Q.
Examining the above example in more detail, we see that our assumption is ‘∀a ∈ R, if
x ≤ a then y ≤ a.’ As mentioned earlier, this does not inform us of the existence of an
a ∈ R satisfying the predicate ‘if x ≤ a then y ≤ a.’ However, since x is a known element
of R, the assumption informs us that the statement ‘if x ≤ x then y ≤ x’ is true. Since this
is an implication, it tells us nothing unless we can establish the truth of the statement x ≤ x
independently. Since x ≤ x is indeed true regardless of our assumption, the assumed fact
that ‘if x ≤ x then y ≤ x’ is true then informs us that y ≤ x is true. This completes the
proof. For another example of these same principles at work, consider the following:
Proposition 1.1.33.
∀x, y ∈ R, if ∀a ∈ R, a < x if and only if a < y, then x = y.
Proof.
Assumption I Suppose ∀a ∈ R, a < x if and only if a < y, and x , y.
Since x , y, by trichotomy, we have either x < y or y < x.
We consider these as cases.
Case 1: x < y.
Demonstration [ Since x ∈ R and x < y, by our assumption we must have x < x.
Contradiction I We then have x = x and x < x, which contradicts trichotomy.
Case 2: y < x.
Demonstration [ Since y ∈ R and y < x, by our assumption we must have y < y.
Contradiction I We then have y = y and y < y, which contradicts trichotomy.
Both cases lead to a contradiction.
Conclusion I Therefore, if ∀a ∈ R, a < x if and only if a < y, then x = y.
Conclusion I Therefore, ∀x, y ∈ R, if ∀a ∈ R, a < x if and only if a < y, then x = y.
A remark about ‘if and only if’

For propositions P and Q, to say ‘P if and only if Q’ means ‘if P, then Q, and if Q, then P’.
That is, it makes the statement that both the implication ‘if P, then Q’ and its converse ‘if
Q, then P’ are true. ‘P if and only if Q’ is sometimes denoted by ‘P ⇔ Q’, which recalls
the similar notation P ⇒ Q which stands for ‘if P, then Q’. On occasion, one also sees ‘P
if and only if Q’ abbreviated as ‘P iff Q’.
The Archimedean Property
The Archimedean property gives us an important connection between the real numbers and
the natural numbers (the positive whole numbers). To understand what the Archimedean
property says, consider the length of a single step to be a real number a, with a > 0.
Next, consider another real number b that represents a great distance that we must travel.
The Archimedean property optimistically guarantees us that it is possible to complete our
journey. That is, there is a number of steps n for which na > b. Although this may seem
intuitively obvious, the proof requires an axiom that we have not yet used in a proof: the
completeness axiom.
The completeness axiom

If A and B are non-empty subsets of R such that for all a ∈ A and all b ∈ B, a < b, then
there is an element x ∈ R such that for all a ∈ A, a ≤ x, and for all b ∈ B, x ≤ b.
Recall the remark earlier, that if we imagine the real numbers as a line and we cut the line
into two pieces A and B, then the cut must fall on a real number. That is, there is a real
number x lying between all the points in piece A and the points in piece B.
To prove the Archimedean property, we use the proof by contradiction technique.

Hence, we are able to assume that there is a distance b that cannot be traveled in any
number of steps of length a. That is, b > na for all n ∈ N. With a view to using the
completeness axiom, we introduce a set A consisting of all possible distances travelable by
taking steps of size n. That is,
A = {x ∈ R | ∃n ∈ N, x = an}.
Since under our assumption, the distance b is larger than every element of A, we define the
set B to be all such distances that are too long to be reached by steps of size a. That is,
B = {y ∈ R | ∀n ∈ N, na ≤ y}.
We then have all of the travelable distances in A and the distances that are too large to be
traversed in B. Hence, we can show that the elements of A are all smaller than the elements
of B.
With this setup, the completeness axiom gives us the existence of a c ∈ R, larger
than all the elements of A and smaller that all of the elements of B. A contradiction arises
when we take one step back from c and consider the length c − a. This length is too small
to be in B, yet if it is reachable by n steps of size a, that is if na ≥ c − a, then one more
step to (n + 1)a will take us beyond c, which is likewise not allowable under the given
circumstances. Thus we can reach a contradiction. The details of this argument are worked
out as follows:
Theorem 1.1.34 (The Archimedean Property).
∀a, b ∈ R, if a > 0 then ∃n ∈ N, na > b.
Proof.
Let a, b ∈ R.
Assume a > 0.
Suppose there does not exist n ∈ N, such that na > b.
Then for all n ∈ N, na ≤ b.
Let A = {x ∈ R | ∃n ∈ N, x = na} and let B = {y ∈ R | ∀n ∈ N, na ≤ y}.
Then, a = 1a ∈ A and b ∈ B, so A and B are non-empty sets.
Let x ∈ A and y ∈ B.
Then, we can choose m ∈ N such that x = ma.
and, by definition of B, x = ma ≤ y.
Thus, x ≤ y.
Hence, ∀x ∈ A, ∀y ∈ B, x ≤ y.
Thus, by the completeness axiom of R, we can choose c ∈ R, such that
for all x ∈ A and y ∈ B, x ≤ c and c ≤ y.
Now, let n ∈ N.
Then, n + 1 ∈ N, so (n + 1)a ∈ A.
Therefore, (n + 1)a ≤ c.
But then, na ≤ c − a.
Therefore, ∀n ∈ N, na ≤ c − a.
Hence, c − a ∈ B, so c ≤ c − a.
This is a contradiction, since a > 0.
Thus, there exists n ∈ N, such that na > b.
Therefore, if a > 0 then ∃n ∈ N, na > b.
Therefore, ∀a, b ∈ R, if a > 0 then ∃n ∈ N, na > b.
The following two corollaries associated with the Archimedean property are also very use-
ful:
Corollary 1.1.35.
∀b ∈ R, ∃n ∈ N, n > b.
Proof.
Let b ∈ R.
Since 1 > 0, we have by the Archimedean property, ∃n ∈ N, n(1) > b.
Therefore, ∀b ∈ R, ∃n ∈ N, n > b.
Corollary 1.1.36.
1
∀a ∈ R, if a > 0 then ∃n ∈ N, < a.
n
Proof.
Let a ∈ R.
Assume a > 0.
By the Archimedean property, since 1 ∈ R, we have ∃n ∈ N, na > 1.
1
For such an n, we have a > .
n
1
Therefore, ∃n ∈ N, < a.
n
1
Therefore, if a > 0 then ∃n ∈ N, < a.
n
1
Therefore, ∀a ∈ R, if a > 0 then ∃n ∈ N, < a.
n
Exercises 1.1.
Notation:
For all x, y ∈ R, x − y means x + (−y).
For all x, y ∈ R with y , 0, yx means x(y−1 ).
For all x, y ∈ R, x ≤ y means x < y or x = y.
For all x, y, z ∈ R, x < y < z means x < y and y < z.
2 is defined by 2 = 1 + 1; 3 is defined by 3 = 2 + 1; 4 is defined by 4 = 3 + 1; and so on.
Prove the following propositions. At each step indicate 22. ∀x ∈ R, if 0 < x − 2, then 3 < 2x.
the axiom or proposition you have used.
23. ∀x, y ∈ R, x < y if and only if −y < −x.
1. ∀x, y ∈ R, −(x + y) = −x − y.
24. ∀x, y ∈ R, if 0 < x < y, then y−1 < x−1 .
2. ∀x, y ∈ R, −(x − y) = y − x.
25. ∀x, y ∈ R, if x < 0 and y < 0, then 0 < xy.
−1 −1
3. ∀x ∈ R \ {0}, (x ) = x.
26. ∀x, y ∈ R, if x < 0 and 0 < y, then xy < 0.
4. ∀x ∈ R, ∀y ∈ R \ {0}, yx(y ) = x. −1
27. ∀x, y ∈ R, if x < y and 0 < y, then 2x < 4y.
5. ∀x, y ∈ R \ {0}, (xy) −1
=y x . −1 −1
28. ∀x, y ∈ R, if x < y, then 2 + x < 4 + y.
−1
6. ∀x, y ∈ R \ {0}, (x y) −1
= y x. −1
29. ∀a, b, x, y ∈ R, if a < b and x < y, then a+ x < b+y.
7. ∀a, b ∈ R, ∀x ∈ R \ {0}, xa+xb
x = a + b. 30. ∀a, b, x, y ∈ R, if 0 < a < x and 0 < b < y, then
ab < xy.
8. ∀a ∈ R, ∀b, x ∈ R \ {0}, xa
xb = ba .
9. ∀a, b ∈ R, ∀x ∈ R \ {0}, a
x + b
x = a+b
x . Prove the following propositions using a proof by con-
ay+bx traposition.
10. ∀a, x ∈ R, ∀b, y ∈ R \ {0}, + = a
b
x
y by .
31. ∀x, y ∈ R, if x ≤ y and y ≤ x, then x = y.
11. ∀a, x ∈ R, ∀b, y ∈ R \ {0}, ab yx = ax
by .
32. ∀x ∈ R, if 2x ≤ 0, then x − 1 ≤ 0.
−1 ay
12. ∀a ∈ R, ∀b, x, y ∈ R \ {0}, a
b
x
y = bx . 33. ∀x ∈ R, if x2 ≤ x, then x ≤ 1.
13. 1 < 3. 34. ∀x, y ∈ R, if 3 + x ≤ 1 + y, then x ≤ y.
14. 1 < 4. 35. ∀x, y ∈ R, if x2 ≤ y2 , then y ≤ 0 or x ≤ y.
15. ∀x ∈ R, x < x + 1. 36. ∀x, y ∈ R, if 2x ≤ y, then x ≤ y or x ≤ 0.
16. ∀x ∈ R, x < x + 2. 37. ∀x, y ∈ R, if x2 − xy ≤ xy − y2 , then x ≤ y.

38. ∀x, y ∈ R, if x2 − xy ≤ xy − y2 , then y ≤ x.
Prove the following propositions using a direct proof.
17. ∀x, y, a ∈ R, if ax = ay and a , 0, then x = y. Prove the following propositions using a proof by con-
tradiction.
18. ∀x ∈ R, if 0 < x, then −x < 0.
39. ∀x, y ∈ R, if x < 0 and 0 < xy, then y < 0.
19. ∀a, x ∈ R, if a + x = 0, then a = −x. (That is, −x is
40. ∀x, y ∈ R, if 0 < x and 0 < xy, then 0 < y.
the only additive inverse of x).
41. ∀x ∈ R, if x , 0, then x−1 , 0.
20. ∀a, x ∈ R, if ax = 1, then a = x−1 . (That is, x−1 is
the only multiplicative inverse of x). 42. ∀x ∈ R, if x < 0, then x−1 < 0.
21. ∀x ∈ R, if 0 < x, then 0 < x + 1. 43. ∀x, y ∈ R, if x < y < 0, then y−1 < x−1 .
44. ∀x, y ∈ R, if xy < 0 and x ≤ 0, then 0 ≤ y−1 . 70. ∀a ∈ R, if ∀x ∈ R, ax ≤ a, then a = 0.
45. ∀a, x, y ∈ R, if ax < ay and y ≤ x, then a ≤ 0. 71. ∀a ∈ R, if ∀x ∈ R, ax ≤ 0, then a = 0.
46. ∀a, x, y ∈ R, if 0 < a and xy < a2 , then x < a or 72. ∀x ∈ R, if ∀a ∈ R, ax ≤ 0, then ∀b ∈ R, 0 ≤ bx.
y < a. 73. ∀x ∈ R, if ∀a ∈ (0, ∞), x ≤ a, then x ≤ 0.
74. ∀x, y ∈ R, if ∀a ∈ (0, ∞), x ≤ a + y, then x ≤ y.
Prove the following propositions.
75. ∀x ∈ R, if ∀a ∈ (0, ∞), x ≤ a, then ∀b ∈ (0, ∞),
47. ∃x ∈ R, x < x2 . x < b.
48. ∃x ∈ R, x2 < x. 76. ∀x ∈ R, if ∀a ∈ (0, ∞), x < a, then ∀b, c ∈ (0, ∞),
cx < b.
49. ∃x ∈ R, x = x2 .
77. ∀x ∈ R, if ∀a ∈ (0, ∞), x < 100a, then ∀b ∈ (0, ∞),
50. ∃x ∈ R, x2 − 3x = 4. x < b.
51. ∀y ∈ R, ∃x ∈ R, y = x + 1. 78. ∀x ∈ R, if ∃m ∈ (0, ∞), ∀a ∈ (0, ∞), x < ma, then
∀b ∈ (0, ∞), x < b.
52. ∀y ∈ R, ∃x ∈ R, y = 3x.
79. ∀x, y ∈ R, if ∀a ∈ R, x ≤ a if and only if y ≤ a, then
53. ∀y ∈ R, ∃x ∈ R, y = 5x − 2.
x = y.
54. ∀y ∈ R, ∃x ∈ R, 2y = 3x + 5. 80. ∀x, y ∈ R, if ∀a ∈ R, x < a if and only if y < a, then
55. ∀a, b ∈ R, if a < b, then ∃x ∈ (0, ∞), a + x = b. x = y.
81. ∀x, y ∈ R, if ∀a ∈ (−∞, x], a < y, then ∃b ∈
56. ∀a, b ∈ R, if a , 0, then ∃x ∈ R, ax = b.
(−∞, y], x < b.
57. ∀x ∈ (0, 1), ∃y ∈ (0, 1), y < x.
82. For a subset S ⊆ R, ∀a, b ∈ S , if ∀x ∈ S , a ≤ x and
58. ∀x ∈ (0, 1), ∃y ∈ (0, 1), x < y. ∀x ∈ S , b ≤ x, then a = b.
(That is, the smallest element of a set is unique.)
59. ∀x, y ∈ R, if x < y, then ∃a ∈ (0, ∞), x + a < y.
60. ∀x, y ∈ R, if x < y, then ∀b ∈ (0, ∞), ∃a ∈ (0, ∞), Prove the following propositions.
x + ab < y. 83. ∀a ∈ R, if ∃x ∈ (0, ∞), x < a, then 0 < a.
61. ∀a, b, x ∈ R, if a < x < b, then ∃t ∈ (0, 1), 84. ∀x, y ∈ R, if ∃a ∈ R, x < a and a ≤ z, then x < z.
x = (1 − t)a + tb.
85. ∀a ∈ R, if ∃x ∈ R, a + x = x, then a = 0.
62. ∀a, b, x ∈ R, if a < b and ∃t ∈ (0, 1), x = (1−t)a+tb,
then a < x < b. 86. ∀a ∈ R, if ∃x ∈ R, x , 0 and ax = x, then a = 1.
63. ∀x, y ∈ R, ∃z ∈ R, x < z and y < z. 87. ∀a ∈ R, if ∃x ∈ R, x , 1 and ax = a, then a = 0.
64. ∀x, y ∈ (0, 1), ∃z ∈ (0, 1), z < x and z < y. 88. ∀a ∈ R, if ∃x ∈ R, x , 0 and ax = 0, then a = 0.
89. ∀a ∈ R, if ∃x ∈ R, ax > 1, then ∃y ∈ R, ay < −1.
Prove the following propositions. 90. ∀a, b ∈ R, if ∃x ∈ (0, ∞), a + x = b, then a < b.
65. ∀a ∈ R, if ∀x ∈ R, a + x = x, then a = 0. (That is, 0 91. ∀x ∈ R, if ∃a ∈ (0, ∞), a ≤ x, then ∃b ∈ (0, ∞),
is the only additive identity). b < x.
66. ∀a ∈ R, if ∀x ∈ R, ax = x, then a = 1. (That is, 1 is 92. ∀x ∈ R, if ∃a, b ∈ (0, ∞), a < bx, then ∃c ∈ (0, ∞),
the only multiplicative identity). c < x.
67. ∀a ∈ R, if ∀x ∈ R, ax = a, then a = 0. 93. ∀x ∈ R, if ∃a ∈ (0, ∞), a < x, then ∀b ∈ (0, ∞),

0 < bx.
68. ∀a ∈ R, if ∀x ∈ R, ax = 0, then a = 0.
94. ∀x ∈ R, if ∃a ∈ (0, ∞), x + a < 0, then ∀b ∈ (0, ∞),
69. ∀a ∈ R, if ∀x ∈ R, ax ≤ x, then a = 1. bx < 0.
95. ∀x ∈ R, if ∃a ∈ (0, ∞), ax < 0, then ∀b ∈ (0, ∞), 118. ∀x, y, a ∈ R, min(x, y) > a if and only if x > a and
bx < 0. y > a.
96. ∀x ∈ R, if ∃a ∈ R, ax < 0, then ∃b ∈ R, 0 < bx. 119. ∀x, y, z ∈ R, if ∀a ∈ R, z ≤ a if and only if x ≤ a
and y ≤ a, then z = max(x, y).
Prove the following inequalities. 120. ∀x, y, z ∈ R, if ∀a ∈ R, a < z if and only if a < x
and a < y, then z = min(x, y).
97. ∀x, y ∈ R, 2xy ≤ x2 + y2 .
98. ∀x, y ∈ R, 4xy ≤ x2 + 4y2 .
Prove the following propositions involving the absolute
99. ∀x, y ∈ R, 4xy ≤ (x + y)2 . value function.
100. ∀x, y ∈ R, 8xy ≤ (x + 2y)2 . 121. ∀x, y ∈ R, |xy| = |x||y|.

101. ∀x, y ∈ R, 4xy ≤ (y + 2x)2 − y2 . 122. ∀x ∈ R, |x|2 = x2 .
102. ∀x, y ∈ R, 4xy ≤ (y + 2x)2 − x2 . 123. ∀x, y ∈ R, ||x| − |y|| ≤ |x + y|.
124. ∀x, y, z ∈ R, |x − y| ≤ |x − z| + |z − y|.

Prove the following propositions involving the max and
min functions. 125. ∀x, y ∈ R, y < |x| if and only if x < −y or y < x.
103. ∀x, y ∈ R, min(x, y) ≤ x and min(x, y) ≤ y. 126. ∀x, y ∈ R, if |x| < y, then x2 < y2 .
104. ∀x, y ∈ R, x + y = max(x, y) + min(x, y). 127. ∀x, y ∈ R, if −y < x < y, then x2 < y2 .
105. ∀a, x, y ∈ R, if x ≤ y, then max(a, x) ≤ max(a, y). 128. ∀x, a ∈ R, if x2 ≤ a2 , then |x| ≤ |a|.
106. ∀a, x, y ∈ R, if x ≤ y, then min(a, x) ≤ min(a, y). 129. ∀x, y ∈ (0, 1), |x − y| < 1.
107. ∀a, b, x, y ∈ R, if x ≤ y and a ≤ b, then max(a, x) ≤
130. ∀a, b, x, y ∈ R, if a < x < b and a < y < b, then
max(b, y).
|x − y| < b − a.
108. ∀a, b, x, y ∈ R, if x ≤ y and a ≤ b, then min(a, x) ≤
131. ∀x ∈ R, if |x − 1| < 1, then x2 + 3x − 4 < 6.
min(b, y).
109. ∀a, x, y ∈ R, if max(a, x) = max(a, y) and 132. ∀x ∈ R, if |x − 2| < 3, then |x2 − 4| < 21.
min(a, x) = min(a, y), then x = y. 133. ∀x, y ∈ R, if ∀a ∈ (0, ∞), |x − y| ≤ a, then x = y.
110. ∀a, x, y ∈ R, if max(a, x) = min(a, y), then x ≤ y. 134. ∀x, y ∈ R, if ∃b ∈ (0, ∞), ∀a ∈ (0, ∞), |x − y| ≤ ab ,
111. ∀x, y, z ∈ R, if x ≤ z, then max(x, min(y, z)) = then x = y.
min(max(x, y), z).
112. ∀x, y, z ∈ R, if x ≤ z, then max(x, max(y, z)) = Prove the following propositions.
max(y, z).
135. ∀x, y ∈ R, max(x, y) = 12 (|x − y| + x + y).
113. ∀a, b, x ∈ R, if a < x < b, then max(b − x, x − a) <
b − a. 136. ∀x, y ∈ R, min(x, y) = 12 (x + y − |x − y|).
114. ∀a, b, x ∈ R, if a < x < b, then 2 min(b − x, x − a) ≤ 137. ∀x, y ∈ R, if min(x, y) < |x − y|, then 2 min(x, y) <
b − a. max(x, y).
115. ∀x, y, a ∈ R, max(x, y) > a if and only if x > a or 138. ∀x, y ∈ R, if |x − y| < min(x, y), then max(x, y) <
y > a. 2 min(x, y).
116. ∀x, y, a ∈ R, min(x, y) < a if and only if x < a or 139. ∀x, y ∈ R, if x > 0 and y > 0, then |x − y| <
y < a. max(x, y).
117. ∀x, y, a ∈ R, max(x, y) < a if and only if x < a and 140. ∀x, y ∈ R, if |x − y| < max(x, y), then x > 0 and
y < a. y > 0.
149. ∀a, x, y ∈ R, if ∀n ∈ N, x + an ≤ y, then a ≤ 0.
Prove the following propositions using the Archimd- 150. ∀x, y ∈ R, if ∀n ∈ N, |x − y| < 1n , then x = y.
edean property.
151. ∀x, y ∈ R, if ∃b ∈ R, ∀n ∈ N, |x − y| < bn , then x = y.
141. ∀x, y ∈ R, if 0 < x, then ∃n ∈ N, y ≤ nx. 152. Let S ⊆ R. If ∃a ∈ R, ∀x ∈ S , x ≤ a, then ∃n ∈ N,
142. ∀x ∈ R, ∃k ∈ Z, k < x. ∀x ∈ S , x ≤ n.
153. ∀x, ε ∈ R, if ε > 0, then ∃n ∈ N, x

< ε.
143. ∀x, y ∈ R, ∃n ∈ N, y ≤ x + n. n
154. ∀x, y ∈ R, if x < y, then ∃n ∈ N, x(n + 1) < y(n − 1).

144. ∀x ∈ R, if 0 < x, then ∃n ∈ N, 3
n < x.
155. ∀a, x, y ∈ R, if x < y, then ∃n ∈ N, x(n + a) <
145. ∀x ∈ R, if ∀n ∈ N, x ≤ 3 + n1 , then x ≤ 3. y(n − a).
146. ∀x ∈ R, if ∀n ∈ N, x ≤ 3 + n5 , then x ≤ 3. 156. ∀x ∈ R, if 1 < x, then ∃n ∈ N, n+1
< x.
n
1
147. ∀x ∈ R, if ∀n ∈ N, 3 − n ≤ x, then 3 ≤ x. 157. ∀x ∈ R, x < 10 if and only if ∃n ∈ N, x + 1
n ≤ 10.
148. ∀x ∈ R, if ∀n ∈ N, 3 − 5
n ≤ x, then 3 ≤ x. 158. ∀x ∈ R, x ≤ 10 if and only if ∀n ∈ N, x < 10 + 1n .
Let x, y, z ∈ [0, 1]. Compare the following properties of the max and min functions to the Boolean algebraic
properties given in section 0.3.
1−0=1 1−1=0 1. Prove the Idempotence properties for max and min.
Idempotence
max(x, x) = x min(x, x) = x
Commutativity 2. Prove the Associativity properties for max and min.

max(x, y) = max(y, x) min(x, y) = min(y, x)
Associativity
max(x, max(y, z)) = max(max(x, y), z)
min(x, min(y, z)) = min(min(x, y), z) 3. Prove the Absorption properties for max and min.
Absorption
max(x, min(x, y)) = x min(x, max(x, y)) = x
Distributivity 4. Prove the Distributivity properties for max and min.

max(x, min(y, z)) = min(max(x, y), max(x, z))
min(x, max(y, z)) = max(min(x, y), min(x, z))
Annihilator
max(x, 1) = 1 min(x, 0) = 0
5. Prove De Morgan’s Laws for max and min.
Identity
max(x, 0) = x min(x, 1) = x
Double Negation
1 − (1 − x) = x 6. Give an example of a value of x ∈ [0, 1] for
which the analogous properties to Complementa-
De Morgan’s Laws tion: min(x, 1 − x) = 0 and max(x, 1 − x) = 1 do
1 − min(x, y) = max(1 − x, 1 − y) not hold.
1 − max(x, y) = min(1 − x, 1 − y)
1.2. THE INTEGERS 69
1.2 The Integers
Definition 1.2.1. The system of integers, denoted Z is defined to be a set, containing con-
stants 0 and 1 with 0 , 1, binary operations given by (x, y) 7→ x + y and (x, y) 7→ xy, a
unary operation given by x 7→ −x, and a relation <, satisfying the following axioms:
A1 For all x, y ∈ Z, x + y = y + x. (Addition is commutative)
A2 For all x, y, z ∈ Z, (x + y) + z = x + (y + z). (Addition is associative)

A3 For every x ∈ Z, x + 0 = x and 0 + x = x. (0 an additive identity)
A4 For every x ∈ Z, x + (−x) = 0 and (−x) + x = 0. (−x is an additive inverse of x)
M1 For all x, y ∈ Z, xy = yx. (Multiplication is commutative)
M2 For all x, y ∈ Z, (xy)z = x(yz). (Multiplication is associative)

M3 For every x ∈ Z, x1 = x and 1x = x. (1 is a multiplicative identity)
FZ For all x, y ∈ Z, if xy = 0 then x = 0 or y = 0. (Zero is not a product of non-zero
factors)
DL For all x, y, z ∈ Z, x(y + z) = (xy) + (xz) and (y + z)x = (yx) + (zx). (Multiplication
distributes over addition)
O1 For all x, y ∈ Z, exactly one of x < y, x = y, or y < x holds. (Trichotomy)
O2 For all x, y, z ∈ Z, if x < y and y < z then x < z. (Transitivity)
O3 For all x, y, z ∈ Z, if x < y then x + z < y + z. (Addition preserves order)

O4 For all x, y, z ∈ Z, if x < y and 0 < z, then xz < yz. (Multiplication by a positive
preserves order)
WOP If S ⊆ Z is a non-empty subset with the property that for all x ∈ S , 0 ≤ x, then there
is an element a ∈ S with the property that for all x ∈ S , a ≤ x. (The Well-Ordering
Property)
Since the integers are defined using many of the same axioms as the real numbers, several
properties we proved for the real numbers hold for the integers as well and can be proven
using the same arguments. One must be cautious however, that since not all non-zero
integers have multiplicative inverses, any results that made use of axiom M4 for the real
numbers may not hold in the integers. In particular, one result we proved for the real
numbers was ∀x, y ∈ R, if x < y then ∃z ∈ R, x < z < y. The proof of this result used the
existence of the multiplicative inverse of 2. This same argument cannot be applied in the
integers, and in fact we will see that this result is false for the integers.
Although we give the definition of the integers independently of the real numbers, it is often
useful to view the integers as a subset of the real numbers. One can show that Z can be
identified with the smallest subset A of R such that 0 ∈ A, and ∀x ∈ A, x + 1 ∈ A and
x − 1 ∈ A. We will elaborate on this construction of the integers throughout the section.
For now, it will suffice to consider the integers as the set defined above and to keep in mind
that in some situations we will find it useful to think of the integers as being a subset of the
real numbers. For example, the Archimedean property and the notion of a rational number
make use of a link between Z and R.
The Well-Ordering Property
On the other hand, there is one axiom of the integers that is not satisfied by the real num-
bers. This is axiom WOP; the well-ordering property. Using axiom WOP, we may prove
statements about the integers that are not true in the real numbers. This section is concerned
with this axiom and its implications.
The Well-Ordering Property

If S ⊆ Z is a non-empty subset with the property that for all x ∈ S , 0 ≤ x, then there is an
element m ∈ S with the property that for all x ∈ S , m ≤ x.
For a ∈ S , the statement ‘for all x ∈ S , a ≤ x’ means a is the smallest element in the set S .
Hence this property is ensuring that every non-empty set of non-negative integers will have
a minimum. That the integers must be non-negative merely ensures that we have a lower
bound on the set S . In fact, the lower bound need not be 0; it can be any integer.
Theorem 1.2.1.
Let A ⊆ Z, with A , ∅. If A is bounded below, then A has a smallest element. That is, if
∃b ∈ Z, ∀x ∈ A, b ≤ x, then ∃a ∈ A, ∀x ∈ A, a ≤ x.
Proof.
Assume ∃b ∈ Z, ∀x ∈ A, b ≤ x.
Choose such a b.
Let S = {x ∈ Z | x + b ∈ A}.
Since A , ∅, we have that ∃y ∈ Z, y ∈ A. Choose such a y.
Let z = y − b.
Then z + b = (y − b) + b = y ∈ A.
Therefore, z ∈ S .
Therefore, ∃z ∈ Z, z ∈ S .
Therefore, S , ∅.
Let x ∈ S .
Hence, x + b ∈ A.
Therefore, b ≤ x + b.
We then have 0 ≤ x.
Therefore, ∀x ∈ S , 0 ≤ x.
We now have S , ∅ and ∀x ∈ S , 0 ≤ x.
By the well-ordering property, we then have ∃m ∈ S , ∀x ∈ S , m ≤ x.
Choose such an m, and set a = m + b.
Since m ∈ S , we have that a = m + b ∈ A.
Let x ∈ A.
Since (x − b) + b = x ∈ A, we have x − b ∈ S .
Therefore, m ≤ x − b.
This gives us m + b ≤ x; hence a ≤ x.
Therefore, ∀x ∈ A, a ≤ x.
Therefore, ∃a ∈ A, ∀x ∈ A, a ≤ x.
Therefore, if ∃b ∈ Z, ∀x ∈ A, b ≤ x, then ∃a ∈ A, ∀x ∈ A, a ≤ x.
An exercise in the previous section asks the reader to write a proof of the fact that
for a subset S ⊆ R, ∀a, b ∈ S , if ∀x ∈ S , a ≤ x and ∀x ∈ S , b ≤ x, then a = b.
Since the order axioms of the integers are the same as those of the real numbers, an
identical proof will suffice to show that the same is true for subsets of the integers.
After a careful reading of this proposition, we can see that the elements a and b
are assumed to be smallest elements of the set S . The conclusion that follows from
this assumption is that a = b. This gives us the rather intuitive result that there can
only be one smallest element of a set. Applied to the topic at hand, this means that the
smallest element of a set, whose existence is given by the well-ordering property, is unique.
In general, given any non-empty set S ⊆ Z, by virtue of the set being non-empty,
we know ∃x ∈ Z, x ∈ S ; hence we can choose an element a ∈ S to work with in our proof.
However, in the case where the non-empty subset of Z is bounded below, the well-ordering
property gives us the existence of a unique smallest element. Hence given a non-empty set
S ⊆ Z that is bounded below, we may choose the smallest element a ∈ S .
We may also prove a reversal of the well-ordering property; that every non-empty set of
integers that is bounded above has a largest element.
Theorem 1.2.2.
Let A ⊆ Z, with A , ∅. If A is bounded above, then A has a largest element. That is, if
∃b ∈ Z, ∀x ∈ A, x ≤ b, then ∃a ∈ A, ∀x ∈ A, x ≤ a.
Proof.
Assume ∃b ∈ Z, ∀x ∈ A, x ≤ b.
Choose such a b.
Let S = {x ∈ Z | − x ∈ A}.
Since A , ∅, choose y ∈ A.
Choose z = −y.
Then −z = −(−y) = y ∈ A.
Therefore, z ∈ S .
Therefore, ∃z ∈ Z, z ∈ S .
Therefore, S , ∅.
Let x ∈ S .
Hence, −x ∈ A.
Therefore, −x ≤ b.
We then have −b ≤ x.
Therefore, ∀x ∈ S , −b ≤ x.
We now have S , ∅ and ∀x ∈ S , −b ≤ x.
By the previous theorem, we can choose m to be the smallest element of S .
Choose a = −m.
Since m ∈ S , we have that a = −m ∈ A.
Let x ∈ A.
Since −(−x) = x ∈ A, we have −x ∈ S .
Therefore, m ≤ −x.
This gives us x ≤ −m; hence x ≤ a.
Therefore, ∀x ∈ A, x ≤ a.
Therefore, ∃a ∈ A, ∀x ∈ A, x ≤ a.
Therefore, if ∃b ∈ Z, ∀x ∈ A, x ≤ b, then ∃a ∈ A, ∀x ∈ A, x ≤ a.
Proofs Using the Well-Ordering Property
A common technique, using the well-ordering property, is often used to prove statements
of the form ‘∀x ∈ Z, P(x),’ where P(x) is an open sentence. Using a proof by contradiction,
we assume ‘∃x ∈ Z, ¬P(x).’ This assumption is identical to the assumption that the set
S = {x ∈ Z | ¬P(x)}
is non-empty. If we can show that S is bounded below, then the well-ordering property
ensures that S has a smallest element. If we are able to construct a smaller element of the
set S then we can establish a contradiction. This argument usually follows the following
structure:
Proof of ∀x ∈ Z, P(x).
Proof.
Assumption I Suppose ∃x ∈ Z, ¬P(x).
Define S I Let S = {x ∈ Z | ¬P(x)}. By our assumption, S , ∅.
..
.


Demonstrate that S is bounded below.

Demonstration  
..
.


Therefore S is bounded below.
Apply WOP I By the WOP, choose a to be the smallest element of S .
Set a value  I Put b = .
..
.


Demonstration  Demonstrate that b ∈ S and b < a.
..
.
Conclusion I Therefore, ∃b ∈ S , b < a.
Contradiction I This is a contradiction, since a is the smallest element of S .
Conclusion I Therefore ∀x ∈ Z, P(x).
We demonstrate this technique by proving one of the most fundamental properties of the
integers. This property may seem intuitively obvious, but since it is not shared by the real
numbers, its proof must use the well-ordering property.
Theorem 1.2.3.
∀x ∈ Z, if x > 0 then x ≥ 1.
Proof.
Assumption I Suppose ∃x ∈ Z, x > 0 and x < 1.
Define S I Let S = {x ∈ Z | 0 < x < 1}. By our assumption, S , ∅.
 Let x ∈ S .

 Then 0 < x < 1.
Demonstration  In particular, 0 < x; hence 0 ≤ x.
Therefore, ∀x ∈ S , 0 ≤ x.


Hence, S is bounded below.
Apply WOP I By the well-ordering property, choose a to be the smallest element of S .
Set a value  I Put b = a2 .
 Since a < 1, we have a2 < (1)a, hence b < a.
Since b < a and a < 1, we have b < 1.


 Since 0 < a, we have 0(a) < a2 , hence 0 < b.
 We now have 0 < b and b < 1, hence b ∈ S .
Contradiction I This is a contradiction, since a is the smallest element of S .
Conclusion I Therefore, ∀x ∈ Z, if x > 0 then x ≥ 1.
Corollary 1.2.4.
∀x, y ∈ Z, if x < y then x + 1 ≤ y.
Proof.
Let x, y ∈ Z.
Assume x < y.
Then y − x > 0.
Therefore, by theorem 1.2.3, we have y − x ≥ 1.
Therefore, x + 1 ≤ y.
Therefore, if x < y, then x + y ≤ y.
Therefore, ∀x, y ∈ Z, if x < y then x + 1 ≤ y.
The above technique is not the only way that one can use the well-ordering property in
a proof. Notice that since the well-ordering property implies the existence of a smallest
element of the set S , it can be used in place of construction to prove statements of the
form ∃x ∈ Z, P(x). If the smallest element of a particular set happens to satisfy the open
sentence P(x), then the existence of such an element can be given by the well-ordering
property. Such an argument will often follow the following structure:
Proof of ∃x ∈ Z, P(x).
Proof.
Define S I Let S = {x ∈ Z | }.
..
.


 Demonstrate that S is non-empty and bounded below.
Demonstration  
..

 .
Therefore S , ∅ and S is bounded below.
Apply WOP I By the WOP, choose a to be the smallest element of S .

 .
..

Demonstration   Demonstrate P(a).
..
.

Conclusion I Therefore, ∃x ∈ Z, P(x).
We demonstrate this technique with several examples:

Proposition 1.2.5.
∀x ∈ R, ∃n ∈ Z, n − 1 < x ≤ n.
Proof.
Let x ∈ R.
Define S I Let S = {k ∈ Z | x ≤ k}.
 By the Archimedean property, we have ∃m ∈ N, x < m.

 For such an m, we have m ∈ S ; hence S , ∅.
 Further, by the Archimedean property, we have ∃p ∈ N, −x < p.
For such a p, we have −p < x.


Choosing b = −p gives us b ∈ Z and b < x.


 Let k ∈ S .
 Then b < x and x < k, hence b < k.

 Therefore, b ≤ k.
 Therefore, ∀k ∈ S , b ≤ k.

 Therefore, ∃b ∈ Z, ∀k ∈ S , b ≤ k.
Hence, S has a lower bound.


Therefore S , ∅ and S is bounded below.
By the well-ordering property, choose n to be the smallest element of S .
Apply WOP  I
 Since n ∈ S , we have x ≤ n.
 Since n − 1 < n, we have n − 1 < S .
 Therefore, n − 1 < x.
Conclusion I Therefore, ∃n ∈ Z, n − 1 < x ≤ n.
Therefore, ∀x ∈ R, ∃n ∈ Z, n − 1 < x ≤ n.
Our next example is Euclid’s division algorithm, which describes the idea of long division;
the expression of a fraction as a quotient and remainder. The proof emerges out of the
interpretation of division as repeated subtraction, along with a guarantee that we will not
14
subtract ad-infinitum. For example, if we wish to reduce to a quotient and remainder,
3
we may repeatedly add or subtract 3 from 14, forming the following sequence of possible
remainders: {. . . , 14, 11, 8, 5, 2, . . . }. The remainder is the smallest non-negative element
in this set (in this case 2), and the quotient is a count of the number of times 3 was sub-
tracted from 14 in order to reach this remainder (in this case 4 times). We then obtain the
14 2
expression 14 = (4)(3) + 2, or = 4 + . In general, the remainder when x is divided by
3 3
y will be the smallest non-negative element of the set of all possible remainders
{a ∈ Z | ∃q ∈ Z, x = yq + a}.
Since the well-ordering property guarantees the existence of a smallest element in any non-
empty set of non-negative integers, its role in the proof of the division algorithm is to secure
the existence of a smallest element in this chain of remainders. Hence, the well-ordering
property ensures that by forming the set of all possible remainders ({. . . , 14, 11, 8, 5, 2, . . . }
in our example), we will inevitably arrive at a smallest non-negative element. Hence, the
remainder will always exist.
Theorem 1.2.6 (The Division Algorithm).
∀x, y ∈ Z, if y > 0 then ∃q, r ∈ Z, x = yq + r and 0 ≤ r < y.
Proof.
Let x, y ∈ Z.
Assume y > 0.
Define S I Let S = {a ∈ Z | 0 ≤ a and ∃q ∈ Z, x = yq + a}.
 We have either x ≥ 0 or x < 0.
In case x ≥ 0, we see that since x = y(0) + x,


∃q ∈ Z, x = yq + x, and 0 ≤ x.


 Therefore, x ∈ S ; hence S , ∅.
< y > 0, we have x = yx + (x − yx).

 In case x 0, since
Therefore, ∃q ∈ Z, x = yq + x(1 − y).

Further, since x < 0 and 1 − y ≤ 0, we have x(1 − y) ≥ 0.



 Therefore, x(1 − y) ∈ S ; hence S , ∅.
 In both cases we have S , ∅.

 Further, for a ∈ S , we have 0 ≤ a; hence S is bounded below by 0.
Therefore, S , ∅ and S is bounded below.

Apply WOP I By the well-ordering property, choose r to be the smallest element of S .
Since r ∈ S , we have that ∃q ∈ Z, x = yq + r and 0 ≤ r.



 Choose q0 ∈ Z such that x = yq0 + r.

 It only remains to show that r < y.
 To this end, suppose y ≤ r.
Then 0 ≤ r − y.


 Put q = q0 + 1.
 Then x = yq0 + r = yq0 + y + r − y = y(q0 + 1) + (r − y) = yq + (r − y).
Therefore, ∃q ∈ Z, x = yq + (r − y) and 0 ≤ r − y.


 Therefore, r − y ∈ S .
However, since y > 0, we have r − y < r.



 Since r is the smallest element of S , this is a contradiction.
Therefore, r < y.
Conclusion I Therefore, ∃q, r ∈ Z, x = yq + r and 0 ≤ r < y.
Therefore, if y > 0 then ∃q, r ∈ Z, x = yq + r and 0 ≤ r < y.
Therefore, ∀x, y ∈ Z, if y > 0 then ∃q, r ∈ Z, x = yq + r and 0 ≤ r < y.
One consequence of the division algorithm is that it partitions the integers into evens and
odds. Since the remainder when an integer is divided by 2 must be strictly less than 2,
it will either be 0 or 1. When the remainder is 0, we say the integer is even. When the
remainder is 1, we say the integer is odd. A precise definition for even and odd can be
given as follows:
Definition 1.2.2. For x ∈ Z, x is even means
∃q ∈ Z, x = 2q.
x is odd means
∃q ∈ Z, x = 2q + 1.
Proposition 1.2.7.
∀x ∈ Z, x is odd if and only if x is not even.
Proof.
Let x ∈ Z.
Suppose x is odd and x is even.
Then ∃q ∈ Z, x = 2q, and ∃q ∈ Z, x = 2q + 1.
Choose a ∈ Z such that x = 2a, and choose b ∈ Z for which x = 2b + 1.
We then have 2a = 2b + 1; hence 2(a − b) = 1.
If a − b ≤ 0, then 2(a − b) ≤ 0, hence 1 ≤ 0 which is a contradiction.
If a − b > 0, then a − b ≥ 1 by theorem 1.2.3; hence 2(a − b) ≥ 2.
Then 1 ≥ 2, which can be reduced to the contradiction 0 ≥ 1.
Therefore, if x is even then x is not odd.
Conversely, assume x is not odd.
Applying the division algorithm, we have ∃q, r ∈ Z, x = 2q + r and 0 ≤ r < 2.
Choose such q and r.
Since r < 2, we have 0 < 2 − r; hence 1 ≤ 2 − r by theorem 1.2.3.
Therefore, r ≤ 1.
Since x is not odd, we have r , 1; hence r < 1.
Therefore, 0 < 1 − r; hence 1 ≤ 1 − r by theorem 1.2.3.
Therefore, r ≤ 0.
We now have 0 ≤ r and r ≤ 0. Hence r = 0.
Therefore, x = 2q + 0; hence x is even.
Therefore, if x is not odd then x is even.
We now have if x is even then x is not odd, and if x is not odd then x is even.
Therefore, x is even if and only if x is not odd.
Therefore, ∀x ∈ Z, x is even if and only if x is not odd.
A remark about the exclusive or

The proposition above has the form ‘P if and only if ¬Q’. Recall that this means ‘if P, then
¬Q, and if ¬Q, then P’. That is,
P ⇔ ¬Q ≡ (P ⇒ ¬Q) and (¬Q ⇒ P).
Note that the negation of the first implication ‘if P, then ¬Q’ is ‘P and Q’. Hence, the first
implication is equivalent to ‘¬(P and Q)’. That is,
P ⇒ ¬Q ≡ ¬(P and Q).
Further, using the fact that A ⇒ B ≡ ¬A or B, we see that the second implication ‘if ¬Q,
then P’ is equivalent to ‘¬(¬Q) or P’. That is,
¬Q ⇒ P ≡ Q or P.
Putting these equivalent forms of the first and second implications together gives us
P ⇔ ¬Q ≡ (P or Q) and ¬(P and Q).
That is, either P is true or Q is true but not both. This is sometimes written ‘P or else Q’
and is known as the exclusive or; it excludes the possibility that both are true. Hence, in
the above proposition, to say ‘x is odd if and only if x is not even’ is equivalent to saying
‘x is either even or odd but not both’. That is, x is even or else x is odd.
Definition 1.2.3. For x, y ∈ Z, x divides y means
∃a ∈ Z, y = xa.
Proposition 1.2.8.
∀x, y ∈ Z, if y , 0 and x divides y then |x| ≤ |y|.
Proof.
Let x, y ∈ Z.
Assume y , 0 and x divides y.
Then ∃a ∈ Z, y = xa. Choose such an a.
Since y , 0, we have a , 0; hence |a| , 0.
Since 0 ≤ |a| and |a| , 0, we have 0 < |a|.
Therefore, 1 ≤ |a|.
Therefore, |x| ≤ |x||a|; hence |x| ≤ |xa|.
Therefore, |x| ≤ |y|.
Therefore, if y , 0 and x divides y, then |x| ≤ |y|.
Therefore, ∀x, y ∈ Z, if y , 0 and x divides y then |x| ≤ |y|.
Definition 1.2.4. For x, y, g ∈ Z, with x , 0 and y , 0, g is the greatest

common divisor of x and y means
1. g > 0,
2. g divides x and g divides y, and

3. ∀a ∈ Z, if a divides x and a divides y, then a ≤ g.
We denote that g is the greatest common divisor of x and y by g =
gcd(x, y).
Definition 1.2.5. For x, y, f ∈ Z, with x , 0 and y , 0, f is the least

common multiple of x and y means
1. f > 0,
2. x divides f and y divides f , and
3. ∀a ∈ Z, if x divides a and y divides a and a , 0, then f ≤ |a|.
We denote that g is the least common multiple of x and y by f =
lcm (x, y).
Theorem 1.2.9.
∀x, y ∈ Z, if x , 0 and y , 0, then ∃g ∈ Z, g = gcd(x, y).
Proof.
Let x, y ∈ Z.
Assume x , 0 and y , 0.
Define S I Let S = {a ∈ Z | a divides both x and y}.
 Since x = (1)x and y = (1)y, we have that 1 divides both x and y.
Therefore, 1 ∈ S .



 Therefore, S , ∅.
 Let a ∈ S .

Demonstration  Then a divides x, and since x , 0, we have |a| ≤ |x|.

 Since a ≤ |a|, we then have a ≤ |x| by transitivity.

 Therefore, ∀a ∈ S , a ≤ |x|.
 Therefore, S is bounded above by |x|.

Therefore, S , ∅, and S is bounded above.
Apply WOP I By theorem 1.2.2, choose g to be the largest element of S .
Since g ∈ S , we have that g divides both x and y.



 Let a ∈ Z, and assume a divides both x and y.
 Then a ∈ S ; hence a ≤ g.
 Therefore, ∀a ∈ Z, if a divides both x and y then a ≤ g.
Finally, since 1 ∈ S , we have 1 ≤ g; hence g > 0.


