CHAPTER 11

Proofs Involving Sets

tudents in their first advanced mathematics or computer science courses

S are often surprised by the extensive role that sets play and by the fact
that most of the proofs they encounter are proofs about sets. Perhaps you’ve
already seen such proofs in your linear algebra course, where a vector
space was defined to be a set of objects (called vectors) that obey certain
properties. Your text proved many things about vector spaces, such as the
fact that the intersection of two vector spaces is also a vector space, and the
proofs used ideas from set theory. As you go deeper into mathematics, you
will encounter more and more ideas, theorems and proofs that involve sets.
The purpose of this chapter is to give you a foundation that will prepare
you for this new outlook.
We will discuss how to show that an object is an element of a set, how to
prove one set is a subset of another and how to prove two sets are equal. As
you read this chapter, you may need to occasionally refer back to Chapter 2
to refresh your memory. For your convenience, the main definitions from
Chapter 2 are summarized below. If A and B are sets, then:
© ª
A×B = (x, y) : x ∈ A, y ∈ B ,
© ª
A∪B = x : (x ∈ A) ∨ (x ∈ B) ,
© ª
A∩B = x : (x ∈ A) ∧ (x ∈ B) ,
© ª
A−B = x : (x ∈ A) ∧ (x ∉ B) ,
A = U − A.

Recall that A ⊆ B means that every element of A is also an element of B.

©ª
Also, the empty set ; = is the unique set with no elements, and ; ⊆ B for
any set B.

11.1 How to Prove a ∈ A

We will begin with a review of set-builder notation, and then review how to
show that a given object a is an element of some set A .
How to Prove a ∈ A 267

© ª
Generally, a set A will be expressed in set-builder notation A = x : P(x) ,
where P(x) is some statement (or open sentence) about x. The set A is
understood to have as elements all those things x for which P(x) is true. For
example,
© ª © ª
x : x is an odd integer = . . . , −5, −3, −1, 1, 3, 5, . . . .
© ª
A common variation of this notation is to express a set as A = x ∈ S : P(x) .
Here it is understood that A consists of all elements x of the (predetermined)
set S for which P(x) is true. Keep in mind that, depending on context, x
could be any kind of object (integer, ordered pair, set, function, etc.). There is
also nothing special about the particular variable x; any reasonable symbol
x, y, k, etc., would do. Some examples follow.

n ∈ Z : n is odd
© ª © ª
= . . . , −5, −3, −1, 1, 3, 5, . . .
x ∈ N : 6| x
© ª © ª
= 6, 12, 18, 24, 30, . . .
(a, b) ∈ Z × Z : b = a + 5
© ª © ª
= . . . , (−2, 3), (−1, 4), (0, 5), (1, 6), . . .
X ∈ P (Z) : | X | = 1
© ª © © ª© ª© ª© ª© ª© ª ª
= ..., −1 , 0 , 1 , 2 , 3 , 4 ,...
© ª
Now it should be clear how to prove that an object a belongs to a set x : P(x) .
© ª
Since x : P(x) consists of all things x for which P(x) is true, to show that
© ª
a ∈ x : P(x) we just need to show that P(a) is true. Likewise, to show
© ª
a ∈ x ∈ S : P(x) , we need to confirm that a ∈ S and that P(a) is true. These
ideas are summarized below. However, you should not memorize these
methods, you should understand them. With contemplation and practice,
using them becomes natural and intuitive.
© ª © ª
How to show a ∈ x : P(x) How to show a ∈ x ∈ S : P(x)
Show that P(a) is true. 1. Verify that a ∈ S .
2. Show that P(a) is true.

Example 11.1 Let’s investigate elements of A = x : x ∈ N and 7 | x . This

© ª

set has form A = x : P(x) where P(x) is the open sentence (x ∈ N)∧(7 | x). Thus
© ª

21 ∈ A because P(21) is true. Similarly, 7, 14, 28, 35, etc., are all elements of


A . But 8 ∉ A (for example) because P(8) is false. Likewise −14 ∉ A because
P(−14) is false.
Example 11.2 Consider the set A = X ∈¯P (N) : | X | = 3 . We know that
© ª

4, 13, 45 ∈ A because 4, 13, 45 ∈ P (N) and ¯ 4, 13, 45 ¯ = 3. Also 1, 2, 3 ∈ A ,