Therefore, g = gcd(x, y).
Conclusion I Therefore, ∃g ∈ Z, g = gcd(x, y).
Therefore, if x , 0 and y , 0, then ∃g ∈ Z, g = gcd(x, y).
Therefore, ∀x, y ∈ Z, if x , 0 and y , 0, then ∃g ∈ Z, g = gcd(x, y).
There is a connection between the common divisors of x and y and the linear combinations
of x and y. By linear combinations of x and y, we mean the numbers that can be written in
the form xs + ty for some s and t. In fact, we will show that the greatest common divisor
is the smallest linear combination. Before we do this, however, we need the following
characterization of the common divisors of x and y.
Lemma 1.2.10.
∀a, x, y ∈ Z, a divides both x and y if and only if ∀s, t ∈ Z, a divides xs + yt.
Proof.
Let a, x, y ∈ Z.
Assume a divides both x and y.
Choose u, v ∈ Z with x = au and y = av.
Let s, t ∈ Z.
Put w = us + vt.
xs + yt = aus + avt = a(us + vt) = aw.
Therefore, ∃w ∈ Z, xs + yt = aw. Hence, a divides xs + yt.
Therefore, ∀s, t ∈ Z, a divides xs + yt.
Therefore, if a divides both x and y, then ∀s, t ∈ Z, a divides xs + yt.
Conversely, assume ∀s, t ∈ Z, a divides xs + yt.
Taking s = 1 and t = 0 gives us that a divides x.
Taking s = 0 and t = 1 gives us that a divides y.
Therefore, a divides both x and y.
Therefore, if ∀s, t ∈ Z, a divides xs + yt, then a divides both x and y.
Therefore, ∀a, x, y ∈ Z, a divides both x and y if and only if ∀s, t ∈ Z, a divides xs + yt.
With this view of common divisors in mind, we will show that the greatest common di-
visor is itself a linear combination of x and y. In fact, it is the smallest positive linear
combination.
Theorem 1.2.11 (Bezout’s Identity).
∀x, y ∈ Z, if x , 0 and y , 0, then ∃s, t ∈ Z, gcd(x, y) = sx + ty.
Proof.
Let x, y ∈ Z.
Let S = {a ∈ Z | a > 0 and ∃s, t ∈ Z, a = sx + ty}.
Since x , 0 and y , 0, we have x2 > 0 and y2 > 0.
Therefore, x2 + y2 > 0 and x2 + y2 = (x)x + (y)y, hence x2 + y2 ∈ S .
Therefore, S , ∅.
Further, for a ∈ S , we have 0 < a; hence S is bounded below by 0.
Therefore, S , ∅ and S is bounded below.
By the well-ordering property, choose g to be the smallest element of S .
Since g ∈ S , g > 0 and ∃s, t ∈ Z, g = sx + ty.
Choose such s and t.
We will show that g = gcd(x, y).
Applying the division algorithm, choose q, r ∈ Z, x = gq + r and 0 ≤ r < g.
Suppose r , 0.
Then r > 0.
Further, r = x − gq = x − (sx + ty)q = (1 − sq)x + (−tq)y.
Therefore r ∈ S .
We then have g ≤ r, which is a contradiction, since r < g.
Therefore, r = 0.
Therefore, x = gq; hence g divides x.
Similarly, g divides y.
Let a ∈ Z, and assume a divides both x and y.
Then, since g = sx + ty, a divides g by the previous lemma.
Therefore, |g| ≥ |a|.
Since g > 0, we have |g| = g; hence g ≥ |a|.
Since |a| ≥ a, we then have g ≥ a.
Therefore, ∀a ∈ Z, if a divides both x and y, then g ≥ a.
Therefore, g = gcd(x, y).
Therefore, gcd(x, y) = sx + ty.
Therefore, ∃s, t ∈ Z, gcd(x, y) = sx + ty.
Therefore, if x , 0 and y , 0, then ∃s, t ∈ Z, gcd(x, y) = sx + ty.
Therefore, ∀x, y ∈ Z, if x , 0 and y , 0, then ∃s, t ∈ Z, gcd(x, y) = sx + ty.
Corollary 1.2.12.
∀x, y, a ∈ Z, if x , 0 and y , 0 and a divides both x and y, then a divides gcd(x, y).
Proof.
Let x, y, a ∈ Z.
Assume x , 0 and y , 0 and a divides both x and y.
By Bezout’s identity, we may choose s, t ∈ Z with gcd(x, y) = sx + ty.
Hence, a divides gcd(x, y) by lemma 1.2.10.
Therefore, if x , 0 and y , 0 and a divides both x and y, then a divides gcd(x, y).
Therefore, ∀x, y, a ∈ Z, if x , 0 and y , 0 and a divides both x and y, then a divides
gcd(x, y).
√
In the next example, we will prove that 2 is irrational. Since this example deals with
the distinction between rational and irrational numbers, we must state a usable definition of
rational and irrational numbers. Roughly speaking, rational numbers are those real numbers
that can be expressed as a fraction of integers. Irrational numbers are those real numbers
that are not rational. However, to give a definition that is usable in a proof, we must make
sure our definition is precise (and preferably in symbolic form). We state the definition as
follows:
Definition 1.2.6. For x ∈ R, x is rational means
∃a, b ∈ Z, a = bx and b , 0.
x is irrational means x is not rational.
By insisting that the denominator b in the above definition is positive, we get a much more
convenient characterization of the rational numbers. That we can place such a demand on
the denominator is the subject of the next example:
Proposition 1.2.13.
∀x ∈ R, x is rational if and only if ∃a, b ∈ Z, a = bx and b > 0.
Proof.
Let x ∈ R.
Assume x is rational.
Then ∃a, b ∈ Z, a = bx and b , 0.
Accordingly, choose s, t ∈ Z with s = tx and t , 0.
Case 1: t > 0.
Since s = tx and t > 0, we have ∃a, b ∈ Z, a = bx and b > 0.
Case 2: t < 0.
Put a = −s and b = −t.
Since s = tx, we have −s = −tx; hence a = bx.
Also, since t < 0, we have −t > 0; hence b > 0.
Therefore, ∃a, b ∈ Z, a = bx and b > 0.
Therefore, if x is rational, then ∃a, b ∈ Z, a = bx and b > 0.
Conversely, assume ∃a, b ∈ Z, a = bx and b > 0.
Choose such a, b ∈ Z.
Since b > 0, we have b , 0.
Therefore, ∃a, b ∈ Z, a = bx and b , 0.
That is, x is rational.
Therefore, if ∃a, b ∈ Z, a = bx and b > 0, then x is rational.
Therefore, x is rational if and only if ∃a, b ∈ Z, a = bx and b > 0.
Therefore, ∀x ∈ R, x is rational if and only if ∃a, b ∈ Z, a = bx and b > 0.
√
Before proving that √2 is irrational, we give a property of the even integers that will be
used in the proof that 2 is irrational.
Lemma 1.2.14.
∀x ∈ Z, if x2 is even, then x is even.
Proof.
Let x ∈ Z.
Assume x is not even.
In this case, x is odd.
Therefore, ∃t ∈ Z, x = 2t + 1. Choose such a t.
Choose s = 2t2 + 2t
x2 = (2t + 1)2 = 4t2 + 4t + 1 = 2(2t2 + 2t) + 1 = 2s + 1.
Therefore, ∃s ∈ Z, x2 = 2s + 1.
Therefore, x2 is odd.
Therefore, x2 is not even.
Therefore, if x is not even, then x2 is not even.
Therefore, if x2 is even, then x is even.
Therefore, ∀x ∈ Z, if x2 is even, then x is even.
√
We are now ready to prove that 2 is irrational. To do this, we will return to the method
discussed earlier for using a proof by contradiction and the well-ordering property.
Proposition 1.2.15.
√
2 is irrational.
Proof. √
Assumption I Suppose 2 is rational. √
Then ∃x, y ∈ Z, y = x 2 and x > 0.
Hence, ∃x, y ∈ Z, y2 = 2x2 and x > 0.
Define S I Let S = {x ∈ Z | ∃y ∈ Z, y2 = 2x2 and x > 0}. By our assumption, S , ∅.
Demonstration [ For all x ∈ S , x > 0, hence S is bounded below.
Apply WOP I By the well-ordering property, choose a to be the smallest element of S .
Then ∃y ∈ Z, y2 = 2a2 and a > 0. Choose such a y.
Since y2 is even, we have that y is even by lemma 1.2.14.
Since y is even, choose z ∈ Z such that y = 2z.
We now have (2z)2 = 2a2 ; hence 4z2 = 2a2 .
This gives us 2z2 = a2 , which means a2 is even, hence a is even.
Set a value  I Choose b ∈ Z such that a = 2b.
 We then have 2z2 = (2b)2 ; hence z2 = 2b2 .
Therefore, ∃z ∈ Z, z2 = 2b2 .


Demonstration  Further, since a > 0 and a = 2b, we have b > 0.
Therefore, b ∈ S .


Further, since 1 < 2, b < 2b; hence b < a.
Contradiction I This is a√contradiction, since a is the smallest element of S .
Conclusion I Therefore, 2 is irrational.
The Principle of Mathematical Induction
We can define the set of natural numbers to be the set of positive integers:
Definition 1.2.7.
N = {x ∈ Z | x > 0}.
The definition of the natural numbers as positive integers may not be the most intuitive
definition we can give, but it allows us the advantage of being able to use the properties
of the integers that we have been able to prove so far. For a more natural definition, we
note that the natural numbers are completely determined by the idea of counting. That is,
if we start with 1, and for each n that has been counted we proceed to its successor n + 1,
then there is no natural number which cannot be reached by this process. The view of
the natural numbers as ‘numbers that can be produced by counting’ is much closer to our
intuitive notion of natural numbers than the definition given as ‘positive integers.’ It seems
very likely that humans come to understand the idea of natural numbers by counting, long
before we entertain any notions of negative numbers or the complete set of integers.
Fortunately, the approach we have taken, of defining the integers by axioms and defining
the natural numbers as a subset of these integers, is consistent with the intuitive notion of
natural numbers as the elements of the counting process. That this is the case, is known as
the principle of mathematical induction.
The Principle of Mathematical Induction

Let A be a set. If 1 ∈ A and ∀n ∈ N, if n ∈ A then n + 1 ∈ A, then N ⊆ A.
The proof of the principle of mathematical induction makes use of the well-ordering prop-
erty, using a proof by contradiction:
Theorem 1.2.16 (The Principle of Mathematical Induction).
Let A be a set. If 1 ∈ A and ∀n ∈ N, if n ∈ A then n + 1 ∈ A, then N ⊆ A.
Proof.
Let A be a set.
Assume 1 ∈ A and ∀n ∈ N, if n ∈ A then n + 1 ∈ A.
Suppose N * A; hence ∃x ∈ N, x < A.
Let S = {x ∈ N | x < A}. By our assumption, S , ∅.
Since S ⊆ N, we have that S is bounded below by 0.
By the well-ordering property, choose a to be the smallest element of S .
Since a ∈ N, we have a > 0; hence a ≥ 1.
Further, since 1 ∈ A, we have 1 < S . Therefore, a , 1.
Therefore, a > 1.
Choose n = a − 1.
We then have n > 0; hence n ∈ N.
Since n + 1 = a and a < A, we have n + 1 < A.
Therefore, n < A.
We now have n ∈ N and n < A, hence n ∈ S .
Further, since a − 1 < a, we have n < a.
Therefore, ∃n ∈ S , n < a
This is a contradiction, since a is the smallest element of S .
Therefore, N ⊆ A.
Therefore, if 1 ∈ A and ∀n ∈ N, if n ∈ A then n + 1 ∈ A, then N ⊆ A.
Although this gives a nice characterization of the natural numbers in terms of counting, the
starting point of 1 is somewhat arbitrary. We could reasonably begin the counting process
at any integer, and the set produced would contain all integers after our chosen starting
point. That is, the set would contain all elements of the set {x ∈ Z | x ≥ a}. To simplify
notation, we use the following:
Notation
For a ∈ Z, denote by Z≥a , the set
Z≥a = {x ∈ Z | x ≥ a}.
Similarly, denote by Z≤a , the set
Z≤a = {x ∈ Z | x ≤ a}.
If we allow the starting point to be flexible, then we obtain the following more general
version of the principle of mathematical induction:
The Principle of Mathematical Induction Starting at a ∈ Z

Let a ∈ Z, and let A be a set. If a ∈ A and ∀n ∈ Z≥a , if n ∈ A then n + 1 ∈ A, then Z≥a ⊆ A.
The proof of this version of the principle of mathematical induction is similar to that of the
version beginning from 1:
Theorem 1.2.17.
Let a ∈ Z, and let A be a set. If a ∈ A and ∀n ∈ {x ∈ Z | x ≥ a}, if n ∈ A then n + 1 ∈ A,

then {x ∈ Z | x ≥ a} ⊆ A.
Proof.
Let a ∈ Z, and let A be a set.
Assume a ∈ A and ∀n ∈ {x ∈ Z | x ≥ a}, if n ∈ A then n + 1 ∈ A.
Suppose {x ∈ Z | x ≥ a} * A; hence ∃x ∈ Z, x ≥ a and x < A.
Let S = {x ∈ Z | x ≥ a and x < A}. By our assumption, S , ∅.
For x ∈ S , we have x ≥ a; hence S is bounded below by a.
By the well-ordering property, choose m to be the smallest element of S .
We then have m ≥ a.
Further, since a ∈ A, we have a < S . Therefore, m , a.
Therefore, m > a; hence m ≥ a + 1.
Choose n = m − 1.
We then have n ≥ a; hence n ∈ {x ∈ Z | x ≥ a}.
Since n + 1 = m and m < A, we have n + 1 < A.
Therefore, n < A.
We now have n ≥ a and n < A, hence n ∈ S .
Therefore, ∃n ∈ S , n < m
This is a contradiction, since m is the smallest element of S .
Therefore, {x ∈ Z | x ≥ a} ⊆ A.
Therefore, if a ∈ A and ∀n ∈ Z, if n ∈ A then n + 1 ∈ A, then {x ∈ Z | x ≥ a} ⊆ A.
Notice that these two versions of the principle of mathematical induction are identical,
only with 1 replaced by a and N replaced by Z≥a . In fact, since N = Z≥1 , we see that
the original principle of mathematical induction is the special case of this more general
version when we take a = 1.
To generalize the principle of mathematical induction further, we will consider how

one might obtain the set of negative integers in a way that is analogous to the construction
of the positive integers by induction. For ease in describing the set of negative integers, we
introduce the notation:
Notation
Denote by −N, the set
−N = {x ∈ Z | x < 0}.
The negative integers can also be arrived at by counting, only rather than proceeding from
a number n to its successor n + 1, we must count backwards, proceeding from a number n
to its predecessor n − 1. In this reversed version of the principle of mathematical induction,
we will start the counting process at the largest negative number −1. We leave it as an
exercise to verify that if this backwards counting process is started at any integer a, it will
completely determine the set Z≤a = {x ∈ Z | x ≤ a}.
Theorem 1.2.18.
Let A be a set. If −1 ∈ A and ∀n ∈ −N, if n ∈ A then n − 1 ∈ A, then −N ⊆ A.
Proof.
Let A be a set.
Assume −1 ∈ A and ∀n ∈ −N, if n ∈ A then n − 1 ∈ A.
Suppose −N * A; hence ∃x ∈ −N, x < A.
Let S = {x ∈ −N | x < A}. By our assumption, S , ∅.
Since S ⊆ −N, we have that S is bounded above by 0.
By theorem 1.2.2, choose a to be the largest element of S .
Since a ∈ −N, we have a < 0; hence a ≤ −1.
Further, since −1 ∈ A, we have −1 < S . Therefore, a , −1.
Therefore, a < −1.
Choose n = a + 1.
We then have n < 0; hence n ∈ −N.
Since n − 1 = a and a < A, we have n − 1 < A.
Therefore, n < A.
We now have n ∈ −N and n < A, hence n ∈ S .
Further, since a + 1 > a, we have n > a.
Therefore, ∃n ∈ S , n > a
This is a contradiction, since a is the largest element of S .
Therefore, −N ⊆ A.
Therefore, if −1 ∈ A and ∀n ∈ −N, if n ∈ A then n − 1 ∈ A, then −N ⊆ A.
Proofs Using the Principle of Mathematical Induction
The principle of mathematical induction provides us with a proof technique that can be very
effective for proving statements of the form ∀x ∈ N, P(x), where P(x) is an open sentence.
The reasoning behind the technique is as follows: If we define the truth set of the open
sentence P(x) as the set A = {x ∈ N | P(x)}, then to prove N ⊆ A is to prove ∀x ∈ N, P(x).
Further, the principle of mathematical induction gives us a way to prove N ⊆ A. That is,
if we can show that 1 ∈ A and that ∀n ∈ N, if n ∈ A then n + 1 ∈ A, then the principle of
mathematical induction ensures us that our set A contains all of the natural numbers. Hence
the truth set of the open sentence P(x) contains all natural numbers, meaning P(x) is true
for all natural numbers x. The structure of such a proof will then be the following:
Proof of ∀x ∈ N, P(x).
Proof.
Define A I Let A = {x ∈ N | P(x)}.
..

.


Demonstrate that 1 ∈ A.

..
.


Therefore 1 ∈ A.
Arbitrary element I Let n ∈ N.
Assumption I Assume n ∈ A.
 ..
 .
Demonstrate that n + 1 ∈ A.

..
.

Conclusion I Therefore, n + 1 ∈ A.
Conclusion I Therefore, if n ∈ A then n + 1 ∈ A.
Therefore, ∀n ∈ N, if n ∈ A then n + 1 ∈ A.
We now have 1 ∈ A and ∀n ∈ N, if n ∈ A then n + 1 ∈ A.
Apply the PMI I Therefore, by the principle of mathematical induction, N ⊆ A.
Therefore, ∀x ∈ N, x ∈ A.
Conclusion I Therefore, ∀x ∈ N, P(x).
Note that to apply the principle of mathematical induction, we must demonstrate that the
two conditions given in the principle are satisfied. The first demonstration must verify the
statement ‘1 ∈ A.’ This is sometimes easy and sometimes difficult, depending on the set A.
The second condition that must be verified in order to apply the principle of mathe-
matical induction is ‘∀n ∈ N, if n ∈ A then n + 1 ∈ A.’ Since this is a statement about all
natural numbers, its proof involves the introduction of an arbitrary variable n. Hence the
line ‘Let n ∈ N’ in the above proof template. Further, the statement that must be demon-
strated about this arbitrary natural number n is ‘if n ∈ A then n+1 ∈ A.’ Since this statement
is an implication of the form ‘if P then Q,’ its proof will use one of the three techniques for
proving implications; those being direct proof, contraposition, and proof by contradiction.
The above template shows a direct proof being used to show ‘if n ∈ A then n + 1 ∈ A.’ That
is, we assume ‘n ∈ A,’ and demonstrate ‘n + 1 ∈ A.’ If we were to use a proof by contrapo-
sition at this stage, then our assumption would be ‘n + 1 < A’ and our conclusion would be
‘n < A.’ Similarly, if we decide to employ a proof by contradiction, our assumption would
be ‘n ∈ A and n+1 < A,’ and from this assumption we would need to derive a contradiction.
If we can successfully demonstrate that both of these conditions hold, then the prin-
ciple of mathematical induction ensures that A = N, which allows us to complete our
proof. The following examples are intended to illustrate how one uses this structure to
prove statements about the natural numbers.
Proposition 1.2.19.
∀x ∈ N, 7 divides 32x−1 + 2 x+1 .
Proof.
Define A I Let A = {x ∈ N | 7 divides 32x−1 + 2 x+1 }.
Choose a = 1.



 32(1)−1 + 21+1 = 7 = 7a.
Demonstration   Therefore, ∃a ∈ Z, 32(1)−1 + 21+1 = 7a.
Therefore, 7 divides 32(1)−1 + 21+1 .


Hence, 1 ∈ A.
Then, 7 divides 32n−1 + 2n+1 .


 That is, ∃t ∈ Z, 32n−1 + 2n+1 = 7t. Choose such a t.
Choose s = 32n−1 + 2t.



32(n+1)−1 + 2(n+1)+1 = 9(32n−1 ) + 2(2n+1 )



 = 7(32n−1 ) + 2(32n−1 ) + 2(2n+1 )

 = 7(32n−1 ) + 2(32n−1 + 2n+1 )
 = 7(32n−1 ) + 2(7t)


 = 7(32n−1 + 2t)

 = 7s



 Therefore, ∃s ∈ Z, 32(n+1)−1 + 2(n+1)+1 = 7s.
Hence, 7 divides 32(n+1)−1 + 2(n+1)+1 .

Conclusion I Therefore, ∀x ∈ N, 7 divides 32x−1 + 2 x+1 .
Proposition 1.2.20.
∀x ∈ N, 2 x > x.
Proof.
Define A I Let A = {x ∈ N | 2 x > x}.
Since 21 = 2 and 2 > 1, we have 21 > 1.
"
Demonstration
Therefore, 1 ∈ A.
Hence, 2n > n.


 Therefore, 2(2n ) > 2n; hence 2n+1 > 2n.
Since n ≥ 1, we have n + n ≥ n + 1. Therefore, 2n ≥ n + 1.


 We now have 2n+1 > 2n and 2n ≥ n + 1.
Hence, by transitivity, 2n+1 > n + 1.
Conclusion I Therefore, ∀x ∈ N, 2 x > x.
In the next example, we will prove the statement
∀x ∈ Z, if ∃a ∈ Z, x = 3a + 1, then ∀n ∈ N, ∃b ∈ Z, xn = 3b + 1.
Since this is a statement about all integers x, we will introduce an arbitrary variable x ∈ Z.
For this x, we will need to demonstrate the implication
if ∃a ∈ Z, x = 3a + 1, then ∀n ∈ N, ∃b ∈ Z, xn = 3b + 1
| {z } | {z } .
P Q
Using a direct proof, we will assume ∃a ∈ Z, x = 3a + 1. We are then required to

demonstrate
∀n ∈ N, ∃b ∈ Z, xn = 3b + 1.
This is a statement about all natural numbers, and we will prove it using induction. We then
see that our technique of using the principle of mathematical induction need not be em-
ployed from the beginning of the proof. We may decide to use other more direct techniques
until we reach a point in our demonstration where a proof by induction is appropriate.
Proposition 1.2.21.
∀x ∈ Z, if ∃a ∈ Z, x = 3a + 1, then ∀n ∈ N, ∃b ∈ Z, xn = 3b + 1.
Proof.
Let x ∈ Z.
Assume ∃a ∈ Z, x = 3a + 1. Choose such an a.
Define A I Let A = {n ∈ N | ∃b ∈ Z, xn = 3b + 1}.
Demonstration [ Since x1 = x = 3a + 1, we have 1 ∈ A.

 Then ∃s ∈ Z, xn = 3s + 1. Choose such an s.

 Choose t = 3as + a + s.


 xn+1 = x(xn )

 = (3a + 1)(3s + 1)
= 9as + 3a + 3s + 1

= 3(3as + a + s) + 1


= 3t + 1





Therefore, ∃t ∈ Z, xn+1 = 3t + 1.

Conclusion I Therefore, ∀n ∈ N, ∃b ∈ Z, xn = 3b + 1.
Therefore, if ∃a ∈ Z, x = 3a + 1, then ∀n ∈ N, ∃b ∈ Z, xn = 3b + 1.
Therefore, ∀x ∈ Z, if ∃a ∈ Z, x = 3a + 1, then ∀n ∈ N, ∃b ∈ Z, xn = 3b + 1.
The next example shows that every finite set of real numbers has a maximum element. That
is, if S ⊆ R is a set with a finite number of elements, then ∃a ∈ S , ∀x ∈ S , x ≤ a. We prove
this using induction on the number of elements in the set.
Proposition 1.2.22.
∀m ∈ N, if S ⊆ R is a set with m elements, then ∃a ∈ S , ∀x ∈ S , x ≤ a.
Proof.
Define A I Let A = {m ∈ N | if S ⊆ R is a set with m elements, then ∃a ∈ S , ∀x ∈ S , x ≤ a}.

 Assume S ⊆ R is a set with 1 element.

 Then S = {b} for some b ∈ R.
 Therefore, ∀x ∈ S , x ≤ b.
 Therefore, ∃a ∈ S , ∀x ∈ S , x ≤ a.

 Therefore, if S ⊆ R is a set with 1 element, then ∃a ∈ S , ∀x ∈ S , x ≤ a.
Therefore, 1 ∈ A.
Assume S ⊆ R is a set with n + 1 elements.



 Since S is not empty, we may choose an element b ∈ S .
 The set S \ {b} has n elements.

 Since n ∈ A, we have ∃a ∈ S \ {b}, ∀x ∈ S \ {b}, x ≤ a.
Choose such an a.



 Choose c = max(a, b).
 Let x ∈ S .
We consider two cases: x = b and x , b.


Case 1: x = b.


Demonstration  In this case, x = b ≤ max(a, b) = c,
x ≤ c.

 hence

 Case 2: x , b.
 Then, x ∈ S \ {b}.
Hence, x ≤ a ≤ max(a, b) = c.


Therefore, x ≤ c.



 In both cases, x ≤ c.

 Therefore, ∀x ∈ S , x ≤ c.
 Therefore, ∃c ∈ S , ∀x ∈ S , x ≤ c.
So, if S ⊆ R is a set with n + 1 elements, then ∃c ∈ S , ∀x ∈ S , x ≤ c.
Conclusion I Thus, ∀m ∈ N, if S ⊆ R is a set with m elements, then ∃a ∈ S , ∀x ∈ S , x ≤ a.
Inductive Subsets of R
Earlier, we considered a generalization of the principle of mathematical induction in which

the starting point is allowed to be any integer, as opposed to the common starting point of
1 used to construct the natural numbers. To extend this idea further, we may also consider
starting points outside of the integers. The common concept that is featured in all of our
generalizations is that the set in question is closed under the counting process. To simplify
the language in our discussion, it will be worth while giving a name to this feature of a set.
Definition 1.2.8. For a subset S ⊆ R, with S , ∅, to say S is inductive

means ∀x ∈ R, if x ∈ S then x + 1 ∈ S .
There are many inductive subsets of the real numbers. For example, the interval (0, ∞) is
inductive. To prove that this set is inductive, we simply show that the defining condition
‘∀x ∈ R, if x ∈ S then x + 1 ∈ S ’ is satisfied when the set S is replaced by the set in
question. Here is the proof:
Proposition 1.2.23.
(0, ∞) is inductive.
Proof.
Let x ∈ R.
Assume x ∈ (0, ∞).
That is, x > 0.
Then, since x > 0, we have x + 1 > 1.
Since x + 1 > 1 and 1 > 0, we have x + 1 > 0 by transitivity.
Therefore, x + 1 ∈ (0, ∞).
Therefore, if x ∈ (0, ∞) then x + 1 ∈ (0, ∞).
Therefore, (0, ∞) is inductive.
Earlier, we showed that the principle of mathematical induction for the natural numbers can
be proven using the well-ordering property. It should be noted again that the well-ordering
property is not satisfied by all bounded subsets of the real numbers. For example, the set
in the previous example, (0, ∞) satisfies the condition that ∀x ∈ (0, ∞), x ≥ 0. However,
the set (0, ∞) does not have a smallest element. However, we will show that for any given
a ∈ R, the smallest inductive subset of the real numbers containing a will always satisfy
the well-ordering property. By the ‘smallest inductive subset containing a,’ we mean that
the set does not contain an proper subsets that are inductive and contain a. More formally:
Definition 1.2.9. Let a ∈ R. The smallest inductive set containing a

is an inductive set Ma containing a, satisfying the property that if A ⊆ R
for which a ∈ A and A is inductive, then Ma ⊆ A.
Letting a ∈ R, and letting Ma be the smallest inductive set containing a, we obtain a proof
technique similar to a proof by induction, to prove statements of the form ∀x ∈ Ma , P(x),
for some predicate P. By defining the set A = {x ∈ Ma | P(x)}, if we can show that a ∈ A
and A is inductive, then the fact that Ma is the smallest inductive set containing a will give
us that Ma ⊆ A. The statement ∀x ∈ Ma , x ∈ A would then be true, which due to the
definition of A means ∀x ∈ Ma , P(x) is true. The proof structure is very similar to that
obtained from the original principle of mathematical induction:
Proof of ∀x ∈ Ma , P(x).
Proof.
Define A I Let A = {x ∈ Ma | P(x)}.
..

.


Demonstrate that a ∈ A.

..
.


Therefore a ∈ A.
Assumption I Assume x ∈ A.
 ..
 .
Demonstrate that x + 1 ∈ A.

..
.

Conclusion I Therefore, x + 1 ∈ A.
Conclusion I Therefore, if x ∈ A then x + 1 ∈ A.
Therefore, A is inductive.
We now have a ∈ A and A is inductive.
Apply Defn of Ma I Therefore, Ma ⊆ A.
Therefore, ∀x ∈ Ma , x ∈ A.
Conclusion I Therefore, ∀x ∈ Ma , P(x).
The following examples make use of the above proof technique.
Proposition 1.2.24.
Let a ∈ R, and let Ma be the smallest inductive set containing a. Then, ∀x ∈ Ma , x ≥ a.
Proof.
Let A = {x ∈ R | x ∈ Ma and x ≥ a}.
We claim that A is an inductive set containing a.
Indeed, since a ∈ Ma and a ≥ a, we have a ∈ A.
Let x ∈ R.
Assume x ∈ A.
Then since x ∈ Ma and Ma is inductive, we have x + 1 ∈ Ma .
Further, since x ≥ a, we have x + 1 ≥ a + 1.
Since x + 1 ≥ a + 1 and a + 1 ≥ a, we have x + 1 ≥ a by transitivity.
We have thus shown that x + 1 ∈ Ma and x + 1 ≥ a. Hence x + 1 ∈ A.
Therefore, if x ∈ A then x + 1 ∈ A.
Since A ⊆ Ma for which a ∈ A and A is inductive, we have Ma ⊆ A.
Let x ∈ Ma .
Then x ∈ A since Ma ⊆ A.
Hence x ≥ a.
Therefore, ∀x ∈ Ma , x ≥ a.
Proposition 1.2.25.
Let a ∈ R, and let Ma be the smallest inductive set containing a. Then, ∀x ∈ Ma , if x > a
then x − 1 ∈ Ma .
Proof.
Let A = {x ∈ Ma | if x > a then x − 1 ∈ Ma }.
Since a > a is false, the statement if a > a then a − 1 ∈ Ma is true.
Let x ∈ R.
Assume x ∈ A.
Assume x + 1 > a.
Since x ∈ A, we have x ∈ Ma ; hence (x + 1) − 1 ∈ Ma .
Therefore, if x + 1 > a then (x + 1) − 1 ∈ Ma .
Therefore, x + 1 ∈ A.
Therefore, if x ∈ A then x + 1 ∈ A.
Hence, A is inductive.
It then follows from the definition of Ma , that Ma ⊆ A.
Therefore, ∀x ∈ Ma , if x > a then x − 1 ∈ Ma .
Notice the similarity between the next example and theorem 1.2.3.
Proposition 1.2.26.
Let a ∈ R, and let Ma be the smallest inductive set containing a. Then, ∀x, y ∈ Ma , if y < x
then y + 1 ≤ x.
Proof.
Let a ∈ R, and let Ma be the smallest inductive set containing a.
Let A = {x ∈ Ma | ∀y ∈ Ma , if y < x then y + 1 ≤ x}.
Let y ∈ Ma .
Suppose y < a and a < y + 1.
By proposition 1.2.24, since y ∈ Ma , we have a ≤ y.
Hence, we have the contradiction y < a and a ≤ y.
Therefore, if y < a then y + 1 ≤ a.
Therefore, a ∈ A.
Let x ∈ R, and assume x ∈ A.
Let y ∈ Ma .
Assume y < x + 1.
In the case where y = a, we have by proposition 1.2.24 that
y ≤ x; hence y + 1 ≤ x + 1.
In the case where y , a, we have that a < y; hence
by the proposition 1.2.25, y − 1 ∈ Ma .
Since y − 1 < x and x ∈ A, we then have (y − 1) + 1 ≤ x.
Therefore, y ≤ x; hence y + 1 ≤ x + 1.
Therefore, if y < x + 1 then y + 1 ≤ x + 1.
Hence x + 1 ∈ A.
Therefore, Ma ⊆ A.
Therefore, ∀x, y ∈ Ma , if y < x then y + 1 ≤ x.
The following theorem shows that for any a ∈ R, the smallest inductive set containing a
satisfies the well-ordering property.
Theorem 1.2.27.
Let a ∈ R, and let Ma be the smallest inductive set containing a. Then, if S ⊆ Ma with
S , ∅, then S has a smallest element. That is, ∃c ∈ S , ∀x ∈ S , c ≤ x.
Proof.
Let a ∈ R, and let Ma be the smallest inductive set containing a.
Assume S ⊆ Ma , with S , ∅.
Further, assume that S does not have a smallest element.
Let A = {c ∈ Ma | ∀x ∈ S , c ≤ x}.
We claim that a ∈ A.
Indeed, let x ∈ S . Then x ∈ Ma , since S ⊆ Ma .
Therefore, a ≤ x.
Therefore, ∀x ∈ S , a ≤ x.
Therefore, a ∈ A.
Next, we claim that A is inductive.
Indeed, let c ∈ A.
Then c ∈ Ma ; hence c + 1 ∈ Ma .
Let x ∈ S .
Then c ≤ x.
Since S does not have a smallest element, we have c < S .
Hence, c , x.
Therefore, c < x; hence c + 1 ≤ x.
Therefore, ∀x ∈ S , c + 1 ≤ x.
Hence, c + 1 ∈ A.
Since a ∈ A and A is inductive, we have Ma ⊆ A.
Since S , ∅, we have ∃c ∈ Ma , c ∈ S .
Since c ∈ Ma and Ma ⊆ A, we have c ∈ A.
Therefore, ∀x ∈ S , c ≤ x.
This is a contradiction, since c ∈ S , and S does not have a smallest element.
Therefore, if S ⊆ Ma with S , ∅, then S has a smallest element.
In fact, the inductive subsets of the real numbers provide an alternate construction of the
integers. One can define the natural numbers to be M1 , the smallest inductive subset of R
containing 1, and the integers to be {x ∈ R | ∃m, n ∈ M1 , x = m − n}. Indeed, all of the
defining axioms of the integers can be proven using this definition. The last axiom, the
well-ordering property, was done above.
The Principle of Complete Induction
An objection that one might have to the principle of mathematical induction is that the one
doing the counting has a very short memory. In only using the fact that n ∈ A to prove
n + 1 ∈ A, one forgets that all of the numbers from 1 through n have already been counted
among the elements of A. In fact, in applying the counting process in the principle of
mathematical induction, one does not know that n ∈ A until having already established that
{1, . . . , n} ⊆ A. This information may be useful if we wish to prove that n + 1 ∈ A, and it
does not need to be forgotten.
Notation
{1, . . . , n} means {k ∈ N | k ≤ n}.
We suggest a version of the principle of mathematical induction with a better memory: To

prove that n + 1 ∈ A, we do not merely assume n ∈ A but rather that {1, . . . , n} ⊆ A. The
statement of the principle of complete induction is as follows:
The Principle of Complete Induction

Let A be a set. If 1 ∈ A and ∀n ∈ N, if {1, . . . , n} ⊆ A then n + 1 ∈ A, then N ⊆ A.
Using the well-ordering property, the proof of the principle of complete induction is very
similar to that of the principle of mathematical induction.
Theorem 1.2.28 (The Principle of Complete Induction).
Let A be a set. If 1 ∈ A and ∀n ∈ N, if {1, . . . , n} ⊆ A then n + 1 ∈ A, then N ⊆ A.
Proof.
Let A be a set.
Assume 1 ∈ A and ∀n ∈ N, if {1, . . . , n} ⊆ A then n + 1 ∈ A.
Suppose N * A; hence ∃x ∈ N, x < A.
Let S = {x ∈ N | x < A}. By our assumption, S , ∅.
Since S ⊆ N, we have that S is bounded below by 0.
By the well-ordering property, ∃a ∈ S , ∀x ∈ S , a ≤ x.
Choose such an a.
Since a ∈ N, we have a > 0; hence a ≥ 1.
Further, since 1 ∈ A, we have 1 < S . Therefore, a , 1.
Therefore, a > 1.
Choose n = a − 1.
We then have n > 0; hence n ∈ N.
Since n + 1 = a and a < A, we have n + 1 < A.
Therefore, {1, . . . , n} * A.
Therefore, ∃x ∈ N, x ∈ {1, . . . , n} and x < A.
For such an x, we have x ∈ S and x ≤ n.
Since n = a − 1, we have x < a.
Therefore, ∃x ∈ S , x < a
This is a contradiction, since a is the smallest element of S .
Therefore, N ⊆ A.
Therefore, if 1 ∈ A and ∀n ∈ N, if {1, . . . , n} ∈ A then n + 1 ∈ A, then N ⊆ A.
Proofs Using the Principle of Complete Induction
Proofs using the principle of complete induction are very similar to those using the principle
of mathematical induction. In both cases we define the set A as the set of all natural numbers
for which the statement we are proving is true. In both cases we must demonstrate that
1 ∈ A. The difference is in the inductive step. When using the principle of mathematical
induction, we must prove the statement
∀n ∈ N, if n ∈ A, then n + 1 ∈ A.
When using the principle of complete induction, we prove the statement
∀n ∈ N, if {1, . . . , n} ⊆ A, then n + 1 ∈ A.
The general form of a proof using the principle of complete induction is as follows:
Proof of ∀x ∈ N, P(x).
Proof.
Define A I Let A = {x ∈ N | P(x)}.
..

.


Demonstrate that 1 ∈ A.

Demonstration  ..
.


Therefore 1 ∈ A.
Assumption I Assume {1, . . . , n} ⊆ A.
 ..
.


Demonstration  Demonstrate that n + 1 ∈ A.
 ..
.
Conclusion I Therefore, if {1, . . . , n} ⊆ A then n + 1 ∈ A.
Therefore, ∀n ∈ N, if {1, . . . , n} ⊆ A then n + 1 ∈ A.
Now, 1 ∈ A and ∀n ∈ N, if {1, . . . , n} ⊆ A then n + 1 ∈ A.
Apply PCI I By the principle of complete induction, N ⊆ A.
Conclusion I Therefore, ∀x ∈ N, P(x).
We demonstrate this technique by proving the following two properties of the natural num-
bers:
Proposition 1.2.29.
∀x ∈ N, ∃m, y ∈ N, x = 2m−1 (2y − 1).
Proof.
Let A = {x ∈ N | ∃m, y ∈ N, x = 2m−1 (2y − 1)}.
Choose m = 1 and y = 1.
1 = 20 (2(1) − 1) = 2m−1 (2y − 1).
Therefore, ∃m, y ∈ N, 1 = 2m−1 (2y − 1).
Hence, 1 ∈ A.
Let n ∈ N.
Assume {1, . . . , n} ⊆ A.
n + 1 is either even or odd, and we consider these possibilities in cases.
Case 1: n + 1 is even.
The ∃x ∈ N, n + 1 = 2x.
Since 1 < 2, we have x < 2x; hence x < n + 1.
Therefore, x ≤ n; hence x ∈ {1, . . . , n}.
We then have that x ∈ A.
Therefore, ∃k, y ∈ N, x = 2k−1 (2y − 1). Choose such k, y.
Choose m = k + 1.
Then n + 1 = 2(2k−1 )(2y − 1) = 2m−1 (2y − 1).
Therefore, ∃m, y ∈ N, n + 1 = 2m−1 (2y − 1).
Therefore, n + 1 ∈ A.
Case 2: n + 1 is odd.
Hence ∃z ∈ N, n + 1 = 2z + 1.
Choose m = 1 and y = z + 1.
n + 1 = 2z + 1 = 2(y − 1) + 1 = 2y − 1 = 20 (2y − 1) = 2m−1 (2y − 1).
Therefore, ∃m, y ∈ N, n + 1 = 2m−1 (2y − 1).
In both cases, n + 1 ∈ A.
Therefore, if {1, . . . , n} ⊆ A, then n + 1 ∈ A.
Therefore, ∀n ∈ N, if {1, . . . , n} ⊆ A, then n + 1 ∈ A.
We now have 1 ∈ A and ∀n ∈ N, if {1, . . . , n} ⊆ A, then n + 1 ∈ A.
By the principle of complete induction, N ⊆ A.
Therefore, ∀x ∈ N, ∃m, y ∈ N, x = 2m−1 (2y − 1).
In our next example, we prove that every natural number other than 1 has a prime factor.
However, since we have not yet seen any examples dealing with prime numbers, we will
need define what it means for a natural number to be prime. The definition of a prime
number that we will use is as follows:
Definition 1.2.10. Let x ∈ N. x is prime means x , 1 and ∀a, b ∈ N, if

x = ab then either a = 1 or b = 1.
Proposition 1.2.30.
∀x ∈ N, either x = 1 or ∃p ∈ N, p is prime and p divides x.
Proof.
Let A = {x ∈ N | either x = 1 or ∃p ∈ N, p is prime and p divides x}.
1 ∈ A is given explicitly in the definition of A.
Let n ∈ N.
Assume {1, . . . , n} ⊆ A.
Either n + 1 is prime or it is not.
Case 1: n + 1 is prime.
Choosing p = n + 1 gives us that p is prime and p divides n + 1.
Therefore, ∃p ∈ N, p is prime and p divides n + 1.
Case 2: n + 1 is not prime.
Therefore, ∃a, b ∈ N, n + 1 = ab and neither a nor b is equal to 1.
Since a , 1, we have a > 1, hence ab > b.
Therefore, b < n + 1, which then gives us b ≤ n.
Therefore, b ∈ {1, . . . , n}; hence b ∈ A.
Since b , 1, this means ∃p ∈ N, p is prime and p divides b.
Choosing such a p, since p divides b, we have ∃t ∈ N, b = tp.
Therefore, n + 1 = (at)p; hence p divides n + 1.
Therefore, ∃p ∈ N, p is prime and p divides n + 1.
In both cases, we have n + 1 ∈ A.
Therefore, if {1, . . . , n} ⊆ A, then n + 1 ∈ A.
Therefore, ∀n ∈ N, if {1, . . . , n} ⊆ A, then n + 1 ∈ A.
We now have 1 ∈ A and ∀n ∈ N, if {1, . . . , n} ⊆ A, then n + 1 ∈ A.
Therefore, ∀x ∈ N, either x = 1 or ∃p ∈ N, p is prime and p divides x.
Recursive Definitions
There are many mathematical concepts that one comes to understand only by an inductive
process. For example, when we first come to understand the concept of exponents, we
identify x1 as x, x2 as (x)x, x3 as ((x)x)x = (x2 )x, and so on. The idea is known through
a repetition of the operation of multiplication, where the exponent is a count of how many
times the multiplication has been repeated. In a similar way, we can view multiplication (in
the natural numbers) as repeated addition and division with remainder as repeated subtrac-
tion. The defining of concepts by a process of repetition (or recursion) can be made precise
in a way that makes these definitions useable in formal proofs. We will explore several
such definitions and show how they can be used in a proof.
Definition 1.2.11. For x ∈ R, define x1 = x and for each n ∈ N, define

xn+1 = (xn )x.
Note that we need not begin with the exponent 1. A compatible definition can be given in
the case where x , 0, by defining x0 = 1 and for each n ∈ Z≥0 , xn+1 = (xn )x. This is indeed
compatible with the definition above since x1 = (x0 )x = (1)x = x. Using this recursive
definition, we will prove some well known properties of exponents.
Proposition 1.2.31.
∀x, y ∈ R, ∀n ∈ N, (xy)n = xn yn .
Proof.
Let x, y ∈ R.
Let A = {n ∈ N | (xy)n = xn yn }.
Since (xy)1 = xy = x1 y1 , we have 1 ∈ A.
Let n ∈ N.
Assume n ∈ A.
Then (xy)n = xn yn .
Now, (xy)n+1 = (xy)n xy = xn yn xy = (xn )x(yn )y = xn+1 yn+1 .
Hence, n + 1 ∈ A.
Therefore, if n ∈ A then n + 1 ∈ A.
By the PMI, N ⊆ A.
Therefore, ∀n ∈ N, (xy)n = xn yn .
Therefore, ∀x, y ∈ R, ∀n ∈ N, (xy)n = xn yn .
Proposition 1.2.32.
∀x ∈ R, ∀m, n ∈ N, xm+n = xm xn .
Proof.
Let x ∈ R.
Let A = {n ∈ N | ∀m ∈ N, xm+n = xm xn }.
Let m ∈ N.
Then xm+1 = (xm )x = xm x1 .
Therefore, ∀m ∈ N, xm+1 = xm x1 .
Therefore, 1 ∈ A.
Let n ∈ N.
Assume n ∈ A.
Let m ∈ N.
Then xm+n = xm xn .
xm+(n+1) = x(m+n)+1 = (xm+n )x = xm xn x = xm xn+1 .
Therefore, ∀m ∈ N, xm+(n+1) = xm xn+1 .
Hence, n + 1 ∈ A.
Therefore, ∀n ∈ N, ∀m ∈ N, xm+n = xm xn .
Therefore, ∀x ∈ R, ∀m, n ∈ N, xm+n = xm xn .
When giving a recursive definition, we are essentially describing a pattern that repeats in-
definitely. We can make this idea yet more precise by introducing the notion of a sequence.
Consider, for example, the pattern 1, 4, 9, 16, 25, 36, 49, . . . . If one recognizes this pattern,
he or she will most likely infer that the next term in the pattern is 64, even though it has
not been stated explicitly. However, our purpose throughout the course has been to re-
move all ambiguity and all dependence on the reader’s intuition. To grasp the meaning of
the notation 1, 4, 9, 16, 25, 36, 49, . . . , our reader must have some mathematical intuition to
recognize the pattern, hence this notation is not precise enough for our purposes. A better
way to describe this pattern is as the sequence (k2 )k∈N . In general, a sequence (ak )k∈N is a
correspondence between the natural numbers N and the universe of discourse to which the
terms of the sequence ak belong. For us, this universe of discourse will usually be R or Z.
By correspondence, we mean that to each k ∈ N corresponds a term ak . In our example,
the sequence (k2 )k∈N is the correspondence k 7→ k2 . That is, to each k ∈ N corresponds the
term k2 . We then know explicitly that the first term is 12 = 1, the second term is 22 = 4,
and so on. Alternatively, one can view a sequence of say real numbers (ak )k∈N as a function
f : N → R, given by f (k) = ak .
The concept of a series is very closely related to idea of sequences. A series is the accu-
mulation of the terms of a sequence. For an example of what this means, imagine that a
person deposits a certain amount of money into her bank account each week. Another way
to say this is that to each week corresponds an amount of money she deposits. This can be
viewed as a sequence (ak )k∈N , where for each k ∈ N, ak represents the amount of money de-
posited in week k. Since this money will accumulate in her bank account, we can imagine
another related sequence: That of her account balance in week k. This accumulation of the
sequence (ak )k∈N is what we call a series. In this example, the first term in the series will
be a1 , the second term will be a1 + a2 , the third a1 + a2 + a3 , and so on. We make this idea
precise by giving the following recursive definition:
Definition 1.2.12. For a sequence of real numbers (ak )k∈N , define

1
X
ak = a1 ,
k=1
and for each n ∈ N, define

n+1
 n 
X X 
ak =  ak  + an+1 .