© ª © ª © ª¯ © ª
© ª © ª ¯© ª¯


10, 854, 3 ∈ A , etc. However 1, 2, 3, 4 ∉ A because ¯ 1, 2, 3, 4 ¯ 6= 3. Further,
− 1, 2, 3 ∉ A because − 1, 2, 3 ∉ P (N).
© ª © ª
268 Proofs Involving Sets

Example 11.3 Consider the set B = (x, y) ∈ Z × Z : x ≡ y (mod 5) . Notice

© ª

(8, 23) ∈ B because (8, 23) ∈ Z × Z and 8 ≡ 23 (mod 5). Likewise, (100, 75) ∈ B,
(102, 77) ∈ B, etc., but (6, 10) ∉ B.
Now suppose n ∈ Z and consider the ordered pair (4n + 3, 9n − 2). Does
this ordered pair belong to B? To answer this, we first observe that (4n +
3, 9n − 2) ∈ Z × Z. Next, we observe that (4n + 3) − (9n − 2) = −5n + 5 = 5(1 − n),
¡ ¢
so 5 | (4n + 3) − (9n − 2) , which means (4n + 3) ≡ (9n − 2) (mod 5). Therefore we


have established that (4n + 3, 9n − 2) meets the requirements for belonging
to B, so (4n + 3, 9n − 2) ∈ B for every n ∈ Z.
Example 11.4 This illustrates another common way of defining a set.
Consider the set C = 3x3 + 2 : x ∈ Z . Elements of this set consist of all the
© ª

values 3x3 + 2 where x is an integer. Thus −22 ∈ C because −22 = 3(−2)3 + 2.

You can confirm −1 ∈ C and 5 ∈ C , etc. Also 0 ∉ C and 12 ∉ C , etc. 
11.2 How to Prove A ⊆ B
In this course (and more importantly, beyond it) you will encounter many
circumstances where it is necessary to prove that one set is a subset of an-
other. This section explains how to do this. The methods we discuss should
improve your skills in both writing your own proofs and in comprehending
the proofs that you read.
Recall (Definition 2.3) that if A and B are sets, then A ⊆ B means that
every element of A is also an element of B. In other words, it means if a ∈ A ,
then a ∈ B. Therefore to prove that A ⊆ B, we just need to prove that the
conditional statement

“If a ∈ A , then a ∈ B”

is true. This can be proved directly, by assuming a ∈ A and deducing a ∈ B.

The contrapositive approach is another option: Assume a ∉ B and deduce
a ∉ A . Each of these two approaches is outlined below.

How to Prove A ⊆ B How to Prove A ⊆ B

(Direct approach) (Contrapositive approach)
Proof. Suppose a ∈ A . Proof. Suppose a ∉ B.
.. ..
. .
Therefore a ∈ B. Therefore a ∉ A .
Thus a ∈ A implies a ∈ B, Thus a ∉ B implies a ∉ A ,
so it follows that A ⊆ B. ■ so it follows that A ⊆ B. ■
How to Prove A ⊆ B 269

In practice, the direct approach usually results in the most straightfor-

ward and easy proof, though occasionally the contrapositive is the most ex-
pedient. (You can even prove A ⊆ B by contradiction: Assume (a ∈ A) ∧ (a ∉ B),
and deduce a contradiction.) The remainder of this section consists of exam-
ples with occasional commentary. Unless stated otherwise, we will use the
direct approach in all proofs; pay special attention to how the above outline
for the direct approach is used.

Prove that x ∈ Z : 18 | x ⊆ x ∈ Z : 6 | x .
© ª © ª
Example 11.5

Proof. Suppose a ∈ x ∈ Z : 18 | x .
© ª

This means that a ∈ Z and 18 | a.

By definition of divisibility, there is an integer c for which a = 18c.
Consequently a = 6(3c), and from this we deduce that 6 | a.
Therefore a is one of the integers that 6 divides, so a ∈ x ∈ Z : 6 | x .
© ª

We’ve shown a ∈ x ∈ Z : 18 | x implies a ∈ n ∈ Z : 6 | x , so it follows that

© ª © ª

x ∈ Z : 18 | x ⊆ x ∈ Z : 6 | x .
© ª © ª
■

Prove that x ∈ Z : 2 | x ∩ x ∈ Z : 9 | x ⊆ x ∈ Z : 6 | x .
© ª © ª © ª
Example 11.6

Proof. Suppose a ∈ x ∈ Z : 2 | x ∩ x ∈ Z : 9 | x .
© ª © ª

By definition of intersection, this means a ∈ x ∈ Z : 2 | x and a ∈ x ∈ Z : 9 | x .