k=1 k=1
Consider, as an example, the sum of all consecutive numbers from 1 to a certain ending
point n. Although it may seem like the expression 1 + 2 + 3 + · · · + n is sufficient to express
the idea of the sum of all consecutive numbers from 1 to n, it contains a flaw that the
Xn
expression k does not: The dots are not categorically defined. That is, the dots ‘. . . ’ in
k=1
1 + 2 + 3 + · · · + n tell the reader that the observed pattern seen in the beginning 1 + 2 + 3 is to
continue until the ending point n. This leaves the recognition of the pattern to the reader’s
own intuition. Since the pattern is not given explicitly, the meaning of the dots is open to
interpretation. In fact, there are many different patterns beginning with 1+2+3: The reader
may wish to verify that the first three terms of the sequence (n3 − 6n2 + 12n − 6)n∈N are 1, 2,
and 3. Thus the expression 1 + 2 + 3 + · · · + n, although it may seem clear enough in what
it seeks to represent, depends on the reader for interpretation and hence does not have the
precision required to be useable in a mathematical proof. On the other hand, the expression
Pn
k=1 k is defined precisely and, provided the reader follows the definition with care, it can
be interpreted in only one way.
The starting point of k = 1 in the sum is somewhat arbitrary. One often sees sums starting at
k = 0, or other integers. To extend this definition to allow for other starting points, we need
only a slight modification. First, we must allow sequences (ak )k∈Z indexed by the integers
Z rather than the natural numbers N. That is, the sequence (ak )k∈Z gives a correspondence
between the integers Z and the universe of discourse to which the terms ak belong. In
fact, we can allow sequences to be indexed by other subsets of the integers as well. For
example, the sequence (3k + 1)k∈Z≥−2 gives a correspondence between the integers k that
are greater than or equal to −2 and the terms 3k + 1. This sequence describes the pattern
−5, −2, 1, 4, 7, 10, 13, . . . . Next, we must modify the definition of a series as follows:
Definition 1.2.13. Let m ∈ Z. For a sequence of real numbers (ak )k∈Z≥m ,

define
Xm
ak = am ,
k=m
and for each n ∈ Z≥m , define

n+1
 n 
X X 
ak =  ak  + an+1 .
k=m k=m
In fact, we can express a sum using any starting point, provided we alter the summand ak
accordingly. As a simple example of this, consider how one might express a sum beginning
from k = 0 as a sum beginning from k = 1:
Proposition 1.2.33.
n
X n+1
X
For a sequence of real numbers (ak )k∈Z≥0 , ∀n ∈ Z≥0 , ak = ak−1 .
k=0 k=1
Proof.
Let (ak )k∈Z≥0 be a sequence
of real numbers.
 x x+1

 X X 
Let A =  ak =
 
x ∈ Z≥0 ak−1 
 

 

k=0 k=1

0
X 1
X
Since ak = a0 = a1−1 = ak−1 , we have that 0 ∈ A.
k=0 k=1
Let n ∈ Z≥0 .
Assume n ∈ A.
n
X n+1
X
Then ak = ak−1 .
k=0 k=1
n+1
 n 
X X 
ak =  ak  + an+1

k=0 k=0
 n+1 
X 
=  ak−1  + an+1

k=1
 n+1 
X 
=  ak−1  + a(n+2)−1

k=1
n+2
X
= ak−1 .
k=1
n+1
X (n+1)+1
X
Therefore, ak = ak−1 ; hence n + 1 ∈ A.
k=0 k=1
Therefore, ∀n ∈ Z≥0 , if n ∈ A then n + 1 ∈ A.
By the principle of mathematical induction, we have Z≥0 ⊆ A.
X n n+1
X
Therefore, ∀n ∈ Z≥0 , ak = ak−1 .
k=0 k=1
We see then that by increasing the range of the index by 1; which is to say by changing
the range from {0, 1, . . . , n} to {1, 2, . . . , n + 1}, we need only compensate by decreasing the
index in the summand by 1. That is, by changing the summand from ak to ak−1 . In this case,
the resulting sum is unchanged. This principle can be generalized quite easily to allow us
to choose any starting point for our sum. Here is the general result:
Proposition 1.2.34.
n
X n+m
X
For a sequence of real numbers (ak )k∈Z≥0 , ∀m ∈ Z, ∀n ∈ Z≥0 , ak = ak−m .
k=0 k=m
Proof.
Let (ak )k∈Z≥0 be a sequence of real numbers.
Let m ∈ Z.  
 x
X x+m
X 
Let A =  ak =
 
x ∈ Z≥0 ak−m 
 

 

k=0 k=m

0
X m
X
Since ak = a0 = am−m = ak−m , we have that 0 ∈ A.
k=0 k=m
Let n ∈ Z≥0 .
Assume n ∈ A.
n
X n+m
X
Then ak = ak−m .
k=0 k=m
n+1
 n 
X X 
ak =  ak  + an+1
k=0 k=0
n+m 
X
=  ak−m  + an+1

k=m
n+m 
X
=  ak−m  + a(n+1+m)−m


k=m
n+1+m
X
= ak−m .
k=m
n+1
X (n+1)+m
X
Therefore, ak = ak−m ; hence n + 1 ∈ A.
k=0 k=m
By the principle of mathematical induction, we have Z≥0 ⊆ A.
X n n+m
X
Therefore, ∀n ∈ Z≥0 , ak = ak−m .
k=0 k=m
n
X n+m
X
Therefore, ∀m ∈ Z, ∀n ∈ Z≥0 , ak = ak−m .
k=0 k=m
Although sums are defined recursively, it is occasionally possible to find explicit formulas
for certain sums. As an example, consider the problem of summing all consecutive natural
numbers up to a certain number n. That is, consider the sum 1 + 2 + 3 + · · · + n, or more
Xn
precisely: k. An interesting observation can be made if we add this sum to itself in
k=1
reverse order.
1 + 2 + 3 + ... + (n − 1) + n
+ n + (n − 1) + (n − 2) + ... + 2 + 1
(n + 1) + (n + 1) + (n + 1) + ... + (n + 1) + (n + 1)
Note that each column adds to n + 1. Counting that there are n such columns  n gives
 us that
X 
the total of the two sums added together is n(n + 1). It then follows that 2  k = n(n + 1);
k=1
hence
n
X n(n + 1)
k= .
k=1
2
Although this argument may seem convincing enough, we should note that it relies on the
reader’s intuition in several places. First, as mentioned earlier, the expression 1 + 2 + 3 +
· · · + n relies on the reader’s interpretation of the pattern beginning with 1, 2, and 3, to give
meaning to the symbol ‘. . . ’. Further, the argument that every column adds to n + 1 is only
explicitly demonstrated for the first three columns and the last two columns. We then rely
on the hope that our reader guesses the same pattern will continue. The intuition may be
reasonable and correct that this pattern of n + 1 resulting in each column will continue.
However, the reasoning involves more than just the definitions of the terms involved; hence
this argument does not have the strength required of a mathematical demonstration. For a
proper proof of this formula, we must turn to the recursive definition of the sum. The proof
is as follows:
Proposition 1.2.35.
n
X n(n + 1)
∀n ∈ N, k= .
k=1
2
Proof.  
x
 X x(x + 1) 
Let A =  k=
 
x∈N
 

2 

 

k=1
1
X 1(1 + 1)
k=1= .
k=1
2
Therefore, 1 ∈ A.
Let n ∈ N.
Assume n ∈ A.
n
X n(n + 1)
Then k= .
k=1
2
n+1
 n 
X X 
k =  k + (n + 1)
k=1 k=1
n(n + 1)
= +n+1
2
n(n + 1) + 2(n + 1)
=
2
(n + 1)(n + 2)
=
2
n
X n(n + 1)
Therefore, ∀n ∈ N, k= .
k=1
2
An interesting application of explicit formulas for sums is to investigate the paradoxes of

Zeno. In one such paradox, Zeno argues that for an object in motion to travel a certain
distance, it must first travel half of that distance. To then travel the remaining distance,
it must travel half of this remaining distance, and so on. Since the distance traveled by
the object can be so divided into an infinite sequence of shorter distances, the object must
traverse an infinite number of distances in order to reach its destination. Since admitting
the possibility of completing an infinite number of tasks is not desirable, Zeno suggests
that the completion of the journey is impossible. Further, since the total distance that the
object had intended to travel was arbitrary, this suggests that no journey of any distance
can be completed; hence motion is impossible. However, since it is abundantly evident that
motion is possible, this is a paradox [2]. Zeno’s paradoxes suggest a discrepancy between
what we observe motion to be and one of the ways in which we reason about motion.
In addition to the philosophical challenges of Zeno’s dichotomy paradox, the argument

also presents some interesting mathematical ideas that are worth investigating. Suppose,
with Zeno, that an object is to travel a certain distance. Since the units of measurement
for distance can be arbitrarily assigned, we may assume without loss of generality that the
object’s intended distance is 1 unit. At the first stage of the journey, Zeno allows the object
1
to travel a distance of . At the second stage, it is allowed to travel an additional distance
2
1 1 1
of , bringing its total distance to + . In general, we have that the distance traveled
4 2 4
after the nth stage in Zeno’s sequence is
n
X 1
k
.
k=1
2
There is in fact an explicit formula for this sum, which is given in the following proposition:
Proposition 1.2.36.
n
X 1 1
∀n ∈ N, k
= 1 − n.
k=1
2 2
Proof.  
x
 X 1 1
Let A =  = 1− x
 
x∈N
 
k

 k=1 2 2 

1
X 1 1 1
k
= = 1 − 1.
k=1
2 2 2
Therefore, 1 ∈ A.
Let n ∈ N.
Assume n ∈ A.
n
X 1 1
Then k
= 1 − n.
k=1
2 2
n+1
 n 
X 1 X 1  1
k
=   + n+1
k
k=1
2 k=1
2 2
1 1
=1− n
+ n+1
2 2 !
1 1
= 1 − n − n+1
2 2
2−1
= 1 − n+1
2
1
= 1 − n+1 .
2
Hence, n + 1 ∈ A.
n
X 1 1
Therefore, ∀n ∈ N, k
= 1 − n.
k=1
2 2
We may be able to shed some more light on Zeno’s dichotomy paradox if we recall two
earlier propositions. The first was given as an early example of a proof by induction:
Proposition 1.2.20 tells us that ∀n ∈ N, n < 2n . This means that for all natural numbers
1 1
n, n < . Next, recall that the Archimedean property gives us that for any ε ∈ R with
2 n
1
ε > 0, there exists a natural number n for which < ε. The combination of these two
n
propositions with the formula above yields the following:
Proposition 1.2.37.
n
X 1
∀ε > 0, ∃n ∈ N, 1 − < ε.
k=1
2k
Proof.
Let ε > 0.
1
By the Archimedean property, ∃n ∈ N, < ε. Choose such an n.
n
By proposition 1.2.20, n < 2n .
1 1
Therefore, n < .
2 n
1 1
Since we also have < ε, by transitivity n < ε.
n 2
n !
X 1 1 1
Since 1 − k
= 1 − 1 − n = n,
k=1
2 2 2
n
X 1
We have 1 − < ε.
k=1
2k
n
X 1
Therefore, ∃n ∈ N, 1 − k
< ε.
k=1
2
n
X 1
Therefore, ∀ε > 0, ∃n ∈ N, 1 − < ε.
k=1
2k
The significance of this result can be seen if we appeal to an earlier result about real
numbers. Proposition 1.1.14 tells us ∀a, b ∈ R, if ∀x > a, b ≤ x, then b ≤ a. Choosing
a = 0 and replacing the arbitrary variable x with ε, this proposition reads
∀b ∈ R, if ∀ε > 0, b ≤ ε, then b ≤ 0.
Suppose the actual distance traveled by Zeno’s object is d units. Since the question
of whether the object does or does not complete its total journey of 1 unit is open for
debate, all that can be said is that d ≤ 1. Further, for any n ∈ N the distance d cannot be
n
X 1
less than the distance traveled at the nth stage. That is, ∀n ∈ N, k
≤ d; hence ∀n ∈ N,
k=1
2
n n
X 1 X 1
1−d ≤ 1− k
. Thus, for any ε > 0, since there is an n ∈ N with 1 − < ε, we
k=1
2 k=1
2k
then have 1 − d < ε. Therefore, ∀ε > 0, 1 − d ≤ ε. By proposition 1.1.14, we then have
1 − d ≤ 0; hence 1 ≤ d. Since d ≤ 1 and 1 ≤ d, we must have d = 1 by trichotomy.
Although this analysis does not address all of the philosophical challenges presented by
Zeno’s paradox, it does provide the foundation for a proper and rigorous study of infinite
sums. Such a study is now indispensable to modern math and physics.
The reduction of the sum to an explicit formula in the example above is a special case of a
more general result concerning the sum of all powers of a given constant:
Proposition 1.2.38.
n
X a − an+1
For all a ∈ R, if a , 1 then ak = .
k=1
1−a
Proof.
Let a ∈ R, and
 assume a , 1. 
x
 X a − an+1 
Let A =  a =
k
 
x ∈ N
 
1 − a

 


k=1
1
a − a1+1
!
X 1−a
ak = a = a = .
k=1
1−a 1−a
Therefore, 1 ∈ A.
Let n ∈ N.
Assume n ∈ A.
n
X a − an+1
Then ak = .
k=1
1−a
n+1
 n 
X X k 
a =  a  + an+1
k
k=1 k=1
a − an+1
= + an+1
1−a
a − an+1 + an+1 − an+2
=
1−a
a − an+2
= .
1−a
Hence, n + 1 ∈ A.
n
X a − an+1
Therefore, ∀n ∈ N, ak = .
k=1
1−a
n
X a − an+1
Therefore, ∀a ∈ R, if a , 1 then ak = .
k=1
1−a
This result can be made more general still:

Proposition 1.2.39.
n
X
For all a, b ∈ R and all n ∈ N, an − bn = (a − b) an−k bk−1 .
k=1
Proof.
Let a, b ∈ R. 
 x Xx 
Let A =  a − b = (a − b)
x x−k k−1
 
x∈N a b 
 

 

k=1

X1
a − b = (a − b)(1) = (a − b)(a b ) = (a − b)
1−1 1−1
a1−k bk−1 .
k=1
Therefore, 1 ∈ A.
Let n ∈ N.
Assume n ∈ A.
n
X
Then an − bn = (a − b) an−k bk−1 .
k=1
an+1 − bn+1 = a(an ) − b(bn )

= a(an ) − abn + abn − b(bn )
= a(an − bn ) + bn (a − b)
X n
= a(a − b) an−k bk−1 + bn (a − b)
k=1
  n  
 X n−k k−1 
= (a − b) a  a b  + b 

n
  
k=1
 n  
X n+1−k k−1 
= (a − b)  a b  + b 
n

  
k=1
 n  
X n+1−k k−1 
= (a − b)  a b  + a (n+1)−(n+1) n 

   b 
k=1
n+1
X
= (a − b) an+1−k bk−1 .
k=1
X n
Therefore, ∀n ∈ N, an − bn = (a − b) an−k bk−1 .
k=1
n
X
Therefore, ∀a, b ∈ R, ∀n ∈ N, a − b = (a − b)
n n
an−k bk−1 .
k=1
Definition 1.2.14. Define 0! = 1, and for each n ∈ N, define n! = n(n −

1)!. Further, for n, k ∈ N with n ≥ k, define the binomial coefficient
!
n n!
= .
k k!(n − k)!
Proposition 1.2.40 (Pascal’s Law).
n+1
! ! !
n n
For all n, k ∈ N with n > k, = + .
k+1 k k+1
Proof.
Let n, k ∈ N with n > k.
The following calculation gives the desired result:
! !
n n n! n!
+ = +
k k+1 k!(n − k)! (k + 1)!(n − k − 1)!
n!(k + 1) n!(n − k)
= +
(k + 1)k!(n − k)! (k + 1)!(n − k)(n − k − 1)!
n!(k + 1) + n!(n − k)
=
(k + 1)!(n − k)!
n!(n + 1)
=
(k + 1)!(n − k)!
(n + 1)!
=
(k + 1)!(n − k)!
(n + 1)!
=
(k + 1)!(n + 1 − (k + 1))!
n+1
!
=
k+1
n+1
! ! !
n n
Therefore, = + .
k+1 k k+1
n+1
! ! !
n n
Therefore, for all n, k ∈ N with n > k, = + .
k+1 k k+1
Proposition 1.2.41 (The Binomial Theorem).

n !
X n k n−k
∀a, b ∈ R with a, b , 0, ∀n ∈ Z≥0 , (a + b) = n
ab .
k=0
k
Proof.
Let a, b ∈ R with a, b , 0. 
x !
 X n k x−k 
Let A =  x ∈ N (a + b) =
x
 
ab 
 

 k=0
k 

! 0 !
0 0 0 X n k 0−k
(a + b)0 = 1 = a b = ab .
0 k=0
k
Therefore, 0 ∈ A.
Let n ∈ Z≥0 .
Assume n ∈ A.
n !
X n k n−k
Then (a + b) = n
ab .
k=0
k
(a + b)n+1 = (a + b)(a + b)n

n !
X n k n−k
= (a + b) ab
k=0
k
n ! n !
X n k n−k X n k n−k
=a a b +b ab
k=0
k k=0
k
n ! n !
X n k+1 n−k X n k n+1−k
= a b + ab
k=0
k k=0
k
n+1 ! n !
X n X n k n+1−k
= abk n−(k−1)
+ ab
k=1
k−1 k=0
k
 n !  n !
X n X n k n+1−k
=   + an+1 + bn+1 +
k n+1−k

ab ab
k=1
k−1 k=1
k
 n 
n + 1 n+1 X n k n+1−k  n + 1 n+1
! ! !! !
n
= b +  + ab  + a
0 k=1
k−1 k n+1
 n 
n + 1 n+1 X n + 1 k n+1−k  n + 1 n+1
! ! !
= b +  ab  + n + 1 a
0 k=1
k
n+1
n + 1 k n+1−k
X !
= ab
k=0
k
By the PMI, Z≥0 ⊆ A.
n !
X n k n−k
Therefore, ∀n ∈ Z≥0 , (a + b) =
n
ab .
k=0
k
n !
X n k n−k
Therefore, ∀a, b ∈ R with a, b , 0, ∀n ∈ Z≥0 , (a + b) = n
ab .
k=0
k
Definition 1.2.15. Define the Fibonacci sequence as follows: f1 = 1,

f2 = 1, and for each n ∈ N with n ≥ 2, fn+1 = fn + fn−1 .
Proposition 1.2.42.
an −bn
Let a and b be the two real solutions of the equation x2 = x + 1. Then ∀n ∈ N, fn = a−b .
Proof.
Let a and( b be the two real solutions of the equation x2 = x + 1.
x x
)
a −b
Let A = x ∈ N f x =
a−b
a1 − b1
Since = 1 = f1 , we have 1 ∈ A.
a−b
Let n ∈ N.
Assume {1, . . . , n} ⊆ A.
We consider the case n = 1 and the case n ≥ 2.
Case 1: n = 1.
an+1 − bn+1 a2 − b2 a + 1 − (b + 1)
Then = = = 1 = f2 = fn+1 .
a−b a−b a−b
Case 2: n ≥ 2.
Then n − 1 ≥ 1.
We then have n ∈ {1, . . . , n} and n − 1 ∈ {1, . . . , n},
hence n ∈ A and n − 1 ∈ A.
an − bn an−1 − bn−1
Therefore, fn = and fn−1 = .
a−b a−b
We then have
fn+1 = fn + fn−1
an − bn an−1 − bn−1
= +
a−b a−b
an−1 (a + 1) − bn−1 (b + 1)
=
a−b
an−1 a2 − bn−1 b2
=
a−b
an+1 − bn+1
=
a−b
In both cases, n + 1 ∈ A.
Therefore, if {1, . . . , n} ⊆ A then n + 1 ∈ A.
Therefore, ∀n ∈ N, if {1, . . . , n} ⊆ A then n + 1 ∈ A.
By the PCI, N ⊆ A.
n
−bn
Therefore, ∀n ∈ N, fn = aa−b .
The roots of the equation x2 = x + 1 are significant for another reason: Two lengths x and
y are said to be in the golden ratio provided the ratio of the larger to the smaller is equal to
the ratio of the total to the larger. That is, two lengths x and y with x > y are in the golden
x x+y x
ratio provided = . In this case, the golden ratio is the ratio of x to y; that is . The
y x y
diagram shows two similar rectangles whose side lengths are in the golden ratio.
x+y x
=
x y
Fixing the smaller of the two as one unit length, i.e. taking y = 1, we see that x is equal to
x x+1
the golden ratio provided = . Hence x2 = x + 1.
1 x
x+1 x
= =⇒ x2 = x + 1
x 1
Exercises 1.2.
Prove the following propositions using the Well- 24. ∀n ∈ N, if ∀x ∈ N, ∃t ∈ N, 2x = nt, then n = 1 or
Ordering Property. n = 2.
1. ∀x ∈ R, if 0 < x, then ∃n ∈ N, n − 1 < x ≤ n. 25. ∀x ∈ R, ∀m, n ∈ Z, if n ≤ x < n + 1 and

m ≤ x < m + 1, then m = n.
2. ∀x ∈ R, if 0 < x < 1, then ∃n ∈ N, 1
n <x≤ 1
n−1 .
26. ∀x, y, q1 , q2 , r1 , r2 ∈ Z, if x = yq1 +r1 and 0 ≤ r1 < y
3. Let S ⊆ N. Let A = {x ∈ R | 1
x ∈ S }. If S , ∅, then and x = yq2 + r2 and 0 ≤ r2 < y, then q1 = q2 and
A has a largest element. r1 = r2 .
(That is, the quotient and remainder given by the
4. Let S ⊆ N. Let A = {x ∈ R | x3 ∈ S }. If S , ∅, then division algorithm are unique.)
A has a smallest element.
5. Let a ∈ N. If a , 1, then the set S = {x ∈ Prove the following propositions.

N | x divides a and x , a} has a largest element.
27. ∀x ∈ Z, if x2 is odd then x is odd.
6. Let a ∈ N. If a , 1, then the set S = {x ∈
28. ∀x, y ∈ Z, if x is even and y is odd, then xy is even.
N | x divides a and x , 1} has a smallest element.
29. ∀x, y ∈ Z, if x is even and x + y is even, then y is
7. ∀x ∈ N, 3 x ≥ 1 + 2 x .
even.
8. ∀a, b ∈ R, if a ≥ 0 and b ≥ 0, then ∀x ∈ N, 30. ∀x, y ∈ Z, if x is odd and x + y is odd, then y is even.
(a + b) x ≥ a x + b x .
31. ∀x, y ∈ Z, if x is odd and x + y is even, then y is odd.
9. ∀x ∈ Z, if x is odd, then ∀n ∈ N, xn is odd.
32. ∀x, y ∈ Z, if x is even and x + y is odd, then y is odd.
10. ∀x ∈ Z, if ∃n ∈ N, xn is odd, then x is odd.
33. ∀x, y ∈ Z, if x is odd and xy is odd, then y is odd.
11. ∀x ∈ R, if ∃n ∈ N, xn < 0, then x < 0.
34. ∀x, y ∈ Z, if x is odd and xy is even, then y is even.
12. ∀x ∈ R, if ∃n ∈ N, xn < x, then x < 1.
35. ∀x, y ∈ Z, if xy is even, then x is even or y is even.
13. ∀x ∈ R, if 1 < x, then ∀n ∈ N, 1 < xn .
36. ∀x, y ∈ Z, if xy is odd, then x is odd and y is odd.
14. ∀x, y ∈ R, if 0 < x < y, then ∀n ∈ N, xn < yn . 37. ∀x, y, z ∈ Z, if x − y is even and y − z is even, then
15. ∀x ∈ R, if 0 < x < 1, then ∀n ∈ N, xn < 1. x − z is even.
16. ∀x, y ∈ R, if 0 < x < 1, then ∀n ∈ N, xn < x. 38. ∀x, y ∈ Z, if x divides x + y then x divides y.
39. ∀x, y, z ∈ Z, if x divides y and y divides z, then x
17. ∀m, n ∈ N, if m < n, then 2m < 2n .
n m divides z.
18. ∀m, n ∈ N, if m < n, then 21 < 12 . 40. ∀x, y ∈ Z, if x divides y and y divides x, then x = y
19. ∀x ∈ R, if 0 < x < 1, then ∀m, n ∈ N, if m < n, then or x = −y.
xn < xm . 41. ∀x ∈ Z, if 3 divides x, then 3 divides 9 − x.
20. ∀x ∈ R, if 1 < x, then ∀m, n ∈ N, if m < n, then 42. ∀x ∈ Z, if 3 divides 9 − x, then 3 divides x.
xm < xn .
43. ∀x, y ∈ Z, if 5 divides 11x + 6y, then 5 divides x + y.
Prove the following propositions using theorem 1.2.3 44. ∀x, y ∈ Z, if 5 divides x + y, then 5 divides 11x + 6y.
or its corollary. 45. ∀x, y ∈ Z, if x divides y, then x divides |y|.
21. ∀x ∈ Z, if x < 0, then x ≤ −1. 46. ∀x, y ∈ Z, if x divides y, then |x| divides y.
22. ∀x, y ∈ Z, if x < y + 1, then x ≤ y. 47. ∀x, y ∈ Z, if |x| divides y, then x divides y.
23. ∀x ∈ N, if x divides 2, then x = 1 or x = 2. 48. ∀x, y ∈ Z, if x divides |y|, then x divides y.
49. ∀x ∈ Z, if 3 divides x and 2 divides x, then 6 divides 66. ∀x, y ∈ R, if x is rational and x , 0 and y is irra-
x. x
tional, then is irrational.
y
50. ∀x ∈ Z, if 6 divides x, then 3 divides x and 2 divides 67. ∃x, y ∈ R, x is irrational and y is irrational and xy is
x. rational.
51. ∀x ∈ Z, if 30 divides x, then 5 divides x and 6 di- 68. ∃x, y ∈ R, x is irrational and y is irrational and x + y
vides x. is rational.
52. ∀x ∈ Z, if 5 divides x and 6 divides x, then 30 di-
vides x. Prove the following propositions. They are analogous
to lemma 1.2.14 and proposition 1.2.15.
Prove the following propositions about greatest com- 69. ∀x ∈ Z, if 3 divides x2 , then 3 divides x.
√
mon divisors and least common multiples. 70. 3 is irrational.
53. ∀x, y ∈ Z, if x , 0 and y , 0, then ∃ f ∈ Z,
f = lcm (x, y). Using the Well-Ordering Property, prove the following
forms of the Principle of Mathematical Induction.
54. ∀a, x, y ∈ N, if x divides a and y divides a, then
lcm (x, y) divides a. (Hint: use the division algo- 71. Let a ∈ Z, and let A ⊆ Z. If a ∈ A and ∀n ∈ Z, if
rithm.) n ∈ A then n − 1 ∈ A, then Z≤a ⊆ A.
72. Let A ⊆ Z. If 0 ∈ A and ∀n ∈ Z, if n ∈ A then
55. ∀a, x, y ∈ Z, if x divides y, then gcd(a, x) divides
n + 1 ∈ A and n − 1 ∈ A, then A = Z.
gcd(a, y).
73. Let A ⊆ Z. If A , ∅ and ∀n ∈ Z, if n ∈ A then
56. ∀a, x, y ∈ Z, if x divides y, then lcm (a, x) divides n + 1 ∈ A and n − 1 ∈ A, then A = Z.
lcm (a, y).
74. Let a ∈ Z, and let A ⊆ Z. If a ∈ A and ∀n ∈ Z≥a , if
57. ∀a, b, x, y ∈ Z, if x divides y and a divides b, then {a, . . . , n} ⊆ A then n + 1 ∈ A, then Z≥a ⊆ A.
gcd(a, x) divides gcd(b, y).
58. ∀a, b, x, y ∈ Z, if x divides y and a divides b, then Prove the following propositions using the Principle of
lcm (a, x) divides lcm (b, y). Mathematical Induction.
75. ∀x ∈ N, 5 divides 8 x + 2(3 x−1 ).
59. ∀m, n ∈ Z, if m, n , 0 and gcd(m, n) = 1, then
∀x ∈ Z, ∃u, v ∈ Z, x = mu + nv. 76. ∀x ∈ N, 3 divides 22x+1 + 1.
60. ∀m, n, x ∈ Z, if m, n , 0 and gcd(m, n) divides x, 77. ∀x, y ∈ Z, ∀m ∈ N, x − y divides xm − ym .

then ∃u, v ∈ Z, x = mu + nv. 78. ∀x ∈ Z, ∀m ∈ N, x2 + x+1 divides (x+1)2m−1 + xm+1 .
61. ∀x, y, a ∈ Z, if x, y , 0 and x divides a and y divides 79. ∀x, y, a, b ∈ Z, if x − y divides a + b, then ∀k ∈ N,
a and gcd(x, y) = 1, then xy divides a. x − y divides axk + byk .
62. ∀x, y, a ∈ Z, if if a, x , 0 and a divides xy and 80. ∀x ∈ N, 5 x ≥ 2 x + 3 x .
gcd(a, x) = 1, then a divides y.
81. ∀a, b ∈ R, if a ≥ 0 and b ≥ 0, then ∀x ∈ N,
(a + b) x ≥ a x + b x .
Prove the following propositions about rational and ir-
82. ∀x ∈ Z, if x is odd, then ∀n ∈ N, xn is odd.
rational numbers.
83. ∀x ∈ Z, if ∃n ∈ N, xn is odd, then x is odd.
63. ∀x, y ∈ R, if x is rational and y is rational, then x + y
is rational. 84. ∀x ∈ R, if ∃n ∈ N, xn is irrational, then x is irra-
tional.
64. ∀x, y ∈ R, if x is rational and y is irrational, then
x + y is irrational. 85. ∀x ∈ R, if x > 1, then ∀n ∈ N, xn > 1.
86. ∀x ∈ R, if ∃n ∈ N, xn < 0, then x < 0.
65. ∀x, y ∈ R, if x , 0 and xy is rational and y is irra-
tional, then x is irrational. 87. ∀x ∈ R, if ∃n ∈ N, xn < x, then x < 1.
88. ∀x ∈ R, if 0 < x < 1, then ∀n ∈ N, 0 < xn < 1. 107. ∀x ∈ R \ {0}, ∀n ∈ N, (x−1 )n = (xn )−1 .
!n
89. ∀x, y ∈ R, if 0 < x < y, then ∀n ∈ N, xn < yn . x xn
108. ∀x ∈ R, ∀y ∈ R \ {0}, ∀n ∈ N, = n.
y y
90. ∀m, n ∈ N, if m < n, then 2m < 2n .
n m 109. ∀x ∈ R, ∀n, m ∈ N, (xn )m = xnm .
91. ∀m, n ∈ N, if m < n, then 21 < 12 .
xn
110. ∀x ∈ R \ {0}, ∀n, m ∈ N with n > m, xn−m = xm .
92. ∀x ∈ R, if 0 < x < 1, then ∀m, n ∈ N, if m < n, then
xn < xm .
Prove the following properties of series.
93. ∀x ∈ R, if 1 < x, then ∀m, n ∈ N, if m < n, then
xm < xn . n
X
111. ∀a ∈ R, ∀n ∈ N, a = na.
an+1 −1 an −1
94. Let a ∈ R with a > 1. Then, ∀n ∈ N, n+1 > n .
k=1
95. Let a ∈ R with 0 < a < 1. Then, ∀n ∈ N, 112. For a sequence of real numbers (ak )k∈N , ∀c ∈ R,
n n
an+1 −1 an −1
n+1 > n .
X X
∀n ∈ N, cak = c ak .
k=1 k=1
96. (Bernoulli’s Inequality) ∀x ∈ R, if −1 ≤ x, then
∀n ∈ N, 1 + nx ≤ (1 + x)n . 113. For sequences of real numbers  (ak)k∈N and (bk )k∈N ,
n
 n n
X X  X
(ak + bk ) =  ak  +  bk .

97. Every finite set of real numbers has a minimum el- ∀n ∈ N,
ement. k=1 k=1 k=1
98. Let f : Z → Z be a function with the property 114. For a sequence of real numbers (ak )k∈Z≥0 , ∀n ∈ N,
n n
that ∀a ∈ Z, f (a) < f (a + 1). Prove ∀x, y ∈ Z, if X X
ak = a0 + ak .
x < y then f (x) < f (y). (Hint: Let x ∈ Z, and let
k=0 k=1
A = {k ∈ N | f (x) < f (x + k)}. Prove by induction
that A = N, and use this to prove the desired result.) 115. For a sequence of real numbers (ak )k∈N , ∀m, n ∈ N,
m
X Xn Xn
99. Let S ⊆ R. If S is closed under addition in the sense if m < n, then ak + ak = ak .

that ∀x, y ∈ S , x + y ∈ S , then ∀a ∈ S , ∀n ∈ N, k=1 k=m+1 k=1
na ∈ S . 116. For a sequence of real numbers (ak )k∈N , ∀n ∈ N,

Xn Xn
100. Let S ⊆ R. If S is closed under multiplication in the ak = an+1−k .
sense that ∀x, y ∈ S , xy ∈ S , then ∀a ∈ S , ∀n ∈ N, k=1 k=1
an ∈ S .
117. For a sequence of real numbers (ak )k∈N , ∀n ∈ N,
Xn
(ak+1 − ak ) = an+1 − a1 . (Such a series is called
Prove the following propositions about inductive sets.
k=1
101. The set of rational numbers Q is an inductive subset a telescoping sum)
of R. 118. For a sequence of real numbers (ak )k∈N , ∀n ∈ N,

n n
102. Let A ⊆ R be an inductive set. Then, ∀a ∈ A, X X
ak ≤ |ak |.
∀n ∈ N, a + n ∈ A. k=1
k=1
103. M0 (the smallest inductive set containing 0) is 119. For a sequence of integers (ak )k∈N , if ∀k ∈ N, ak is
closed under addition. Xn
even, then ∀n ∈ N, ak is even.
104. For all a ∈ Z, Ma is closed under addition. k=1
105. Let a ∈ R, and let Ma be the smallest inductive set 120. For a sequence of real numbers (ak )k∈N , if ∀k ∈ N,
n
containing a. ∀x, y ∈ Ma , if x < y, then y − x ∈ N. X
ak is rational, then ∀n ∈ N, ak is rational.
106. Let a ∈ Z, and let Ma be the smallest inductive set k=1
containing a. Then Ma = Z≥a .

n
Prove the following propositions using the recursive
X n(n + 1)(2n + 1)
121. ∀n ∈ N, k2 = .
definition of exponents. k=1
6
n !2
X n(n + 1)
122. ∀n ∈ N, k = 3
.
2 For each of the following recursively defined functions,
k=1
guess an explicit formula for f (x) and prove that your
n
X n(n + 1)(2n + 1)(3n2 + 3n − 1) formula is true for all x ∈ N.
123. ∀n ∈ N, k4 = .
k=1
30 127. f : N → N given by: f (1) = 2 and for each n ∈ N,
n
X f (n + 1) = f (n) + 2.
124. ∀n ∈ N, (k)(k!) = (n + 1)! − 1.
128. f : N → N given by: f (1) = 1, f (2) = 3 and for
k=1
each n ≥ 2, f (n + 1) = 2 f (n) − f (n − 1).
n
X k 1
125. ∀n ∈ N, =1− . 129. f : N → N given by: f (1) = 1, f (2) = 4 and for
(k + 1)! (n + 1)!
k=1 each n ≥ 2, f (n + 1) = 2( f (n) + 1) − f (n − 1).
n
130. f : N → N given by: f (1) = 1, f (2) = 8 and for
X
126. ∀n ∈ N, 2k − 1 = n2 .
k=1
each n ≥ 2, f (n + 1) = 2( f (n) + 3n) − f (n − 1).
Let n ∈ N and let N|(n) be the set of divisors of n. That is,
N|(n) = {x ∈ N | x divides n}.
Let x, y, z ∈ N|(n). Compare the following properties of the gcd and lcm functions to the Boolean algebraic
properties given in section 0.3.
n
=n n
=1 1. Prove the Idempotence properties for gcd and lcm .
1 n
Idempotence
lcm (x, x) = x gcd(x, x) = x 2. Prove the Annihilator properties for gcd and lcm .
Commutativity
lcm (x, y) = lcm (y, x) gcd(x, y) = gcd(y, x) 3. Prove the Identity properties for gcd and lcm .
Associativity
lcm (x, lcm (y, z)) = lcm (lcm (x, y), z) 4. Prove the Absorption properties for gcd and lcm .
gcd(x, gcd(y, z)) = gcd(gcd(x, y), z)
Absorption 5. Prove the Associativity properties for gcd and lcm .

lcm (x, gcd(x, y)) = x gcd(x, lcm (x, y)) = x (Hint: use the result proven in exercise 54.)
Distributivity
lcm (x, gcd(y, z)) = gcd(lcm (x, y), lcm (x, z)) 6. Prove ∀a, x, y ∈ N|(n), both nx and n
y divide n
a if and
lcm (x, gcd(y, z)) = lcm (gcd(x, y), gcd(x, z)) only if a divides both x and y.
Annihilator
lcm (x, n) = n gcd(x, 1) = 1 7. Prove ∀a, x, y ∈ N|(n), an divides both n
x and n
y if and
only if both x and y divide a.
Identity
lcm (x, 1) = x gcd(x, n) = x
8. Prove De Morgan’s Laws for gcd and lcm .
Double Negation
n
=x 9. Give an example of a value of n ∈ N and x ∈ N|(n)
n
x
for which the analogous properties to Complemen-
tation: gcd(x, nx ) = 1 and lcm (x, nx ) = n do not hold.
De Morgan’s Laws
gcd(x,y) = lcm x , y
n n n
lcm (x,y) = gcd x , y

n n n
Chapter 2
Sets
We began discussing the common notation used to describe sets and set related concepts in
section 0.4. We now return to this topic with a view toward applying the proof techniques
of chapter 1 to propositions involving sets. We will see in this chapter that the language
of sets can be used to simplify and generalize a number of mathematical concepts. In one
important example, the concepts of greatest common divisor and least common multiple can
be viewed as simple operations on the ideals of Z defined in section 0.4 (definition 0.4.2).
In fact, since the language and notation used in math can often become bulky and unwieldy
as the theory develops, and sets offer us a convenient way to organize and simplify this
language, their use has become indispensable in all areas of mathematics.
2.1 Relations and Operations
The primary concept with which the language of sets is concerned is the relation between
element and set; that is, the relation ‘∈’. To explore the linguistic purpose served by this re-
lation, recall the earlier discourse on sentence structure in chapter 0, section 0.1. We began
then by identifying the subject x and predicate P(x) in a given proposition. For example, in
the proposition ‘2 is even,’ the subject is 2 and the predicate is ‘x is even.’ Recall that when
the subject is a free variable, the predicate ‘x is even’ is called an open sentence. Further,
the possible subjects for which a given predicate is intelligible we called the universe of
discourse (for the variable x in the open sentence ‘x is even,’ the universe of discourse is Z).
Now, given any universe of discourse U and predicate P(x), we can specify a set
A = {x ∈ U | P(x)}.
Having specified such a set, we see that for x ∈ U, the statements P(x) and x ∈ A have
the same meaning. That is, if one is conversing about subjects that belong to a universe
of discourse U (for example, when discussing real numbers or integers, the universe of
discourse would be R or Z respectively), then any predicate P(x) can be replaced with a
predicate of the form x ∈ A for an appropriate set A. For example, for x ∈ Z, rather than
saying ‘x is even,’ one can say x ∈ h2i.
125
126 CHAPTER 2. SETS
Recall that for a ∈ Z, the ideal of Z generated by a is the set
hai = {x ∈ Z | ∃t ∈ Z, x = at}.
Similarly, for x ∈ R, rather than saying x > 0, one can say x ∈ (0, ∞). In this way, all
predicates can be phrased in terms of sets.
Consider now a statement involving a universal quantifier,
‘∀x ∈ A, P(x),’ or equivalently ‘∀x ∈ U, if x ∈ A, then P(x)’.
By specifying B = {x ∈ U | P(x)}, this statement becomes
‘∀x ∈ U, if x ∈ A, then x ∈ B’. That is, ‘A ⊆ B’.
Hence statements with universal quantifiers can be phrased in terms of the subset
relation. Similarly, for the same open sentence P(x), the existentially quantified statement
‘∃x ∈ U, P(x)’ can be written in the language of sets as ‘B , ∅’.
As an example, consider the statement ‘∀x ∈ R, ∃y ∈ R, x < y’. This can be written in
set terms as ‘{x ∈ R | (x, ∞) , ∅} = R.’ For a more extreme example, we can rewrite the
statement ‘∀x ∈ R, if ∀ε > 0, x < ε, then x ≤ 0’ as ‘{x ∈ R | (0, ∞) ⊆ (x, ∞)} ⊆ (−∞, 0].’
It should be noted that these translations yield no new insight into the meaning or truth of
the propositions in question. They simply say the same thing in different notation. In the
examples given above, there is not necessarily any practical advantage to be gained from
one notation or the other. The examples are only intended to illustrate that the element-set,
set-subset, and equality of sets, relations comprise the core of the structure of mathematical
language. Hence, these relations deserve further discussion.
Subsets and Set Equality
Recall that given two sets A and B, whose elements belong to a common universe of dis-
course U, to say that A is a subset of B means that all of the elements of A are also elements
of B. More precisely:

A is a subset of B, denoted A ⊆ B, means
∀x ∈ U, if x ∈ A, then x ∈ B.
Equivalently, we may use the set A as the universe of discourse and write
A ⊆ B as
∀x ∈ A, x ∈ B.
2.1. RELATIONS AND OPERATIONS 127
Notice that the relation ⊆ is defined in terms of an implication:
‘∀x ∈ U, if x ∈ A, then x ∈ B.’
This means that to prove A ⊆ B, for given sets A and B, we must prove the impli-
cation
‘if x ∈ A, then x ∈ B’
for an arbitrary constant x. Recall that we have three structures that can be used to
prove implications. These are direct proof, proof by contraposition, and proof by contra-
diction. A proof that A ⊆ B will hence follow one of these three structures. To examine
some of the basic properties of the subset relation we will look at three fundamental results
that hold for every set. First, when a set is defined using a universe of discourse and an
open sentence:
{x ∈ U | P(x)},
we require that the universe of discourse U is itself a set. This leads to the very intuitive
result that every set is a subset of the universe of discourse in which it is defined. Second,
we have the similarly intuitive result that every set is a subset of itself. Third, we have the
perhaps less intuitive result that the empty set ∅ is a subset of every set. We will use a direct
proof for the first two and a proof by contradiction for the third.
Proposition 2.1.1.
Let A be a set whose elements belong to the universe U. Then
1. A ⊆ U,
2. A ⊆ A
3. ∅ ⊆ A.
Proof.
(1) Let x ∈ U.
Assume x ∈ A.
Therefore, x ∈ U.
Therefore, if x ∈ A, then x ∈ U.
Therefore, ∀x ∈ U, if x ∈ A, then x ∈ U.
Therefore, A ⊆ U.
(2) Let x ∈ U.
Assume x ∈ A.
Then x ∈ A.
Therefore, if x ∈ A, then x ∈ A.
Therefore, ∀x ∈ U, if x ∈ A, then x ∈ A.
Therefore, A ⊆ A.
(3) Suppose ∃x ∈ U, x ∈ ∅ and x < A.
Choose such an x.
Then x ∈ ∅, which is a contradiction.
Therefore, ∀x ∈ U, if x ∈ ∅, then x ∈ A.
Therefore, ∅ ⊆ A.
128 CHAPTER 2. SETS
All three proofs above are trivial in the sense that they do not involve an argument or
demonstration. They are evident simply by their sentence structure. In the first case, since
for any x ∈ U, the proposition x ∈ U is true, the statement
‘if x ∈ A, then x ∈ U’ is of the form ‘if P then T RUE’
which is a tautology (it is true regardless of the truth value of P). Similarly, in the
second case, the statement
‘if x ∈ A, then x ∈ A’ is a tautology of the form ‘if P, then P’.
Finally, since for any x ∈ U, the statement x ∈ ∅ is false, the third statement
‘if x ∈ ∅, then x ∈ A’ is of the form ‘if FALS E, then P’,
which again is true regardless of the truth or falsity of P. Hence all three proposi-
tions are true by virtue of their sentence structure, without regards to the meaning of the
predicate x ∈ A.
Proofs of abstract statements that apply to all sets will generally exhibit this behaviour.
Since the basic predicate x ∈ A has no determined meaning when A is an arbitrary set,
it is only the logical sentence structure of whatever proposition we are trying to prove
that contributes to its proof. For this reason, proofs involving abstract sets give us a good
opportunity to practice structuring our mathematical proofs, without the distraction of cal-
culations or demonstrations. For another example of a proof that applies to all sets, we
prove that the subset relation is transitive:
Proposition 2.1.2.
Let A, B, and C be sets whose elements belong to a common universe of discourse U. If

A ⊆ B and B ⊆ C, then A ⊆ C.
Proof.
Assume A ⊆ B and B ⊆ C.
Let x ∈ U.
Assume x ∈ A
Since x ∈ A and A ⊆ B, we have x ∈ B.
Since x ∈ B and B ⊆ C, we have x ∈ C.
Therefore, x ∈ C.
Therefore, if x ∈ A, then x ∈ C.
Therefore, ∀x ∈ U, if x ∈ A, then x ∈ C.
That is, A ⊆ C.
Therefore, if A ⊆ B and B ⊆ C, then A ⊆ C
Recall from section 0.4 that equality of sets is given in terms of the subset relation:

A = B means
A ⊆ B and B ⊆ A.
This means that to prove sets A and B are equal, we prove that A ⊆ B and that B ⊆ A.
Thus a proof of equality is simply two proofs that a subset relation holds. We demonstrate
this with two examples, one abstract and one more concrete. Both examples involve set
complements, which were defined in section 0.4. So that the notation in these examples is
understood, recall definition 0.4.4:
Let A be a set whose elements are in the universe of discourse U. We

define the complement of A, denoted Ac , to be
Ac = {x ∈ U | x < A}.
Further, given any two sets A and B, we define the complement of A

relative to B, denoted B \ A to be
B \ A = {x ∈ U | x ∈ B and x < A}.
Proposition 2.1.3.
Let A and B be sets within a common universe of discourse U. If B ⊆ Ac , then A \ B = A.
Proof.
Assume B ⊆ Ac .
Let x ∈ U.
Assume x ∈ A \ B.
Then x ∈ A and x < B.
In particular, we have x ∈ A.
Therefore, if x ∈ A \ B, then x ∈ A.
Therefore, A \ B ⊆ A.
Let x ∈ U.
Assume x ∈ A.
Suppose x ∈ B, seeking a contradiction.
Since x ∈ B and B ⊆ Ac , we have x ∈ Ac .
Therefore, x < A.
We now have the contradiction x ∈ A and x < A.
Therefore, x < B.
Then x ∈ A and x < B; hence x ∈ A \ B.
Therefore, if x ∈ A, then x ∈ A \ B.
Therefore, A ⊆ A \ B.
Hence, we have A \ B ⊆ A and A ⊆ A \ B.
Therefore, A = A \ B.
Therefore, if B ⊆ Ac , then A \ B = A.
130 CHAPTER 2. SETS
Next, for a concrete example, we consider the relative complement of two real intervals:
(0, 2) \ (1, 3). If one can imagine these two sets drawn on a number line and consider
those points on the line that lie within the interval (0, 2) and outside of the interval (1, 3), it
will soon become clear that the set (0, 2) \ (1, 3) is the interval (0, 1]. To confirm that our
intuition is correct, we prove the equality of these sets:
Example 2.1.4.
(0, 2) \ (1, 3) = (0, 1].
Proof.
Let x ∈ R.
Assume x ∈ (0, 2) \ (1, 3).
Then x ∈ (0, 2) and x < (1, 3).
Since x ∈ (0, 2), we have that 0 < x and x < 2.
Since x < 2 and 2 < 3, we have by transitivity, x < 3.
Further, since x < (1, 3), we have either x ≤ 1 or x ≥ 3.
However, since x < 3, it is not the case that x ≥ 3.
Therefore, it must be the case that x ≤ 1.
We now have 0 < x and x ≤ 1
Therefore, x ∈ (0, 1]
Therefore, if x ∈ (0, 2) \ (1, 3), then x ∈ (0, 1].
Therefore, (0, 2) \ (1, 3) ⊆ (0, 1].
Let x ∈ R.
Assume x ∈ (0, 1].
Then 0 < x and x ≤ 1.
Since x ≤ 1 and 1 < 2, we have x < 2.
We now have 0 < x and x < 2, hence x ∈ (0, 2).
Suppose x ∈ (1, 3).
Then 1 < x and x < 3.
In this case, we have the contradiction 1 < x and x ≤ 1.
Therefore, x < (1, 3)
We have thus shown that x ∈ (0, 2) and x < (1, 3).
Therefore, x ∈ (0, 2) \ (1, 3).
Therefore, if x ∈ (0, 1], then x ∈ (0, 2) \ (1, 3).
Therefore, (0, 1] ⊆ (0, 2) \ (1, 3).
Therefore, (0, 2) \ (1, 3) = (0, 1].
Binary Operations on Sets
In addition to the relative complement, we define two other binary operations on sets. Given
two sets A and B, the intersection is the set of all elements common to both A and B, and
the union is the set of elements that appear in at least one of A or B. The proper definitions
are:
Definition 2.1.1. Let A and B be sets. We define the intersection of A

and B, denoted A ∩ B, to be
A ∩ B = {x | x ∈ A and x ∈ B}.
We define the union of A and B, denoted A ∪ B to be
A ∪ B = {x | x ∈ A or x ∈ B}.
Venn diagrams, although not used in formal proofs, can give us a convenient way
of visualizing these binary operations. If we draw the universe of discourse as an all-
encompassing rectangle and represent sets as circles within this universe, we can picture
the regions captured by the definitions of intersection, union, complement, and relative
complement.
A∩B A∪B B\A Ac
One quick observation that we can see in the above diagram is that A ∩ B is in some sense
smaller than each of A and B, while A ∪ B is in some sense larger than each of A and B. In
fact, the sense in which these are smaller and larger is in terms of the subset relation. We
have A ∩ B ⊆ A and A ∩ B ⊆ B. Likewise, A ⊆ A ∪ B and B ⊆ A ∪ B. However, since an
appeal to the diagram does not constitute a mathematical proof, we should prove that these
relations do in fact hold as stated.
Proposition 2.1.5.
Let A and B be sets. Then A ∩ B ⊆ A and A ⊆ A ∪ B.
Proof.
Let x ∈ A ∩ B.
Then x ∈ A and x ∈ B.
In particular, x ∈ A.
Therefore, A ∩ B ⊆ A.
Let x ∈ A.
Then it is true that x ∈ A or x ∈ B.
Therefore, x ∈ A ∪ B.
Therefore, A ⊆ A ∪ B.
Therefore, A ∩ B ⊆ A and A ⊆ A ∪ B.
132 CHAPTER 2. SETS
The binary operations ∩ and ∪ satisfy all of the same Boolean algebraic properties satisfied
by the logical operators ∧ and ∨ (see section 0.3). Thus, the family of sets in a given
universe of discourse form a boolean algebra:
Boolean Algebraic Properties of Sets

For sets A, B, and C in a common universe of discourse U, the following are true:
Uc = ∅ ∅c = U
Idempotence
A∩A= A A∪A= A
Commutativity
A∩B= B∩A A∪B= B∪A
Associativity
A ∩ (B ∩ C) = (A ∩ B) ∩ C A ∪ (B ∪ C) = (A ∪ B) ∪ C
Absorption
A ∩ (A ∪ B) = A A ∪ (A ∩ B) = A
Distributivity
A ∩ (B ∪ C) = (A ∩ B) ∪ (A ∩ C) A ∪ (B ∩ C) = (A ∪ B) ∩ (A ∪ C)
Annihilator
A∩∅=∅ A∪U =U
Identity
A∩U = A A∪∅= A
Complementation
A ∩ Ac = ∅ A ∪ Ac = U
Double Negation
(Ac )c = A
De Morgan’s Laws
(A ∪ B)c = Ac ∩ Bc (A ∩ B)c = Ac ∪ Bc
Unlike the axioms of the real numbers, which establish an algebraic system by defining the
real numbers as such, the algebraic properties of sets are not given by definition. Rather,
each of the above properties can be deduced using the definitions of intersection, union, and
complement. The proofs of the boolean algebraic properties of sets are all fairly straight
forward and do not lead to any proof structures or techniques that we have not already
discussed at length. It would therefore be tedious and of very little pedagogical value to
recount all of their proofs here. However, the reader is certainly invited to attempt proofs
of these properties at his or her leisure.
With all of our definitions in place, for the sake of practice we will give several examples
of proofs involving intersections, unions, complements, subsets, and set equality. For each
of the following examples, let A, B, C, and D be sets (if needed, assume that the elements
of these sets belong to a common universe of discourse U).
Proposition 2.1.6.
If A ⊆ B, then A ∩ C ⊆ B ∩ C.
Proof.
Assume A ⊆ B.
Let x ∈ A ∩ C.
Then x ∈ A and x ∈ C.
We now have x ∈ B and x ∈ C.
Therefore, x ∈ B ∩ C.
Therefore, A ∩ C ⊆ B ∩ C.
Therefore, if A ⊆ B, then A ∩ C ⊆ B ∩ C.
Proposition 2.1.7.
If A ⊆ B and C ⊆ D, then A ∪ C ⊆ B ∪ D.
Proof.
Assume A ⊆ B and C ⊆ D.
Let x ∈ A ∪ C.
Then x ∈ A or x ∈ C.
Case 1: x ∈ A.
Since x ∈ B and B ⊆ B ∪ D, we have x ∈ B ∪ D.
Case 2: x ∈ C.
Since x ∈ C and C ⊆ D, we have x ∈ D.
Since x ∈ D and D ⊆ B ∪ D, we have x ∈ B ∪ D.
In both cases we have x ∈ B ∪ D.
Therefore, A ∪ C ⊆ B ∪ D.
Therefore, if A ⊆ B and C ⊆ D, then A ∪ C ⊆ B ∪ D.
134 CHAPTER 2. SETS
Proposition 2.1.8.
If A ⊆ B, then C \ B ⊆ C \ A.
Proof.
Assume A ⊆ B.
Let x ∈ U.
Suppose x ∈ C \ B and x < C \ A.
Since x ∈ C \ B, we have x ∈ C and x < B.
Also, since x < C \ A, we have x < C or x ∈ A.
Considering that x ∈ C, it must then be the case that x ∈ A.
This gives us the contradiction x ∈ B and x < B.
Therefore, if x ∈ C \ B, then x ∈ C \ A.
Therefore, C \ B ⊆ C \ A.
Therefore, if A ⊆ B, then C \ B ⊆ C \ A.
Proposition 2.1.9.
If A \ B ⊆ C, then A \ C ⊆ B.
Proof.
Suppose A \ B ⊆ C and A \ C * B.
Since A \ C * B, we have that ∃x ∈ U, x ∈ A \ C and x < B.
Choose such an x.
We then have x ∈ A and x < C and x < B.
Since x ∈ A and x < B, we have x ∈ A \ B.
Therefore, x ∈ C, since A \ B ⊆ C.
We now have the contradiction x ∈ C and x < C.
Therefore, if A \ B ⊆ C, then A \ C ⊆ B.
Proposition 2.1.10.
If A ∩ B ⊆ C, then (A \ C) ∩ B = ∅.
Proof.
Suppose A ∩ B ⊆ C and (A \ C) ∩ B , ∅.
Since (A \ C) ∩ B , ∅, we can choose x ∈ (A \ C) ∩ B.
For such an x, we have x ∈ A and x < C and x ∈ B.
Since x ∈ A and x ∈ B, we have x ∈ A ∩ B.
Therefore, x ∈ C, since A ∩ B ⊆ C.
This gives us the contradiction x ∈ C and x < C.
Therefore, if A ∩ B ⊆ C, then (A \ C) ∩ B = ∅.
A remark about proving emptiness

Generally, for sets A and B, to prove A = B we show A ⊆ B and B ⊆ A.
To prove A = ∅, this amounts to showing ∅ ⊆ A, which is always true,
and A ⊆ ∅, which presents some stylistic concerns: To show A ⊆ ∅ by
a direct proof involves assuming x ∈ A and demonstrating x ∈ ∅. Since
x ∈ ∅ is categorically false, it is not likely that such a demonstration
will be feasible. In fact, demonstrating that x ∈ ∅ amounts to finding
a contradiction. Because of this, in practically all cases, the statement
A = ∅ is best proved using a proof by contradiction. That is, we assume
A , ∅, which allows us to choose a nonspecific constant x ∈ A. From
this assumption, we then attempt to derive a contradiction.
For some more concrete examples, we will look at unions and intersections involving real
intervals and ideals of Z.
Example 2.1.11.
(0, 1] ∩ [2, 3) = ∅.
Proof.
Suppose (0, 1] ∩ [2, 3) , ∅.
We may then choose x ∈ (0, 1] ∩ [2, 3).
For such an x, we have x ∈ (0, 1] and x ∈ [2, 3).
This gives us four inequalities: 0 < x, x ≤ 1, 2 ≤ x, and x < 3.
Of particular interest are the inequalities x ≤ 1 and 2 ≤ x.
Since x ≤ 1 and 1 < 2, we have x < 2, which contradicts 2 ≤ x.
We thus have a contradiction.
Therefore, (0, 1] ∩ [2, 3) = ∅.
136 CHAPTER 2. SETS
Example 2.1.12.
(0, 2] ∪ [1, 3) = (0, 3).
Proof.
Let x ∈ (0, 2] ∪ [1, 3).
Then x ∈ (0, 2] or x ∈ [1, 3).
Case 1: x ∈ (0, 2].
In this case, we have 0 < x and x ≤ 2.
Since x ≤ 2 and 2 < 3, we have x < 3.
This gives us 0 < x and x < 3; hence, x ∈ (0, 3).
Case 2: x ∈ [1, 3).
In this case, we have 1 ≤ x and x < 3.
Since 0 < 1 and 1 ≤ x, we have 0 < x.
We then have 0 < x and x < 3; hence, x ∈ (0, 3).
In both cases, x ∈ (0, 3).
Therefore, (0, 2] ∪ [1, 3) ⊆ (0, 3).
Conversely, let x ∈ (0, 3).
We consider two cases: x ≤ 2 and 2 < x.
Case 1: x ≤ 2.
In this case, we have 0 < x and x ≤ 2.
Therefore, x ∈ (0, 2].
Therefore, x ∈ (0, 2] ∪ [1, 3), since (0, 2] ⊆ (0, 2] ∪ [1, 3).
Case 2: 2 < x.
In this case, we have 1 < 2 and 2 < x, hence 1 < x.
Therefore, 1 ≤ x.
We then have 1 ≤ x and x < 3, hence x ∈ [1, 3).
Therefore, x ∈ (0, 2] ∪ [1, 3).
In both cases, x ∈ (0, 2] ∪ [1, 3).
Therefore, (0, 3) ⊆ (0, 2] ∪ [1, 3).
Therefore, (0, 2] ∪ [1, 3) = (0, 3).
Example 2.1.13.
h2i ∩ h3i = h6i.
Proof.
Let x ∈ h2i ∩ h3i.
Then x ∈ h2i and x ∈ h3i.
This means, ∃t ∈ Z, x = 2t and ∃s ∈ Z, x = 3s.
Choose such s and t, and set r = t − s.
x = 3x − 2x = 3(2t) − 2(3s) = 6(t − s) = 6r.
Therefore, ∃r ∈ Z, x = 6r; hence, x ∈ h6i.
Therefore, h2i ∩ h3i ⊆ h6i.
Conversely, let x ∈ h6i.
Then ∃a ∈ Z, x = 6a. Choose such an a.
Put b = 3a.
x = 6a = 2(3a) = 2b.
Therefore, ∃b ∈ Z, x = 2b; hence, x ∈ h2i.
Put c = 2a.
x = 6a = 3(2a) = 3c.
Therefore, ∃c ∈ Z, x = 3c; hence, x ∈ h3i.
We now have x ∈ h2i and x ∈ h3i; hence, x ∈ h2i ∩ h3i.
Therefore, h6i ⊆ h2i ∩ h3i.
Therefore, h2i ∩ h3i = h6i.
Extended Set Operations
Notice that both the union and the intersection are associative operations. That is, A ∩ (B ∩
C) = (A ∩ B) ∩ C, and A ∪ (B ∪ C) = (A ∪ B) ∪ C. For this reason, the notation A ∪ B ∪ C
can be used without ambiguity. Further, given a sequence of four sets A, B, C, and D, we
can use the notation A ∩ B ∩ C ∩ D to denote (A ∩ B ∩ C) ∩ D. The same is true for unions,
and it is easy to see the idea of an intersection or union of any number of sets emerging
from this convention. In fact, we can make this precise by using a recursive definition, as
discussed in section 1.2. However, we will quickly run out of letters if we choose to denote
our sets by uppercase letters A, B, C, etc. A better choice of notation would be something
along the lines of A1 , A2 , A3 , etc. To make this notation more precise, we denote by (Ak )k∈N
a sequence of sets. Formally, this represents a correspondence between natural numbers
and sets. That is, to each natural number k corresponds a set Ak . An example of such a
sequence is given by: ∀k ∈ N, Ak = [0, k). In this case, to each k ∈ N corresponds the real
interval [0, k). Thus A1 = [0, 1), A2 = [0, 2), and so on. With this idea in mind, we can
extend the definitions of union and intersection to any finite number of sets.
138 CHAPTER 2. SETS
Definition 2.1.2. Let (Ak )k∈N be a sequence of sets. Define

1 n+1
 n 
[ [ [ 
Ak = A1 and for each n ∈ N, Ak =  Ak  ∪ An+1 .
k=1 k=1 k=1
Define
1 n+1
 n 
\ \ \ 
Ak = A1 and for each n ∈ N, Ak =  Ak  ∩ An+1 .

k=1 k=1 k=1
Since these extensions of the union and intersection operations are defined recursively,
proofs involving the union or intersection of a sequence of sets very often require induction.
We give the following two proofs as examples.
Proposition 2.1.14.
n
[ m
[
Let (Ak )k∈N be a sequence of sets. ∀m, n ∈ N, if n ≤ m, then Ak ⊆ Ak .
k=1 k=1
Proof.
Let A = {x ∈ N | ∀n ∈ N, if n ≤ x, then nk=1 Ak ⊆ k=1
S Sx
Ak }.
Let n ∈ N.
Assume n ≤ 1.
Then n = 1; so nk=1 Ak = 1k=1 Ak , and hence nk=1 Ak ⊆ 1k=1 Ak .
S S S S
Sn S1
Therefore, if n ≤ 1, then k=1 Ak ⊆ k=1 Ak .
Therefore, ∀n ∈ N, if n ≤ 1, then nk=1 Ak ⊆ 1k=1 Ak .
S S
That is, 1 ∈ A.
Let m ∈ N and assume m ∈ A.
Then ∀n ∈ N, if n ≤ m, then nk=1 Ak ⊆ m
S S
k=1 Ak .
Let n ∈ N.
Assume n ≤ m + 1.
We consider two cases: n < m + 1 and n = m + 1.
Case 1: n < m + 1.
In this case, n ≤ m, hence nk=1 Ak ⊆ m
S S
k=1 Ak .
Since k=1 Ak ⊆ Am+1 ∪ k=1 Ak = k=1 Ak , we have
Sm Sm Sm+1
Sn Sm+1
k=1 Ak ⊆ k=1 Ak by transitivity.
Case 2: n = m + 1.
In this case, nk=1 Ak = m+1 Ak ; thus nk=1 Ak ⊆ m+1
S S S S
Sn k=1 S k=1 Ak .
In both cases, we have k=1 Ak ⊆ m+1 k=1 Ak .
Therefore, if n ≤ m + 1, then nk=1 Ak ⊆ m+1
S S
k=1S Ak .
Therefore, ∀n ∈ N, if n ≤ m + 1, then k=1 Ak ⊆ m+1
Sn
k=1 Ak .
That is, m + 1 ∈ A.
Therefore, ∀m ∈ N, if m ∈ A, then m + 1 ∈ A.
Hence, by the PMI, N ⊆ A.
Therefore, ∀m, n ∈ N, if n ≤ m, then nk=1 Ak ⊆ m
S S
k=1 Ak .
Proposition 2.1.15.
n
\
Let (Ak )k∈N be a sequence of sets. ∀m, n ∈ N, if m ≤ n, then Ak ⊆ Am .
k=1
Proof.
Let m ∈ N.
Let A = {x ∈ Z≥m | k=1
Tx
Ak ⊆ Am }.
To show that m ∈ A, we consider two cases: m = 1 and m > 1.
Case 1: m = 1.
k=1 Ak = k=1 Ak = A1 = Am .
T1
In this case, we have m
T
Tm
Therefore, k=1 Ak ⊆ Am .
Case 2: m > 1.
Then m − 1 > 0; hence m − 1 ∈ N.
We then have m A = Am ∩ m−1
T T
Ak .
Tm−1k=1 k k=1 T
Since Am ∩ k=1 Ak ⊆ Am , we have m k=1 Ak ⊆ Am .
In both cases, m
T
k=1 Ak ⊆ Am ; hence m ∈ A.
Let n ∈ Z≥m .
Assume n ∈ A.
Then nk=1 Ak ⊆ Am .
T
Since n+1 A = An+1 ∩ nk=1 Ak ⊆ nk=1 Ak ,
T T T
Tn+1k
k=1
We have k=1 Ak ⊆ Am by transitivity.
Hence, n + 1 ∈ A.
Therefore, if n ∈ A, then n + 1 ∈ A.
Therefore, ∀n ∈ Z≥m , if n ∈ A, then n + 1 ∈ A.
Hence, by the PMI, Z≥m ⊆ A.
n
\
Therefore, ∀m, n ∈ N, if m ≤ n, then Ak ⊆ Am .
k=1
140 CHAPTER 2. SETS
Exercises 2.1.
Let A, B, C, and D be sets (assume the elements of (b) If A ∪ E = A for all sets E, then A = U.
these sets belong to a common universe of discourse (c) If A ∩ E = E for all sets E, then A = U.
U). Prove the following propositions.
(d) If E ⊆ A for all sets E, then A = U.
1. U \ A = Ac .
19. If A ∩ B = ∅ and A ∪ B = U, then A = Bc .
2. A \ B = A ∩ B .c
20. If A ⊆ B \ A, then A = ∅.
3. (a) If A ⊆ B, then A ∩ B = A.
21. If U \ A ⊆ A, then A = U.
(b) If A ∩ B = A, then A ∪ B = B.
22. If A ∩ B ⊆ A \ B, then A ∩ B = ∅.
(c) If A ∪ B = B, then A \ B = ∅.
(d) If A \ B = ∅, then A ⊆ B. 23. If C ⊆ A ∪ B, then C \ A ⊆ B.
4. (a) If A ∩ B = ∅, then A ⊆ Bc . 24. If C ⊆ A ∪ Bc , then (C \ A) ∩ B = ∅.

(b) If A ⊆ Bc , then (A ∪ B) \ B = A. 25. If A ∩ B ∩ C = ∅, then C ⊆ Ac ∪ Bc .
(c) If (A ∪ B) \ B = A, then B ⊆ Ac .
26. If A ⊆ B and B ∩ C = ∅, then A ∩ C = ∅.
(d) If B ⊆ Ac , then A \ B = A.
(e) If A \ B = A, then B \ A = B.
Let (Ak )k∈N be a sequence of sets, and let B be a set.
(f) If B \ A = B, then A ∩ B = ∅. Prove the following propositions.
5. (A \ B) \ C = (A \ C) \ (B \ C). n
[
27. ∀m, n ∈ N, if m ≤ n, then Am ⊆ Ak .
6. A \ (A \ B) = A ∩ B. k=1
m n
7. A \ (B ∪ C) = (A \ B) ∩ (A \ C). \ \
28. ∀m, n ∈ N, if n ≤ m, then Ak ⊆ Ak .
8. A \ (B ∩ C) = (A \ B) ∪ (A \ C). k=1 k=1
n n
9. A = (A \ B) ∪ (A ∩ B).
\ \
29. ∀n ∈ N, B ∪ Ak = (B ∪ Ak ).
k=1 k=1
10. (A \ B) ∪ (B \ A) = (A ∪ B) \ (A ∩ B).
n
[ n
[
11. If A ⊆ B, then A ∪ C ⊆ B ∪ C. 30. ∀n ∈ N, B ∩ Ak = (B ∩ Ak ).
k=1 k=1
12. If A ⊆ B, then A \ C ⊆ B \ C.
n
[ n
\
13. if A ⊆ B then C \ B ⊆ C \ A. 31. ∀n ∈ N, B \ Ak = (B \ Ak ).
k=1 k=1
14. If A ⊆ B and C ⊆ D, then A ∩ C ⊆ B ∩ D. n n
\ [
15. If A ⊆ B and C ⊆ D, then A \ D ⊆ B \ C. 32. ∀n ∈ N, B \ Ak = (B \ Ak ).
k=1 k=1
16. (a) If A ∩ E ⊆ B ∩ E for all sets E, then A ⊆ B.
[ c \
 n  n
(b) If A ∪ E ⊆ B ∪ E for all sets E, then A ⊆ B. 33. ∀n ∈ N,  Ak  = (Ak )c .
k=1 k=1
(c) If A \ E ⊆ B \ E for all sets E, then A ⊆ B.
\ c [
 n  n
(d) If E \ B ⊆ E \ A for all sets E, then A ⊆ B. 34. ∀n ∈ N,  Ak  =
 (Ak )c .
17. (a) If A ∩ E = ∅ for all sets E, then A = ∅. k=1 k=1
n
(b) If A ∩ E = A for all sets E, then A = ∅. [
35. If B ⊆ A1 , then ∀n ∈ N, B ⊆ Ak .
(c) If A ∪ E = E for all sets E, then A = ∅. k=1
(d) If A ⊆ E for all sets E, then A = ∅. n
\
36. If A1 ⊆ B, then ∀n ∈ N, Ak ⊆ B.
18. (a) If A ∪ E = U for all sets E, then A = U. k=1
n
\ 46. (−2, 1] ∩ [0, 3) = [0, 1].
37. If ∀m ∈ N, B ⊆ Am , then ∀n ∈ N, B ⊆ Ak .
k=1 47. (−2, 1] \ [0, 3) = (−2, 0).
n
48. [0, 3) \ (−2, 1] = (1, 3).
[
38. If ∀m ∈ N, Am ⊆ B, then ∀n ∈ N, Ak ⊆ B.
k=1 49. (−∞, 3) \ (−2, 1] = (−∞, −2] ∪ (1, 3).
39. If ∀k ∈ N, Ak+1 ⊆ Ak , then ∀m, n ∈ N, if m ≤ n,
50. (−∞, 1) ∩ (−2, 3] = (−2, 1).
then An ⊆ Am .
51. h5i ∩ h6i = h30i.
40. If ∀k ∈ N, Ak ⊆ Ak+1 , then ∀m, n ∈ N, if m ≤ n,
then Am ⊆ An . 52. h5i ∩ h2i = h10i.
n
\ 53. ∀a, b ∈ Z, if gcd(a, b) = 1, then hai ∩ hbi = habi.
41. If ∀m ∈ N, Am+1 ⊆ Am , then ∀n ∈ N, Ak = An .
k=1 54. h12i ∩ h18i = h36i.
n
[ 55. Let A = {x ∈ Z | ∃t ∈ Z, x = 15t + 7}, B = {x ∈
42. If ∀m ∈ N, Am ⊆ Am+1 , then ∀n ∈ N, Ak = An .
Z | ∃s ∈ Z, x = 3s + 1}, and C = {x ∈ Z | ∃r ∈ Z, x =
k=1
5r + 2}. Then A = B ∩ C.
n
56. Let A = {x ∈ Z | ∃t ∈ Z, x = 6t + 5}, B = {x ∈
[
43. If ∀m ∈ N, Am+1 ⊆ Am , then ∀n ∈ N, Ak = A1 .
k=1 Z | ∃s ∈ Z, x = 8s + 3}, and C = {x ∈ Z | ∃r ∈
n
Z, x = 24r + 11}. Then A ∩ B = C.
\
44. If ∀m ∈ N, Am ⊆ Am+1 , then ∀n ∈ N, Ak = A1 . n
[
k=1 57. ∀n ∈ N, (0, k] = (0, n].
k=1
Prove the following propositions. n
\ 1 1
58. ∀n ∈ N, (0, ] = (0, ].
45. (−2, 1] ∪ [0, 3) = (−2, 3). k=1
k n
142 CHAPTER 2. SETS
2.2 Real Intervals
Our two most frequently used examples of sets are the real intervals and the ideals of the
integers. The reason for this focus is to encourage practice with the techniques associated
with the order relation in the real numbers and the divides relation in the integers.
These relations and the techniques we use to study them are essential in many areas of
mathematics and indispensable for any future student of analysis or algebra. To explore
these two examples in more detail, we devote the current section to examining properties
of real intervals and the section that follows to the ideals of the integers.
To begin, we distinguish between two types of real intervals: Those that are bounded, and
those that are unbounded. We dealt briefly with upper and lower bounds in the section on
the well-ordering property, since the well-ordering property states that sets of integers that
are bounded below have and smallest element and those that are bounded above have a
largest element. Although the well-ordering property does not hold in the real numbers,
the concept of upper and lower bounds is still very useful. For the sake of building our
inquiry on firm ground, we give the precise definition of a bounded set:
Definition 2.2.1. For a subset A of the real numbers R, A is bounded

below means
∃a ∈ R, ∀x ∈ A, a ≤ x.
A is bounded above means
∃a ∈ R, ∀x ∈ A, x ≤ a.
A is bounded means A is both bounded above and bounded below.
When a set is bounded, it will generally have more than one upper bound and more than one
lower bound. This allows us some flexibility in selecting upper and lower bounds we wish
to focus on, for any given bounded set. The following proposition shows that it is always
possible to select the upper and lower bounds to be of the same magnitude. That is, if A
is any bounded set, we may select k ∈ R such that A is bounded above by k and bounded
below by −k. It is left as an exercise to show that due to the Archimedean property, we may
also ask that such a k be a natural number.
2.2. REAL INTERVALS 143
Proposition 2.2.1.
For every subset A of R, A is bounded if and only if ∃k ∈ R, ∀x ∈ A, |x| ≤ k.
Proof.
Let A ⊆ R.
Assume A is bounded.
Choose a ∈ R such that ∀x ∈ A, a ≤ x.
Choose b ∈ R such that ∀x ∈ A, x ≤ b.
Choose k = max(−a, b).
Then −a ≤ k and b ≤ k.
This can be written as −k ≤ a and b ≤ k.
Let x ∈ A.
Then −k ≤ a ≤ x ≤ b ≤ k.
By transitivity, −k ≤ x ≤ k; hence |x| ≤ k.
Therefore, ∃k ∈ R, ∀x ∈ A, |x| ≤ k.
Conversely, assume ∃k ∈ R, ∀x ∈ A, |x| ≤ k.
For such a k, we have ∀x ∈ A, −k ≤ x and ∀x ∈ A, x ≤ k.
Therefore, A is bounded below by −k and above by k.
Thus A is bounded.
Therefore, A is bounded if and only if ∃k ∈ R, ∀x ∈ A, |x| ≤ k.
Therefore, for any A ⊆ R, A is bounded if and only if ∃k ∈ R, ∀x ∈ A, |x| ≤ k.
We classify the nine types of intervals into those that are bounded and those that are not
bounded:
For a, b ∈ R,
Bounded Intervals Unbounded Intervals

(a, b) = {x ∈ R | a < x and x < b} (a, ∞) = {x ∈ R | a < x}
(a, b] = {x ∈ R | a < x and x ≤ b} [a, ∞) = {x ∈ R | a ≤ x}
[a, b) = {x ∈ R | a ≤ x and x < b} (−∞, b) = {x ∈ R | x < b}
[a, b] = {x ∈ R | a ≤ x and x ≤ b} (−∞, b] = {x ∈ R | x ≤ b}
(−∞, ∞) = R.
We note that all of the bounded intervals can be viewed as intersections of the unbounded
intervals. That is, (a, b) = (a, ∞) ∩ (−∞b), [a, b) = [a, ∞) ∩ (−∞, b), (a, b] = (a, ∞) ∩
(−∞, b], and [a, b] = [a, ∞) ∩ (−∞, b]. The proofs of each of these are similar to one
another, and so we only present the details for the first. To produce the other three results,
the proof need only be modified by replacing < by ≤ wherever appropriate.
144 CHAPTER 2. SETS
Proposition 2.2.2.
∀a, b ∈ R, (a, b) = (a, ∞) ∩ (−∞, b).
Proof.
Let a, b ∈ R.
Let x ∈ (a, b).
Then a < x and x < b.
Since a < x, we have x ∈ (a, ∞).
Since x < b, we have x ∈ (−∞, b).
Therefore, x ∈ (a, ∞) ∩ (−∞, b).
Therefore, (a, b) ⊆ (a, ∞) ∩ (−∞, b).
Conversely, let x ∈ (a, ∞) ∩ (−∞, b).
Then x ∈ (a, ∞) and x ∈ (−∞, b).
Therefore, a < x and x < b; hence x ∈ (a, b).
Thus, (a, ∞) ∩ (−∞, b) ⊆ (a, b); hence (a, b) = (a, ∞) ∩ (−∞, b).
Therefore, ∀a, b ∈ R, (a, b) = (a, ∞) ∩ (−∞, b).
It is worth calling attention to the fact that the two directions in the above proof consist of
exactly the same series of steps, performed in reverse order. In the rare occasions where
this occurs, we can restructure the proof to execute both directions simultaneously. Here is
an alternate version of the proof above:
Proof.
Let a, b ∈ R.
x ∈ (a, b) if and only if a < x and x < b.
a < x and x < b if and only if x ∈ (a, ∞) and x ∈ (−∞, b).
x ∈ (a, ∞) and x ∈ (−∞, b) if and only if x ∈ (a, ∞) ∩ (−∞, b).
Therefore, x ∈ (a, b) if and only if x ∈ (a, ∞) ∩ (−∞, b).
Thus, (a, b) = (a, ∞) ∩ (−∞, b).
Therefore, ∀a, b ∈ R, (a, b) = (a, ∞) ∩ (−∞, b).
Here is another example of intervals of a certain type intersecting to form other intervals.
In this proof, we use the following two theorems:
∀x, y, a ∈ R, a < min(x, y) if and only if a < x and a < y.
∀x, y, a ∈ R, max(x, y) < a if and only if x < a and y < a.

These are exercises 118 and 117 from section 1.1.
Proposition 2.2.3.
∀a, b, c, d ∈ R, (a, b) ∩ (−∞, c) = (a, u) where u = min(b, c),

and (a, b) ∩ (d, ∞) = (v, b) where v = max(a, d).
Proof.
Let a, b, c, d ∈ R.
Set u = min(b, c), and let x ∈ R.
x ∈ (a, b) ∩ (−∞, c) if and only if a < x and x < b and x < c.
a < x and (x < b and x < c) if and only if a < x and x < min(b, c) = u.
a < x and x < u if and only if x ∈ (a, u).
Therefore, x ∈ (a, b) ∩ (−∞, c) if and only if x ∈ (a, u).
Therefore, (a, b) ∩ (−∞, c) = (a, u), where u = min(b, c).
Next, set v = max(a, d), and let x ∈ R.
x ∈ (a, b) ∩ (d, ∞) if and only if a < x and x < b and d < x.
(a < x and d < x) and x < b if and only if v = max(a, d) < x and x < b.
v = max(a, d) < x and x < b if and only if x ∈ (v, b).
Therefore, x ∈ (a, b) ∩ (d, ∞) if and only if x ∈ (v, b).
Therefore, (a, b) ∩ (d, ∞) = (v, b), where v = max(a, d).
Now, ∀a, b, c, d ∈ R, (a, b) ∩ (−∞, c) = (a, u) where u = min(b, c),
and (a, b) ∩ (d, ∞) = (v, b) where v = max(a, d).
It is left as an exercise to show that the intersection of any two intervals (no matter the type)
will always produce another interval. However, the same cannot be said for the union or
the relative complement. For example, (0, 1) ∪ [2, 3) is not an interval, and (0, 3) \ [1, 2) is
not an interval (actually, (0, 1) ∪ [2, 3) = (0, 3) \ [1, 2)). For another example:
Proposition 2.2.4.
∀c, d ∈ R, R \ [c, d] = (−∞, c) ∪ (d, ∞).
Proof. Let c, d ∈ R.
Let x ∈ R \ [c, d].
Then x < [c, d].
This means that x < c or d < x.
In case x < c, we have x ∈ (−∞, c); hence x ∈ (−∞, c) ∪ (d, ∞).
In case d < x, we have x ∈ (d, ∞; so again x ∈ (−∞, c) ∪ (d, ∞).
Therefore, R \ [c, d] ⊆ (−∞, c) ∪ (d, ∞).
Conversely, let x ∈ (−∞, c) ∪ (d, ∞).
Then x ∈ (−∞, c) or x ∈ (d, ∞).
That is, x < c or d < x.
Therefore, it is not the case that c ≤ x and x ≤ d.
Hence, x < [c, d].
Therefore, x ∈ R \ [c, d].
Therefore, R \ [c, d] ⊆ (−∞, c) ∪ (d, ∞).
Thus, ∀c, d ∈ R, R \ [c, d] = (−∞, c) ∪ (d, ∞).
146 CHAPTER 2. SETS
There are some situations, however, when the relative complement of two intervals does
produce another interval. For example:
Proposition 2.2.5.
∀a, b, x, y ∈ R, if x ∈ (a, b) and b ∈ (x, y), then (a, b) \ (x, y) = (a, x].
Proof.
Let a, b, x, y ∈ R.
Assume x ∈ (a, b) and b ∈ (x, y).
Let t ∈ (a, b) \ (x, y).
Then t ∈ (a, b) and t < (x, y).
Since t ∈ (a, b), we have t < b.
Also, since b ∈ (x, y), we have b < y.
With t < b and b < y, we have by transitivity, t < y.
Now, since t < (x, y), we have t ≤ x or y ≤ t.
Since t < y, it is not the case that y ≤ t.
Therefore, it is the case that t ≤ x.
Finally, since t ∈ (a, b), we have a < t.
Now, a < t and t ≤ x. Hence t ∈ (a, x].
Therefore, (a, b) \ (x, y) ⊆ (a, x].
Conversely, let t ∈ (a, x].
Then a < t and t ≤ x.
Since x ∈ (a, b), we have x < b.
Now, since t ≤ x and x < b, we have by transitivity, t < b.
We now have a < t and t < b; hence t ∈ (a, b).
Further, since t ≤ x, it is not the case that x < t.
Therefore, t < (x, y).
We then have t ∈ (a, b) and t < (x, y). Hence t ∈ (a, b) \ (x, y).
Therefore, (a, x] ⊆ (a, b) \ (x, y); hence (a, b) \ (x, y) = (a, x].
Therefore, if x ∈ (a, b) and b ∈ (x, y), then (a, b) \ (x, y) = (a, x].
Q.E.D.
A remark about the use of Q.E.D.

Q.E.D. stands for the Latin ‘quod erat demonstrandum,’ meaning ‘that which was to have
been demonstrated.’ Since it is understood that the last line of a proper formal proof is
always an exact restatement of the theorem that was to be proven, the line Q.E.D. can
stand as a substitute for this last line. When reading a proof that ends with Q.E.D., we
mentally insert the exact statement of the theorem in place of the letters Q.E.D. (in the
previous exercise this would be ‘For all a, b, x, y ∈ R, if x ∈ (a, b) and b ∈ (x, y), then
(a, b) \ (x, y) = (a, x].’) We should then verify that this line follows from the proof that
appears above it. You may also come across the acronym Q.E.F., which stands for ‘quod
erat faciendum’ meaning ‘that which was to have been done.’ This is used in the same way
as Q.E.D. but for proofs that are pure constructions, such as proposition 1.1.10. However,
the preferred practice is always to include, as the last line of the proof, a full and exact
restatement of the property being proven.
Let us consider in more detail the criteria under which the relative complement or union of
two intervals is itself an interval. In fact, the case of relative complements is exactly the
same as the case of unions. The reason is that the relative complement of two intervals is
always equal to a union of two intervals. We will prove this in the special case (a, b) \ [c, d].
To simplify the proof, we will make use of some basic properties of sets from section 2.1.
In particular, we will use the fact that A\ B = A∩ Bc = A∩(U \ B), as well as the distributive
law A ∩ (B ∪ C) = (A ∩ B) ∪ (A ∩ C).
Proposition 2.2.6.
For all a, b, c, d ∈ R, (a, b) \ [c, d] = (a, u) ∪ (v, b), where u = min{b, c}

and v = max{a, d}.
Proof.
Set u = min{b, c} and v = max{a, d}.
(a, b) \ [c, d] = (a, b) ∩ (R \ [c, d])

= (a, b) ∩ ((−∞, c) ∪ (d, ∞)) by proposition 2.2.4
= ((a, b) ∩ (−∞, c)) ∪ ((a, b) ∩ (d, ∞))
= (a, u) ∪ (v, b) by proposition 2.2.3.
Therefore, (a, b) \ [c, d] = (a, u) ∪ (v, b), where u = min{b, c} and v = max{a, d}.
Therefore, for all a, b, c, d ∈ R, (a, b) \ [c, d] = (a, u) ∪ (v, b), where u = min{b, c}
and v = max{a, d}.
Note that in the above proof, we could have one or both of the intervals (a, u) and (v, b)
being empty. For example, the above proposition tells us that
(1, 3) \ [2, 5] = (1, 2) ∪ (5, 3) = (1, 2) ∪ ∅ = (1, 2).
This is indeed the proper result.
We now consider the circumstances under which the union of two intervals is itself
an interval. The answer is that the two intervals must overlap. We prove this in the case of
bounded open intervals:
148 CHAPTER 2. SETS
Proposition 2.2.7.
For all a, b, c, d ∈ R, if (a, b) ∩ (c, d) , ∅, then (a, b) ∪ (c, d) = (u, v), where u = min(a, c)
and v = max(b, d).
Proof.
Assume (a, b) ∩ (c, d) , ∅.
Set u = min(a, c) and v = max(b, d), and let x ∈ (a, b) ∪ (c, d).
Then x ∈ (a, b) or x ∈ (c, d).
In case x ∈ (a, b), we have u ≤ a < x < b ≤ v; hence u < x < v.
In case x ∈ (c, d), we have u ≤ c < x < d ≤ v and again u < x < v.
Therefore, x ∈ (u, v).
Therefore, (a, b) ∪ (c, d) ⊆ (u, v)
Conversely, let x ∈ (u, v).
Since (a, b) ∩ (c, d) , ∅, choose an element t ∈ (a, b) ∩ (c, d).
This gives us a < t < b and c < t < d. Therefore, c < b and a < d.
Case 1: u = a and v = b
Then x ∈ (a, b), so x ∈ (a, b) ∪ (c, d).
Case 2: u = a and v = d
Then a < x < d.
If x < b, then a < x < b, in which case x ∈ (a, b).
Hence, in the case where x < b, we have x ∈ (a, b) ∪ (c, d).
Otherwise, if b ≤ x, then since c < b we have c < x by transitivity.
In this case, c < x < d, which means x ∈ (c, d).
So, again x ∈ (a, b) ∪ (c, d).
Case 3: u = c and v = b
Then c < x < b.
If x < d, then c < x < d; hence x ∈ (c, d).
In this case, x ∈ (a, b) ∪ (c, d).
Otherwise, if d ≤ x, then since a < d we have a < x by transitivity.
In this case, a < x < b; hence x ∈ (a, b).
So, again x ∈ (a, b) ∪ (c, d).
Case 4: u = c and v = d
Then x ∈ (c, d), so x ∈ (a, b) ∪ (c, d).
Therefore, (u, v) ⊆ (a, b) ∪ (c, d); hence (a, b) ∪ (c, d) = (u, v).
Therefore, if (a, b) ∩ (c, d) , ∅, then (a, b) ∪ (c, d) = (u, v), where u = min(a, c)
and v = max(b, d).
Q.E.D.
Nesting of Bounded Intervals
It is a fairly simple task to show that the four different types of bounded intervals are nested
in the following ways:
(a, b) ⊆ (a, b] ⊆ [a, b] and (a, b) ⊆ [a, b) ⊆ [a, b].
The proofs of these are left as exercises. It is also trivially true that
(a, b) ⊆ (a, b) ⊆ [a, b] and (a, b) ⊆ [a, b] ⊆ [a, b].
Moreover, it is useful to note that the bounded intervals are the only sets nested between
(a, b) and [a, b]. That is, the only sets J for which (a, b) ⊆ J ⊆ [a, b] are the four bounded
intervals (a, b), (a, b], [a, b), and [a, b]:
150 CHAPTER 2. SETS
Theorem 2.2.8.
∀a, b ∈ R, If J is a non-empty subset of R and (a, b) ⊆ J ⊂ [a, b], then J = (a, b) or

J = (a, b] or J = [a, b] or J = [a, b).
Proof.
Let a, b ∈ R.
Assume J is a non-empty subset of R and (a, b) ⊆ J ⊂ [a, b].
Case 1: a < J and b < J.
Let x ∈ J.
Then x ∈ [a, b] since J ⊆ [a, b]; hence a ≤ x ≤ b.
Since a, b < J, we have x , a and x , b.
Therefore, a < x < b; hence x ∈ (a, b).
Therefore, J ⊆ (a, b); hence J = (a, b).
Case 2: a, b ∈ J.
Let x ∈ [a, b].
Then either x = a, or x = b, or x ∈ (a, b).
In all cases, x ∈ J, since a ∈ J, b ∈ J, and (a, b) ⊆ J.
Therefore, [a, b] ⊆ J; hence J = [a, b].
Case 3: a ∈ J and b < J.
For x ∈ [a, b), we have either x = a or x ∈ (a, b).
In both of these cases, x ∈ J, since a ∈ J and (a, b) ⊆ J.
Therefore, [a, b) ⊆ J.
Conversely, for x ∈ J, we have x ∈ [a, b] and hence a ≤ x ≤ b.
Since b < J, we have x , b, hence a ≤ x < b.
Therefore, x ∈ [a, b).
Therefore, J ⊆ [a, b); hence J = [a, b).
Case 4: a < J and b ∈ J.
Letting x ∈ (a, b] gives us x = b or x ∈ (a, b).
In both of these cases, x ∈ J.
Therefore, (a, b] ⊆ J.
Conversely, letting x ∈ J, we have x ∈ [a, b] and hence a ≤ x ≤ b.
Since a < J, x , a; hence a < x ≤ b.
Therefore, x ∈ (a, b].
So, in this case, J = (a, b].
Therefore, if J is a non-empty subset of R and (a, b) ⊆ J ⊂ [a, b],
then J = (a, b) or J = (a, b] or J = [a, b] or J = [a, b).
Q.E.D.
We have results similar to the above for the unbounded intervals: Also true are:
if (a, ∞) ⊆ J ⊆ [a, ∞), then J = (a, ∞) or J = [a, ∞)
and
if (−∞, b) ⊆ J ⊆ (−∞, b], then J = (−∞, b) or J = (−∞, b).
The proofs of these are left as exercises.
Characterization of Intervals
One convenient characteristic of the intervals is that since both the < relation and the ≤
relation satisfy the same axioms (transitivity, preserving addition, and preserving multipli-
cation by positives), we see that the different types of intervals behave in essentially the
same ways. That is, a result such as
∀a, b, x, y ∈ R, if a ≤ x and y ≤ b, then (x, y) ⊆ (a, b).
also holds for other types of intervals. For example,
∀a, b, x, y ∈ R, if a ≤ x and y ≤ b, then [x, y) ⊆ [a, b),
and
∀a, b, x, y ∈ R, if a ≤ x, then [x, ∞) ⊆ [a, ∞)
are also true (the proofs of these are left as exercises).
One downside of this is that to prove these statements for all nine types of interval would
require nine separate proofs, the details of which would be virtually identical to one another.
To circumvent this monotony, we want a characterization of what it means to be an interval.
That is, we look for one single property that is held by all nine types of intervals and no
other sets. One such common property that is shared by all intervals, no matter the type, is
that intervals (and no other sets) contain all of the real numbers that fall between any two
of their elements. More precisely, for a subset J ⊆ R, J is an interval if and only if
∀x, y ∈ J, ∀z ∈ R, if x < z < y, then z ∈ J.
The proof of this theorem is not trivial, and it will take a fair amount of discussion to bring
to light.
However, it is quite easy to see how useful the above property can be. For example, consider
the proposition ‘the intersection of any two intervals (of any type) is itself an interval.’
Immediately, we see that to prove this from the definition of an interval would require
(9)(9) = 81 cases considering each of the different types that the two intervals might be.
It gets worse: Even if the types of the two intervals are fixed, the intersection can be of
several different types, depending on the circumstances. For example, (0, 2]∩[1, 3) = [1, 2],
(1, 3] ∩ [0, 2) = (1, 2), (0, 3] ∩ [1, 2) = [1, 2), (1, 2] ∩ [0, 3) = (1, 2]. We see that even in
the case where the types of intervals are fixed as (a, b] ∩ [x, y), the intersection might be
any of the four types of bounded interval. Hence, our 81 initial cases will contain further
subcases. We have no intention of working through the proof as such, or even asking the
reader to do so. Rather, to prove the proposition ‘the intersection of any two intervals (of
any type) is itself an interval,’ we can choose merely to show that the intersection of any
two sets satisfying the property
∀x, y ∈ J, ∀z ∈ R, if x < z < y, then z ∈ J
will itself satisfy this property. This is enough, to prove the desired proposition, since the
intervals are exactly the sets that satisfy this property. Note: The proof that the intersection
of intervals is an interval is left as an exercise. The reader is strongly encouraged to make
use of the characterization of intervals as the sets satisfying the property ∀x, y ∈ J, ∀z ∈ R,
if x < z < y, then z ∈ J.
152 CHAPTER 2. SETS
The remainder of the section will be devoted to proving the theorem: For a subset J ⊆ R, J
is an interval if and only if
∀x, y ∈ J, ∀z ∈ R, if x < z < y, then z ∈ J.
We begin with the forward direction: If J is an interval, then ∀x, y ∈ J, ∀z ∈ R, if x < z < y,
then z ∈ J. We make use of the nesting properties of intervals described in the preceding
section.
Theorem 2.2.9.
If J is an interval of R (one of the nine types), then ∀x, y ∈ J, ∀z ∈ R, if x < z < y, then
z ∈ J.
Proof.
Let J be an interval of R.
Then J is one of 9 forms (a, b), [a, b), etc.,
We consider the four bounded intervals first:
For some a, b ∈ R, J = [a, b], J = (a, b], J = [a, b), or J = (a, b).
Choose a, b ∈ R so that J is given by one of the above four cases.
In all cases, we have (a, b) ⊆ J ⊆ [a, b].
Let x, y ∈ J and z ∈ R. Assume x < z < y.
Since J ⊆ [a, b], we have x, y ∈ [a, b].
We then have a ≤ x < z < y ≤ b, so a < z < b by transitivity.
Then z ∈ (a, b); hence z ∈ J.
Therefore, ∀x, y ∈ J, ∀z ∈ R, if x < z < y, then z ∈ J.
Next, we consider the two types of interval bounded below:
For some a ∈ R, J = (a, ∞) or J = [a, ∞).
Choose a ∈ R for which J is one of these two intervals.
In either case, we have (a, ∞) ⊆ J ⊆ [a, ∞).
Since J ⊆ [a, ∞), we have x ∈ [a, ∞).
Then a ≤ x < z < y; hence a < z.
This means z ∈ (a, ∞); hence z ∈ J.
Next, if J is bounded above but not below:
For some b ∈ R, J = (−∞, b) or J = (−∞, b].
Choose such a b ∈ R.
In either case, we have (−∞, b) ⊆ J ⊆ (−∞, b].
Since J ⊆ (−∞, b], we have y ≤ b.
Then x < z < y ≤ b; hence z < b.
This means z ∈ (−∞, b); hence z ∈ J.
Finally, if J = (−∞, ∞), then J = R, and the result is trivially true.
Thus, if J is an interval of R, then ∀x, y ∈ J, ∀z ∈ R, if x < z < y, then z ∈ J.
The following quick but useful corollary is one half of a proposition stating that if I is an
interval and b < I, then b is either an upper or lower bound of I. Hence, all real numbers
will fall into at least one of these three categories: the upper bounds of I, the lower bounds
of I, and I itself. We give the upper bound case here. The lower bound case is an exercise.
Corollary 2.2.10.
Let I ⊆ R be an interval. For all b ∈ R, if b < I and ∃a ∈ I, a < b, then ∀x ∈ I, x < b.
Proof.
Let b ∈ R.
Suppose b < I, ∃a ∈ I, a < b, and ∃x ∈ I, b ≤ x.
Choose a ∈ I with a < b and x ∈ I with b ≤ x.
Since x ∈ I and b < I, we have b , x.
Therefore, a < b < x.
Since a, x ∈ I, we then have b ∈ I by theorem 2.2.9.
We now have the contradiction b ∈ I and b < I.
Therefore, if b < I and ∃a ∈ I, a < b, then ∀x ∈ I, x < b.
Therefore, for all b ∈ R, if b < I and ∃a ∈ I, a < b, then ∀x ∈ I, x < b.
Recall that our goal in this section is to characterize the intervals as follows: For J ⊆ R,
J is an interval if and only if ∀x, y ∈ J, ∀z ∈ R, if x < z < y, then z ∈ J.
Theorem 2.2.9 gives us one direction of this theorem, and so it only remains to show that
the converse also holds:
Theorem 2.2.11.
If J is a subset of R such that ∀x, y ∈ J, ∀z ∈ R, x < z < y implies z ∈ J, then J is an

interval. (This is the converse of Theorem 2.2.9.)
In the exercises that follow, the reader is expected to know and understand the statements of
both theorems 2.2.9 and 2.2.11 and to use these results to prove statements about intervals.
However, the proof of theorem 2.2.11 involves techniques that we have not sufficiently
practiced (in particular its use of the completeness axiom). As such, the reproduction of
this proof is beyond the expectations of a student in this course. Nevertheless, we have
no intention of leaving the reader speculating about a mysterious and supposedly difficult
proof. We include the proof here along with an explanation of the steps involved. The
reader is invited to read and verify the proof, paying close attention to the use of complete-
ness. The techniques involved can be compared to those used to prove the Archimedean
property (theorem 1.1.34) which also makes use of completeness.
154 CHAPTER 2. SETS
The Proof of Theorem 2.2.11
We of course begin the proof by assuming J is a subset of R satisfying the property that
∀x, y ∈ J, ∀z ∈ R, x < z < y implies z ∈ J. We must then show that J is an interval of one of
the nine types. Which type will naturally depend on whether J is bounded or unbounded,
and so we consider several different cases depending on the boundaries of the set J:
Proof.
Let J be a subset of R such that ∀x, y ∈ J, ∀z ∈ R, x < z < y implies z ∈ J.
Then, there are 5 cases:
1. J = ∅.
2. J is non-empty and bounded above and below.