© ª © ª

Since a ∈ x ∈ Z : 2 | x we know 2 | a, so a = 2c for some c ∈ Z. Thus a is even.

© ª

Since a ∈ x ∈ Z : 9 | x we know 9 | a, so a = 9d for some d ∈ Z.

© ª

As a is even, a = 9d implies d is even. (Otherwise a = 9d would be odd.)

Then d = 2e for some integer e, and we have a = 9d = 9(2e) = 6(3e).
From a = 6(3e), we conclude 6 | a, and this means a ∈ x ∈ Z : 6 | x .
© ª

We have shown that a ∈ x ∈ Z : 2 | x ∩ x ∈ Z : 9 | x implies a ∈ x ∈ Z : 6 | x ,

© ª © ª © ª

so it follows that x ∈ Z : 2 | x ∩ x ∈ Z : 9 | x ⊆ x ∈ Z : 6 | x .
© ª © ª © ª
■

Example 11.7 Show (x, y) ∈ Z×Z : x ≡ y (mod 6) ⊆ (x, y) ∈ Z×Z : x ≡ y (mod 3) .

© ª © ª

Proof. Suppose (a, b) ∈ (x, y) ∈ Z × Z : x ≡ y (mod 6) .

© ª

This means (a, b) ∈ Z × Z and a ≡ b (mod 6).

Consequently 6 |(a − b), so a − b = 6c for some integer c.
It follows that a − b = 3(2c), and this means 3 |(a − b), so a ≡ b (mod 3).
Thus (a, b) ∈ (x, y) ∈ Z × Z : x ≡ y (mod 3) .
© ª

We’ve now seen that (a, b) ∈ (x, y) ∈ Z × Z : x ≡ y (mod 6) implies (a, b) ∈

© ª

(x, y) ∈ Z × Z : x ≡ y (mod 3) , so it follows that (x, y) ∈ Z × Z : x ≡ y (mod 6) ⊆

© ª © ª

(x, y) ∈ Z × Z : x ≡ y (mod 3) .
© ª
■
270 Proofs Involving Sets

Some statements involving subsets are transparent enough that we

often accept (and use) them without proof. For example, if A and B are any
sets, then it’s very easy to confirm A ∩ B ⊆ A . (Reason: Suppose x ∈ A ∩ B.
Then x ∈ A and x ∈ B by definition of intersection, so in particular x ∈ A .
Thus x ∈ A ∩ B implies x ∈ A , so A ∩ B ⊆ A .) Other statements of this nature
include A ⊆ A ∪ B and A − B ⊆ A , as well as conditional statements such as
¡ ¢
(A ⊆ B) ∧ (B ⊆ C) ⇒ (A ⊆ C) and (X ⊆ A) ⇒ (X ⊆ A ∪ B). Our point of view in
this text is that we do not need to prove such obvious statements unless
we are explicitly asked to do so in an exercise. (Still, you should do some
quick mental proofs to convince yourself that the above statements are true.
If you don’t see that A ∩ B ⊆ A is true but that A ⊆ A ∩ B is not necessarily
true, then you need to spend more time on this topic.)
The next example will show that if A and B are sets, then P (A) ∪ P (B) ⊆
P (A ∪ B). Before beginning our proof, let’s look at an example to see if this
© ª © ª
statement really makes sense. Suppose A = 1, 2 and B = 2, 3 . Then

P (A) ∪ P (B) =
© © ª© ª© ªª © © ª © ª © ªª
;, 1 , 2 , 1, 2 ∪ ;, 2 , 3 , 2, 3
© © ª© ª© ª© ª© ªª
= ;, 1 , 2 , 3 , 1, 2 , 2, 3 .

Also P (A ∪ B) = P ( 1, 2, 3 ) = ;, 1 , 2 , 3 , 1, 2 , 2, 3 , 1, 3 , 1, 2, 3 . Thus,
© ª © © ª© ª© ª© ª© ª© ª© ªª

even though P (A) ∪ P (B) 6= P (A ∪ B), it is true that P (A) ∪ P (B) ⊆ P (A ∪ B)

for this particular A and B. Now let’s prove P (A) ∪ P (B) ⊆ P (A ∪ B) no
matter what sets A and B are.