3. J is non-empty and bounded below, unbounded above.
4. J is non-empty and bounded above, unbounded below.
5. J is non-empty and unbounded above and below.
We then exhaust these five cases one by one. The first case is easy:
Case 1: J = ∅.
Then J can be viewed as the degenerate interval (0, 0).
That is, since (0, 0) = ∅, we have J = (0, 0).
Therefore, J is an interval.
The second case requires more work. Since there are four types of bounded interval ((a, b),
(a, b], [a, b), and [a, b]), in the case where J is non-empty and bounded, we must prove that
J is of one of these four types. Due to the nesting of bounded intervals given in theorem
2.2.8, we can prove that J must be one of these four types by showing that
(a, b) ⊆ J ⊆ [a, b]
for some a, b ∈ R. The task then becomes to find real numbers a and b for which (a, b) ⊆
J ⊆ [a, b]. To find such a and b, we turn to the completeness axiom.
The completeness axiom

If A and B are non-empty subsets of R such that for all a ∈ A and all b ∈ B, a < b, then
there is an element x ∈ R such that for all a ∈ A, a ≤ x and for all b ∈ B, x ≤ b.
Recall that the completeness axiom ensures that if we cut the real line into two pieces A
and B, the cut must fall on a real number. That is, there is a real number x lying between
all the points in piece A and the points in piece B.
To prove the existence of real numbers a and b for which (a, b) ⊆ J ⊆ [a, b], we make
two cuts: One separating J from its lower bounds, and another separating J from its upper
bounds. That is, if A is the set of all lower bounds of J and B is the set of all upper bounds
of J, we apply the completeness axiom to create the following pictures:
Combining these pictures, we have
For the a and b chosen in this way, we can then prove that (a, b) ⊆ J ⊆ [a, b]; hence J must
be an interval of one of the four bounded types. The details of this case are as follows:
156 CHAPTER 2. SETS
Case 2: J is non-empty and bounded above and below.

Let A be the set of lower bounds of J
and let B be the set of upper bounds of J; that is, let
A = {` ∈ R : ∀x ∈ J, ` ≤ x} and B = {u ∈ R : ∀x ∈ J, x ≤ u}.
Then, for all ` ∈ A and all x ∈ J, ` ≤ x.

Therefore, by completeness we can choose a ∈ R such that
∀` ∈ A, ` ≤ a and ∀x ∈ J, a ≤ x.
In the same way, ∀x ∈ J, ∀u ∈ B, x ≤ u, so we can choose b ∈ R such that
∀x ∈ J, x ≤ b and ∀u ∈ B, b ≤ u.
We will prove that (a, b) ⊆ J ⊆ [a, b].

Indeed, we established that ∀x ∈ J, a ≤ x and ∀x ∈ J, x ≤ b,
so ∀x ∈ J, a ≤ x ≤ b; that is, J ⊂ [a, b].
It remains to show that (a, b) ⊆ J.
To this end, let t ∈ (a, b).
Then a < t and t < b.
But, ∀` ∈ A, ` ≤ a, so t < A.
Since t is not a lower bound for J, we can choose x ∈ J with x < t,
Similarly, since t < b, t is not an upper bound for J,
so we can choose y ∈ J with t < y.
Thus, x < t < y, with x, y ∈ J; hence, by hypothesis t ∈ J.
This shows for all t ∈ (a, b), t ∈ J. In other words, (a, b) ⊆ J.
In summary, (a, b) ⊆ J ⊆ [a, b].
Therefore, J = (a, b) or J = (a, b] or J = [a, b] or J = [a, b), by theorem 2.2.8.
Case 3 (J is non-empty and bounded below, unbounded above) is somewhat similar to case
2. In this case, we find a real number a for which (a, ∞) ⊆ J ⊆ [a, ∞); hence J = (a, ∞) or
J = [a, ∞). To find such an a, we apply completeness, making a cut between J and the set
A consisting of all lower bounds of J.
The details of this case are as follows:

Case 3: J is non-empty and bounded below, unbounded above.

As in Case 2, put A = {` ∈ R : ∀x ∈ J, ` ≤ x}.
Then, for all ` ∈ A and all x ∈ J, ` ≤ x.
By completeness we can choose a ∈ R such that
∀` ∈ A, ` ≤ a and ∀x ∈ J, a ≤ x.
We will show (a, ∞) ⊆ J ⊆ [a, ∞).

Indeed, since ∀x ∈ J, a ≤ x, we have J ⊆ [a, ∞).
Next, let t ∈ (a, ∞); that is, a < t.
As before, we have t < A, so we may choose x ∈ J, with x < t.
But J is not bounded above, so we can choose y ∈ J with t < y.
Thus, x, y ∈ J and x < t < y, so t ∈ J by hypothesis.
Therefore, (a, ∞) ⊆ J.
We have shown (a, ∞) ⊆ J ⊆ [a, ∞).
It follows that J = (a, ∞) or J = [a, ∞).
Thus, in case J is non-empty, bounded below and not above, J is an interval.
Case 4 (J is non-empty and bounded above, unbounded below) is very similar to case 3,
only using the set B of upper bounds of J to obtain a b such that J = (−∞, b) or J = (−∞, b].
Case 4: J is non-empty and bounded above, unbounded below.

This time, put B = {u ∈ R : ∀x ∈ J, x ≤ u}.
Then, for all x ∈ j and all u ∈ B, x ≤ u.
By completeness we can choose b ∈ R such that
∀x ∈ J, x ≤ b and ∀u ∈ B, b ≤ x.
We will show (−∞, b) ⊆ J ⊆ (−∞, b].

Indeed, since ∀x ∈ J, x ≤ b, we have J ⊆ (−∞, b].
Next, let t ∈ (−∞, b); hence, t < b.
We then have t < B, so we may choose y ∈ J, with t < y.
Further, since J is not bounded below, we can choose x ∈ J with x < t.
Thus, x, y ∈ J and x < t < y, so t ∈ J by hypothesis.
Therefore, (−∞, b) ⊆ J.
We have shown (−∞, b) ⊆ J ⊆ (−∞, b].
It follows that J = (−∞, b) or J = (−∞, b].
Finally, the case in which J is neither bounded above nor below is somewhat easier, in that
it does not require completeness. The details are as follows:
158 CHAPTER 2. SETS
Case 5: In case J is non-empty and neither bounded above nor bounded below, let t ∈
(−∞, +∞) = R.
Since J is not bounded below, we can choose x ∈ J with x < t.
Since J is not bounded above, we can choose y ∈ J with t < y.
Then, x, y ∈ J and x < t < y, so t ∈ J also.
This shows that t ∈ (−∞, ∞) ⊆ J.
But J is a subset of R = (−∞, +∞), so J = (−∞, ∞), an interval.
Thus, in all 5 cases J is an interval.
Therefore, if J is a subset of R such that ∀x, y ∈ J, ∀z ∈ R, x < z < y implies z ∈ J, then J
is an interval.
Exercises 2.2.
Prove the following propositions. 26. ∀a, b ∈ R, R \ (a, b) = (−∞, a] ∪ [b, ∞).
1. For every subset A of R, A is bounded above if and 27. ∀a, b, c ∈ R, [a, b] ∩ (−∞, c] = [a, u], where u =
only if ∃k ∈ N, ∀x ∈ A, x ≤ k. min(b, c).
2. For every subset A of R, A is bounded if and only if 28. ∀a, b, c ∈ R, [a, b] ∩ [c, ∞) = [v, b], where v =
∃k ∈ N, ∀x ∈ A, |x| ≤ k. max(a, c).
3. ∀a, b ∈ R, (a, b) ⊆ (a, b] ⊂ [a, b]. 29. ∀a, b, c, d ∈ R, [a, b] \ (c, d) = [a, u] ∪ [v, b], where
4. ∀a, b ∈ R, (a, b) ⊆ [a, b) ⊂ [a, b]. u = min(b, c) and v = max(a, d).
5. ∀a, b ∈ R, if b ≤ a, then (a, b] = ∅. (So the empty 30. ∀a, b ∈ R, [a, ∞) \ (b, ∞) = [a, b].
set is an interval.) 31. ∀a, b, c ∈ R, if (−∞, a) ∩ (b, c) , ∅, then (−∞, a) ∪
6. ∀a, b ∈ R, if b < a, then [a, b] = ∅. (b, c) = (−∞, u), where u = max(a, c).
7. ∀a, b ∈ R, if a = b, then [a, b] = {a}. 32. ∀a, b, c ∈ R, if (a, b) ∩ (c, ∞) , ∅, then (a, b) ∪
(c, ∞) = (v, b), where v = min(a, c).
8. ∀a, b ∈ R, if a ≤ b, then [a, b] = (a, b) ∪ {a, b}.
33. ∀a ∈ R, if J ⊆ R with (a, ∞) ⊆ J ⊆ [a, ∞), then
9. ∀x, y ∈ R, if (−∞, x) = (−∞, y), then x = y. J = (a, ∞) or J = [a, ∞).
10. ∀x, y ∈ R, if [x, ∞) = [y, ∞), then x = y. 34. ∀b ∈ R, if J ⊆ R with (−∞, b) ⊆ J ⊆ (−∞, b], then
11. ∀a, b, x, y ∈ R, if a ≤ x and y ≤ b, then (x, y) ⊆ J = (−∞, b) or J = (−∞, b].
(a, b).
Prove the following propositions using theorems 2.2.9
12. ∀a, x ∈ R, if a ≤ x, then [x, ∞) ⊆ [a, ∞).
and 2.2.11.
13. ∀a, b ∈ R≥0 , if [0, a) ⊆ [0, b), then a ≤ b.
35. Let I be an interval of R. For all a ∈ R, if a < I and
14. ∀a ∈ R, if (0, ∞) ⊆ [a, ∞), then a ≤ 0. ∃b ∈ I, a < b, then ∀x ∈ I, a < x.
15. ∀a, b ∈ R≥0 , if [0, a] = [0, b], then a = b. 36. Let A and B be intervals in R with A ∩ B = ∅. If
∃a ∈ A, ∃b ∈ B, a < b, then ∀x ∈ A, ∀y ∈ B, x < y.
16. ∀a, b ∈ R≥0 , if [0, a) = [0, b), then a = b.
37. The intersection of any two intervals is an interval.
17. ∀a ∈ R, if (0, ∞) ⊆ [a, ∞), then (0, ∞) ⊆ (a, ∞).
38. Let (Ik )k∈N be a sequence of intervals. Then ∀n ∈ N,
18. ∀a, b, c, d ∈ R, if (a, b) ⊆ [c, d], then (a, b) ⊆ (c, d). \n
Ik is an interval.
k=1
Prove the following propositions about unions, inter-
sections, and complements of intervals. Prove the following propositions characterizing
19. ∀a, b, x, y ∈ R, if x ∈ (a, b) and b ∈ (x, y), then bounded intervals.
(a, b) ∩ (x, y) = (x, b). 39. ∀a, b ∈ R, if a < b, then (a, b) = {x ∈ R | ∃t ∈
20. ∀a, b, x, y ∈ R, if x ∈ (a, b) and b ∈ (x, y), then (0, 1), x = (1 − t)a + tb}.
(a, b) ∪ (x, y) = (a, y). 40. ∀a, b ∈ R, if a ≤ b, then [a, b] = {x ∈ R | ∃t ∈
21. ∀a, b, c, d ∈ R, (a, b) ∩ (c, d) = (e, f ), where e = [0, 1], x = (1 − t)a + tb}.
max(a, c) and f = min(b, d). 41. ∀a, b ∈ R, if a , b, then (a, b) ∪ (b, a) = {x ∈
22. ∀a, b, c, d ∈ R, [a, b] ∩ [c, d] = [e, f ], where e = R | ∃t ∈ (0, 1), x = (1 − t)a + tb}.
max(a, c) and f = min(b, d). 42. ∀a, b ∈ R, [a, b] ∪ [b, a] = {x ∈ R | ∃t ∈ [0, 1], x =
23. ∀a, b ∈ R, (a, ∞) ∩ (b, ∞) = (c, ∞), where c = (1 − t)a + tb}.
max(a, b). 43. ∀a, b ∈ R, if a < b, then (a, b] = {x ∈ R | ∃t ∈
24. ∀a, b ∈ R, (−∞, a) ∩ (−∞, b) = (−∞, c), where (0, 1], x = (1 − t)a + tb}.
c = min(a, b). 44. ∀a, b ∈ R, if a < b, then [a, b) = {x ∈ R | ∃t ∈
25. ∀a ∈ R, R \ (a, ∞) = (−∞, a]. [0, 1), x = (1 − t)a + tb}.
160 CHAPTER 2. SETS
2.3 Ideals of the Integers
Recall that for a ∈ Z, the ideal of Z generated by a is the set of multiples of a. That is,
For a ∈ Z:
hai = {x ∈ Z | ∃t ∈ Z, x = at}.
These sets provide a convenient notation for expressing ideas related to the integers,
including such ideas as the greatest common divisor, least common multiple, and the
distinction between prime and composite integers. In many cases, the notation becomes
less cumbersome and the ideas become more lucid and manageable when expressed in
terms of ideals.
We first note that the set of all integers is an ideal. In fact, Z is the ideal generated
by 1.
Proposition 2.3.1.
Z = h1i.
Proof.
Since Z is the universe of discourse in which the ideals are defined, we have h1i ⊆ Z.
Conversely, let x ∈ Z.
Put t = x.
Then x = (1)x = (1)t.
Therefore, ∃t ∈ Z, x = (1)t.
Hence, x ∈ h1i.
Therefore, Z ⊆ h1i.
Hence Z = h1i.
One particular complication that is avoided by speaking in terms of ideals is the distinction
between positive and negative integers. Notice that gcd(15, 20) is the same as gcd(−15, 20),
gcd(−15, −20), and so on. The same is true for the least common multiple. Likewise,
we see that 5 is prime and −5 is also prime, in fact for a ∈ Z, a is prime if and only
if −a is prime. In general, we see that the concepts of prime, composite, gcd, and lcm
are unaffected by the sign of the integer. Considering positive and negative integers as
different cases will then only add needless repetition to our proofs. The reader may already
be familiar with the following notation:
Notation
The notation ‘x = ±y’ means ‘x = y or x = −y’.
2.3. IDEALS OF THE INTEGERS 161
This is often used when one wants to speak about results that are true for an integer and
its corresponding negative. For example, one may write ‘if x = ±y, then x2 = y2 ,’ rather
than ‘if x = y or x = −y then x2 = y2 .’ Another common way of simplifying this notation
is to write |x| = |y|, which we have seen is true if and only if x = ±y. Our example could
then be written as ‘if |x| = |y|, then x2 = y2 .’ This notation may get us out of writing a long
series of disjunctions when we express the result, but to ensure that the result holds for both
the number and its negative will still require multiple cases in its proof. Thus it is only a
shorthand notation used to state the result; it does not add clarity to an argument in support
of the result. On the other hand, when we speak in terms of ideals, the issue of positives
and negatives is avoided altogether. This is due to the following theorem:
Theorem 2.3.2.
For all x, y ∈ Z, hxi = hyi if and only if x = ±y.
Proof.
Let x, y ∈ Z.
Assume hxi = hyi.
Since x = 1(x), we have x ∈ hxi.
Therefore, x ∈ hyi.
Similarly, we have y ∈ hxi.
It follows that ∃a ∈ Z, x = ay and ∃b ∈ Z, y = bx.
Therefore, x = abx; hence x(1 − ab) = 0
By axiom FZ, we then have either x = 0 or 1 − ab = 0.
Case 1: x = 0.
Then y = bx = b(0) = 0.
In this case, we have x = y; hence x = ±y.
Case 2: 1 − ab = 0.
Therefore, ab = 1; hence |a| ≤ 1 by proposition 1.2.8.
Further, since ab = 1 , 0, we have a , 0 and hence |a| , 0.
We now have, 0 < |a| ≤ 1; hence |a| = 1.
Therefore, a = 1 or a = −1.
Since x = ay, we then have either x = (1)y or x = (−1)y.
Therefore, x = ±y.
Therefore, if hxi = hyi, then x = ±y.
Conversely, assume x = ±y.
In the case x = y we have hxi = hyi,
hence it only remains to consider the case x = −y.
Let a ∈ hxi.
Then ∃t ∈ Z, a = xt. Choose such a t.
Put s = −t.
a = xt = (−y)t = (−t)y = sy.
Therefore, ∃s ∈ Z, a = sy.
Therefore, a ∈ hyi.
Hence, hxi ⊆ hyi.
Similarly, since y = −x, we have hyi ⊆ hxi.
Therefore, hxi = hyi.
Therefore, if x = ±y, then hxi = hyi.
Therefore, for all x, y ∈ Z, hxi = hyi if and only if x = ±y.
162 CHAPTER 2. SETS
A remark about the word ‘similarly’

The use of the word ‘similarly’ in a mathematical proof indicates a very particular circum-
stance: ‘Similarly’ means that the proof of the statement is exactly the same as the proof
above, with only a suitable alteration of the notation. In the example above, since the fact
that x = −y implies that y = −x, the letters x and y can be interchanged without affecting the
truth of the assumption. Hence there is no logical difference between the role of x and the
role of y in our proof; there is only a notational difference. The word ‘similarly’ indicates
that if the roles of x and y are reversed in the argument that x = −y implies hxi ⊆ hyi, the
result will be a correct proof of the statement that y = −x implies hyi ⊆ hxi. The careful
reader should verify that this is the case, whenever the word ‘similarly’ is encountered in a
mathematical proof.
The previous proposition shows that there is no difference between the ideal generated by
a positive number and that generated by its corresponding negative. That is, hxi = h−xi.
Further, the ideal is generated only by this pair. Therefore, to speak of an ideal hai is to
speak simultaneously about a and −a.
The ideals also have a convenient property: that membership in the set implies a subset
relation. Hence, any idea that can be expressed in terms of membership in an ideal can also
be expressed in terms of subsets, and vice versa.
Theorem 2.3.3.
For all x, y ∈ Z, x ∈ hyi if and only if hxi ⊆ hyi.
Proof.
Let x, y ∈ Z.
Assume x ∈ hyi.
Then ∃t ∈ Z, x = ty.
Let a ∈ hxi.
Then ∃s ∈ Z, a = sx.
Put r = st.
Then a = sx = sty = ry.
Hence, ∃r ∈ Z, a = ry.
Therefore, a ∈ hyi.
Therefore, hxi ⊆ hyi.
Therefore, if x ∈ hyi, then hxi ⊆ hyi.
Conversely, assume hxi ⊆ hyi.
Since x = (1)x, we have x ∈ hxi.
Since x ∈ hxi and hxi ⊆ hyi, we have x ∈ hyi.
Therefore, if hxi ⊆ hyi, then x ∈ hyi.
Therefore, for all x, y ∈ Z, x ∈ hyi if and only if hxi ⊆ hyi.
This means that for ideals the element-set relation x ∈ hyi can always be expressed as a
subset relation hxi ⊆ hyi; hence all concepts that are expressible in terms of ideals can be
expressed simply by the subset relations between the ideals themselves. To examine the
significance of such subset relations in the integers, we look at an example of two integers
for which this relation holds.
Proposition 2.3.4.
h6i ⊆ h3i.
Proof.
Let x ∈ h6i.
Then ∃t ∈ Z, x = 6t.
Put s = 2t.
x = 6t = 3(2t) = 3s.
Therefore, ∃s ∈ Z, x = 3s.
Hence, x ∈ h3i.
Therefore, h6i ⊆ h3i.
Many students find this counter-intuitive at first, since the order of the ideals is an inversion
of the order of the numbers generating them. That is h6i ⊆ h3i looks like a reversal of the
well-known ordering 3 ≤ 6. This is in fact almost always the case. Indeed, if hyi ⊆ hxi and
y , 0, then |x| ≤ |y|. The reason for this is in part due to the following theorem:
Theorem 2.3.5.
For all x, y ∈ Z, x divides y if and only if hyi ⊆ hxi.
Proof.
Let x, y ∈ Z.
Assume x divides y.
Then ∃t ∈ Z, y = xt.
Hence, y ∈ hxi.
Therefore, hyi ⊆ hxi by theorem 2.3.3.
Therefore, if x divides y then hyi ⊆ hxi.
Conversely, assume hyi ⊆ hxi.
Then y ∈ hxi by theorem 2.3.3.
That is, ∃t ∈ Z, y = xt.
Hence, x divides y.
Therefore, if hyi ⊆ hxi, then x divides y.
Therefore, for all x, y ∈ Z, x divides y if and only if hyi ⊆ hxi.
164 CHAPTER 2. SETS
Recall from proposition 1.2.8 that if x divides y and y , 0, then |x| ≤ |y|. We then see that
since the statement about ideals ‘hyi ⊆ hxi,’ being equivalent to the statement ‘x divides
y,’ must also imply that |x| ≤ |y| in the case where y , 0. Hence, in all but the case where
y = 0, we have that hyi ⊆ hxi implies |x| ≤ |y|. Thus it is a common occurrence that the
ordering of the ideals is contrary to the ordering of the numbers that generate them.
The above theorem is important for another reason: It shows that the set theoretic statement
hyi ⊆ hxi means the same thing as the statement x divides y in the integers. Hence, when
speaking about integers dividing other integers, which is the fundamental concept behind
such ideas as prime numbers, composite numbers, gcd, lcm , and factorization, we can use
the language of ideals.
Algebraic Characterization of Ideals
Ideals, in addition to their descriptive power, have a number of other properties that make
them particularly easy to work with. For example, they are closed under addition; that is,
∀a, x, y ∈ Z, if x ∈ hai and y ∈ hai, then x + y ∈ hai.
They are also closed under the unary operation given by x 7→ −x, in the sense that
∀a, x ∈ Z, if x ∈ hai, then − x ∈ hai.
Further, since x ∈ hai implies hxi ⊆ hai, we see that the ideals contain all multiples of their
elements. That is,
∀x, a ∈ Z, if x ∈ hai, then ∀t ∈ Z, xt ∈ hai.
The proofs of the above three properties are left as exercises. However, we will show here
that these three properties completely characterize the ideals of the integers, in the sense
that any non-empty set satisfying these properties must be equal to hai for some a ∈ Z.
To begin, we will show that the third property discussed above follows from the first two.
That is, if a non-empty subset of the integers is closed under addition and negation, then
it necessarily contains all multiples of its elements. To prove this, we recall a form of the
principle of mathematical induction that was proved as an exercise in section 1.2: For a
subset A ⊆ Z,
if 0 ∈ A and ∀n ∈ Z, n ∈ A ⇒ n + 1 ∈ A and n − 1 ∈ A, then A = Z.

Proposition 2.3.6.
Let S ⊆ Z, with S , ∅. If ∀x, y ∈ S , x + y ∈ S and ∀x ∈ S , −x ∈ S , then ∀x ∈ S , ∀a ∈ Z,

ax ∈ S .
Proof.
Let S ⊆ Z, with S , ∅.
Assume ∀x, y ∈ S , x + y ∈ S and ∀x ∈ S , −x ∈ S .
Let x ∈ S .
Let A = {a ∈ Z | ax ∈ S }.
Since x ∈ S , we have −x ∈ S ; hence x + (−x) ∈ S .
This gives us 0x ∈ S ; hence 0 ∈ A.
Let n ∈ Z, and assume n ∈ A.
Then nx ∈ S .
Since x ∈ S as well, we have nx + x ∈ S .
Hence, (n + 1)x ∈ S , which means n + 1 ∈ A.
Further, since x ∈ S , we have −x ∈ S .
Therefore, nx + (−x) ∈ S ; hence (n − 1)x ∈ S .
This gives us n − 1 ∈ A.
Therefore, ∀n ∈ Z, if n ∈ A, then n + 1 ∈ A and n − 1 ∈ A.
Therefore, by the PMI, A = Z.
Therefore, ∀a ∈ Z, ax ∈ S .
Therefore, ∀x ∈ S , ∀a ∈ Z, ax ∈ S .
Q.E.D.
The above proposition refers to a certain abstract type of subset of the integers. The subset
S in question, being closed under addition and negation (and hence also containing all
multiples of its elements), in some sense contains much of the algebraic structure of the
integers themselves. Yet, at this point the set is merely a hypothetical entity. Exactly what
such sets look like remains to be determined. Applying the same argument that was used to
prove Bezout’s identity (theorem 1.2.11), we can show that any such set must be an ideal
hai for some a ∈ Z. The proof, as in the case of Bezout’s identity, uses the well-ordering
property to select the smallest element of the set S , then uses the division algorithm to show
that S is exactly the ideal generated by its smallest element.
166 CHAPTER 2. SETS
Theorem 2.3.7.
Let S ⊆ Z, with S , ∅. If ∀x, y ∈ S , x + y ∈ S and ∀x ∈ S , −x ∈ S , then ∃g ∈ Z, S = hgi.
Proof.
Let S ⊆ Z, with S , ∅.
Assume ∀x, y ∈ S , x + y ∈ S and ∀x ∈ S , −x ∈ S .
First consider the case where S = {0}.
Letting x ∈ h0i gives us that ∃t ∈ Z, x = 0t.
In this case, x = 0; hence x ∈ S .
Therefore, h0i ⊆ S .
Conversely, letting x ∈ {0} gives us x = 0; hence x ∈ h0i.
Therefore, S ⊆ h0i.
Therefore, in this case, we have S = h0i.
Next, we consider the case where S , {0}.
Since S , ∅ and S , {0}, we may choose x ∈ S with x , 0.
If x > 0, then x ∈ S ∩ N.
If x < 0, then −x ∈ S ∩ N.
In both cases, we have S ∩ N , ∅.
Applying the well-ordering property,
choose g to be the smallest element of S ∩ N.
We claim that S = hgi.
Indeed, let x ∈ S .
Applying the division algorithm, choose q, r ∈ Z with x = qg + r
and 0 ≤ r < g.
Since g ∈ S , we have −qg ∈ S by proposition 2.3.6.
Since x ∈ S as well, we have x + (−qg) ∈ S .
Since r = x − qg, this means that r ∈ S .
Since r < g and g is the smallest element of S ∩ N, we have r < S ∩ N.
Therefore, either r < S or r < N.
Since we have shown that r ∈ S , it must be the case that r < N.
Therefore, r ≤ 0.
Since 0 ≤ r and r ≤ 0, we have that r = 0.
Therefore, x = qg + 0; hence ∃q ∈ Z, x = qg.
Therefore, x ∈ hgi.
Therefore, S ⊆ hgi.
Conversely, let x ∈ hgi.
Then, ∃t ∈ Z, x = gt.
Since g ∈ S and t ∈ Z, we have gt ∈ S by proposition 2.3.6.
Therefore, x ∈ S .
Hence, hgi ⊆ S .
Therefore, S = hgi.
In both cases, ∃g ∈ Z, S = hgi.
Therefore, if ∀x, y ∈ S , x + y ∈ S and ∀x ∈ S , −x ∈ S , then ∃g ∈ Z, S = hgi.
Set Operations on Ideals
An example in section 2.1 (example 2.1.13) showed that h2i ∩ h3i = h6i. Hence, the inter-
section of the ideals h2i and h3i yields another ideal. For another such example, consider
the following:
Example 2.3.8.
h9i ∩ h15i = h45i.
Proof.
Let x ∈ h9i ∩ h15i.
Then x ∈ h9i and x ∈ h15i.
Choose a ∈ Z with x = 9a and choose b ∈ Z with x = 15b.
Put c = 2b − a.
x = 6x − 5x = 6(15b) − 5(9a) = 45(2b) − 45a = 45(2b − a) = 45c.
Therefore, ∃c ∈ Z, x = 45c.
Hence x ∈ h45i.
Therefore, h9i ∩ h15i ⊆ h45i.
Conversely, let x ∈ h45i.
Choose s ∈ Z with x = 45s.
Put t = 5s.
x = 45s = 9(5s) = 9t.
Therefore, ∃t ∈ Z, x = 9t.
Hence, x ∈ h9i.
Put r = 3s.
x = 45s = 15(3s) = 15r.
Therefore, ∃t ∈ Z, x = 15r.
Therefore, x ∈ h15i.
We now have x ∈ h9i and x ∈ h15i.
Therefore, x ∈ h9i ∩ h15i.
Therefore, h45i ⊆ h9i ∩ h15i.
Therefore, h9i ∩ h15i = h45i.
It is in fact always true that the intersection of two ideals is an ideal. This can be shown
quite easily if we appeal to the algebraic characterization of ideals as those sets that are
closed under addition and negation and contain the multiples of all of their elements.
168 CHAPTER 2. SETS
Proposition 2.3.9.
For all x, y ∈ Z, ∃g ∈ Z, hxi ∩ hyi = hgi.
Proof.
Let x, y ∈ Z.
Let a, b ∈ hxi ∩ hyi.
Then a ∈ hxi and a ∈ hyi and b ∈ hxi and b ∈ hyi.
Choose q, r, s, t ∈ Z with a = qx, a = ry, b = sx, and b = ty.
Put u = q + s.
Then a + b = qx + sx = (q + s)x = ux.
Therefore, a + b ∈ hxi.
Put v = r + t.
Then a + b = ry + ty = (r + t)y = vy.
Therefore, a + b ∈ hyi.
Hence, a + b ∈ hxi ∩ hyi.
Therefore, ∀a, b ∈ hxi ∩ hyi, a + b ∈ hxi ∩ hyi.
Next, let a ∈ hxi ∩ hyi.
Once again, choose q, r ∈ Z with a = qx and a = ry.
This time, put u = −q.
Then −a = −qx = ux.
Hence −a ∈ hxi.
Put v = −r.
Then, −a = −ry = vy.
Hence, −a ∈ hyi.
Therefore, −a ∈ hxi ∩ hyi.
Therefore, ∀a ∈ hxi ∩ hyi, −a ∈ hxi ∩ hyi.
By theorem 2.3.7, we then have ∃g ∈ Z, hxi ∩ hyi = hgi.
In the above proposition, the existence of the integer a that generates the intersection hxi ∩
hyi is ultimately known only via the well-ordering property. That is, the only clue we have
as to the identity of a, in relation to x and y, is that it is the smallest non-negative element
of hxi ∩ hyi. We can do better than this. If we look closely at the connection between
ideals and ‘division’ in the integers, we see that elements of the intersection hxi ∩ hyi are
elements of both hxi and hyi. We also know that the elements of hxi are the multiples of x
and that the elements of hyi are the multiples of y. Hence, the elements of hxi ∩ hyi must
be the common multiples of both x and y. The generator a in the proposition above, being
the smallest non-negative such element, must then be the least common multiple of x and
y. Recall the definition of the least common multiple (definition 1.2.5):
For x, y, f ∈ Z, with x , 0 and y , 0, f is the least common multiple of

x and y means
1. f > 0,
2. x divides f and y divides f , and

3. ∀a ∈ Z, if x divides a and y divides a, then f ≤ |a|.
We denote that g is the least common multiple of x and y by f =
lcm (x, y).
Theorem 2.3.10.
For all x, y ∈ Z, if x , 0 and y , 0, then hxi ∩ hyi = hlcm (x, y)i.
Proof.
Let x, y ∈ Z.
Then hxi ∩ hyi , {0}, since xy ∈ hxi ∩ hyi, and xy < {0}.
By the previous proposition, we have ∃g ∈ Z, hxi ∩ hyi = hgi.
Since for every a ∈ Z, we have hai = h−ai, we may choose f ∈ Z with
hxi ∩ hyi = h f i and f ≥ 0.
In fact, since hxi ∩ hyi , {0}, we have f > 0. (1)
Since f ∈ hxi ∩ hyi, we have f ∈ hxi and f ∈ hyi.
Therefore, x divides f and y divides f . (2)
Let a ∈ Z, and assume x divides a and y divides a.
Then a ∈ hxi and a ∈ hyi.
Hence, a ∈ hxi ∩ hyi.
Therefore, a ∈ h f i.
This means that f divides a; hence | f | ≤ |a|.
Since f > 0, this means f ≤ |a|.
Therefore, ∀a ∈ Z, if x divides a and y divides a, then f ≤ |a|. (3)
By (1), (2), and (3), we have that f = lcm (x, y).
Therefore, hxi ∩ hyi = hlcm (x, y)i.
Therefore, if x , 0 and y , 0, then hxi ∩ hyi = hlcm (x, y)i.
Therefore, for all x, y ∈ Z, if x , 0 and y , 0, then hxi ∩ hyi = hlcm (x, y)i.
Unlike the intersection, the union of two ideals is not necessarily an ideal. For example,
consider the union h2i ∪ h3i. It is easy to see that 2 ∈ h2i ∪ h3i and 3 ∈ h2i ∪ h3i, but
5 < h2i ∪ h3i. This shows that the union h2i ∪ h3i is not closed under addition. However, all
ideals are closed under addition (the proof of this is an exercise). It follows that h2i ∪ h3i
cannot possibly be an ideal. There is however another way to combine ideals that will
always result in an ideal:
170 CHAPTER 2. SETS
Definition 2.3.1. For a, b ∈ Z, we define the sum of ideals hai + hbi to

be the set
hai + hbi = {x ∈ Z | ∃y ∈ hai, ∃z ∈ hbi, x = y + z}.
We can view the set hai + hbi as the collection of all sums of elements from hai and hbi.
That is, the set of all numbers that can be expressed as the sum of a multiple of a and
a multiple of b. It should be noted that unlike the intersection, the sum of ideals is not
purely set-theoretic. That is, it cannot be defined for arbitrary sets, since it makes use of
the addition operation that is defined for the integers. In fact, it is the elements of the sets
that are actually being added, not the sets themselves. Hence, it is only because we can
add integers together that we can have a sensible definition of the addition of ideals of the
integers. We could not for example extend the notion of addition of sets to collections of
apples or people, since we do not have the binary operation of addition defined on such
objects. The intersection and union of sets, on the other hand, are defined for all sets
regardless of the universe of discourse.
That the sum of two ideals is itself an ideal can be shown using the algebraic characteri-
zation of ideals given in theorem 2.3.7. That is, any set that is closed under addition and
negation is an ideal hai for some integer a.
Proposition 2.3.11.
For all x, y ∈ Z, ∃g ∈ Z, hxi + hyi = hgi.
Proof.
Let x, y ∈ Z.
Let a, b ∈ hxi + hyi.
Choose q, r, s, t ∈ Z with a = qx + ry, b = sx + ty.
Put u = q + s and v = r + t.
Then a + b = qx + ry + sx + ty = (q + s)x + (r + t)y = ux + vy.
Therefore, a + b ∈ hxi + hyi.
Therefore, ∀a, b ∈ hxi + hyi, a + b ∈ hxi + hyi.
Next, let a ∈ hxi + hyi.
Once again, choose q, r ∈ Z with a = qx + ry.
This time, put u = −q and v = −r.
Then, −a = −(qx + ry) = −qx + (−ry) = ux + vy.
Hence, −a ∈ hxi + hyi.
Therefore, ∀a ∈ hxi + hyi, −a ∈ hxi + hyi.
By theorem 2.3.7, we then have ∃g ∈ Z, hxi + hyi = hgi.
Just as we did for the intersection of two ideals, by examining the relationship between
ideals and the concept of division in the integers, we can determine the identity of the
generator a in the above proposition. The result is quite nice; the sum of ideals hxi + hyi
is generated by the greatest common divisor of x and y. Recall the definition (definition
1.2.4):
For x, y, g ∈ Z, with x , 0 and y , 0, g is the greatest common divisor

of x and y means
1. g > 0,
2. g divides x and g divides y, and

3. ∀a ∈ Z, if a divides x and a divides y, then g ≥ a.
We denote that g is the greatest common divisor of x and y by g =
gcd(x, y).
Theorem 2.3.12.
For all x, y ∈ Z, if x , 0 and y , 0, then hxi + hyi = hgcd(x, y)i.
Proof.
Let x, y ∈ Z.
Then hxi + hyi , {0}, since x = (1)x + (0)y ∈ hxi + hyi, and x < {0}.
By the previous proposition, we have ∃g ∈ Z, hxi + hyi = hgi.
Since for every a ∈ Z, we have hai = h−ai, we may choose g ∈ Z with
hxi + hyi = hgi and g ≥ 0.
Since hxi + hyi , {0}, we have g > 0. (1)
Since x = (1)x + (0)y, we have x ∈ hxi + hyi.
Therefore, x ∈ hgi and hence g divides x.
Similarly, y ∈ hxi + hyi, which means y ∈ hgi and hence g divides y.
Therefore, g divides x and g divides y. (2)
Let a ∈ Z, and assume a divides x and a divides y.
Choose s, t ∈ Z with x = as and y = at.
Since g ∈ hgi, we have g ∈ hxi + hyi.
Therefore, we can choose q, r ∈ Z, g = xq + yr.
Then g = asq + atr = a(sq + tr); hence g ∈ hai.
Therefore, a divides g; hence |a| ≤ |g|.
Since a ≤ |a| and g = |g|, this gives us a ≤ g.
Therefore, ∀a ∈ Z, if a divides x and a divides y, then a ≤ g. (3)
By (1), (2), and (3), we have that g = gcd(x, y).
Therefore, hxi + hyi = hgcd(x, y)i.
Therefore, if x , 0 and y , 0, then hxi + hyi = hgcd(x, y)i.
Therefore, for all x, y ∈ Z, if x , 0 and y , 0, then hxi + hyi = hgcd(x, y)i.
172 CHAPTER 2. SETS
Characterization of Prime Numbers
Recall the definition of a prime natural number (definition 1.2.10):
For x ∈ N, x is prime means x , 1 and ∀a, b ∈ N, if x = ab, then either

a = 1 or b = 1.
Since we have already seen that h1i = Z, the statement x , 1 can be written as hxi , Z.
Further, it is fairly clear that in the event that x = ab, to say b = 1 is equivalent to saying
a = x in the case where x , 0. That is, the characterization of x being prime given above
could be equivalently expressed as
hxi , Z and ∀a, b ∈ Z, if x = ab, then hai = Z or hai = hxi.
The expression x = ab also has an interpretation in terms of ideals. Since it indicates that x
is a multiple of a, the equation x = ab implies that hxi ⊆ hai. This suggests the following
interpretation of the statement ‘x is prime’:
hxi , Z and ∀a ∈ Z, if hxi ⊆ hai, then hai = Z or hai = hxi.
For example, consider the following proposition:
Proposition 2.3.13.
∀a ∈ Z, if h2i ⊆ hai, then hai = Z or hai = h2i.
Proof.
Let a ∈ Z.
Assume h2i ⊆ hai.
Since 2 ∈ h2i, we then have 2 ∈ hai.
Therefore, a divides 2; hence |a| ≤ 2.
Therefore, a = ±1 or a = ±2.
Case 1: a = ±1.
In this case, hai = h1i = Z.
Therefore, hai = Z or hai = h2i.
Case 2: a = ±2.
In this case, hai = h2i.
Therefore, hai = Z or hai = h2i.
Therefore, if h2i ⊆ hai, then hai = Z or hai = h2i.
Therefore, ∀a ∈ Z, if h2i ⊆ hai, then hai = Z or hai = h2i.
The fact that the only divisors of 2 are −1, 1, −2, and 2 (that is, 2 is prime) certainly plays
a key role in the above proof, and one may be able to see that the truth of the set-theoretic
concept in the above proposition is in some way related to the primeness of 2. It can be
somewhat enlightening to examine exactly what the statement
∀a ∈ Z, , if h2i ⊆ hai, then hai = Z or hai = h2i.
indicates about the ideal h2i. Essentially, this is saying that the ideal h2i is not contained
in any strictly larger ideal, except the all encompassing ideal that is Z itself. That is, one
cannot fit an ideal hai between h2i and Z in the sense that h2i ( hai ( Z. In this way, h2i
is as large as an ideal can be without being all of Z; the next larger ideal is the entire set Z.
The proof that this maximality of ideals is entwined with the notion of prime in the integers
is no more difficult than the proof of the special case h2i given above:
Proposition 2.3.14.
∀x ∈ Z, if x is prime, then ∀a ∈ Z, if hxi ⊆ hai, then hai = Z or hai = hxi.
Proof.
Let x ∈ Z.
Assume x is prime.
Let a ∈ Z.
Assume hxi ⊆ hai.
Since x ∈ hxi, we then have x ∈ hai.
Therefore, ∃b ∈ Z, x = ab. Choose such a b.
Therefore, a = ±1 or b = ±1, since x is prime.
Case 1: a = ±1.
In this case, hai = h1i = Z.
Therefore, hai = Z or hai = hxi.
Case 2: b = ±1.
In this case, a = ±x; hence hai = hxi.
Therefore, hai = Z or hai = hxi.
Therefore, if hxi ⊆ hai, then hai = Z or hai = hxi.
Therefore, ∀a ∈ Z, if hxi ⊆ hai, then hai = Z or hai = hxi.
Therefore, if x is prime, then ∀a ∈ Z, if hxi ⊆ hai, then hai = Z or hai = hxi.
Hence, ∀x ∈ Z, if x is prime, then ∀a ∈ Z, if hxi ⊆ hai, then hai = Z or hai = hxi.
The converse of the above proposition is true in the case where x , ±1 and x , 0. The
proof of this is left as an exercise.
174 CHAPTER 2. SETS
Exercises 2.3.
Prove the following propositions. 18. ∀a, x, y ∈ Z \ {0}, if hai ⊆ hxi and hai ⊆ hyi and
gcd(x, y) = 1, then hai ⊆ hxyi.
1. ∀a, x, y ∈ Z, if x ∈ hai and y ∈ hai, then x + y ∈ hai.
19. ∀a, x, y ∈ Z \ {0}, if xy ∈ hai and gcd(a, x) = 1, then
2. ∀a, x ∈ Z, if x ∈ hai, then −x ∈ hai.
y ∈ hai.
3. ∀a, x ∈ Z, if x ∈ hai, then ∀t ∈ Z, xt ∈ hai.
20. ∀a, b, n ∈ Z, hbi ⊆ hai + hni if and only if ∃x ∈ Z,
4. ∀a, x, y ∈ Z, if x + y ∈ hai and y ∈ hai, then x ∈ hai. b − ax ∈ hni.
5. h4i ∩ h6i = h12i. 21. ∀a ∈ Z, if a is prime, then ∀x ∈ Z, hai + hxi = Z or

hai + hxi = hai.
6. h4i + h6i = h2i.
22. ∀x ∈ Z \ {−1, 0, 1}, if ∀a ∈ Z, if hxi ⊆ hai, then
7. ∀a ∈ Z, hai + hai = hai. hai = Z or hai = hxi, then x is prime.
8. ∀a, b ∈ Z, hbi ⊆ hai + hbi. 23. (Euclid’s Lemma) ∀a ∈ Z, if a is prime, then
9. ∀a, b ∈ Z, if hai + hbi = hbi, then hai ⊆ hbi. ∀x, y ∈ Z, if xy ∈ hai, then x ∈ hai or y ∈ hai.
(Hint: Combine the result of problem 21, with those
10. ∀a, b ∈ Z, if hai ⊆ hbi, then hai + hbi = hbi. of problems 9 and 19.)
11. ∀a, b ∈ Z, hai ∩ (hai + hbi) = hai. 24. ∀a ∈ Z \ {0}, if hai , Z and ∀x, y ∈ Z, if xy ∈ hai,
then x ∈ hai or y ∈ hai, then a is prime.
12. ∀a, b ∈ Z, hai + (hai ∩ hbi) = hai.
13. ∀a, b, c ∈ Z, if hci ⊆ hai, then hai ∩ (hbi + hci) =
Let (Ik )k∈N be a sequence of ideals. That is, for each
(hai ∩ hbi) + hci.
k ∈ N, ∃ak ∈ Z, Ik = hak i. Prove the following proposi-
14. ∀a, x, y ∈ Z, if hxi ⊆ hai and hyi ⊆ hai, then tions.
hxi + hyi ⊆ hai.
25. ∀x ∈ Z, if ∀k ∈ N, ak ∈ hxi, then ∀n ∈ N,
n
15. ∀a, x, y ∈ Z, if hxi ⊆ hyi, then hai + hxi ⊆ hai + hyi.
X
ak ∈ hxi.
16. ∀a, b, x, y ∈ Z, if hxi ⊆ hyi and hai ⊆ hbi, then k=1
hai + hxi ⊆ hbi + hyi. \n