Example 11.8 Prove: If A and B are sets, then P (A) ∪ P (B) ⊆ P (A ∪ B).

Proof. Suppose X ∈ P (A) ∪ P (B).

By definition of union, this means X ∈ P (A) or X ∈ P (B).
Therefore X ⊆ A or X ⊆ B (by definition of power sets). We consider cases.
Case 1. Suppose X ⊆ A . Then X ⊆ A ∪ B, and this means X ∈ P (A ∪ B).
Case 2. Suppose X ⊆ B. Then X ⊆ A ∪ B, and this means X ∈ P (A ∪ B).
(We do not need to consider the case where X ⊆ A and X ⊆ B because that
is taken care of by either of cases 1 or 2.) The above cases show that
X ∈ P (A ∪ B).
Thus we’ve shown that X ∈ P (A) ∪ P (B) implies X ∈ P (A ∪ B), and this
completes the proof that P (A) ∪ P (B) ⊆ P (A ∪ B). ■

In our next example, we prove a conditional statement. Direct proof is

used, and in the process we use our new technique for showing A ⊆ B.
How to Prove A = B 271

Example 11.9 Suppose A and B are sets. If P (A) ⊆ P (B), then A ⊆ B.

Proof. We use direct proof. Assume P (A) ⊆ P (B).

Based on this assumption, we must now show that A ⊆ B.
To show A ⊆ B, suppose that a ∈ A .
Then the one-element set a is a subset of A , so a ∈ P (A).
© ª © ª

But then, since P (A) ⊆ P (B), it follows that a ∈ P (B).

© ª
© ª
This means that a ⊆ B, hence a ∈ B.
We’ve shown that a ∈ A implies a ∈ B, so therefore A ⊆ B. ■

11.3 How to Prove A = B

In proofs it is often necessary to show that two sets are equal. There is a
standard way of doing this. Suppose we want to show A = B. If we show
A ⊆ B, then every element of A is also in B, but there is still a possibility
that B could have some elements that are not in A , so we can’t conclude
A = B. But if in addition we also show B ⊆ A , then B can’t contain anything
that is not in A , so A = B. This is the standard procedure for proving A = B:
Prove both A ⊆ B and B ⊆ A .

How to Prove A = B
Proof.
[Prove that A ⊆ B.]
[Prove that B ⊆ A .]
Therefore, since A ⊆ B and B ⊆ A ,
it follows that A = B. ■

Prove that n ∈ Z : 35 | n = n ∈ Z : 5 | n ∩ n ∈ Z : 7 | n .
© ª © ª © ª
Example 11.10

Proof. First we show n ∈ Z : 35 | n ⊆ n ∈ Z : 5 | n ∩ n ∈ Z : 7 | n . Suppose

© ª © ª © ª

a ∈ n ∈ Z : 35 | n . This means 35 | a, so a = 35c for some c ∈ Z. Thus a = 5(7c)

© ª

and a = 7(5c). From a = 5(7c) it follows that 5 | a, so a ∈ n ∈ Z : 5 | n . From

© ª

a = 7(5c) it follows that 7 | a, which means a ∈ n ∈ Z : 7 | n . As a belongs to

© ª

both n ∈ Z : 5 | n and n ∈ Z : 7 | n , we get a ∈ n ∈ Z : 5 | n ∩ n ∈ Z : 7 | n . Thus

© ª © ª © ª © ª

we’ve shown that n ∈ Z : 35 | n ⊆ n ∈ Z : 5 | n ∩ n ∈ Z : 7 | n .

© ª © ª © ª

Next we show n ∈ Z : 5 | n ∩ n ∈ Z : 7 | n ⊆ n ∈ Z : 35 | n . Suppose that

© ª © ª © ª

a ∈ n ∈ Z : 5 | n ∩ n ∈ Z : 7 | n . By definition of intersection, this means that

© ª © ª

a ∈ n ∈ Z : 5 | n and a ∈ n ∈ Z : 7 | n . Therefore it follows that 5 | a and 7 | a.

© ª © ª

By definition of divisibility, there are integers c and d with a = 5c and a = 7d .

Then a has both 5 and 7 as prime factors, so the prime factorization of a
272 Proofs Involving Sets

must include factors of 5 and 7. Hence 5 · 7 = 35 divides a, so a ∈ n ∈ Z : 35 | n .