26. ∀n ∈ N, ∃g ∈ Z, Ik = hgi. (Hint: use proposition
17. ∀x, y ∈ Z\{0}, gcd(x, y) = 1 if and only if hxi+hyi = k=1
Z. 2.3.9).
2.4. FAMILIES OF SETS 175
2.4 Families of Sets
We begin this section by drawing a distinction between statements about the members of
a set and statements about the set itself. For example, when we say ‘humans are mortal,’
the statement is about each individual human. That is, we are speaking about the members
of the class of humans. On the other hand, statements such as ‘humans are diverse,’ ‘hu-
mans are widespread throughout the world,’ or ‘humans are numerous,’ refer to humans
collectively rather than individually. These are statements about the set of humans rather
than about the individual members of that set. For an example in mathematics, consider the
statements ‘the natural numbers are positive’ and ‘the natural numbers are infinite.’ The
former statement refers to the members of the set N, while the latter refers to the set N itself.
To make this precise, consider these statements in terms of their subjects and predicates: in
the one case, when saying the natural numbers are positive, we use the predicate ‘x > 0’
where the subject x is ranges over the elements of the set N. In the other case, when saying
the natural numbers are infinite, we use the predicate ‘X is infinite’ with the subject being
the set N itself.
In fact, there are many predicates whose subjects are the sets themselves rather than the
elements of those sets. Two further examples are ‘S has a smallest element’ and ‘S is
bounded above.’ Recall that for a universe of discourse U and a predicate P(x), the notation
{x ∈ U | P(x)}
is used to define the set of all subjects x for which the statement P(x) is true. Since it is
common to have predicates whose subjects are sets, it is just as common to see sets whose
elements are sets themselves. To indicate that a set has sets as elements, we will often call
a set of sets a family of sets. To illustrate the concept of a family of sets, consider the fact
that some subsets of the real numbers have smallest elements, while others do not. For
example, the interval [0, 1) has a smallest element but the interval (0, 1) does not. There
may be situations in which we want to speak about the subsets of R that do have smallest
elements. In such a situation, we would be speaking about the family of sets
S = {S ⊆ R | S has a smallest element}.
Another example was encountered in the previous section: We found that any subset of the
integers that is closed under addition and negation is an ideal. In this case, we saw that
{S ⊆ Z | S , ∅ and ∀x, y ∈ S , x + y ∈ S and − x ∈ S } = {S ⊆ Z | ∃g ∈ Z, S = hgi}.
These are all examples in which the members of a set are sets themselves.
The Powerset
The notation used above ({S ⊆ R | S has a smallest element}) is not quite proper and ought
to be corrected. The reason is that set specification notation {x ∈ U | P(x)} requires that
we indicate the universe of discourse U to which the elements of the set belong. Saying
that S ⊆ R indicates that the element S is a subset of R, but this is not quite the same thing
as identifying the universe of discourse to which S belongs. To be proper, the universe of
discourse in this case should be the set of all subsets of R, which we call the powerset of
R:
176 CHAPTER 2. SETS
Definition 2.4.1. Let A be a set. The powerset of A, denoted P(A) is the

set of all subsets of A. That is, X ∈ P(A) means X ⊆ A.
Example 2.4.1.
For a simple example, list all elements of the powerset of A = {0, 1}.
P(A) = {∅, {0}, {1}, {0, 1}}.
In practice, when working with the powerset, we keep in mind that the statements X ∈ P(A)
and X ⊆ A are interchangeable. Hence, to prove X ∈ P(A), one must prove X ⊆ A.
Likewise, if we assume X ∈ P(A), then it is known that X ⊆ A. Keep this in mind when
examining the structure of the following proof:
Proposition 2.4.2.
Let A and B be sets whose elements belong to a common universe of discourse U. Then
A ⊆ B if and only if P(A) ⊆ P(B).
Proof.
Assume A ⊆ B.
Let X ∈ P(U).
Assume X ∈ P(A).
Then X ⊆ A.
Let x ∈ U.
Assume x ∈ X.
Then x ∈ A, since X ⊆ A.
Hence x ∈ B, since A ⊆ B.
Therefore, if x ∈ X, then x ∈ B.
Therefore, X ⊆ B.
Equivalently, X ∈ P(B).
Therefore, if X ∈ P(A), then X ∈ P(B).
Therefore, P(A) ⊆ P(B).
Therefore, if A ⊆ B, then P(A) ⊆ P(B).
Conversely, assume P(A) ⊆ P(B).
Since A ⊆ A, we have A ∈ P(A).
Therefore, A ∈ P(B), since P(A) ⊆ P(B).
Hence, A ⊆ B.
Therefore, if P(A) ⊆ P(B), then A ⊆ B.
Therefore, A ⊆ B if and only if P(A) ⊆ P(B).
A common error when working with powersets is to confuse the element relation ∈ with
the subset relation ⊆. For example, X ⊆ A ∪ B does NOT mean X ⊆ A or X ⊆ B. The union
A ∪ B is defined in terms of elements rather than subsets. That is, x ∈ A ∪ B means x ∈ A
or x ∈ B. To replace the ∈ symbol with the ⊆ symbol is not valid and in this case results in
a statement that is not true. A particularly frustrating aspect of this error is the fact that, in
the case of intersections, the crime is committed without penalty. In fact, X ⊆ A ∩ B does
imply that X ⊆ A and X ⊆ B. The converse is also true. This property can be phrased as
follows:
Proposition 2.4.3.
Let A and B be sets whose elements belong to a common universe of discourse U. Then
P(A) ∩ P(B) = P(A ∩ B).
Proof.
Let X ∈ P(A) ∩ P(B).
Then X ∈ P(A) and X ∈ P(B).
That is, X ⊆ A and X ⊆ B.
Let x ∈ X.
Then x ∈ A, since X ⊆ A.
Likewise, x ∈ B, since X ⊆ B.
We now have x ∈ A and x ∈ B; hence x ∈ A ∩ B.
Therefore, if x ∈ X, then x ∈ A ∩ B.
Therefore, X ⊆ A ∩ B.
Then X ∈ P(A ∩ B).
Therefore, if X ∈ P(A) ∩ P(B), then X ∈ P(A ∩ B).
Therefore, P(A) ∩ P(B) ⊆ P(A ∩ B).
Conversely, let X ∈ P(A ∩ B).
Then X ⊆ A ∩ B.
Since X ⊆ A ∩ B and A ∩ B ⊆ A, we have X ⊆ A by transitivity.
Therefore, X ∈ P(A).
Similarly, since X ⊆ A ∩ B and A ∩ B ⊆ B, we know X ⊆ B.
Therefore, X ∈ P(B).
We now have X ∈ P(A) and X ∈ P(B), hence X ∈ P(A) ∩ P(B).
Therefore, if X ∈ P(A ∩ B), then X ∈ P(A) ∩ P(B).
Hence P(A ∩ B) ⊆ P(A) ∩ P(B).
Therefore, P(A) ∩ P(B) = P(A ∩ B).
Despite this convenient property of intersections, the reader is cautioned that the same
does not hold for unions, complements, or relative complements. That is, P(A) ∪ P(B) ,
P(A ∪ B), P(A) \ P(B) , P(A \ B), and (P(A))c , P(Ac ). The connections between these
sets will be further investigated in the exercises.
178 CHAPTER 2. SETS
When working with sets, sets of sets, and even sets of sets of sets, it can be important to
keep track of the different levels in which our mathematical objects sit. To do this, we
will try wherever possible to denote elements at the lowest level of our discourse (such
as individual real numbers or integers) using lowercase letters (such as a), sets containing
these elements using uppercase letters (such as A), and sets containing these sets using
calligraphic letters (such as A). Hence, we will usually see notation such as a ∈ A, A ⊆ B,
A ∈ A, and A ⊆ B. There is no formal rule in mathematics saying that sets must be
denoted by uppercase letters and their elements by lowercase letters. In fact, one often
encounters the notation x ∈ y, where the letter y is used to denote a set and the letter x is
used to denote one of its elements. However, in the interest of avoiding confusion, we will
stick to our notational convention wherever we can.
To practice identifying the levels in which our mathematical objects sit, the reader
is invited to try the following example first without looking at the answers:
Example 2.4.4.
Decide which of the following are true and which are false:
1. Z ⊆ P(Z).
2. Z ∈ P(Z)
3. 1 ∈ P(Z)
4. {1} ⊆ P(Z)
5. {1} ∈ P(Z)
6. ∅ ∈ P(Z)
7. ∅ ∈ Z
8. ∅ ⊆ P(Z)
9. {∅} ∈ P(Z)
10. {∅} ⊆ P(Z)
Answers: 1. False; 2. True; 3. False; 4. False; 5. True; 6. True; 7. False; 8. True; 9. False; 10. True.
Set Operations
In section 2.1, we defined the operations ∩ and ∪. These operations combine two sets into
one. Using a recursive definition, we extended these operations to allow the combination
\n n
[
of any finite number of sets into one set: Ak in the case of intersection and Ak for
k=1 k=1
union. With the concept of a family of sets in place, we can extend these operations even
further.
n
\
First, let us consider the intersection of a sequence of sets Ak in another way: the
k=1
sequence of sets (Ak ) can be viewed as a family of sets {Ak | 1 ≤ k ≤ n}. We have seen that
n
\
even though it is defined recursively, the intersection Ak can be thought of intuitively as
k=1
consisting of those elements that belong to every set Ak for 1 ≤ k ≤ n. In other words, the
intersection consists of those elements that belong to every set in the family {Ak | 1 ≤ k ≤ n}.
[n
Likewise, the union Ak can be thought of as consisting of all elements that belong to at
k=1
least one set in the family {Ak | 1 ≤ k ≤ n}.
Now, consider the family of sets I = {hki | k ∈ Z}. Since this family consists of an
infinite number of sets, our current definitions are insufficient if we want to talk about the
intersection or union of the sets in this family. However, the idea given above, which says
the intersection consists of those elements that are common to all sets in the family and the
union consists of those elements that appear in at least one set in the family, can easily be
extended to the family of sets I. In fact, the only integer that is common to every ideal is
0, and every integer appears in at least one ideal. It seems reasonable then to say that the
intersection of all of the ideals is the set {0}, and the union of all of the ideals is Z.
In general, we define the union and intersection of a family of sets as follows:
Definition 2.4.2. Let A ⊆ P(U) be a family of sets. The intersection of

the family of sets A is defined to be
\
S = {x ∈ U | ∀S ∈ A, x ∈ S }.
S ∈A
The union of A is defined to be

[
S = {x ∈ U | ∃S ∈ A, x ∈ S }.
S ∈A
This definition is consistent with our earlier definitions, since the intersection A ∩ B can
be viewed as consisting of those elements that are common to all sets in the family {A, B}.
Likewise, the union of two sets A ∪ B can be viewed as those elements appearing in at least
one set in the family {A, B}. Moreover, as mentioned earlier, the intersection and union of a
\n [n
sequence of sets Ak and Ak can be viewed as the intersection and union of the family
k=1 k=1
of sets {Ak | 1 ≤ k ≤ n}.
180 CHAPTER 2. SETS
Indexed Families of Sets
Notice that in some of our examples, such as any sequence of sets {Ak | 1 ≤ k ≤ n}, the
sets in the family are labeled with an index. In this case, the index is the variable k which
ranges in value from 1 through n. That is, to each number k ∈ {1, . . . , n}, there corresponds
a set Ak in our family if sets. In general, we say a family of sets A is indexed by a set I
provided to each element k ∈ I, there corresponds a set in the family A. In this way, the
sequence of sets {Ak | 1 ≤ k ≤ n} is indexed by the set {1, . . . , n}.
Likewise, we can say that the family of ideals I = {hki | k ∈ Z} is indexed by the
set Z. This is because to each element k ∈ Z, there corresponds a set hki in the family
I. When a family of sets is indexed, we often adjust the notation used to describe
the intersection and union by referring to the sets by their index. A restatement of the
definition of the intersection and union of an indexed family of sets, using this alternate
notation, is as follows:
Definition 2.4.3. Let A = {S k | k ∈ I} be a family of sets indexed by the

set I. The intersection of A is defined to be
\
S k = {x ∈ U | ∀k ∈ I, x ∈ S k }.
k∈I
The union of A is defined to be

[
S k = {x ∈ U | ∃k ∈ I, x ∈ S k }.
k∈I
\
S k is often read as ‘the intersection of the sets S k over I’, similarly
k∈I
for the union.
For a slightly more involved example of the intersection and union of a family of sets, for
each n ∈ N let
m
S n = x ∈ R ∃m ∈ Z, x = .
n
Let A = {S n | n ∈ N}. In this case, A is a family of sets indexed by N. That is, to each
n ∈ N, there corresponds a set S n in the family A. Writing down the first few members of
this family: ( )
−3 −2 −1 0 1 2 3 4
S1 = ..., , , , , , , , ,...
1 1 1 1 1 1 1 1
( )
−3 −2 −1 0 1 2 3 4
S2 = ..., , , , , , , , ,...
2 2 2 2 2 2 2 2
( )
−3 −2 −1 0 1 2 3 4
S3 = ..., , , , , , , , ,...
3 3 3 3 3 3 3 3
..
.
we can guess that the union of this family will be the set of all rational numbers and that
the intersection will be the set of all integers. Indeed, since we have a precise definition of
the intersection and union over a family of sets, we can prove that this is the case:
Example 2.4.5.
[
For each n ∈ N, let S n = {x ∈ R | ∃m ∈ Z, x = m
n }. Then S n = Q.
n∈N
Proof.
S
Let x ∈ n∈N S n .
Then ∃n ∈ N, x ∈ S n . Choose such an n.
We then have ∃m ∈ Z, x = mn . Choose such an m.
Since n ∈ N, we have n , 0.
We have nx = m and n , 0; hence x ∈ Q.
S
Therefore, n∈N S n ⊆ Q.
Conversely, let x ∈ Q. By proposition 1.2.13, choose a, b ∈ Z, bx = a and b > 0.
Since b > 0, we have b ∈ N.
Further, since x = ba , we have x ∈ S b .
Therefore, ∃n ∈ N, x ∈ S n .
S
Hence, x ∈ n∈N S n
S
Therefore, Q ⊆ n∈N S n .
Therefore, n∈N S n = Q.
S

182 CHAPTER 2. SETS
Example 2.4.6.
\
For each n ∈ N, let S n = {x ∈ R | ∃m ∈ Z, x = m
n }. Then S n = Z.
n∈N
Proof.
T
Let x ∈ n∈N S n .
Then ∀n ∈ N, x ∈ S n .
Since 1 ∈ N, we have x ∈ S 1 .
Therefore, ∃m ∈ Z, x = m1 = m; hence x ∈ Z.
T
Therefore, n∈N S n ⊆ Z.
Conversely, let x ∈ Z.
Let n ∈ N.
Put m = xn.
Then x = xnn = n.
m
Therefore, ∃m ∈ Z, x = mn ; thus, x ∈ S n .
Therefore, ∀n ∈ N, x ∈ S n .
T
That is, x ∈ n∈N S n
T
Therefore, Z ⊆ n∈N S n .
Therefore, n∈N S n = Z.
T

For practice, here are two more examples of proofs involving indexed families of sets:
Example 2.4.7.
#
[ 1
0, 1 − = (0, 1).
n∈N
n
Proof.
Let x ∈ n∈N (0, 1 − 1n ].
S
Then ∃n ∈ N, x ∈ (0, 1 − 1n ]. Choose such an n.
We then have 0 < x and x < 1 − n1 .
Since n > 0, we have 1n > 0; hence − 1n < 0. Therefore, 1 − 1n < 1.
Since x < 1 − 1n and 1 − 1n < 1, we have x < 1.
Therefore, 0 < x and x < 1; hence x ∈ (0, 1).
Therefore, n∈N (0, 1 − 1n ] ⊆ (0, 1).
S
Conversely, let x ∈ (0, 1).
Since x < 1, we have 0 < 1 − x.
By the Archimedean property, we can choose n ∈ N with n1 ≤ 1 − x.
For such an n, we have x ≤ 1 − 1n ; hence x ∈ (0, 1 − 1n ].
Therefore, ∃n ∈ N, x ∈ (0, 1 − 1n ].
That is, x ∈ n∈N (0, 1 − n1 ].
S
Therefore, (0, 1) ⊆ n∈N (0, 1 − 1n ].
S
Therefore, n∈N (0, 1 − 1n ] = (0, 1).
S

Example 2.4.8.
\
[0, a] = {0}.
a∈(0,∞)
Proof.
T
Let x ∈ a∈(0,∞) [0, a].
Then ∀a ∈ (0, ∞), we have x ∈ [0, a].
Since 1 ∈ (0, ∞), we have x ∈ [0, 1]; hence 0 ≤ x.
Suppose x > 0.
Taking a = 2x gives us a ∈ (0, ∞); hence x ∈ [0, a].
Then x ≤ a, which means x ≤ 2x .
Therefore, 2x ≤ x; hence x ≤ 0 which is a contradiction.
Therefore, x ≤ 0, and since 0 ≤ x, it must be the case that x = 0; hence x ∈ {0}.
T
Therefore, a∈(0,∞) [0, a] ⊆ {0}.
Conversely, let x ∈ {0}.
Then x = 0; hence 0 ≤ x.
Let a ∈ (0, ∞).
Then 0 < a, which means x < a. Hence, x ≤ a.
We now have 0 ≤ x ≤ a; hence x ∈ [0, a].
Therefore, ∀a ∈ (0, ∞), x ∈ [0, a].
T
That is, x ∈ a∈(0,∞) [0, a].
T
Therefore, {0} ⊆ a∈(0,∞) [0, a].
Hence, a∈(0,∞) [0, a] = {0}.
T

184 CHAPTER 2. SETS
Exercises 2.4.
[
Let A, B, and C be sets whose elements belong to a 15. If S k = ∅, then ∀m ∈ I, S m = ∅.
common universe of discourse U. Prove the following k∈I
propositions. \
16. If ∃m ∈ I, S m = ∅, then S k = ∅.
1. If P(A) ⊆ P(A) ∩ P(B), then A ⊆ B. k∈I
[ [
2. if P(A) ∩ P(Bc ) = {∅}, then A ⊆ B. 17. ∀J ∈ P(I), Sk ⊆ S k.
k∈J k∈I
3. (a) P(A) ∪ P(B) ⊆ P(A ∪ B). \ \
(b) Give examples of sets A and B for which 18. ∀J ∈ P(I), Sk ⊆ S k.
k∈I k∈J
P(A) ∪ P(B) , P(A ∪ B).
(c) if P(A ∪ B) ⊆ P(A) ∪ P(B), then A ⊆ B or 19. For all n ∈ N, if I = {k ∈ N | k ≤ n}, then
\ \n
B ⊆ A. Sk = S k.
(d) if A ⊆ B or B ⊆ A, then P(A ∪ B) ⊆ P(A) ∪ k∈I k=1
P(B). 20. For all n ∈ N, if I = {k ∈ N | k ≤ n}, then
[ [n
4. (a) There are no sets A and B for which P(A\B) ⊆ Sk = S k.
P(A) \ P(B). k∈I k=1
(b) if A ∩ Bc = ∅, then P(A) \ P(B) ⊆ P(A \ B). \
21. If ∀k ∈ I, S k ⊆ R is an interval, then S k is an
(c) if A ∩ B = ∅, then P(A) \ P(B) ⊆ P(A \ B). k∈I
(d) if A∩Bc , ∅ and A∩B , ∅, then P(A)\P(B) * interval. (Hint: Use theorems 2.2.9 and 2.2.11).
\
P(A \ B). 22. If ∀k ∈ I, S k ⊆ Z is an ideal, then S k is an ideal.
5. If (P(A))c ⊆ P(Ac ), then A = U or A = ∅. (Hint: Use theorem 2.3.7).
k∈I
6. P(Ac ) * (P(A))c . sequence of ideals of Z. If ∀k ∈ N,

23. Let (S k )k∈N be a [
\
7. Let S = {A, B}. Then S = A ∩ B. S k ⊆ S k+1 , then S k is an ideal.
S ∈S k∈N
[ 24. Let (S k )k∈N be a sequence[ of intervals of R. If
8. Let S = {A, B}. Then S = A ∪ B.
∀k ∈ N, S k ⊆ S k+1 , then S k is an interval.
S ∈S
k∈N
Let A = {S k | k ∈ I} be a family of sets with index set

I , ∅, and let B be a set. Prove the following proposi- [
tions. 25. hki = Z.
[ [ k∈Z
9. B ∩ Sk = (B ∩ S k ). \
k∈I k∈I 26. hki = {0}.
\ \ k∈Z
10. B ∪ Sk = (B ∪ S k ). \
k∈I k∈I 27. [n, ∞) = ∅.
[ \ n∈N
11. B \ Sk = (B \ S k ). [
k∈I k∈I 28. [n, ∞) = [1, ∞).
n∈N
\ [
12. B \ Sk = (B \ S k ). [
k∈I k∈I 29. [0, n) = [0, ∞).
 c n∈N
[  \
13.  S k  = (S k )c . \
k∈I k∈I
30. [0, n) = [0, 1).
n∈N
\ c [
 
\
14.  S k  =
 (S k )c . 31. (−∞, x) = ∅.
k∈I k∈I x∈R
[ [
32. (−∞, x) = R. 52. (1 − a, 2 − a] = (0, 2).
x∈R a∈(0,1)
[ [" !
33. [1, 3n) = [1, ∞). 53.
1
0, 2 + = [0, 3).
n∈N
n∈N
n
\
34. [1, n + 1) = [1, 2). \" !
1
n∈N 54. 0, 2 + = [0, 2].
n∈N
n
#
\ 1
= (−∞, 0]. \" #
35. −∞, 1
n∈N
n 55. 0, 2 − = [0, 1].
n∈N
n
[
36. (−∞, n) = R. [" 1
#
n∈N 56. 0, 2 − = [0, 2).
[ n∈N
n
37. [0, 2 + a) = [0, 3). [ "1 #
a∈(−∞,1) 57. , 2 = (0, 2].
\ n∈N
n
38. [0, 2 + a) = [0, 2].
\ "1 #
a∈(0,∞)
58. , 2 = [1, 2].
\ n∈N
n
39. [0, 2 − a] = [0, 1]. #
a∈(−∞,1)
\ 1
59. 1 − , 2 = [1, 2].
[ n
40. [0, 2 − a] = [0, 2). n∈N
#
a∈(0,∞) [ 1
60. 1 − , 2 = (0, 2].
[ n
41. [a, 2] = (0, 2]. n∈N
a∈(0,∞)
!
[ 1 1
61. 1 − ,2 + = (0, 3).
n n
\
42. [a, 2] = [1, 2]. n∈N
a∈(−∞,1) !
\ 1 1
\ 62. 1 − ,2 + = [1, 2].
43. (1 − a, 2] = [1, 2]. n∈N
n n
a∈(0,∞)
\ "1 1
#
[ 63. ,2 − = {1}.
44. (1 − a, 2] = (0, 2]. n∈N
n n
a∈(−∞,1)
[ "1 1
#
64. ,2 − = (0, 2).
[
45. (1 − a, 2 + a) = (0, 3). n n
n∈N
a∈(−∞,1)
[ "1 1
!
,2 + = (0, 3).
\
46. (1 − a, 2 + a) = [1, 2]. 65.
n∈N
n n
a∈(0,∞)
\ \ "1 1
!
47. [a, 2 − a] = {1}. 66. ,2 + = [1, 2].
n∈N
n n
a∈(−∞,1)
#
[ \ 1 1
48. [a, 2 − a] = (0, 2). 67. 1 − ,2 − = {1}.
a∈(0,∞) n∈N
n n
[ #
49. [a, 2 + a) = (0, 3).
[ 1 1
68. 1 − ,2 − = (0, 2).
a∈(0,1) n∈N
n n
\ [
50. [a, 2 + a) = [1, 2]. 69. hki = h6i.
a∈(0,1) k∈h6i
\ \
51. (1 − a, 2 − a] = {1}. 70. hki = h6i.
a∈(0,1) k∈{n∈Z | 6∈hni}
186 CHAPTER 2. SETS
\
71. (−∞, k] ∪ [k + 1, ∞) = Z. 77. Let A = {A ∈ P(R) | 1 ∈ A and A is inductive}.
k∈Z
[ \
72. (−∞, k + 1) ∩ (k, ∞) = R \ Z. (a) A is inductive.
k∈Z A∈A
\
Let U and V be sets. Let A = {Ay | y ∈ V} be a family (b) N ∩ A is inductive.
of subsets of U, indexed by V. For each x ∈ U, let A∈A
Bx = {y ∈ V | x ∈ Ay }, and let B = {Bx | x ∈ U}. \

\ (c) A = N.
73. Prove ∀x ∈ U, x ∈ Ay if and only if Bx = V. A∈A
y∈V
[ 78. Let a ∈ R. Let M = {A ∈ P(R) | a ∈
74. Prove ∀x ∈ U, x ∈ Ay if and only if Bx , ∅. A and A is inductive}.
y∈V
[
75. Prove ∀y ∈ V, y ∈ Bx if and only if Ay , ∅. (a) M is non-empty.
x∈U \
\ (b) A is inductive.
76. Prove ∀y ∈ V, y ∈ Bx if and only if Ay = U. A∈M
x∈U \
(c) A is the smallest inductive set containing
Prove the following propositions about families of in- A∈M
ductive sets. a (See definition 1.2.9).
Chapter 3
Relations
Relations are statements about two subjects. For example, statements such as ‘x causes y’,
‘x precedes y’, ‘x is beside y’, or for math related examples, ‘x > y’, ‘x = y’, ‘x divides
y’, ‘x ∈ Y’, etc. are all examples of statements relating two objects. In fact, the predicates
‘causes’, ‘precedes’, ‘is beside’, are only sensible when applied to two subjects.
To deal with relations in a set theoretic way, we must accept that the truth set of a
relation is not a collection of single elements. Rather, it is a collection of pairs of elements.
The proposition 0 < 1 is not a statement about the subject 0 or the subject 1 individually;
it is a statement about both subjects 0 and 1 simultaneously. That is, it is a statement about
the pair (0, 1). A very simple way of dealing with this is as follows: Suppose P(x, y) is an
open sentence with two variables (for example, one can imagine that P is the relation <, in
which case P(x, y) represents the statement ‘x < y’). We view the two subjects x and y as
being one pair of subjects. That is, we view P as a predicate requiring a single subject that
is an ordered pair (x, y). In doing so, we can specify the truth set of such a predicate as
{(x, y) | P(x, y)}. In the case where P(x, y) is the statement ‘x < y’, our set is {(x, y) | x < y}.
This set would contain for example the ordered pair (0, 1), meaning that 0 < 1. To make
this precise, we turn to a familiar mathematical construction: The Cartesian plane.
Definition 3.0.1. Let A and B be sets. The Cartesian product of A and

B, denoted A × B, is the set consisting of all ordered pairs (x, y) for which
x ∈ A and y ∈ B. That is,
A × B = {(x, y) | x ∈ A and y ∈ B}.
For a set A, we denote A × A by A2 .
Under this definition, the familiar Cartesian plane is the set R×R, which is more commonly
denoted as R2 . We naturally extend this concept not only to sets of ordered pairs but to sets
of ordered triples, quadruples, or any other number of objects:
187
188 CHAPTER 3. RELATIONS
Definition 3.0.2. Let A be a set. Define A1 = A, and for each n ∈ N

define An+1 = An × A.
With this in mind, we can begin to discuss relations in a set theoretic way. As a starting
point, we state the definition of a relation in these terms:
Definition 3.0.3. Let A and B be sets. A relation R from A to B is any

non-empty subset of A × B. A non-empty subset of A × A is called a
relation on A. When R is a relation from A to B, for a ∈ A and b ∈ B,
we use the notation aRb to denote (a, b) ∈ R.
In this sense, the relation < is viewed as a set of ordered pairs; formally, <⊆ R × R. For
a, b ∈ R, we use the notation a < b to indicate that (a, b) ∈<.
3.1 Equivalence Relations
Equivalence relations are those relations that indicate two subjects are the same in some
way. For example, ‘x is of the same colour as y’ and ‘x has the same shape as y’ are both
examples of predicates that define equivalence relations. Different equivalence relations
merely describe different senses in which two things can be the same. Equivalence relations
are characterized by three properties: First, every object is the same as itself in every sense
of the word ‘same’. For example, every object is of the same colour as itself, and every
object has the same shape as itself. Second, if the object x is the same as the object y, then
we can also say that the object y is the same as the object x. That is, the order in which
the subjects are mentioned does not change the meaning of the statement ‘x is the same
as y’. Again, this is true for any sense of the word ‘same’. Third, x being the same as y
is a transitive relation, in the sense that if x and y are the same and y and z are the same,
then we can expect x to be the same as z. We take these three properties together to be the
definition of an equivalence relation.
Definition 3.1.1. For a relation R on a set U:

1. R is reflexive means ∀x ∈ U, xRx.
2. R is symmetric means ∀x, y ∈ U, if xRy, then yRx.
3. R is transitive means ∀x, y, z ∈ U, if xRy and yRz, then xRz.

A relation that is reflexive, symmetric, and transitive is called an equiv-
alence relation.
3.1. EQUIVALENCE RELATIONS 189
For a math example of an equivalence relation, consider the parity relation. By parity, we
mean whether a number is even or odd. Odd numbers are said to have odd parity and
even numbers are said to have even parity. The relation ‘x is of the same parity as y’ is an
equivalence relation on the integers. This relation can be described formally as follows:
≡2 = {(x, y) ∈ Z × Z | x − y ∈ h2i}.
That is, for x, y ∈ Z, x ≡2 y if and only if x − y ∈ h2i. To prove that this is an equivalence
relation, we must show that ≡2 is reflexive, symmetric, and transitive. That is, we must
prove
1. ∀x ∈ Z, x ≡2 x.
2. ∀x, y ∈ Z, if x ≡2 y, then y ≡2 x.
3. ∀x, y, z ∈ Z, if x ≡2 y and y ≡2 z, then x ≡2 z.
Proposition 3.1.1.
≡2 is an equivalence relation.
Proof.
Let x ∈ Z.
Since x − x = 0 = (0)(2) ∈ h2i, we have x ≡2 x.
Therefore, ∀x ∈ Z, x ≡2 x.
Hence, ≡2 is reflexive.
Next, let x, y ∈ Z.
Assume x ≡2 y.
Then x − y ∈ h2i.
Choose a ∈ Z such that x − y = 2a.
Then y − x = −2a = 2(−a) ∈ h2i.
Therefore, y ≡2 x.
This shows that if x ≡2 y, then y ≡2 x.
Therefore, ≡2 is symmetric.
Finally, let x, y, z ∈ Z.
Assume x ≡2 y and y ≡2 z.
That is, x − y ∈ h2i, and y − z ∈ h2i.
Choose a, b ∈ Z such that x − y = 2a and y − z = 2b.
Then x − z = (x − y) + (y − z) = 2a + 2b = 2(a + b) ∈ h2i.
Therefore, x ≡2 z.
Therefore, if x ≡2 y and y ≡2 z, then x ≡2 z.
Thus ≡2 is transitive.
Since ≡2 is reflexive, symmetric, and transitive, ≡2 is an equivalence relation.
Equivalence Classes
To see precisely how the equivalence relation ≡2 corresponds to the partition of the integers
into evens and odds, we notice that every equivalence relation on a set A implies a grouping
of the elements of A into classes of like objects. For example, consider the following set of
nine coloured shapes:
{ , , , , N, , N, N, }
The equivalence relation given by ‘x is of the same colour as y’ suggests the following
grouping of the elements in A:
{{ , N}, { , }, {, , N}, {, N}}
On the other hand, the equivalence relation given by ‘x has the same shape as y’ implies
we partition the set A into classes of shapes as follows:
{{ , , }, {, , }, {N, N, N}}
To make precise this idea of grouping elements of a set into classes of similar elements, we
give the following definition:
Definition 3.1.2. For an equivalence relation R on a set U, for each a ∈

U, we define the equivalence class of a relative to R to be the set
[a]R = {x ∈ U | xRa}.
We denote by U/R, the family of equivalence classes:
U/R = {[a] | a ∈ U}.
In practice, it is not always necessary to include a reference to the equivalence relation R

in the notation denoting the equivalence class [a]R . If there is no ambiguity and the reader
of our proof can reasonably be expected to know the equivalence relation relative to which
the class [a]R is defined, then we can denote the equivalence class of a simply by [a]. Other
common notation for the equivalence class of a includes a and ȧ. In summary:
Notation
When the relation R is understood in the context in which we are working, we commonly
denote the equivalence class of a relative to R by [a], a, or ȧ.
For an example of equivalence classes, we return to the relation ≡2 . Recall that this relation
on Z is given by: x ≡2 y if and only if x − y ∈ h2i. We can now show that the equivalence
classes relative to this relation are the even integers and the odd integers.
Proposition 3.1.2.
For the equivalence relation ≡2 , [0] = E and [1] = O.
Proof.
Let x ∈ Z.
x ∈ [0] if and only if x ≡2 0.
x ≡2 0 if and only if x − 0 ∈ h2i.
x − 0 ∈ h2i if and only if ∃a ∈ Z, x = 2a.
∃a ∈ Z, x = 2a if and only if x ∈ E.
Therefore, x ∈ [0] if and only if x ∈ E.
Therefore, [0] = E.
To show that [1] = O, we again let x ∈ Z.
x ∈ [1] if and only if x ≡2 1.
x ≡2 1 if and only if x − 1 ∈ h2i.
x − 1 ∈ h2i if and only if ∃a ∈ Z, x − 1 = 2a.
∃a ∈ Z, x − 1 = 2a if and only if ∃a ∈ Z, x = 2a + 1.
∃a ∈ Z, x = 2a + 1 if and only if x ∈ O.
Therefore, x ∈ [1] if and only if x ∈ O.
Therefore, [1] = O.
Although we speak about the equivalence class of a single object [a], the choice of which
object represents its equivalence class is arbitrary. Any element in the equivalence class
can stand as its representative. For example, in the case of the equivalence relation given
by ‘x is the same shape as y’ partitioning the set of coloured shapes into classes:
{{ , , }, {, , }, {N, N, N}}
The class of triangles could be represented by any of the three triangles. That is, the equiv-
alence class of the red triangle is the same as the equivalence class of the green triangle, as
well as that of the blue triangle. That any two related objects can stand as representatives
for the same equivalence class is the main point of the following theorem:
Theorem 3.1.3.
Let R be an equivalence relation on a set U. Let a, b ∈ U. The following statements are

equivalent:
1. aRb
2. a ∈ [b]
3. [a] ⊆ [b]
4. [a] = [b]
5. [a] ∩ [b] , ∅.
Proof.
Let a, b ∈ U.
(1) ⇐⇒ (2) is the definition so we use them interchangeably.
We will prove (1) ⇒ (3), (3) ⇒ (4), (4) ⇒ (5), and (5) ⇒ (1).
(1) ⇒ (3):
Assume aRb.
Let x ∈ [a].
Then xRa.
Since xRa and aRb, we have xRb by transitivity.
Hence, x ∈ [b].
Therefore, [a] ⊆ [b].
Therefore, if aRb, then [a] ⊆ [b].
(3) ⇒ (4):
Assume [a] ⊆ [b].
Since R is reflexive, a ∈ [a]; hence a ∈ [b].
Then aRb, and so bRa by symmetry.
Thus, [b] ⊆ [a], since (1) ⇒ (3).
Now, [a] ⊆ [b] and [b] ⊆ [a], which means [a] = [b].
Therefore, if [a] ⊆ [b], then [a] = [b].
(4) ⇒ (5):
Assume [a] = [b].
Then, [a] ∩ [b] = [a] , ∅, since a ∈ [a], by reflexivity.
Therefore, if [a] = [b], then [a] ∩ [b] , ∅.
(5) ⇒ (1):
Assume [a] ∩ [b] , ∅.
Choose x ∈ [a] ∩ [b].
Then, x ∈ [a] and x ∈ [b], so xRa and xRb.
But R is symmetric, so aRx and xRb.
Hence, aRb by transitivity.
Therefore, if [a] ∩ [b] , ∅, then aRb.
Q.E.D.
Returning to our example ≡2 , recall that [0] = E and [1] = O. The point of the theorem
above is that the class of even integers need not be represented by 0, it could in fact be
represented by any even integer. Likewise, the class of odd integers could be represented
by any odd integer; not only the odd integer 1. We make this precise as follows:
Proposition 3.1.4.
For the equivalence relation ≡2 , ∀x ∈ Z, if x is even, then [x] = E, and if x is odd, then
[x] = O.
Proof.
Let x ∈ Z.
Assume x is even.
Then x ∈ [0], since [0] = E.
Therefore, [x] = [0]; hence [x] = E.
Therefore, if x is even, then [x] = E.
Assume x is odd.
Then x ∈ [1], since [1] = O.
Therefore, [x] = [1]; hence [x] = O.
Therefore, if x is odd, then [x] = O.
Therefore, ∀x ∈ Z, if x is even, then [x] = E, and if x is odd, then [x] = O.
Partitions
To examine in more detail this separation of the underlying set into classes, we can make
the sense in which we use the word partition more precise:
Definition 3.1.3. Let U be a set. A family of sets A ⊆ P(U) is called a

partition of U provided:
1. ∀S ∈ A, S , ∅.
2. ∀S , T ∈ A, if S , T , then S ∩ T = ∅.
[
3. S = U.
S ∈A
In the event that our family of sets is indexed, we restate this definition using the indexed
form:
Definition 3.1.4. Let U be a set. An indexed family of sets A = {S k | k ∈

I} ⊆ P(U) is called a partition of U provided:
1. ∀k ∈ I, S k , ∅.
2. ∀ j, k ∈ I, if S j , S k , then S j ∩ S k = ∅.
[
3. S k = U.
k∈I
It is in this sense of the word partition that equivalence classes form a partition of underly-
ing set:
Theorem 3.1.5.
Let R be an equivalence relation on a set U. The family of equivalence classes relative to

R, denoted U/R = {[x] | x ∈ U}, is a partition of U.
Proof.
Let x ∈ U.
Since R is reflexive, we have xRx; hence x ∈ [x].
Therefore, [x] , ∅.
Therefore, ∀x ∈ U, [x] , ∅.
Let x, y ∈ U.
Assume [x] ∩ [y] , ∅.
Then [x] = [y] by theorem 3.1.3.
Therefore, if [x] ∩ [y] , ∅, then [x] = [y].
Therefore, ∀x, y ∈ U, if [x] , [y], then [x] ∩ [y] = ∅.
S
Since U is the universe of discourse, we have a∈U [a] ⊆ U.
Conversely, let x ∈ U.
Then, since R is reflexive, we have x ∈ [x].
S
Therefore, ∃a ∈ U, x ∈ [a]; hence x ∈ a∈U [a].
Therefore, a∈U [a] = U.
S
Therefore, U/R = {[x] | x ∈ U} is a partition of U.
Although at first it may seem like a novel way of thinking, in some sense the converse of
theorem 3.1.5 is also true. Where theorem 3.1.5 says that whenever one has an equivalence
relation, one also has a partition; it is also true that whenever one has a partition, one
acquires an equivalence relation. In fact, the possible partitions of a set and the possible
equivalence relations on that set are in one-to-one correspondence with one another. For
an example of an equivalence relation derived from a partition, consider that due to the
high enrollment in first year calculus at many universities, it is necessary to offer several
sections of the same course to accommodate all of the students. We thus partition the set
of all first year calculus students into different first year calculus classes. The grouping of
students into these classes may be completely arbitrary, in that the students in a particular
class may not share any other similarity than just that they happen to have been placed in
the same class. However, by thus being placed into the same class, the students acquire the
similarity of belonging to the same class. This is true in general; as soon as one partitions
any set into classes, the elements of each class acquire the relation of belonging to the same
class. This relation is an equivalence relation.
For a partition A of a set U, define the relation ∼A on U by
∼A = {(x, y) ∈ U × U | ∃S ∈ A, x ∈ S and y ∈ S }.
That is, for x, y ∈ U, x ∼A y if and only if ∃S ∈ A, x ∈ S and y ∈ S .

We first verify that for a partition A, the relation ∼A is an equivalence relation:
Theorem 3.1.6.
Let A be a partition of a set U. Then ∼A is an equivalence relation.
Proof.
Let x ∈ U.
Since S ∈A S = U, we have ∃S ∈ A, x ∈ S .
S
Therefore, ∃S ∈ A, x ∈ S and x ∈ S , which means x ∼A x.
Therefore, ∼A is reflexive.
Let x, y ∈ U.
Assume x ∼A y.
Then ∃S ∈ A, x ∈ S and y ∈ S .
Therefore, ∃S ∈ A, y ∈ S and x ∈ S ,
and so y ∼A x.
Therefore, if x ∼A y, then y ∼A x.
Therefore, ∼A is symmetric.
Let x, y, z ∈ U.
Assume x ∼A y and y ∼A z.
Choose S ∈ A such that x ∈ S and y ∈ S .
Choose T ∈ A such that y ∈ T and z ∈ T .
Then y ∈ S ∩ T ; hence S ∩ T , ∅.
Therefore, S = T ; hence z ∈ S .
Therefore, ∃S ∈ A, x ∈ S and z ∈ S , hence x ∼A z.
Therefore, if x ∼A y and y ∼A z, then x ∼A z.
Therefore, ∼A is transitive.
Thus, ∼A is an equivalence relation.
Together, theorems 3.1.5 and 3.1.6 tell us that to each equivalence relation corresponds a
partition and to each partition corresponds an equivalence relation. In fact, the correspon-
dences are inverses of one another, in the following sense:
Theorem 3.1.7.
Let A be a partition of a set U, and let ∼ be ∼A , the corresponding equivalence relation.

Then U/∼ = A.
Proof.
First notice that for all a ∈ U and S ∈ A, if a ∈ S , then S = [a].
Indeed, let S ∈ A, assume a ∈ S .
Let x ∈ S .
Then a, x ∈ S , so x ∼ a
Hence, x ∈ [a].
Thus, S ⊆ [a].
Conversely, let x ∈ [a].
Then, x ∼ a, so we can choose T ∈ A with x, a ∈ T .
But now a ∈ S ∩ T , so S = T , by property 2 of partition.
Therefore, x ∈ S .
Thus, [a] ⊆ S , and we have S = [a].
Therefore, for all a ∈ U and S ∈ A, if a ∈ S , then S = [a].
Now, to prove U/∼ = A,

let S ∈ A.
Then S , ∅, so we can choose a ∈ U with a ∈ S ;
Hence S = [a] as shown above, so S ∈ U/∼.
Therefore, A ⊆ U/ ∼.
Conversely, let S ∈ U/∼.
Choose a ∈ U so that S = [a].
Since U is the union of A, we choose T ∈ A with a ∈ T ,
so that again S = [a] = T ∈ A,
Thus, the families U/∼ and A are equal.
Modular Arithmetic
To put the theory of equivalence relations and their classes in a more concrete setting,
we draw attention to one specific type of equivalence relation on the integers. Modular
arithmetic is a very common and useful bit of mathematics, and some students may already
have some exposure to this concept. However, familiarity with modular arithmetic is not
necessary for this section. Our purpose here is not to develop a comprehensive theory of
modular math. That would take too long and take us too far off course. Rather, we are only
interested in considering modular arithmetic as an example of equivalence relations and to
show how the theory of modular arithmetic is built on the foundation laid by the theory of
equivalence relations.
Definition 3.1.5. For n ∈ Z, we define the relation congruence modulo

n to be
≡n = {(x, y) ∈ Z × Z | x − y ∈ hni}.
That is, for x, y ∈ Z, we say x and y are congruent modulo n, denoted
x ≡n y, if and only if x − y ∈ hni.
Notation
For x, y, n ∈ Z, the notation x ≡ y (mod n) is often used to denote x ≡n y.
Proposition 3.1.8.
For each n ∈ Z, ≡n is an equivalence relation on Z.
Proof.
Let n ∈ Z.
Let x ∈ Z.
Since x − x = 0 = (0)(n) ∈ hni, we have x ≡n x.
Hence, ≡n is reflexive.
Next, let x, y ∈ Z.
Assume x ≡n y.
Then x − y ∈ hni.
Choose a ∈ Z such that x − y = na.
Then y − x = −na = n(−a) ∈ hni.
Therefore, y ≡n x.
This shows that if x ≡n y, then y ≡n x.
Therefore, ≡n is symmetric.
Finally, let x, y, z ∈ Z.
Assume x ≡n y and y ≡n z.
That is, x − y ∈ hni and y − z ∈ hni.
Choose a, b ∈ Z such that x − y = na and y − z = nb.
Then x − z = (x − y) + (y − z) = na + nb = n(a + b) ∈ hni.
Therefore, x ≡n z.
Therefore, if x ≡n y and y ≡n z, then x ≡n z.
Thus ≡n is transitive.
Since ≡n is reflexive, symmetric, and transitive, ≡n is an equivalence relation.
Therefore, for all n ∈ Z, ≡n is an equivalence relation on Z.
Proposition 3.1.9.
For all n, a ∈ Z, the equivalence class of a relative to the relation ≡n is
[a]≡n = {x ∈ Z | ∃t ∈ Z, x = a + nt}.
Proof.
Let n, a ∈ Z.
To simplify notation, let A = {x ∈ Z | ∃t ∈ Z, x = a + nt}.
Let x ∈ Z.
x ∈ [a]≡n if and only if x ≡n a.
x ≡n a if and only if x − a ∈ hni.
x − a ∈ hni if and only if ∃t ∈ Z, x − a = nt.
∃t ∈ Z, x − a = nt if and only if ∃t ∈ Z, x = a + nt.
∃t ∈ Z, x = a + nt if and only if x ∈ A.
Therefore, x ∈ [a]≡n if and only if x ∈ A.
Therefore, [a]≡n = A.
Therefore, for all n, a ∈ Z, [a]≡n = {x ∈ Z | ∃t ∈ Z, x = a + nt}.
Notation
When working with congruence modulo n, we often allow ourselves the following notation:
for n, a ∈ Z, a + hni is defined to be {x ∈ Z | ∃t ∈ Z, x = a + nt}.
The notation a + hni is not in any way meant to indicate that an integer can be added to a
set. It is simply meant as notation representing the set indicated.
Also, when the relation is congruence modulo n, we will simplify the notation for
the equivalence class of a as follows:
for n, a ∈ Z, [a]n is defined to be [a]≡n .
Using this notation, proposition 3.1.9 becomes: For all n, a ∈ Z, [a]n = a + hni.
Theorem 3.1.10.
Let n, a, b ∈ Z. The following are equivalent:
1. a − b ∈ hni
2. a ∈ b + hni
3. a + hni ⊆ b + hni
4. a + hni = b + hni
5. (a + hni) ∩ (b + hni) , ∅.
Proof.
For the relation ≡n , by theorem 3.1.3, we have equivalence of the propositions
1. a ≡n b
2. a ∈ [b]
3. [a]n ⊆ [b]n
4. [a]n = [b]n
5. [a]n ∩ [b]n = ∅.
Since [a]n = a + hni, [b]n = b + hni, and a ≡n b means a − b ∈ hni, we have equivalence of
the propositions
1. a − b ∈ hni
2. a ∈ b + hni
3. a + hni ⊆ b + hni
4. a + hni = b + hni
5. (a + hni) ∩ (b + hni) , ∅.
We have already seen that for the relation ≡2 , [0]2 = E and [1]2 = O. Further, this is a
complete list of distinct equivalence classes, since for any even x, [x]2 = [0]2 and for any
odd x, [x]2 = [1]2 . In general, for any n ∈ N, there will be n distinct equivalence classes
relative to the relation ≡n . In fact, {[0]n , [1]n , . . . , [n − 1]n } is a complete list of equivalence
classes relative to ≡n . The proof of this claim uses the division algorithm:
Theorem 3.1.11.
For all n ∈ N, Z/ ≡n = {[x]n | 0 ≤ x < n}.
Proof.
Let n ∈ N.
{[x]n | 0 ≤ x < n} ⊆ Z/ ≡n ,
since Z/ ≡n is the universe of discourse in which {[x]n | 0 ≤ r < n} is defined.
Next, let A ∈ Z/ ≡n . That is, A = [a]n for some a ∈ Z.
Choose such an a ∈ Z.
Applying the division algorithm, choose q, r ∈ Z, a = qn + r and 0 ≤ r < n.
Then a − r = qn ∈ hni; hence a ≡n r.
Therefore, [a]n = [r]n .
Since 0 ≤ r < n, we have [r]n ∈ {[x]n | 0 ≤ x < n}.
Therefore, [a]n ∈ {[x]n | 0 ≤ x < n}.
Therefore, Z/ ≡n ⊆ {[x]n | 0 ≤ x < n}.
Therefore, for all n ∈ N, Z/ ≡n = {[x]n | 0 ≤ x < n}.
In fact, not only is {[x]n | 0 ≤ x < n} a complete list of all equivalence classes relative to
≡n , but these equivalence classes are also distinct. That is, if 0 ≤ r1 < n and 0 ≤ r2 < n
and r1 , r2 , then [r1 ]n , [r2 ]n . The reason is that the quotient and remainder given by the
division algorithm are unique for a given numerator and denominator. The proof of this
uniqueness was included as an exercise in section 1.2 (exercise 26). We prove this result
again in this setting:
Theorem 3.1.12.
For all n ∈ N, if 0 ≤ r1 < n and 0 ≤ r2 < n and r1 , r2 , then [r1 ]n , [r2 ]n . Where the
equivalence classes are relative to ≡n .
Proof.
Let n ∈ N
Suppose 0 ≤ r1 < n and 0 ≤ r2 < n and r1 , r2 and [r1 ]n = [r2 ]n .
Then r1 ≡n r2 , which means r1 − r2 ∈ hni.
Therefore, n divides r1 − r2 .
Now, since r1 , r2 , we have r1 − r2 , 0.
Therefore, by Proposition 1.2.8, we have n ≤ |r1 − r2 |.
Also, by Proposition 1.1.30, we have |r1 − r2 | ≤ max(r1 , r2 ).
By transitivity, we now have n ≤ max(r1 , r2 ).
However, this presents a problem: since r1 < n and r2 < n, we have max(r1 , r2 ) < n.
Thus, have the contradiction n ≤ max(r1 , r2 ) and max(r1 , r2 ) < n.
Therefore, if 0 ≤ r1 < n and 0 ≤ r2 < n and r1 , r2 , then [r1 ]n , [r2 ]n .
Therefore, for all n ∈ N, if 0 ≤ r1 < n and 0 ≤ r2 < n and r1 , r2 , then [r1 ]n , [r2 ]n .
Definition 3.1.6. For n ∈ N, we denote by Zn the set of equivalence

classes relative to the relation ≡n . That is,
Zn = Z/ ≡n = {[x]n | 0 ≤ x < n}.
For example, consider the relation ≡5 and its equivalence classes Z5 =

{[0]5 , [1]5 , [2]5 , [3]5 , [4]5 }. Recall that [a]5 denotes the set of all integers congruent
to a modulo 5. That is,
[0]5 = {. . . , −15, −10, −5, 0, 5, 10, 15, 20, . . . }
[1]5 = {. . . , −14, −9, −4, 1, 6, 11, 16, 21, . . . }

[2]5 = {. . . , −13, −8, −3, 2, 7, 12, 17, 22, . . . }
[3]5 = {. . . , −12, −7, −2, 3, 8, 13, 18, 23, . . . }
[4]5 = {. . . , −11, −6, −1, 4, 9, 14, 19, 24, . . . }
The equivalence classes relative to ≡n have an interesting property that is not necessar-
ily shared by equivalence classes relative to other relations; they preserve addition and
multiplication. What we mean by this can be seen in the following example: Select any
element from the equivalence class [2]5 and any element from the equivalence class [4]5 ,
say −8 ∈ [2]5 and 14 ∈ [4]5 . If we add these numbers together, we get −8 + 14 = 6,
which is in the equivalence class [1]5 . We claim that this result is independent of the choice
of elements −8 and 14. That is, had we chosen say 17 ∈ [2]5 and −1 ∈ [4]5 , we have
17 + (−1) = 16, which is still in the equivalence class [1]5 . Trying a few more examples,
we begin to suspect that for any x ∈ [2]5 and any y ∈ [4]5 , we will have x + y ∈ [1]5 .
Multiplication of elements from two equivalence classes is similarly independent of the
choice of which elements to multiply. For example, taking −3 ∈ [2]5 and 4 ∈ [4]5 gives
us (−3)(4) = −12 ∈ [3]5 . Selecting different elements, 7 ∈ [2]5 and −1 ∈ [4]5 gives us
(7)(−1) = −7 ∈ [3]5 . In general, we find that for any x ∈ [2]5 and any y ∈ [4]5 , we will
have xy ∈ [3]5 . This works for equivalence classes relative to ≡n , for any n:
Theorem 3.1.13.
Let n ∈ Z. For all a, b, x, y ∈ Z, if x ∈ [a]n and y ∈ [b]n , then x + y ∈ [a + b]n and xy ∈ [ab]n .
Proof.
Let n ∈ Z.
Let a, b, x, y ∈ Z, and assume x ∈ [a]n and y ∈ [b]n .
Then x ≡n a and y ≡n b.
That is, x − a ∈ hni and y − b ∈ hni.
Choose s, t ∈ Z such that x − a = ns and y − b = nt.
Then (x + y) − (a + b) = x − a + y − b = ns + nt = n(s + t) ∈ hni.
Therefore, (x + y) ≡n (a + b); hence x + y ∈ [a + b]n .
Also, xy − ab = xy − ay + ay − ab = (x − a)y + a(y − b) = nsy + nta.
Thus, xy − ab = n(sy + ta) ∈ hni.
Therefore, xy ≡n ab; hence xy ∈ [ab]n .
Therefore, for all a, b, x, y ∈ Z, if x ∈ [a]n and y ∈ [b]n , then x + y ∈ [a + b]n and
xy ∈ [ab]n . Where the equivalence classes are relative to the relation ≡n .
Q.E.D.
Notice that given the equivalence of the statement x ∈ [a]n to the statement [x]n = [a]n , the
above theorem may be rewritten as
For all a, b, x, y ∈ Z, if [x]n = [a]n and [y]n = [b]n , then [x+y]n = [a+b]n and [xy]n = [ab]n .
That is, the result of addition and multiplication is independent of the choice of rep-
resentative for the equivalence class. This enables us to define addition and multiplication
of the equivalence classes themselves. For example, since modulo 5 we see that any
element of [2]5 added to any element of [4]5 results in an element of [1]5 , we can define
without ambiguity [2]5 + [4]5 = [1]5 . Similarly, since any element of [2]5 multiplied by
any element of [4]5 produces an element of [3]5 , we may define [2]5 [4]5 = [3]5 . In general,
we define the following addition and multiplication of equivalence classes:
Definition 3.1.7. For n ∈ N, relative to the relation ≡n , we define for

x, y ∈ Z, [x]n + [y]n = [x + y]n and [x]n [y]n = [xy]n .
We encourage caution when working with addition and multiplication of equivalence

classes, since the addition and multiplication as defined here do not necessarily satisfy
all of the same axioms that addition and multiplication in the integers of real numbers sat-
isfy. For example, in the integers, if we know that 2x = 6, we can deduce that x = 3. The
same deduction cannot be made in, for example, in the integers modulo 8. The integers
modulo 8 have the following multiplication table:
Z8 [0]8 [1]8 [2]8 [3]8 [4]8 [5]8 [6]8 [7]8

[0]8 [0]8 [0]8 [0]8 [0]8 [0]8 [0]8 [0]8 [0]8
[1]8 [0]8 [1]8 [2]8 [3]8 [4]8 [5]8 [6]8 [7]8
[2]8 [0]8 [2]8 [4]8 [6]8 [0]8 [2]8 [4]8 [6]8
[3]8 [0]8 [3]8 [6]8 [1]8 [4]8 [7]8 [2]8 [5]8
[4]8 [0]8 [4]8 [0]8 [4]8 [0]8 [4]8 [0]8 [4]8
[5]8 [0]8 [5]8 [2]8 [7]8 [4]8 [1]8 [6]8 [3]8
[6]8 [0]8 [6]8 [4]8 [2]8 [0]8 [6]8 [4]8 [2]8
[7]8 [0]8 [7]8 [6]8 [5]8 [4]8 [3]8 [2]8 [1]8
We see that [2]8 [x]8 = [6]8 has two solutions: [x]8 = [3]8 and [x]8 = [7]8 . The reason we
cannot solve linear equations in the integers modulo 8 is that Z8 does not satisfy the factors
of zero axiom. In the integers, we have that ∀x, y ∈ Z, if xy = 0, then x = 0 or y = 0.
Whereas in Z8 , there are many non-zero equivalence classes that multiply to produce the
zero equivalence class. For example, [2]8 [4]8 = [0]8 , [4]8 [4]8 = [0]8 , and [4]8 [6]8 = [0]8 .
Exercises 3.1.
Let A and B be sets in a universe of discourse U, and let 21. The relation R on the set R2 given by:
X and Y be sets in a universe of discourse V. Let S and ∀(x1 , x2 ), (y1 , y2 ) ∈ R2 , (x1 , x2 )R(y1 , y2 ) if and only
T be subsets of U ×V. Prove the following propositions. if y2 + 2x1 = x2 + 2y1 .
1. If A ⊆ B, then A × X ⊆ B × X. 22. The relation R on the set U = {(x1 , x2 ) ∈ R2 | x2 ,
0} given by ∀(x1 ,"x2 ), (y1 , y2#) ∈ U, (x1 , x2 )R(y1 , y2 )
2. If A ⊆ B and X ⊆ Y, then A × X ⊆ B × Y. x1 y1
if and only if det = 0.
3. (A × X) ∩ (B × Y) = (A ∩ B) × (X ∩ Y). x2 y2
4. (A × X) ∪ (B × Y) = (A ∪ B) × (X ∪ Y). 23. The relation R on the set R3 given by

∀(x1 , x2 , x3 ), (y1 , y2 , y3 ) ∈ R3 , (x1 , x2 , x3 )R(y1 , y2 , y3 )
5. If S ⊆ A × X and T ⊆ B × Y, then S ∩ T ⊆ if and only if ∃t ∈ R\{0}, (x1 , x2 , x3 ) = (ty1 , ty2 , ty3 ).
(A ∩ B) × (X ∩ Y).
24. The relation R on the set U = {A | A is an invertible 2×
6. If S ⊆ A × X and T ⊆ B × Y, then S ∪ T ⊆ 2 matrix} given by: ∀A, B ∈ U, ARB if and only if
(A ∪ B) × (X ∪ Y). ∃P ∈ U, AP = PB.
7. (A \ B) × X = (A × X) \ (B × X).
Prove the following propositions. These are analogous
8. A × (X \ Y) = (A × X) \ (A × Y). to theorem 3.1.13.
9. (A \ B) × (X \ Y) ⊆ (A × X) \ (B × Y). 25. For the relation ≡Q defined in question 15, prove
c c c ∀a, b, x, y ∈ R, if x ∈ [a]≡Q and y ∈ [b]≡Q , then
10. A × X ⊆ (A × X) .
x + y ∈ [a + b]≡Q .
11. (A × X)c = (U × X c ) ∪ (Ac × V).
26. For the relation ≡Z defined in question 14,
12. (A × X) \ (B × Y) = (A × (X \ Y)) ∪ ((A \ B) × X). ∀a, b, x, y ∈ R, if x ∈ [a]≡Z and y ∈ [b]≡Z , then
x + y ∈ [a + b]≡Z .
Prove that each of the following are equivalence rela-

tions. Describe the equivalence classes. Prove the following propositions about partitions.
13. The relation R on the set R \ {0} given by: ∀x, y ∈ 27. Let A, B ⊆ Z with A , B. Let P = {A, B}. If P is a
R \ {0}, xRy if and only if xy > 0. partition of Z, then B , Z.
14. The relation ≡Z on the set R given by: ∀x, y ∈ R, 28. Let A, B ⊆ Z with A , B. Let P = {A, B}. If P is a
x ≡Z y if and only if x − y ∈ Z. partition of Z, then A = Z \ B.
15. The relation ≡Q on the set R given by: ∀x, y ∈ R, 29. For each n ∈ N, let S n = {x ∈ R | n − 1 ≤ x2 < n}.
x ≡Q y if and only if x − y ∈ Q. Then A = {S n | n ∈ N} is a partition of R.
30. A = {(k − 1, k] | k ∈ Z} is a partition of R.
16. The relation ≡Q∗ on the set R \ {0} given by: ∀x, y ∈
x
R \ {0}, x ≡Q∗ y means ∈ Q. 31. For each y ∈ R, let Ay = {x ∈ R | y = x2 }. Then
y A = {Ay | y ∈ [0, ∞)} is a partition of R.
17. The relation R on the set Z given by: ∀x, y ∈ Z, xRy
32. For the sets Ay as defined in exercise 31, is B =
if and only if x2 = y2 .
{Ay | y ∈ R} a partition of R? Prove or disprove.
33. The family of sets A = {h3i, 1 + h3i, 2 + h3i} is a
if and only if ∃a, b ∈ O, ax = by.
partition of Z.
34. For each n ∈ N, the family of sets A = {r + hni | r ∈
if and only if ∃n ∈ Z, x = 2n y.
Z and 0 ≤ r < n} is a partition of Z.
20. The relation R on the set R given by: ∀x, y ∈ R, 35. A = {h4i, 2 + h4i} is a partition of h2i.
xRy if and only if ∃n ∈ Z, n − 1 < x ≤ n and
n − 1 < y ≤ n. 36. A = {h6i, 3 + h6i} is a partition of h3i.
37. For each n ∈ Z≥0 , let An = 2n + h2n+1 i. Then 49. Prove your answer to exercise 48.
A = {An | n ∈ Z≥0 } is a partition of Z \ {0}.
50. Use the statement in exercise 42 to prove ∀x, y, z ∈
38. For each k ∈ Z, let Ak = {x ∈ Z | ∃n ∈ Z≥0 , x = Z, if [x]5 [z]5 = [y]5 [z]5 and [z]5 , [0]5 , then
2n + 2n+1 k}. Then A = {Ak | k ∈ Z} is a partition of [x]5 = [y]5 .
Z \ {0}.
51. Use the statement in exercise 46 to prove ∀x, y, z ∈
Z, if [x]5 [z]5 = [y]5 [z]5 and [z]5 , [0]5 , then
The following exercises investigate multiplication of
[x]5 = [y]5 .
congruence classes.
39. Write a complete multiplication table for the family 52. Disprove the statement ∀x, y, z ∈ Z, if [x]6 [z]6 =
of equivalence classes Z5 . [y]6 [z]6 and [z]6 , [0]6 , then [x]6 = [y]6 .
40. Write a complete multiplication table for the family

of equivalence classes Z4 . Prove the following propositions about congruence
classes.
41. Write a complete multiplication table for the family
of equivalence classes Z6 . 53. Let n ∈ N. ∀x, y, z ∈ Z, x − y ∈ [z]n if and only if
x ∈ [y + z]n .
42. Decide whether the following statement is true or
false: ∀x, y ∈ Z, if [x]5 [y]5 = [0]5 , then [x]5 = [0]5 54. ∀n, m ∈ Z, (x + hni) ∩ (x + hmi) = x + hni ∩ hmi.
or [y]5 = [0]5 . To which axiom of the integers is
this similar? 55. h2i \ h4i = 2 + h4i.
43. Prove your answer to exercise 42. 56. h2i \ (2 + h4i) = h4i.
57. h3i \ (3 + h6i) = h6i.
false: ∀x, y ∈ Z, if [x]6 [y]6 = [0]6 , then [x]6 = [0]6
or [y]6 = [0]6 . 58. h3i \ h6i = 3 + h6i.
45. Prove your answer to exercise 44. 59. Let n ∈ N. ∀a, x ∈ Z, if a ∈ [x]n , then −a ∈ [−x]n .
false: ∀x ∈ Z, if [x]5 , [0]5 , then ∃y ∈ Z, 60. Let n ∈ N. ∀a, x ∈ Z, if a ∈ [x]n , then ∀m ∈ N,
[x]5 [y]5 = [1]5 . To which axiom of the real num- am ∈ [xm ]n .
bers is this similar? 61. ∀n ∈ N, if n , 1 and n is not prime, then ∃x, y ∈ Z,
47. Prove your answer to exercise 46. [x]n [y]n = [0]n but [x]n , [0]n and [y]n , [0]n .
48. Decide whether the following statement is true or 62. ∀n ∈ N, if n is prime, then ∀x, y ∈ Z, if [x]n [y]n =
false: ∀x ∈ Z, if [x]6 , [0]6 , then ∃y ∈ Z, [0]n , then [x]n = [0]n or [y]n = [0]n . (Hint: use the
[x]6 [y]6 = [1]6 . result proven in exercise 23 of section 2.3).
3.2 Order Relations
Order relations are those that express, in some sense, that one element of a set is smaller
than another element in a set, or that one element precedes another element. Simple ex-
amples of order relations in everyday life are ‘object x is no heavier than object y,’ ‘event
x occurred no later than event y,’ ‘statement x implies statement y,’ ‘event x caused event
y,’ and so on. Each of these expresses a different notion of something being smaller than
or preceding something else. For common examples of order relations in mathematics, we
have the relation ≤ on the set R as well as on the set Z, and the relation ⊆ on the power set
of any set U.
To study order relations in general, we look for properties that are common to all order re-
lations and study all relations having those properties. Consider, for example, the relations
≤ on R and ⊆ on P(U) for some set U. We notice that both relations are reflexive:
∀x ∈ R, x ≤ x and ∀A ∈ P(U), A ⊆ A.
Likewise, both relations are transitive:
∀x, y, z ∈ R, if x ≤ y and y ≤ z, then x ≤ z
and ∀A, B, C ∈ P(U), if A ⊆ B and B ⊆ C, then A ⊆ C.

Moreover, although neither relation is symmetric, both fail to be symmetric in the same
way: the only time x ≤ y and y ≤ x is when x = y. Likewise, the only time A ⊆ B and
B ⊆ A is when A = B. This lack of symmetry in all cases except when the two subjects are
equal is called antisymmetry:
Definition 3.2.1. Let R be a relation on a set U. R is antisymmetric

means
∀x, y ∈ U, if xRy and yRx, then x = y.
Thus, we say that the relations ≤ and ⊆ are both antisymmetric. To study these two rela-
tions simultaneously (along with any other similar order relations), we define the class of
relations satisfying these properties:
Definition 3.2.2. Let R be a relation on a set U. R is a partial ordering

means R is reflexive, antisymmetric, and transitive.
3.2. ORDER RELATIONS 207
There is one important difference between our two motivating examples ≤ and ⊆: There
are sets A and B for which neither A ⊆ B nor B ⊆ A. For example, {0} * {1} and {1} * {0}.
This is actually a very common occurrence for order relations. For example, consider the
common relation ‘event x caused event y.’ It is possible that events can be so remote that
neither event can be said to have caused the other. That is, there are events x and y for
which x did not cause y and y did not cause x. However, no such analogy exists for the
relation ≤ on R. In fact, by trichotomy, we have for all x, y ∈ R, at least one of x ≤ y or
y ≤ x is always true. To make this distinction, we refer to the relation ≤ and others like it
as total orderings.
Definition 3.2.3. Let R be a relation on a set U. R is a total ordering

means R is a partial ordering that also satisfies the property
∀x, y ∈ U, xRy or yRx.
The phrase linear ordering is often used as a synonym for total ordering.
The ‘divides’ relation
Another very common example of a partial ordering in mathematics is the divides relation
on the natural numbers:
Let | denote the relation on N given by ∀a, b ∈ N, a|b if and only if

∃t ∈ N, b = at. (that is, a divides b).
Proposition 3.2.1.
The divides relation | on N is a partial ordering.
Proof.
Let x ∈ N.
Putting t = 1 gives us x = xt; hence x|x.
Therefore, | is reflexive.
Let x, y ∈ N.
Assume x|y and y|x.
Choose s, t ∈ N such that y = xs and x = yt.
Then x = xst. Since x , 0, we then have st = 1.
Since t divides 1, we have t ≤ 1; hence t = 1 since t ∈ N.
Therefore, x = y(1), which means x = y.
Therefore, if x|y and y|x, then x = y.
Hence, | is antisymmetric.
Let x, y, z ∈ N.
Assume x|y and y|z.
Choose s, t ∈ N such that y = xs and z = yt.
Then z = x(st); hence x|z.
Therefore, if x|y and y|z, then x|z.
Therefore, | is transitive.
Since | is reflexive, symmetric, and transitive, | is a partial ordering.
Notice that | is not a total ordering, since for example 2 - 3 and 3 - 2.
The divides relation remains a partial ordering if we allow 0 to be in the set. That is, the
divides relation is also a partial ordering on the set Z≥0 . However, if we include negative
numbers, the relation fails to be antisymmetric. That is, the divides relation is not a partial
ordering on the set Z. The proofs of these facts are left as exercises.
Extreme Elements of a Set
When working with different partial orderings, it is often useful to identify certain extreme
elements of a set. In fact, we will see that many common concepts in mathematics (in-
cluding unions, intersections, greatest common divisors, and least common multiples) are
simply the extreme elements of sets considered under different partial orderings. We begin
the the most obvious type of extreme elements: the greatest and least elements of the set (if
they exist).
Note: throughout this section, we will use the symbol 4 to denote an

arbitrary partial ordering.
Definition 3.2.4. Let 4 be a partial ordering on a set U, and let A be a

subset of U. For a ∈ U, a is the least element of A means
a ∈ A and ∀x ∈ A, a 4 x.
a is the greatest element of A means
a ∈ A and ∀x ∈ A, x 4 a.
It is important to note that not every set has a least or greatest element. For example, the
real interval (0, 1) has neither a greatest nor least element under the usual ordering ≤. Under
the divides relation, the number 1 is the least element of the set N, since 1 ∈ N and ∀x ∈ N,
1|x. Yet, N has no greatest element under the partial ordering |, since there is no natural
number a for which ∀x ∈ N, x|a. However, if we extend the divides relation to allow 0
(that is if it is considered as a relation on the set Z≥0 ), then the set Z≥0 has both a least and
greatest element under the relation |.
Proposition 3.2.2.
Under the partial ordering |, 1 is the least element of Z≥0 , and 0 is the greatest element of
Z≥0 .
Proof.
Both 0 and 1 belong to the set Z≥0 , so it only remains to show:
∀x ∈ Z≥0 , 1|x, and ∀x ∈ Z≥0 , x|0.
To this end, let x ∈ Z≥0 .
Since x = x(1), we have that 1 divides x.
Since 0 = 0(x), we have that x divides 0.
Therefore, ∀x ∈ Z≥0 , 1|x, and ∀x ∈ Z≥0 , x|0.
Hence, 1 is the least element of Z≥0 , and 0 is the greatest element of Z≥0 .
To see how to interpret the idea of greatest element and least element under different partial
orderings, we give a few more examples.
Proposition 3.2.3.
Let U be a set. Under the partial ordering ⊆, ∅ is the least element of P(U), and U is the
greatest element of P(U).
Proof.
Since ∅ ⊆ U, we have ∅ ∈ P(U).
Let X ∈ P(U).
Then ∅ ⊆ X.
Therefore, ∀X ∈ P(U), ∅ ⊆ X.
Hence, ∅ is the least element of P(U).
Since U ⊆ U, we have U ∈ P(U).
Further, for any X ∈ P(U), we have X ⊆ U.
Therefore, U is the greatest element of P(U).
Proposition 3.2.4.
Under the partial ordering ⊆ on the family of sets I = {hni | n ∈ Z}, h0i is the least element
of I, and h1i is the greatest element of I.
Proof.
Since 0 ∈ Z, h0i ∈ {hni | n ∈ Z} = I.
Let X ∈ I.
Then X = hni for some n ∈ Z. Choose such an n.
Let x ∈ h0i.
Since we can choose t ∈ Z with x = 0t, we have x = 0.
Therefore, x = 0(n) ∈ hni.
Therefore, h0i ⊆ hni.
Therefore, ∀X ∈ I, h0i ⊆ X.
Hence, h0i is the least element of I.
Likewise, since 1 ∈ Z, we have h1i ∈ {hni | n ∈ Z} = I.
Again, let X ∈ I and choose n ∈ Z with X = hni.
Let x ∈ hni.
Then x = (x)(1) ∈ h1i.
Therefore, hni ⊆ h1i.
Now, ∀X ∈ I, X ⊆ h1i,
which together with the fact that h1i ∈ I means
h1i is the greatest element of I.
Supremum and Infimum of a Set
Besides the greatest and least elements of a set, there are other notions of extreme elements
that have important applications. We begin by relaxing the requirement that an extreme
element must itself be in the set.

subset of U. For a ∈ U, a is a lower bound of A means
∀x ∈ A, a 4 x.
a is an upper bound of A means
∀x ∈ A, x 4 a.
The definitions of lower bounds and upper bounds appear similar to those of least and
greatest elements, in that both require the extreme element a to satisfy ∀x ∈ A, a 4 x and
∀x ∈ A, x 4 a respectively. The difference is in the fact that in the case of least and greatest
elements, the extreme element a must itself be a member of the set A, whereas to be a lower
or upper bound, a need not be in the set A.
For example, the real number 0 is a lower bound of the set (0, 1), but since 0 itself is not
in the set, it is not the least element. For another example, under the partial ordering |, the
natural number 12 is an upper bound of the set {1, 2, 3, 4, 6}, but again it is not called the
greatest element since 12 is not in the set. Likewise, even though 0 < N, 0 is still an upper
bound of the set N under the partial ordering |, since for all x ∈ N, x|0.
Much like least and greatest elements, there are sets for which lower and upper bounds do
not exist. For example, the set R has no upper bounds and no lower bounds under the usual
ordering ≤. However, unlike least and greatest elements, upper and lower bounds (if they
do exist) are not necessarily unique. That is, one set can have multiple lower bounds and
multiple upper bounds. For example, for the real interval (0, 1), the numbers 1 and 2 are
both upper bounds, since ∀x ∈ (0, 1), x ≤ 1 and ∀x ∈ (0, 1), x ≤ 2. In fact, any number
that is greater than or equal to 1 is an upper bound of (0, 1). Likewise, any number that is
less than or equal to 0 is a lower bound of (0, 1). For another example, under the partial
ordering |, both 36 and 72 are upper bounds of the set {6, 12, 18}, and both 2 and 3 are lower
bounds of the set {6, 12, 18}. In fact, for the set {6, 12, 18}, any multiple of 36 is an upper
bound, and any divisor of 6 is a lower bound. In situations where there are many lower or
upper bounds, the lower bound that is greatest among all other lower bounds, and likewise
the upper bound that is least among all other upper bounds, are often of interest. We take
the time to distinguish these particular lower and upper bounds:
Definition 3.2.6. Let R be a partial ordering on a set U, and let A be

a subset of U. For a ∈ U, a is the infimum of A, denoted a = inf(A),
means a is a lower bound of A and ∀x ∈ U, if x is a lower bound of A,
then xRa. The infimum is often called by the more descriptive name the
greatest lower bound of A.
a is the supremum of A, denoted a = sup(A), means a is an up-

per bound of A and ∀x ∈ U, if x is an upper bound of A, then aRx. The
supremum is often referred to as the least upper bound of A.
Consider the following examples of suprema and infima for sets of real numbers under the
usual ordering ≤.
Example 3.2.5.
1 = inf(1, 3), under the partial ordering ≤.
Proof.
For any x ∈ (1, 3), we have 1 < x; hence 1 ≤ x.
Therefore, ∀x ∈ (1, 3), 1 ≤ x, which means 1 is a lower bound of the set (1, 3).
Now, let x be any lower bound of the set (1, 3).
Suppose 1 < x.
Since, for example, 2 ∈ (1, 3), we have x ≤ 2 which means x < 3.
Choosing y = 1+x2 gives us 1 < y < x < 3.
Then y ∈ (1, 3). Since x is a lower bound of (1, 3), this means x ≤ y.
This gives us the contradiction y < x and x ≤ y.
Therefore, x ≤ 1.
Therefore, if x is a lower bound of (1, 3), then x ≤ 1.
Therefore, 1 is the greatest lower bound of (1, 3).
That is, 1 = inf(1, 3).
Example 3.2.6.
3 = sup(1, 3), under the partial ordering ≤.
Proof.
We first prove that 3 is an upper bound of (1, 3):
Let x ∈ (1, 3).
Then x < 3, which implies x ≤ 3.
Therefore, ∀x ∈ (1, 3), x ≤ 3.
Hence, 3 is an upper bound of (1, 3).
Next we will show that if x is an upper bound of (1, 3), then 3 ≤ x.
To this end, suppose x is an upper bound of (1, 3) but x < 3.
Put y = x+3
2 , so that x < y < 3.
Since 2 ∈ (1, 3) and x is an upper bound of (1, 3), we have 2 ≤ x.
By transitivity, 1 < x; hence 1 < x < y < 3.
Therefore, y ∈ (1, 3), which implies y ≤ x, since x is an upper bound.
Then x < y and y ≤ x, which is a contradiction.
Therefore, if x is an upper bound of (1, 3), then 3 ≤ x.
Thus, 3 = sup(1, 3).
Example 3.2.7.
0 = inf{x ∈ R | ∃n ∈ N, x = 1n }, under the partial ordering ≤.
Proof.
To simplify notations, let A = {x ∈ R | ∃n ∈ N, x = 1n }.
We first prove that 0 is a lower bound of A:
Let x ∈ A.
Choose n ∈ N with x = 1n .
Since 0 < n, we have 0 < n1 ; hence 0 ≤ x.
Therefore, ∀x ∈ A, 0 ≤ x.
That is, 0 is a lower bound of A.
Next we will show that if x is a lower bound of A, then x ≤ 0.
Indeed, suppose x is a lower bound of A but 0 < x.
By the Archimedean property, choose n ∈ N with 1n < x.
Then 1n ∈ A, which is a contradiction since x is a lower bound of A.
Therefore, if x is a lower bound of A, then x ≤ 0.
Thus, 0 = inf A.
Note that even in cases where a set has upper and lower bounds, it is not necessary that an
infimum or supremum of the set exist. For example, the set {x ∈ Q | x2 < 2} has many
upper and lower bounds, but it has neither an infimum nor a supremum in Q. The proofs
of these statements are left as exercises. However, when the infimum or supremum of a set
do exist, they are unique. That is, if there is an infimum, then there is only one. Likewise,
if there is a supremum, then there is only one. We give the proof for infimum and leave the
similar proof for supremum as an exercise.
Proposition 3.2.8.
Let 4 be a partial ordering on a set U, and let A ⊆ U. If a and b are both infima of the set
A, then a = b.
Proof.
Assume both a and b are infima of the set A.
That is, a is a lower bound of A and ∀x ∈ U, if x is a lower bound of A, then x 4 a.
Likewise, b is a lower bound of A and ∀x ∈ U, if x is a lower bound of A, then x 4 b.
Now, since b is a lower bound of A and ∀x ∈ U, if x is a lower bound of A, then x 4 a,
we have b 4 a.
Similarly, since a is a lower bound of A and ∀x ∈ U, if x is a lower bound of A, then
x 4 b,
we have a 4 b.
Now, a 4 b and b 4 a; hence a = b by antisymmetry.
Therefore, if a and b are both infima of the set A, then a = b.
Suprema and Infima for Common Orderings
Many common mathematical concepts can be viewed as suprema and infima under different
orderings. We give a few examples here and leave some others as exercises.
Proposition 3.2.9.
Let U be a set. Under the partial ordering ⊆ on the set P(U), ∀A, B ∈ P(U), A ∩ B =
inf{A, B}.
Proof.
Let A, B ∈ P(U).
Since A ∩ B ⊆ A and A ∩ B ⊆ B,
we have that A ∩ B is a lower bound of the set {A, B}.
Now, suppose C is a lower bound of the set {A, B}.
Then C ⊆ A and C ⊆ B.
Let x ∈ C.
Since C ⊆ A, we have x ∈ A.
Since C ⊆ B, we have x ∈ B.
Since x ∈ A and x ∈ B, we have x ∈ A ∩ B.
Therefore, C ⊆ A ∩ B.
Therefore, if C is a lower bound of the set {A, B}, then C ⊆ A ∩ B.
Therefore, A ∩ B = inf{A, B}.
Therefore, ∀A, B ∈ P(U), A ∩ B = inf{A, B}.
Proposition 3.2.10.
Let U be a set. Under the partial ordering ⊆ on the set P(U), ∀A, B ∈ P(U), A ∪ B =
sup{A, B}.
Proof.
Let A, B ∈ P(U).
Since A ⊆ A ∪ B and B ⊆ A ∪ B,
we have that A ∪ B is an upper bound of the set {A, B}.
Now, suppose C is an upper bound of the set {A, B}.
Then A ⊆ C and B ⊆ C.
Then A ∪ C ⊆ C ∪ C = C and A ∪ B ⊆ A ∪ C.
By transitivity, we have A ∪ B ⊆ C.
Therefore, if C is an upper bound of the set {A, B}, then A ∩ B ⊆ C.
Therefore, A ∪ B = sup{A, B}.
Therefore, ∀A, B ∈ P(U), A ∪ B = sup{A, B}.
Proposition 3.2.11.
Under the partial ordering | on the set N, ∀a, b ∈ N, gcd(a, b) = inf{a, b}.
Proof.
Let a, b ∈ N, and let g = gcd(a, b).
By the definition of gcd(a, b) (definition 1.2.4), we have g|a and g|b.
Therefore, g is a lower bound of the set {a, b}.
Next, let x be any lower bound of the set {a, b}.
Hence, x|a and x|b.
Then, again by the definition of gcd(a, b), we have that x|g.
Therefore, if x is a lower bound of {a, b}, then x|g.
Thus, g is the greatest lower bound of the set {a, b}.
Therefore, ∀a, b ∈ N, gcd(a, b) = inf{a, b}.
Proposition 3.2.12.
Under the partial ordering | on the set N, ∀a, b ∈ N, lcm (a, b) = sup{a, b}.
Proof.
Let a, b ∈ N, and let f = lcm (a, b).
By the definition of lcm (a, b) (definition 1.2.5), we have a| f and b| f .
Therefore, f is an upper bound of the set {a, b}.
Now, let x be any upper bound of the set {a, b}.
This means a|x and b|x.
Then, by the definition of lcm (a, b), we have that f |x.
Therefore, if x is an upper bound of {a, b}, then f |x.
Thus, f is the least upper bound of the set {a, b}.
Therefore, ∀a, b ∈ N, lcm (a, b) = sup{a, b}.
Maximal and Minimal Elements of a Set
There is one final notion of an extreme element of a set that warrants some discussion. To
illustrate what we intend to mean by ‘maximal’ or ‘minimal’ elements of a set, we will
begin with some non-mathematical examples. First, imagine a river that flows East to West
and a city that occupies the Southern shore of the river. In this way, if a person in the city
travels due North from any point, he or she is bound to eventually reach the river. A partial
ordering can be defined on the locations in the city by saying x 4 y provided x = y or
location x is directly South of location y. Thus, a path due North toward the river is an
increasing path under this ordering. Now, the points that lie at the edge of the river bank
are extreme points in some sense, because they are the furthest points one can reach under
this partial ordering. However, they are neither greatest elements nor upper bounds. This is
because no point along the river bank is greater than any other point along the river bank,
as these points lie East or West of one another. Rather, the points on the edge of the river
bank are greatest in the sense that one can go no further than these points; there are no
points beyond them.
For another example, consider the partial ordering on humans given by x 4 y means x = y
or x is a direct descendant of y. That is, under this relation, a person is ordered below
his or her mother and father, grandmother and grandfather, great-grandmother and great-
grandfather, and so on, but does not stand in any order relation with his or her cousins,
aunts, uncles, nephews, siblings, etc. The individuals that are without children are in some
sense extreme elements under this relation. It is at these people that each family tree ends.
However, since they are not related to their siblings, they are not greatest or least ele-
ments of their family trees, nor are they upper or lower bounds. Rather, people without
descendants are extreme elements in the sense that there are none beneath them under this
ordering.
The extreme elements described in the two preceding paragraphs are examples of elements
that can either not be increased or not be decreased. In other words, they are either nonex-
tensible or irreducible. This is what we mean by maximal and minimal elements. The
definitions is as follows.

subset of U. For a ∈ U, a is a minimal element of A means
a ∈ A and ∀x ∈ A, if x 4 a, then x = a.
a is a maximal element of A means
a ∈ A and ∀x ∈ A, if a 4 x, then x = a.
Note that to say a is minimal means that there are no elements other than a itself for which
x 4 a. That is, there is nothing strictly smaller than a. Likewise, to say a is maximal means
that there are no elements but a itself for which a 4 x. In this way, there is nothing strictly
larger than a.
The notions of maximal and minimal are particularly well suited for describing irreducible
or nonextensible mathematical objects. For example, the reader who is familiar with linear
algebra may know that a basis for a vector space is not contained in any strictly larger lin-
early independent set and does not contain and strictly smaller spanning sets. Thus, though
the proof is beyond the scope of this book, a basis for a vector space may be described
as a maximal linearly independent set or as a minimal spanning set. For an example that
is within the scope of this course, the following two propositions show that the minimal
elements of N under the divides relation are what we commonly think of as irreducible
numbers.
Proposition 3.2.13.
For the partial ordering | on the set N and the subset A = {n ∈ N | n ≥ 2}, ∀a ∈ A, if a is
minimal in A, then a is prime.
Proof.
Let a ∈ A, and assume a is minimal in A.
Let p, q ∈ N, and assume a = pq.
Then p|a.
If p ≥ 2, then p ∈ A, in which case p = a since a is minimal in A.
Thus, if p ≥ 2, then p = a and hence q = 1.
Otherwise, if p < 2, then p = 1 since p ∈ N.
Therefore, either q = 1 or p = 1.
This shows, if a = pq, then either q = 1 or p = 1, which means a is prime.
Therefore, ∀a ∈ A, if a is minimal in A, then a is prime.
The converse of the above proposition is also true:
Proposition 3.2.14.
For the partial ordering | on the set N and the subset A = {n ∈ N | n ≥ 2}, ∀a ∈ A, if a is
prime, then a is minimal in A.
Proof.
Let a ∈ A, and assume a is prime.
Since a , 1, we have a ≥ 2; hence a ∈ A.
Next, let x ∈ A, and assume x|a.
Choose y ∈ N such that a = xy.
Since a is prime, we have either x = 1 or y = 1.
In fact, since x ∈ A, it is not the case that x = 1,
hence it must be the case that y = 1.
Therefore, a = x(1).
Therefore, ∀x ∈ A, if x|a, then a = x.
Thus, a is minimal in A.
Therefore, ∀a ∈ A, if a is prime, then a is minimal in A.
Exercises 3.2.
For each of the following, prove that the relation is a 14. ∀a, b ∈ R, if a < b, then sup(a, b) = b.
partial ordering. If the relation is also a total ordering,
prove it. Otherwise, prove that it is not a total order- 15. sup{x ∈ R | ∃n ∈ N, x = n−1
n } = 1.
ing. 16. inf{x ∈ R | ∃n ∈ N, x = n+1
= 1.
n }
1. The relation 4 on Q given by: ∀x, y ∈ Q, x 4 y if 17. inf{x ∈ R | ∃n ∈ N, x = 2n +1
= 1.
2n }
and only if y − x ∈ Z≥0 .
2n −1
18. sup{x ∈ R | ∃n ∈ N, x = 2n } = 1.
2. The relation 4 on R given by: ∀x, y ∈ R, x 4 y if
and only if y − x ∈ Q≥0 . 19. sup{x ∈ R | ∃n ∈ N, x = 3n−2
n } = 3.
3. The relation 4 on Z given by: ∀x, y ∈ Z, x 4 y if 20. inf{x ∈ R | ∃n ∈ N, x = 3n+2
n } = 3.
and only if ∃a ∈ Z≥0 , y = x + 3a.
21. sup{x ∈ R | ∃a ∈ (−∞, 0), x = 1 + a} = 1.
4. The relation 4 on Z given by: ∀x, y ∈ Z, x 4 y if
and only if ∃n ∈ Z≥0 , y = 2n x. 22. inf{x ∈ R | ∃a ∈ (0, ∞), x = 1 + a} = 1.
5. The relation 4 on R2 given by: ∀(x1 , x2 ), (y1 , y2 ) ∈ 23. inf{x ∈ R | ∃a ∈ (−∞, 0), x = 1 − a} = 1.
R2 , (x1 , x2 ) 4 (y1 , y2 ) if and only if x1 ≤ y1 and 24. sup{x ∈ R | ∃a ∈ (0, ∞), x = 1 − a} = 1.
x2 ≤ y2 .
25. sup{x ∈ R | ∃a ∈ (−2, 2), x = a2 } = 4.
6. The relation 4 on R2 given by: ∀(x1 , x2 ), (y1 , y2 ) ∈
R2 , (x1 , x2 ) 4 (y1 , y2 ) if and only if x1 < y1 or 26. inf{x ∈ R | ∃a ∈ (2, ∞), x = a2 } = 4.
(x1 = y1 and x2 ≤ y2 ). (Note: This is known as
27. inf{x ∈ R | ∃a ∈ [0, 1), x2 = a} = −1.
lexicographic order because it is analogous to the
ordering used to arrange words in a dictionary.) 28. sup{x ∈ R | ∃a ∈ [0, 1), x2 = a} = 1.
7. The relation 4 on R × (R \ {0}) given by:
∀(x1 , x2 ), (y1 , y2 ) ∈ R × (R \ {0}), (x1 , x2 ) 4 (y1 , y2 ) Prove the following propositions about the subset or-
if and only if x1 y2 = x2 y1 and x2 ≤ y2 . dering.
\
8. The relation 4 on R × (R \ {0}) given by: 29. For a family of sets A ⊆ P(U), S = inf(A)
∀(x1 , x2 ), (y1 , y2 ) ∈ R × (R \ {0}), (x1 , x2 ) 4 (y1 , y2 ) S ∈A
if and only if x1 y2 ≤ x2 y1 and x2 = y2 . under the partial ordering ⊆.
[
30. For a family of sets A ⊆ P(U), S = sup(A)
S ∈A
Prove the following propositions about extensions of
under the partial ordering ⊆.
the divides relation.
31. Under the partial ordering ⊆ on the family of sets
9. The relation | on Z, given by: ∀x, y ∈ Z, x|y if and I = {hni | n ∈ Z}, ∀a, b ∈ Z, hai + hbi =
only if ∃q ∈ Z, y = xq, is not a partial ordering. sup{hai, hbi}.
10. The relation | on Z≥0 , given by: ∀x, y ∈ Z≥0 , x|y if 32. Under the partial ordering ⊆ on the family of sets
and only if ∃q ∈ Z≥0 , y = xq, is a partial ordering. I = {hni | n ∈ Z}, ∀a, b ∈ Z, hai∩hbi = inf{hai, hbi}.
Prove the following propositions where the ordering is Let 4 be a partial ordering on a set U for which
the usual ordering ≤ on R. sup{x, y} and inf{x, y} exist for all x, y ∈ U. Prove the
following propositions.
11. The set A = {x ∈ Q | x < 2} has both an upper
2
bound and a lower bound in Q. 33. (Idempotence) ∀x ∈ U, sup{x, x} = x.