© ª

We’ve now shown that n ∈ Z : 5 | n ∩ n ∈ Z : 7 | n ⊆ n ∈ Z : 35 | n .

© ª © ª © ª

At this point we’ve shown that n ∈ Z : 35 | n ⊆ n ∈ Z : 5 | n ∩ n ∈ Z : 7 | n

© ª © ª © ª

and n ∈ Z : 5 | n ∩ n ∈ Z : 7 | n ⊆ n ∈ Z : 35 | n , so we’ve proved n ∈ Z : 35 | n =

© ª © ª © ª © ª

n ∈ Z : 5| n ∩ n ∈ Z : 7| n .
© ª © ª
■

You know from algebra that if c 6= 0 and ac = bc, then a = b. The next
example shows that an analogous statement holds for sets A, B and C . The
example asks us to prove a conditional statement. We will prove it with
direct proof. In carrying out the process of direct proof, we will have to use
the new techniques from this section.
Example 11.11 Suppose A , B, and C are sets, and C 6= ;. Prove that if
A × C = B × C , then A = B.

Proof. Suppose A × C = B × C . We must now show A = B.

First we will show A ⊆ B. Suppose a ∈ A . Since C 6= ;, there exists
an element c ∈ C . Thus, since a ∈ A and c ∈ C , we have (a, c) ∈ A × C , by
definition of the Cartesian product. But then, since A × C = B × C , it follows
that (a, c) ∈ B × C . Again by definition of the Cartesian product, it follows
that a ∈ B. We have shown a ∈ A implies a ∈ B, so A ⊆ B.
Next we show B ⊆ A . We use the same argument as above, with the roles
of A and B reversed. Suppose a ∈ B. Since C 6= ;, there exists an element
c ∈ C . Thus, since a ∈ B and c ∈ C , we have (a, c) ∈ B × C . But then, since
B × C = A × C , we have (a, c) ∈ A × C . It follows that a ∈ A . We have shown
a ∈ B implies a ∈ A , so B ⊆ A .
The previous two paragraphs have shown A ⊆ B and B ⊆ A , so A = B. In
summary, we have shown that if A × C = B × C , then A = B. This completes
the proof. ■

Now we’ll look at another way that set operations are similar to oper-
ations on numbers. From algebra you are familiar with the distributive
property a · (b + c) = a · b + a · c. Replace the numbers a, b, c with sets A, B, C,
and replace · with × and + with ∩. We get A × (B ∩ C) = (A × B) ∩ (A × C). This
statement turns out to be true, as we now prove.
Example 11.12 Given sets A , B, and C , prove A × (B ∩ C) = (A × B) ∩ (A × C).

Proof. First we will show that A × (B ∩ C) ⊆ (A × B) ∩ (A × C).

Suppose (a, b) ∈ A × (B ∩ C).
By definition of the Cartesian product, this means a ∈ A and b ∈ B ∩ C .
By definition of intersection, it follows that b ∈ B and b ∈ C .
How to Prove A = B 273

Thus, since a ∈ A and b ∈ B, it follows that (a, b) ∈ A × B (by definition of ×).

Also, since a ∈ A and b ∈ C , it follows that (a, b) ∈ A × C (by definition of ×).
Now we have (a, b) ∈ A × B and (a, b) ∈ A × C , so (a, b) ∈ (A × B) ∩ (A × C).
We’ve shown that (a, b) ∈ A × (B ∩ C) implies (a, b) ∈ (A × B) ∩ (A × C) so we have
A × (B ∩ C) ⊆ (A × B) ∩ (A × C).
Next we will show that (A × B) ∩ (A × C) ⊆ A × (B ∩ C).
Suppose (a, b) ∈ (A × B) ∩ (A × C).
By definition of intersection, this means (a, b) ∈ A × B and (a, b) ∈ A × C .
By definition of the Cartesian product, (a, b) ∈ A × B means a ∈ A and b ∈ B.
By definition of the Cartesian product, (a, b) ∈ A × C means a ∈ A and b ∈ C .
We now have b ∈ B and b ∈ C , so b ∈ B ∩ C , by definition of intersection.
Thus we’ve deduced that a ∈ A and b ∈ B ∩ C , so (a, b) ∈ A × (B ∩ C).
In summary, we’ve shown that (a, b) ∈ (A × B)∩(A × C) implies (a, b) ∈ A ×(B ∩ C)
so we have (A × B) ∩ (A × C) ⊆ A × (B ∩ C).
The previous two paragraphs show that A × (B ∩ C) ⊆ (A × B) ∩ (A × C) and
(A × B) ∩ (A × C) ⊆ A × (B ∩ C), so it follows that (A × B) ∩ (A × C) = A × (B ∩ C). ■