12. The set A = {x ∈ Q | x2 < 2} has neither a supre- 34. (Idempotence) ∀x ∈ U, inf{x, x} = x.
mum nor an infimum in Q.
35. (Associativity) ∀x, y, z ∈ U, inf{inf{x, y}, z} =
13. ∀a, b ∈ R, if a < b, then inf(a, b) = a. inf{x, inf{y, z}}.
36. (Associativity) ∀x, y, z ∈ U, sup{sup{x, y}, z} = 53. If inf A < A, then A does not have a least element.
sup{x, sup{y, z}}.
54. If sup A < A, then A does not have a greatest ele-
37. (Absorption) ∀x, y ∈ U, sup{x, inf{x, y}} = x. ment.
38. (Absorption) ∀x, y ∈ U, inf{x, sup{x, y}} = x. 55. If A ⊆ B, then sup A 4 sup B.
39. (Identity) If b is the greatest element of U, then 56. If A ⊆ B, then inf B 4 inf A.
∀x ∈ U, inf{x, b} = x.
57. If A , ∅, then inf A 4 sup A.
40. (Identity) If a is the least element of U, then ∀x ∈
U, sup{x, a} = x. 58. ∀a, b ∈ U, if a and b are both suprema of A, then
a = b.
41. (Annihilator) If b is the greatest element of U, then
∀x ∈ U, sup{x, b} = b. 59. ∀a ∈ U, if 4 is a total ordering and a is maximal in
42. (Annihilator) If a is the least element of U, then A, then a is the greatest element of A.
∀x ∈ U, inf{x, a} = a.
60. ∀a ∈ U, if 4 is a total ordering and a is minimal in
43. ∀a, x, y ∈ U, if x 4 y, then inf{a, x} 4 inf{a, y}. A, then a is the least element of A.
44. ∀a, x, y ∈ U, if x 4 y, then sup{a, x} 4 sup{a, y}. 61. If ∀x ∈ B, ∀y ∈ A, x 4 y, then sup B 4 inf A.
45. If 4 is a total ordering, then ∀x, y ∈ U, sup{x, y} = x 62. If sup B 4 inf A, then ∀x ∈ B, ∀y ∈ A, x 4 y.
or sup{x, y} = y.
46. If 4 is a total ordering, then ∀x, y ∈ U, inf{x, y} = x For the partial ordering 4Z on the set Q, defined in ex-
or inf{x, y} = y. ercise 1, prove the following propositions.
63. 1
2 is a maximal element of the set A = {x ∈ Q | 0 ≤
Let 4 be a partial ordering on a set U, and let A and B x ≤ 1}.
be subsets of U for which sup A, inf A, sup B, and inf B
exist. Prove the following propositions. 64. 1
2 is a minimal element of the set A = {x ∈ Q | 0 ≤
x ≤ 1}.
47. ∀a, b ∈ U, if a and b are both greatest elements of
A, then a = b.
Let U be a non-empty set with at least two elements,
48. ∀a, b ∈ U, if a and b are both least elements of A,
and let A = P(U) \ {∅, U}. That is, A is the family of
then a = b.
non-empty, non-trivial subsets of U. Prove the follow-
49. ∀a ∈ U, if a ∈ A and a = sup A, then a is the great- ing propositions for the partial ordering ⊆.
est element of A.
65. ∀a ∈ U, {a} is a minimal element of A.
50. ∀a ∈ U, if a ∈ A and a = inf A, then a is the least
element of A. 66. ∀a ∈ U, U \ {a} is a maximal element of A.
51. ∀a ∈ U, if a is the least element of A, then a = 67. ∀S ∈ A, if S is a maximal element of A, then

inf A. ∃a ∈ U, S = U \ {a}.
52. ∀a ∈ U, if a is the greatest element of A, then 68. ∀S ∈ A, if S is a minimal element of A, then

a = sup A. ∃a ∈ U, S = {a}.
3.3 Functions
Functions are often described as rules that relate each element of one set to exactly one
element of another set. Informally, we can think of functions as describing a sense in
which the elements of one set can be made to correspond to certain elements of another
set. For example, the fact that 5-cent candies cost 5 cents each gives us a correspondence
between amounts of candies and amounts of money. The different amounts of candy one
can purchase can be thought of as the set of natural numbers N (if we allow the number 1
to represent one candy and so on). Likewise, different amounts of money can be thought of
as elements in the set of all real numbers having up to two digits after the decimal. After
a moment’s thought, we may think to describe this set as M = {x ∈ R | 100x ∈ N}. Now,
the fact that 5-cent candies cost 5 cents tells us that to each number of candies x ∈ N
corresponds an amount of money 0.05x ∈ M. The relation thus described between the
elements of N and the elements of M is an example of a function.
Formally, we define a function as a type of relation; that is, as a non-empty set of ordered
pairs. More specifically, a function is a relation for which each element of the first set is
related to one and only one element of the second set. The formal definition is as follows:
Definition 3.3.1. Let A and B be sets. A function f from A to B, denoted

f : A → B, is defined to be a non-empty subset of A × B satisfying the
properties
1. ∀x ∈ A, ∃y ∈ B, (x, y) ∈ f .
2. ∀x ∈ A, ∀y1 , y2 ∈ B, if (x, y1 ) ∈ f and (x, y2 ) ∈ f , then y1 = y2 .
In this case, we denote (x, y) ∈ f by f (x) = y. We call the set A the

domain of f and the set B the codomain of f .
Notice that the first property
∀x ∈ A, ∃y ∈ B, (x, y) ∈ f
tells us that every element of A is related to some element of B. The second property
∀x ∈ A, ∀y1 , y2 ∈ B, if (x, y1 ) ∈ f and (x, y2 ) ∈ f , then y1 = y2
tells us that no element of A is related to more than one element of B. Taken together,
these two properties say that every element of A is related to one and only one element of
B. Because of this, for x ∈ A and y ∈ B, the notation y = f (x) (indicating that (x, y) ∈ f )
makes sense. Indeed, in this case, since x ∈ A, there is exactly one element of B that is
related to x, and that element is y. Thus, what we mean by y = f (x) is that y is the one and
only one element that f relates to x. In this sense, for any element x ∈ A, we can think of
the symbol f (x) as representing that one and only one element in the set B that the function
f relates to x. To make sense of this notation, consider the following simple example:
3.3. FUNCTIONS 221
Example 3.3.1.
Let A = {1, 2, 3} and B = {1, 2, 3, 4, 5}. The relation f = {(1, 1), (2, 3), (3, 5)} ⊆ A × B
is a function, since each element of A is related to one and only one element of B. In
function notation, we can describe the pairs in f as: f (1) = 1, f (2) = 3, and f (3) = 5.
Once we understand the definition of a function, the next step is learning to use it. For
example, if we want to prove that a given relation is a function, we are required to prove
that the two defining properties of a function are true for that relation. For an example of
this, we will prove that the relation f = {(x, n) ∈ R × Z | n ≤ x < n + 1} is a function.
Example 3.3.2.
The relation f : R → Z given by f = {(x, n) ∈ R × Z | n ≤ x < n + 1} is a function.
Proof.
Let x ∈ R.
Let S = {n ∈ Z | n ≤ x}.
By the Archimedean property, choose k ∈ N with −x ≤ k.
Then −k ≤ x; hence −k ∈ S . Thus, S , ∅.
Also, by the Archimedean property, we can choose m ∈ N with x < m.
Now, for any s ∈ S , we have s ≤ x < m; hence s < m.
Therefore, S is bounded above by m.
Applying the well-ordering property, choose n ∈ Z to be the largest element of S .
Since n ∈ S , we have n ≤ x.
Since n + 1 < S , we have x < n + 1.
Therefore, n ≤ x < n + 1; hence (x, n) ∈ f .
Therefore, ∃n ∈ Z, (x, n) ∈ f .
Therefore, ∀x ∈ R, ∃n ∈ Z, (x, n) ∈ f .
Next, let x ∈ R, let n1 , n2 ∈ Z, and assume (x, n1 ) ∈ f and (x, n2 ) ∈ f .
Then n1 ≤ x < n1 + 1 and n2 ≤ x < n2 + 1.
Now, n1 < n2 + 1 and n2 < n1 + 1 by transitivity.
Therefore, n1 ≤ n2 and n2 ≤ n1 .
Thus, n1 = n2 .
Therefore, ∀x ∈ R, ∀n1 , n2 ∈ Z, if (x, n1 ) ∈ f and (x, n2 ) ∈ f , then n1 = n2 .
Therefore, f is a function.
Since functions are defined to be sets of ordered pairs, to say two functions f and g are
equal means f ⊆ g and g ⊆ f . However, the following theorem proves that functions
are completely determined by the elements corresponding to each x ∈ A. That is, two
functions f and g are the same function if and only if they relate each element of A to the
same element of B. That is, if and only if ∀x ∈ A, f (x) = g(x).
Theorem 3.3.3.
Let f : A → B and g : A → B be functions. f = g if and only if ∀x ∈ A, f (x) = g(x).
Proof.
Assume f = g, and let x ∈ A.
Since f and g are functions, by property (1), choose y1 , y2 ∈ B with (x, y1 ) ∈ f and
(x, y2 ) ∈ g.
That is, f (x) = y1 and g(x) = y2 .
Since f = g and (x, y2 ) ∈ g, we have (x, y2 ) ∈ f .
Therefore, (x, y1 ) ∈ f and (x, y2 ) ∈ f . So, by property (2), y1 = y2 .
That is, f (x) = g(x).
Therefore, if f = g, then ∀x ∈ A, f (x) = g(x).
Conversely, assume ∀x ∈ A, f (x) = g(x).
Let (x, y) ∈ f .
Then y = f (x); hence y = g(x) since f (x) = g(x).
Hence, (x, y) ∈ g.
Therefore, f ⊆ g.
Similarly, if (x, y) ∈ g, then y = g(x) so y = f (x); hence (x, y) ∈ f .
Therefore, g ⊆ f .
Therefore, if ∀x ∈ A, f (x) = g(x), then f = g.
Thus, f = g if and only if ∀x ∈ A, f (x) = g(x).
The above theorem may seem obvious, but it says something very important about func-
tions: If we know the values f (x) for all x ∈ A, then we know the function, because the
function is uniquely determined by these values. We can therefore describe a function
simply by stating the value of f (x) for each x ∈ A. Consider, for example, the function
f : R → R defined by
f = {(x, y) ∈ R × R | y = 4x}.
Rather than describing this function explicitly as sets of ordered pairs (as we have done
here), we could also describe it simply by stating ∀x ∈ R, f (x) = 4x. Indeed, this statement
completely determines the function.
Another interesting observation due to the theorem above is that two functions f and g
could be described in different ways, for example using two different formulas, and yet if
those formulas happen to result in the same value f (x) = g(x) for every choice of x ∈ A,
then the two functions will be equal. Consider the following two examples:
3.3. FUNCTIONS 223
Example 3.3.4.
Let f : R → R and g : R → R be given by ∀x ∈ R, f (x) = 4x, and ∀x ∈ R, g(x) =

(x + 1)2 − (x − 1)2 . Then f = g.
Proof.
Let x ∈ R.
Then g(x) = (x + 1)2 − (x − 1)2 = (x2 + 2x + 1) − (x2 − 2x + 1) = 4x = f (x).
Therefore, ∀x ∈ R, f (x) = g(x).
Therefore, f = g.
Example 3.3.5.
Let f : N → R and g : N → R be given by ∀n ∈ N, f (n) =

Pn
k=1 2k−1 , and ∀n ∈ N,
g(n) = 2n − 1. Then f = g.
Proof.
Let A = {n ∈ N | f (n) = g(n)}.
Since f (1) = 1k=1 2k−1 = 20 = 1 = 21 − 1 = g(1), we have 1 ∈ A.
P
Let n ∈ N and assume n ∈ A.
Then f (n) = g(n).
n+1
X
f (n + 1) = 2k−1
k=1
n
X
= 2k−1 + 2(n+1)−1
k=1
= f (n) + 2n
= g(n) + 2n
= 2n − 1 + 2n
= 2n+1 − 1
= g(n + 1)
Therefore, ∀n ∈ N, if n ∈ A, then n + 1 ∈ A.
By the principle of mathematical induction, ∀n ∈ N, f (n) = g(n).
Therefore, f = g.
Composition of Functions
The composition of two functions can be thought of intuitively as the application of two
functions in succession. That is, the result of applying one function after another. To make
this clear, suppose A, B, and C are sets, and that f : A → B is a function from A to B, and
g : B → C is a function from B to C. The composition of f and g, denoted g ◦ f , is the
function that results by applying f and g in succession.
Beginning with an element x ∈ A, we can apply the function f resulting in an ele-

ment y = f (x) ∈ B. We then apply the function g to this element y ∈ B, resulting in an
element z = g(y) ∈ C. The net result of this process is a relation between the element x ∈ A
and the element z ∈ C.
Definition 3.3.2. Let f : A → B and g : B → C be functions. The

composition of f and g, denoted g ◦ f , is the function from A to C given
by ∀x ∈ A, g ◦ f (x) = g( f (x)). In set notation:
g ◦ f = {(x, z) ∈ A × C | ∃y ∈ B, (x, y) ∈ f and (y, z) ∈ g}.
It often happens that the domain and codomain of a function are the same, meaning f : A →
A for some set A. In this case, it is possible to consider the composition of the function f
with itself. i.e. f ◦ f , which is again a function from A to A. In this case, we can recursively
define “powers” of the function f as follows:
Definition 3.3.3. Let f : A → A be a function. We define f 1 = f and for

each n ∈ N, f n+1 = f n ◦ f .
3.3. FUNCTIONS 225
Algebra of Functions
Notice that the composition of two functions g ◦ f is another function. This means that
the composition ◦ can be viewed as a binary operation on functions. Just as the binary
operations + and · each combine two real numbers into another real number, and the binary
operations ∩ and ∪ each combine two sets into another set, the binary operation ◦ combines
two functions into another function.
The composition operation shares some characteristics with other binary operations we
have studied. For example, the operation ◦ is associative:
Given functions f : A → B, g : B → C, and h : C → D, we have
(h ◦ g) ◦ f = h ◦ (g ◦ f ).
Indeed, for any x ∈ A, we have
(h ◦ g) ◦ f (x) = h ◦ g( f (x)) = h(g( f (x)) = h(g ◦ f (x)) = h ◦ (g ◦ f )(x).
However, ◦ is not generally commutative. That is, it is not always the case that f ◦g = g◦ f .
In fact, when g ◦ f is defined (i.e. when the codomain of f is equal to the domain of g) the
function f ◦ g may not be defined, because the codomain of g may not be the same as the
domain of f .
There are also functions that play the role of an identity under composition. Recall that in
the real numbers, we have ∀x ∈ R, x + 0 = x and 0 + x = x, making 0 an identity element
under the binary operation +. Similarly, the number 1 is an identity under multiplication,
since ∀x ∈ R, x1 = x and 1x = x. We have also seen that ∅ and the universe of discourse
U are identity elements in P(U) under the operations ∪ and ∩ respectively. Indeed, for any
A ∈ P(U), we have A ∪ ∅ = A and A ∩ U = A. In the case of composition, the following
function fills this role:
Definition 3.3.4. For a set A, the identity function on A, iA : A → A, is

the function given by ∀x ∈ A, iA (x) = x.
Proposition 3.3.6.
Let f : A → B be a function. Then f ◦ iA = f and iB ◦ f = f .
Proof.
Let x ∈ A.
Then f ◦ iA (x) = f (iA (x)) = f (x).
Likewise, iB ◦ f (x) = iB ( f (x)) = f (x)
Therefore, ∀x ∈ A, f ◦ iA (x) = f (x), and iB ◦ f (x) = f (x).
It follows from theorem 3.3.3 that f ◦ iA = f and iB ◦ f = f .
Some functions (not all) also have inverses under composition. Recall that in the real
numbers, to each x ∈ R corresponds an additive inverse −x, with the property that x+(−x) =
0 and (−x) + x = 0. Similarly, to each non-zero x ∈ R corresponds a multiplicative inverse
x−1 , with the property that xx−1 = 1 and x−1 x = 1. The analogous idea for a function
f : A → B under the composition operation would be a function whose composition with
f results in the identity function. This suggests the following definition:
Definition 3.3.5. For a function f : A → B, f is invertible means there

is a function g : B → A such that g ◦ f = iA and f ◦ g = iB . In this case,
we write g = f −1 and call g the inverse of f .
Notice that the definition of invertibility requires the existence of a function that serves as
an inverse in the appropriate way. Thus, if we are to use the definition to prove that a given
function is invertible, we must prove the existence of such an inverse function. This means
our proof must include a construction of the inverse function. For example:
3.3. FUNCTIONS 227
Example 3.3.7.
x−1
The function f : R → R given by f (x) = is invertible.
2
Proof.
Define the function g : R → R by ∀x ∈ R, g(x) = 2x + 1.
Let x ∈ R.
g ◦ f (x) = g( f (x))
!
x−1
=g
2
!
x−1
=2 +1
2
= (x − 1) + 1
=x
= iR (x)
Therefore, ∀x ∈ R, g ◦ f (x) = iR (x), which means g ◦ f = iR .
Proofs like the example above are not always obvious since an easily visible formula for the
inverse of a function is not always apparent. With a view toward developing our knowledge
of invertibility and the tools available for proofs of invertibility, we prove the following
theorems.
Theorem 3.3.8.
For any set A, iA is invertible and (iA )−1 = iA .
Proof.
By proposition 3.3.6, iA ◦ iA = iA , which means iA = (iA )−1 .
Theorem 3.3.9.
Let f : A → B be a function. If f is invertible, then f −1 is invertible and ( f −1 )−1 = f .
Proof. Let f : A → B be a function.

Assume f is invertible.
Then f −1 : B → A is a function for which f −1 ◦ f = iA and f ◦ f −1 = iB .
Put another way, f is a function for which f ◦ f −1 = iB and f −1 ◦ f = iA ,
which by definition makes f −1 invertible and ( f −1 )−1 = f .
Therefore, if f is invertible, then f −1 is invertible and ( f −1 )−1 = f .
Theorem 3.3.10.
Let f : A → B and g : B → C be functions. If f and g are invertible, then g ◦ f is invertible

and (g ◦ f )−1 = f −1 ◦ g−1 .
Proof. Let f : A → B and g : B → C be functions.

Assume f and g are invertible.
Then f −1 : B → A is a function for which f −1 ◦ f = iA and f ◦ f −1 = iB .
Likewise, g−1 : C → B is a function for which g−1 ◦ g = iB and g ◦ g−1 = iC .
Let x ∈ A.
( f −1 ◦ g−1 ) ◦ (g ◦ f )(x) = f −1 ◦ g−1 (g ◦ f (x))

= f −1 (g−1 (g( f (x))))
= f −1 (g−1 ◦ g( f (x)))
= f −1 (iB ( f (x))
= f −1 ( f (x))
= iA (x)
Therefore, ∀x ∈ A, ( f −1 ◦ g−1 ) ◦ (g ◦ f )(x) = iA (x). Hence, ( f −1 ◦ g−1 ) ◦ (g ◦ f ) = iA .

Let z ∈ C.
(g ◦ f ) ◦ ( f −1 ◦ g−1 )(x) = g( f ( f −1 (g−1 (z))))

= g( f ◦ f −1 (g−1 (z)))
= g(iB (g−1 (z)))
= g(g−1 (z)
= iC (z)
Therefore, ∀z ∈ C, (g ◦ f ) ◦ ( f −1 ◦ g)(z) = iC (z). Hence, (g ◦ f ) ◦ ( f −1 ◦ g−1 ) = iC .

Now, since ( f −1 ◦ g−1 ) ◦ (g ◦ f ) = iA and (g ◦ f ) ◦ ( f −1 ◦ g−1 ) = iC , we have that g ◦ f is
invertible and (g ◦ f )−1 = f −1 ◦ g−1 .
Criteria for Invertibility
The following two properties together give a necessary and sufficient condition for a func-
tion to be invertible.
3.3. FUNCTIONS 229
Definition 3.3.6. For a function f : A → B, f is injective (or one-to-one)

means
∀x1 , x2 ∈ A, if f (x1 ) = f (x2 ), then x1 = x2 .
f is surjective (or onto) means
∀y ∈ B, ∃x ∈ A, f (x) = y.
f is bijective means f is both injective and surjective.
Example 3.3.11.
x−1
The function f : R → R given by f (x) = is bijective.
2
Proof.
Let x1 , x2 ∈ R, and assume f (x1 ) = f (x2 ).
x1 − 1 x2 − 1
Then = .
2 2
Multiplying by 2 gives us x1 − 1 = x2 − 1.
Adding 1 gives us x1 = x2 .
Therefore, ∀x1 , x2 ∈ R, if f (x1 ) = f (x2 ), then x1 = x2 .
That is, f is injective.
Next, let y ∈ R.
Put x = 2y + 1.
x − 1 (2y + 1) − 1 2y
Then f (x) = = = = y.
2 2 2
Therefore, ∃x ∈ R, f (x) = y.
Therefore, ∀y ∈ R, ∃x ∈ R, f (x) = y.
That is, f is surjective.
Since f is both injective and surjective, f is bijective.
Theorem 3.3.12.
Let f : A → B be a function. f is invertible if and only if f is bijective.

Assume f is invertible.
Then there is a function f −1 : B → A for which f −1 ◦ f = iA and f ◦ f −1 = iB .
Let x1 , x2 ∈ A, and assume f (x1 ) = f (x2 ).
Then f −1 ( f (x1 )) = f −1 ( f (x2 )); hence f −1 ◦ f (x1 ) = f −1 ◦ f (x2 ).
This means iA (x1 ) = iA (x2 ); hence x1 = x2 .
Therefore, ∀x1 , x2 ∈ A, if f (x1 ) = f (x2 ), then x1 = x2 .
Hence, f is injective.
Next, let y ∈ B.
Put x = f −1 (y).
Then f (x) = f ( f −1 (y)) = f ◦ f −1 (y) = iB (y) = y.
Therefore, ∃x ∈ A, f (x) = y.
Therefore, f is surjective; thus f is bijective.
Conversely, assume f is bijective.
Let g = {(y, x) ∈ B × A | y = f (x)}. We will prove that g : B → A is a function.
Let y ∈ B.
Since f is surjective, we can choose x ∈ A with y = f (x).
For such an x, we have (y, x) ∈ g.
Therefore, ∀y ∈ B, ∃x ∈ A, (y, x) ∈ B.
Next, let y ∈ B and x1 , x2 ∈ A, and assume (y, x1 ) ∈ g and (y, x2 ) ∈ g.
Then y = f (x1 ) and y = f (x2 ).
Since f (x1 ) = f (x2 ) and f is injective, we have x1 = x2 .
Therefore, ∀y ∈ B, ∀x1 , x2 ∈ A, if (y, x1 ) ∈ g and (y, x2 ) ∈ g, then x1 = x2 .
Therefore, g is a function.
Now, for any x ∈ A, letting y = f (x) gives us g(y) = x.
Then g ◦ f (x) = g( f (x)) = g(y) = x = iA (x).
Thus, g ◦ f = iA .
Likewise, for any y ∈ B, letting x = g(y) gives us y = f (x).
Then f ◦ g(y) = f (g(y)) = f (x) = y = iB (y).
Thus, f ◦ g = iB .
Therefore, f is invertible and g = f −1 .
Therefore, f is invertible if and only if f is bijective.
The theorem above suggests an alternate approach to proving a function is invertible.

Rather than finding a formula for the inverse function, as we did previously (and which
is sometimes difficult), we can instead prove a function is invertible by proving it is bijec-
tive. Consider, for example, the following function (the inverse of which would be very
difficult to describe with a formula).
3.3. FUNCTIONS 231
Example 3.3.13.
The function f : N2 → N given by f (m, n) = 2m−1 (2n − 1) is invertible.
Proof. We will show that f is bijective.

Let (m1 , n1 ), (m2 , n2 ) ∈ N2 .
Suppose f (m1 , n1 ) = f (m2 , n2 ) and m1 , m2 .
Then 2m1 −1 (2n1 − 1) = 2m2 −1 (2n2 − 1).
Without loss of generality, assume m1 < m2 . Then 0 < m2 − m1 .
Now, 2n1 − 1 = 2m2 −m1 (2n2 − 1), which is a contradiction, since the left side is odd
and the right is even.
Therefore, if f (m1 , n1 ) = f (m2 , n2 ), then m1 = m2 .
Therefore, f is injective.
Next, let A = {y ∈ N | ∃(m, n) ∈ N2 , f (m, n) = y}.
Since f (1, 1) = 21−1 (2(1) − 1) = 1, we have 1 ∈ A.
Let y ∈ N, and assume {1, . . . , y} ⊆ A.
Case 1: y + 1 is odd.
Choose n ∈ N with y + 1 = 2n − 1.
f (1, n) = 21−1 (2n − 1) = (1)(y + 1) = y + 1.
Therefore, y + 1 ∈ A.
Case 2: y + 1 is even.
Choose q ∈ N with y + 1 = 2q.
Since 1 ≤ q ≤ y, we have q ∈ {1, . . . , y}; hence q ∈ A.
Choose (m, n) ∈ N2 with f (m, n) = q; that is, 2m−1 (2n − 1) = q.
Now, f (m + 1, n) = 2(m+1)−1 (2n − 1) = 2(2m−1 )(2n − 1) = 2q = y + 1.
Therefore, y + 1 ∈ A.
Therefore, if {1, . . . , y} ⊆ A, then y + 1 ∈ A.
Therefore, ∀y ∈ N, ∃(m, n) ∈ N2 , f (m, n) = y.
Therefore, f is surjective; thus f is bijective.
Now, since f is bijective, it is invertible.
A remark about ‘without loss of generality’

In general, if one introduces an assumption P into a proof, it affects the proposition that
is ultimately proven. That is, our conclusion will then only be valid if P is true. This
means that we can only make an assumption that is consistent with the implication we
wish to prove. There is, however, one very important exception to this rule: When our
assumption only refers to the notation used to express our proposition and has no bearing
on the meaning of our proposition, then such an assumption can be introduced without
loss of generality. For example, in the previous proof, we have arbitrary constants m1 and
m2 and the knowledge that m1 , m2 . By trichotomy, this means that either m1 < m2 or
m2 < m1 . However, there is no logical distinction between the natural number denoted
by m1 and the natural number denoted by m2 . The only discernable difference is in the
notation that we have chosen to represent them. Hence the assumption that we have chosen
to denote the smaller of the numbers by m1 and the larger by m2 , is an assumption about
our notation only; it does not interfere with the meaning of the proposition. In fact m1 and
m2 can be interchanged in the proposition with no impact on its meaning. This means that
the additional assumption that m1 < m2 is only an assumption about our choice of notation
and can therefore be made without loss of generality. This is so common that one often
simply writes ‘WLOG, assume...’, in place of ‘without loss of generality, assume...’.
Images and Pre-Images
We now consider not only how a function f : A → B relates elements of the domain A to
elements of the codomain B, but also how f relates the subsets of the domain A to subsets
of the codomain B.
Definition 3.3.7. Let f : A → B be a function, and let S ⊆ A. The image

of the set S under the function f is the set
f (S ) = {y ∈ B | ∃x ∈ S , f (x) = y}.
The set f (A) is called the image (or range) of f .
Caution
The notation f (S ) is not meant to suggest that the function f is applied to the set S in the
same sense that it is applied to an element x. It should be noted that the notation f (S ) when
S ⊆ A stands for a very different concept than the notation f (x) when x ∈ A. Indeed, when
S ⊆ A, f (S ) is a set; it is {y ∈ B | ∃x ∈ S , f (x) = y}. On the other hand, when x ∈ A, you
will recall from the beginning of this section that f (x) stands for the unique element y ∈ B
for which (x, y) ∈ f . Thus, we must be cautious to consider the context when interpreting
this notation.
Proposition 3.3.14.
Let f : A → B be a function. f is surjective if and only if f (A) = B.

Assume f is surjective.
Since B is the universe of discourse in which f (A) is defined, we have f (A) ⊆ B.
Let y ∈ B.
Since f is surjective, ∃x ∈ A, f (x) = y.
That is, y ∈ f (A).
Therefore, B ⊆ f (A); thus f (A) = B.
Conversely, assume f (A) = B.
Let y ∈ B.
Then y ∈ f (A), since f (A) = B.
That is, ∃x ∈ A, f (x) = y.
Therefore, ∀y ∈ B, ∃x ∈ A, f (x) = y.
Therefore, f is surjective.
Therefore, f is surjective if and only if f (A) = B.
3.3. FUNCTIONS 233
Example 3.3.15.
Let f : R → R be the function given by ∀x ∈ R, f (x) = x2 . Then f ((−1, 2)) = [0, 4).
Proof.
Let y ∈ f ((−1, 2)).
Then ∃x ∈ (−1, 2), y = f (x). For such an x, we have y = f (x) = x2 and −1 < x < 2.
Now, since 0 ≤ x2 , we have 0 ≤ y.
Since −2 < −1 < x, we have −2 < x. Then −2 < x < 2, and by exercise 127 in section
1.1, we then have x2 < 4.
Now, 0 ≤ y and y < 4, so y ∈ [0, 4).
Therefore, f ((−1, 2)) ⊆ [0, 4).
Conversely, let y ∈ [0, 4).
√
Putting x = y, gives us y = x2 = f (x). Then 0 ≤ x2 < 4.
√
Now, since x = y, we have 0 ≤ x; hence −1 < x.
Suppose 2 ≤ x.
Then 4 ≤ 2x and 2x ≤ x2 ; hence 4 ≤ x2 .
Since this is a contradiction, we must have x < 2.
Therefore, x ∈ (−1, 2).
Therefore, ∃x ∈ (−1, 2), y = f (x), which means y ∈ f ((−1, 2)).
Therefore, [0, 4) ⊆ f ((−1, 2)). Thus, f ((−1, 2)) = [0, 4).
Example 3.3.16.
Consider the function f : Z → Z given by

(
x if x is even
∀x ∈ Z, f (x) = .
2x if x is odd
Then f (h3i) = h6i.
Proof.
Let y ∈ f (h3i).
Accordingly, choose x ∈ h3i with y = f (x), and choose t ∈ Z with x = 3t.
In the case where x is odd, we have y = f (x) = 2x = 6t ∈ h6i.
In the case where x is even, choose s ∈ Z with x = 2s.
Then y = f (x) = x = 3x − 2x = 3(2s) − 2(3t) = 6(s − t) ∈ h6i.
Thus, in both cases, y ∈ h6i.
Therefore, f (h3i) ⊆ h6i.
Conversely, let y ∈ h6i.
Choose q ∈ Z with y = 6q.
Put x = y.
Since x = y = 6q = 2(3q), we have that x is even.
Thus, f (x) = x = y.
Moreover, x = 6q = 3(2q) ∈ h3i.
Therefore, ∃x ∈ h3i, y = f (x); hence y ∈ f (h3i).
Therefore, h6i ⊆ f (h3i), and so f (h3i) = h6i.
Another relationship between subsets of the domain and codomain is given by what we call
the pre-image.
Definition 3.3.8. Let f : A → B be a function, and let V ⊆ B. The

pre-image of the set V under the function f is the set
f −1 (V) = {x ∈ A | f (x) ∈ V}.
Example 3.3.17.
Let f : R → R be the function given by ∀x ∈ R, f (x) = x2 . Then f −1 ([0, 4)) = (−2, 2).
Proof.
Let x ∈ f −1 ([0, 4)).
Then f (x) ∈ [0, 4), which means 0 ≤ x2 < 4.
Therefore, |x| < 2 by exercise 128 of section 1.1.
Therefore, −2 < x < 2 by proposition 1.1.27.
Thus, x ∈ (−2, 2).
Therefore, f −1 ([0, 4)) ⊆ (−2, 2).
Conversely, let x ∈ (−2, 2)
Since 0 ≤ x2 , we have 0 ≤ f (x).
Since −2 < x < 2, by exercise 127 in section 1.1, we have x2 < 4.
Therefore, 0 ≤ f (x) < 4; hence f (x) ∈ [0, 4).
Therefore, x ∈ f −1 ([0, 4)).
Therefore, (−2, 2) ⊆ f −1 ([0, 4)).
Caution
The notation f −1 (V) is not meant to suggest that the function f is invertible, or that the
inverse of the function f , as discussed previously, is involved in any way. The notation
f −1 (V) when V ⊆ B stands only for the pre-image of the set V as defined. This notation
is also not meant to suggest that the image a pre-image cancel one another out. In many
cases, when S ⊆ A, f −1 ( f (S )) , S . Likewise, when V ⊆ B, we often have f ( f −1 (V)) , V.
For an example illustrating the fact that for a subset S ⊆ A, f −1 ( f (S )) need not be equal to
S (thus the pre-image and image are not inverse operations), recall that for f : R → R given
by ∀x ∈ R, f (x) = x2 , we have f ((−1, 2)) = [0, 4) (see example 3.3.15) and f −1 ([0, 4)) =
(−2, 2) (see example 3.3.18). Combining these, we have
f −1 ( f ((−1, 2))) = f −1 ([0, 4)) = (−2, 2) , (−1, 2).
For an example of a situation in which f ( f −1 (V)) , V for a subset V ⊆ B, consider

the following:
3.3. FUNCTIONS 235
Example 3.3.18.
Let f : R → R be the function given by ∀x ∈ R, f (x) = x2 . Then f ( f −1 ((−4, 4))) = [0, 4).
Proof.
Let y ∈ f ( f −1 ((−4, 4))).
Then ∃x ∈ f −1 ((−4, 4)), y = f (x). Choose such an x.
Since x ∈ f −1 ((−4, 4)), we have f (x) ∈ (−4, 4). Thus, y ∈ (−4, 4).
In particular, we have y < 4.
Moreover, since y = f (x) = x2 , we have 0 ≤ y.
Thus, y ∈ [0, 4).
Therefore, f ( f −1 ((−4, 4))) ⊆ [0, 4)
Conversely, let y ∈ [0, 4)
√
Put x = y.
Then y = x2 = f (x); hence 0 ≤ f (x) < 4, since y ∈ [0, 4).
Now, since −4 < 0 ≤ f (x), we have −4 < f (x); thus, −4 < f (x) < 4.
We now have f (x) ∈ (−4, 4), which means x ∈ f −1 ((−4, 4)).
Therefore, ∃x ∈ f −1 ((−4, 4)), y = f (x).
Therefore, y ∈ f ( f −1 ((−4, 4))).
Therefore, [0, 4) ⊆ f ( f −1 ((−4, 4))).
Thus, f ( f −1 ((−4, 4))) = [0, 4).
It should be noted that the above examples are made possible by the fact that the function
f : R → R, given by ∀x ∈ R, f (x) = x2 , is neither injective nor surjective. For a bijective
function f : A → B, we indeed have f −1 ( f (S )) = S whenever S ⊆ A, and f ( f −1 (V)) = V
whenever V ⊆ B. This fact is demonstrated in the following two theorems.
Theorem 3.3.19.
Let f : A → B be a function, and let S ⊆ A. If f is injective, then f −1 ( f (S )) = S .
Proof.
Assume f is injective.
Let a ∈ f −1 ( f (S )).
Then f (a) ∈ f (S ).
Therefore, ∃x ∈ S , f (a) = f (x). Choose such an x.
Since f is injective, we have a = x; hence a ∈ S .
Therefore, f −1 ( f (S )) ⊆ S .
Conversely, let a ∈ S .
Then f (a) ∈ f (S ).
Thus, a ∈ f −1 ( f (S )).
Therefore, S ⊆ f −1 ( f (S )).
Thus, f −1 ( f (S )) = S .
Therefore, if f is injective, then f −1 ( f (S )) = S .
Theorem 3.3.20.
Let f : A → B be a function, and let V ⊆ B. If f is surjective, then f ( f −1 (V)) = V.
Proof.
Assume f is surjective.
Let y ∈ f ( f −1 (V)).
Then ∃x ∈ f −1 (V), y = f (x).
Since x ∈ f −1 (V), we have f (x) ∈ V; hence y ∈ V.
Therefore, f ( f −1 (V)) ⊆ V.
Conversely, let y ∈ V.
Since f is surjective, ∃x ∈ A, y = f (x). Choose such an x.
Since f (x) = y ∈ V, we have x ∈ f −1 (V).
Therefore, ∃x ∈ f −1 (V), y = f (x); hence y ∈ f ( f −1 (V)).
Therefore, V ⊆ f ( f −1 (V)).
Thus, f ( f −1 (V)) = V.
Therefore, if f is surjective, then f ( f −1 (V)) = V.
3.3. FUNCTIONS 237
Exercises 3.3.
For each of the following, prove that the relation f is a 14. f : Q → Q, given by ∀x ∈ Q, f (x) = 6x − 5.
function. State the domain and codomain of f .
15. f : R2 → R2 , given by
1. f = {(x, n) ∈ (0, 1] × N | 1
n+1 <x≤ 1
n }.
∀(x1 , x2 ) ∈ R2 , f (x1 , x2 ) = (x1 + x2 , x1 − x2 ).
2. f = {(x, n) ∈ R × Z | n − 1 < x ≤ n}.
16. f : Z3 → Z3 , given by
3. Let S = {x ∈ Z | 0 ≤ x < 5}.
f = {(x, r) ∈ N × S | ∃q ∈ Z, x = 5q + r}. ∀(x1 , x2 , x3 ) ∈ Z3 , f (x1 , x2 , x3 ) = (x1 +x2 +x3 , x2 +x3 , x3 ).
4. Let n ∈ N and let S = {x ∈ Z | 0 ≤ x < n}.
f = {(x, r) ∈ N × S | ∃q ∈ Z, x = nq + r}.
Let A be a set, and let f : A → A be a function. Prove
the following.
For each of the following, find expressions for the func-
17. If f ◦ f is injective, then f is injective.
tions g ◦ f and f ◦ g.
18. If f ◦ f is surjective, then f is surjective.
5. f : Z → Z, given by ∀x ∈ Z, f (x) = 2x + 1.
g : Z → Z, given by 19. If f is surjective, then f ◦ f is surjective.
x + 1 if x is even
(
20. If f is injective, then f ◦ f is injective.
∀x ∈ Z, g(x) = .
x − 1 if x is odd
21. If f is injective, then ∀n ∈ N, f n is injective.
6. f : Z → Z, given by ∀x ∈ Z, f (x) = x2 − 2x + 1. 22. If f is surjective, then ∀n ∈ N, f n is surjective.
g : Z → Z, given by
( 23. If ∃n ∈ N, f n is surjective, then f is surjective.
x − 1 if x is even
∀x ∈ Z, g(x) = . 24. If ∃n ∈ N, f n is injective, then f is injective.
x + 1 if x is odd
Let A, B, and C be sets, and let f : A → B and

Let A, B, and C be sets, and let f : A → B and
g : B → C be functions. Prove the following.
g : B → C be functions. Prove the following.
25. If g ◦ f is injective, then f is injective.
7. If α : B → A and β : B → A are functions for which
α ◦ f = iA and f ◦ β = iB , then α = β. (Thus, the 26. If g ◦ f is surjective, then g is surjective.
inverse of a function is unique).
27. If f and g are surjective, then g ◦ f is surjective.
8. If f is invertible, then ∀x ∈ A, ∀y ∈ B, f (x) = y if
and only if f −1 (y) = x. 28. If f and g are injective, then g ◦ f is injective.
9. Suppose A = B, so that f : A → A. If f ◦ f is 29. If g ◦ f is injective and f is surjective, then g is

invertible, then f is invertible. injective.
10. Suppose A = B, so that f : A → A. If f is invert- 30. If g ◦ f is surjective and g is injective, then f is

ible, then f ◦ f is invertible. surjective.
11. If g ◦ f is invertible and g is invertible, then f is

For the given function f and set S , find f (S ). Prove
invertible.
your result.
12. If g ◦ f is invertible and f is invertible, then g is
31. f : R → R, given by ∀x ∈ R, f (x) = 2x + 1.
invertible.
(a) S = [−1, 1).
For each of the following, prove the function is bijec- (b) S = (−3, 5].
tive. 32. f : R → R, given by ∀x ∈ R, f (x) = 1 − 2x.
5x − 7
13. f : R → R, given by ∀x ∈ R, f (x) = . (a) S = (−3, 5].
2
(b) S = [−1, 1). (b) V = [−1, 5).

33. f : R → R, given by ∀x ∈ R, f (x) = x2 . 42. f : R → R, given by ∀x ∈ R, f (x) = 1 − 2x.
(a) S = [−1, 1). (a) V = (0, 3].
(b) S = (−3, 5]. (b) V = [−1, 5].
34. f : R → R, given by ∀x ∈ R, f (x) = 1 − x2 . 43. f : R → R, given by ∀x ∈ R, f (x) = x2 .
(a) S = (−3, 5]. (a) V = (0, 9].
(b) S = [−1, 1). (b) V = [0, 4) ∪ (9, 16].
35. f : Z → Z, given by ∀x, y ∈ Z, f (x, y) = x + y.
2
44. f : R → R, given by ∀x ∈ R, f (x) = 1 − x2 .
(a) S = E . 2
(a) V = (1, 2].
(b) S = O2 .
(b) V = [−3, 2).
36. f : Z3 → Z3 , given by
45. f : Z2 → Z, given by ∀x, y ∈ Z, f (x, y) = x + y.
∀(x1 , x2 , x3 ) ∈ Z , f (x1 , x2 , x3 ) = (x1 +x2 +x3 , x2 +x3 , x3 ).
3
(a) V = E.
(a) S = O . 3
(b) V = O.
(b) S = N3 .
46. f : Z → Z, given by
37. f : Z → Z, given by (
x−1 if x is even
∀x ∈ Z, f (x) = .
x+1
(
x − 1 if x is even if x is odd
∀x ∈ Z, f (x) = .
x + 1 if x is odd
(a) V = h4i.
(a) S = h3i. (b) V = h3i.
(b) S = Z.
47. f : Z → Z, given by
38. f : Z → Z, given by (
2x − 2 if x is even
∀x ∈ Z, f (x) = .
(
2x − 2 if x is even 2x if x is odd
∀x ∈ Z, f (x) = .
2x if x is odd
(a) V = {6, 12}.
(a) S = Z.
(b) V = 6 + h8i.
(b) S = h3i.
48. f : Z → Z15 , given by ∀x ∈ Z, f (x) = [x]15 .
39. f : Z × N → Q, given by
m (a) V = {[0]15 }.
∀(m, n) ∈ Z × N, f (m, n) = .
n (b) V = {[0]15 , [5]15 , [10]15 }.
(a) S = E × (E ∩ N).
(b) S = O × (E ∩ N). Let f : A → B be a function. Let S and T be subsets of
A, and let V and W be subsets of B. Prove the following.
40. f : Z → Z6 , given by ∀x ∈ Z, f (x) = [x]6 .
49. f (S ∪ T ) ⊆ f (S ) ∪ f (T ).
(a) S = h3i.
(b) S = h2i. 50. f (S ) ∪ f (T ) ⊆ f (S ∪ T ).
51. f (S ∩T ) ⊆ f (S )∩ f (T ). Give an example of a func-

−1
For the given function f and set V, find f (V). Prove tion f : R → R and subsets S and T of R for which
your result. f (S ) ∩ f (T ) * f (S ∩ T ).
41. f : R → R, given by ∀x ∈ R, f (x) = 2x + 1. 52. If f is injective, then f (S ) ∩ f (T ) ⊆ f (S ∩ T ).
(a) V = (0, 3]. 53. If f is injective, then f (S \ T ) ⊆ f (S ) \ f (T ).

3.3. FUNCTIONS 239
54. f (S ) \ f (T ) ⊆ f (S \ T ). Give an example of a func- 65. Let f : A → B be a function. Let R be the relation
tion f : R → R and subsets S and T of R for which on A given by: For all a, b ∈ A, aRb if and only if
f (S \ T ) * f (S ) \ f (T ). f (a) = f (b). Prove R is an equivalence relation.
55. If f is injective, then f (S c ) ⊆ ( f (S ))c . 66. Let R be an equivalence relation on a set A. Let
f : A → A/R be the function given by ∀x ∈ A,
56. If f is surjective, then ( f (S ))c ⊆ f (S c ).
f (x) = [x]R . Prove ∀a, b ∈ A, aRb if and only if
57. f −1 (V ∪ W) = f −1 (V) ∪ f −1 (W). f (a) = f (b).
58. f −1 (V ∩ W) = f −1 (V) ∩ f −1 (W). 67. Let f : A → B be a function. Let R be the relation

on A given by: For all a, b ∈ A, aRb if and only if
59. f −1 (V \ W) = f −1 (V) \ f −1 (W). f (a) = f (b). Prove the relation
60. f −1 (V c ) = ( f −1 (V))c .
g = {([x]R , f (x)) ∈ A/R × B | x ∈ A}
61. If ∀E ∈ P(B), E ⊆ f ( f −1 (E)), then f is surjective.
is an injective function.
62. If ∀Z ∈ P(A), f −1 ( f (Z)) ⊆ Z, then f is injective.
68. Let R be an equivalence relation on a set A. Let
63. If f is bijective, then f (S ) = V if and only if f : A → A/R be the function given by ∀x ∈ A,
f −1 (V) = S . f (x) = [x]R . Let g : A/R → B be an injec-
64. If f is invertible, then tive function. Prove ∀a, b ∈ A, aRb if and only if
g ◦ f (a) = g ◦ f (b).
{x ∈ A | f (x) ∈ V} = {x ∈ A | ∃y ∈ V, x = f −1 (y)}.
69. Let f : A → B be a function, and let P be a par-
−1
(Hence when f is invertible, the notation f (V) is tition of A. Let Q = { f (S ) | S ∈ P}. Prove if f is
not ambiguous). bijective, then Q is a partition of B.
70. Let f : A → B be a function, and let Q be a parti-

The following exercises combine the topics in this section of B. Let P = { f −1 (V) | V ∈ Q}. Prove if f is
tion with those in the section 3.1. surjective, then P is a partition of A.
Bibliography
[1] L. Carroll, H. Haughton, and J. Tenniel. Alice’s Adventures in Wonderland and Through the Looking Glass.
Penguin classics. Penguin Books Limited, 1998.
[2] H.W. Eves. An introduction to the history of mathematics. The Saunders series. Saunders College Pub., 1983.
241

Foundations Notes

Uploaded by

Copyright:

Available Formats

Foundations Notes

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Foundations Notes

Uploaded by

Copyright:

Available Formats

Mathematical Foundations

Justin Lariviere and Tim Traynor

1 Structure of a Mathematical Proof 33

0.1 Sentence Structure

To begin our discussion of precision in language, we identify two components of a state-

An open sentence or predicate is a declarative statement containing one

7. ‘2x + 1 = 4.’ (x)

Making an Open Sentence into a Proposition

1. We may assign one specific value to the variable, as in ‘3 is posi-

‘∀x ∈ R, x > 0.’

Read: ‘For all x in R, x is positive.’

2. We write the proposition ‘Some real numbers are positive’ using

‘∃x ∈ R, x > 0.’

Read: ‘There exists an x in R such that x is positive.’

∀x ∈ A, P(x) ⇔ ∀x ∈ U, if x ∈ A, then P(x)

∀x, y ∈ U, P(x, y) means ∀x ∈ U, ∀y ∈ U, P(x, y)

The Benefits of Symbolic Form

7. There is no largest prime number.

Common Notation for Sets of Numbers

3. ∃x ∈ R, x < 0. Write the following propositions in symbolic form by

5. ∀x ∈ Z, if 0 ≤ x, then x ∈ N. 27. Every natural number is positive

6. ∀x ∈ R, 0 < x2 . 28. The negative of any even integer is even.

7. ∃x ∈ R, x2 < 0. 29. The negative of any odd integer is odd.

8. ∀x ∈ R, ∃y ∈ R, x = 2y. 30. The sum of any two odd integers is even.

9. ∀x ∈ Z, ∃y ∈ Z, x = 2y. 31. Some integers have even squares.

For a proposition P, the negation (or denial) of P, denoted ¬P, is the

Write the negation of each of the following propositions.

Negating Statements with Quantified Variables

¬(∀x ∈ O, x is prime ) ⇔ ∃x ∈ B, x is not prime.

Write the negations of the following propositions.

1. All integers are divisible by 2 or divisible by 3.

3. If an integer is divisible by 2 then it is divisible by 3.

1. Some integers are not divisible by 2 and not divisible by 3.

2. Some integers are not even or not positive.

3. Some integers are divisible by 2 and not divisible by 3.

2. ∀x ∈ Z, x is even and x is positive.

1. Negating a statement with an ‘or’ connective results in a statement

2. Negating a statement with an ‘and’ connective results in a state-

1. ¬(P or Q) ⇔ ¬P and ¬Q.

3. ¬(if P then Q) ⇔ P and ¬Q.

7. ∀x ∈ R, if ∃y ∈ R, xy > 0 then x > 0.

1. ∀x ∈ R, x ≤ x. 15. ∀x, y ∈ Q, if x < y, then ∃z ∈ R, z < Q and

11. ∀x, y ∈ Z, if x − y is even, then x + y is even. 22. ∀n ∈ N, if ∀x ∈ N, ∃t ∈ N, 7x = nt then n = 1 or

0.3 Logical Operators

‘16 is divisible by 4 and 16 is divisible by 3.’

A Boolean value is one of either False or True. The Boolean value of

The logical operator ∧ (read ‘and’) is the function accepting two

The logical operator ⇒ (read ‘implies’ or ‘if, . . . then’) is the function

Equivalent Logical Operators

¬(P and Q) ⇔ ¬P or ¬Q.

¬(if P, then Q) ⇔ P and ¬Q

Changing the Form of a Proposition

Since the logical operator x ⇒ (y ∨ z) is equivalent to the logical operator (x ∧ ¬y) ⇒ z,

if P and not Q, then R.

In fact, any compound proposition of the form

can be rewritten in the form

2. ∀x, y ∈ R, if xy is irrational, then x is irrational or y is irrational.

4. ∀x, y ∈ R, if x2 = y2 , then x = y or x = −y.

Boolean Algebraic Properties of Logical Operators

Boolean Algebraic Properties of Logical Operators