Occasionally you can prove two sets are equal by working out a series of
equalities leading from one set to the other. This is analogous to showing two
algebraic expressions are equal by manipulating one until you obtain the
other. We illustrate this in the following example, which gives an alternate
solution to the previous example. You are cautioned that this approach
is sometimes difficult to apply, but when it works it can shorten a proof
dramatically.
Before beginning the example, a note is in order. Notice that any state-
ment P is logically equivalent to P ∧ P . (Write out a truth table if you are in
doubt.) At one point in the following example we will replace the expression
x ∈ A with the logically equivalent statement (x ∈ A) ∧ (x ∈ A).
Example 11.13 Given sets A , B, and C , prove A × (B ∩ C) = (A × B) ∩ (A × C).

Proof. Just observe the following sequence of equalities.

© ª
A × (B ∩ C) = ©(x, y) : (x ∈ A) ∧ (y ∈ B ∩ C) ª (def. of ×)
= ©(x, y) : (x ∈ A) ∧ (y ∈ B) ∧ (y ∈ C) ª (def. of ∩)
= ©(x, y) : (x ∈ A) ∧ (x ∈ A) ∧ (y ∈ B) ∧ (y ∈ C) ª (P = P ∧ P)
= ©(x, y) : ((x ∈ A) ∧ (y ∈ B))
ª ∧©((x ∈ A) ∧ (y ∈ C)) ª (rearrange)
= (x, y) : (x ∈ A) ∧ (y ∈ B) ∩ (x, y) : (x ∈ A) ∧ (y ∈ C) (def. of ∩)
= (A × B) ∩ (A × C) (def. of ×)

The proof is complete. ■

274 Proofs Involving Sets

The equation A × (B ∩ C) = (A × B) ∩ (A × C) just obtained is a fundamental

law that you may actually use fairly often as you continue with mathematics.
Some similar equations are listed below. Each of these can be proved with
this section’s techniques, and the exercises will ask that you do so.
¾
A∩B = A∪B DeMorgan’s laws for sets
A∪B = A∩B o
A ∩ (B ∪ C) = (A ∩ B) ∪ (A ∩ C) Distributive laws for sets
A ∪ (B ∩ C) = (A ∪ B) ∩ (A ∪ C) o
A × (B ∪ C) = (A × B) ∪ (A × C) Distributive laws for sets
A × (B ∩ C) = (A × B) ∩ (A × C)

It is very good practice to prove these equations. Depending on your

learning style, it is probably not necessary to commit them to memory. But
don’t forget them entirely. They may well be useful later in your mathemat-
ical education. If so, you can look them up or re-derive them on the spot. If
you go on to study mathematics deeply, you will at some point realize that
you’ve internalized them without even being cognizant of it.

11.4 Case Study: Perfect Numbers

Sometimes it takes a good bit of work and creativity to show that one set
is a subset of another or that they are equal. We illustrate this now with
examples from number theory involving what are called perfect numbers.
Even though this topic is quite old, dating back more than 2000 years, it
leads to some questions that are unanswered even today.
The problem involves adding up the positive divisors of a natural number.
To begin the discussion, consider the number 12. If we add up the positive
divisors of 12 that are less than 12, we obtain 1 + 2 + 3 + 4 + 6 = 16, which is
greater than 12. Doing the same thing for 15, we get 1 + 3 + 5 = 9 which is
less than 15. For the most part, given a natural number p, the sum of its
positive divisors less than itself will either be greater than p or less than p.
But occasionally the divisors add up to exactly p. If this happens, then p is
said to be a perfect number.

Definition 11.1 A number p ∈ N is perfect if it equals the sum of its

positive divisors less than itself. Some examples follow.

• The number 6 is perfect since 6 = 1 + 2 + 3.

• The number 28 is perfect since 28 = 1 + 2 + 4 + 7 + 14.
• The number 496 is perfect since 496 = 1 + 2 + 4 + 8 + 16 + 31 + 62 + 124 + 248.
Case Study: Perfect Numbers 275

Though it would take a while to find it by trial-and-error, the next perfect

number after 496 is 8128. You can check that 8128 is perfect. Its divisors
are 1, 2, 4, 8, 16, 32, 64, 127, 254, 508, 1016, 2032, 4064 and indeed

8128 = 1 + 2 + 4 + 8 + 16 + 32 + 64 + 127 + 254 + 508 + 1016 + 2032 + 4064.

Are there other perfect numbers? How can they be found? Do they obey
any patterns? These questions fascinated the ancient Greek mathemati-
cians. In what follows we will develop an idea—recorded by Euclid—that
partially answers these questions. Although Euclid did not use sets, we will
nonetheless phrase his idea using the language of sets.
Since our goal is to understand what numbers are perfect, let’s define
the following set:
P = p ∈ N : p is perfect .
© ª

© ª
Therefore P = 6, 28, 496, 8128, . . . , but it is unclear what numbers are in P
other than the ones listed. Our goal is to gain a better understanding of
just which numbers the set P includes. To do this, we will examine the
following set A . It looks more complicated than P , but it will be very helpful
for understanding P , as we will soon see.

A = 2n−1 (2n − 1) : n ∈ N, and 2n − 1 is prime

© ª

In words, A consists of every natural number of form 2n−1 (2n − 1), where
2n − 1 is prime. To get a feel for what numbers belong to A , look at the
following table. For each natural number n, it tallies the corresponding
numbers 2n−1 and 2n − 1. If 2n − 1 happens to be prime, then the product
2n−1 (2n − 1) is given; otherwise that entry is labeled with an ∗.

n 2n−1 2n − 1 2n−1 (2n − 1)

1 1 1 ∗
2 2 3 6
3 4 7 28
4 8 15 ∗
5 16 31 496
6 32 63 ∗
7 64 127 8128
8 128 255 ∗
9 256 511 ∗
10 512 1023 ∗
11 1024 2047 ∗
12 2048 4095 ∗
13 4096 8191 33, 550, 336
276 Proofs Involving Sets

Notice that the first four entries of A are the perfect numbers 6, 28, 496
and 8128. At this point you may want to jump to the conclusion that A = P .
But it is a shocking fact that in over 2000 years no one has ever been able
to determine whether or not A = P . But it is known that A ⊆ P , and we will
now prove it. In other words, we are going to show that every element of A
is perfect. (But by itself, that leaves open the possibility that there may be
some perfect numbers in P that are not in A .)
The main ingredient for the proof will be the formula for the sum of a
geometric series with common ratio r . You probably saw this most recently
in Calculus II. The formula is
n r n+1 − 1
rk =
X
.
k=0 r−1

We will need this for the case r = 2, which is

n
2k = 2n+1 − 1.
X
(11.1)
k=0

(See the solution for Exercise 19 in Section 12.4 for a proof of this formula.)
Now we are ready to prove our result. Let’s draw attention to its significance
by calling it a theorem rather than a proposition.

Theorem 11.1 If A = 2n−1 (2n − 1) : n ∈ N, and 2n − 1 is prime and P =

© ª

p ∈ N : p is perfect , then A ⊆ P .
© ª

Proof. Assume A and P are as stated. To show A ⊆ P , we must show that

p ∈ A implies p ∈ P . Thus suppose p ∈ A . By definition of A , this means

p = 2n−1 (2n − 1) (11.2)

for some n ∈ N for which 2n − 1 is prime. We want to show that p ∈ P , that is,
we want to show p is perfect. Thus, we need to show that the sum of the
positive divisors of p that are less than p add up to p. Notice that since 2n − 1
is prime, any divisor of p = 2n−1 (2n − 1) must have the form 2k or 2k (2n − 1)
for 0 ≤ k ≤ n − 1. Thus the positive divisors of p are as follows:

20 , 21 , 22 , . . . 2n−2 , 2n−1 ,
20 (2n − 1), 21 (2n − 1), 22 (2n − 1), . . . 2n−2 (2n − 1), 2n−1 (2n − 1).

Notice that this list starts with 20 = 1 and ends with 2n−1 (2n − 1) = p.
Case Study: Perfect Numbers 277

If we add up all these divisors except for the last one (which equals p)
we get the following:

nX
−1 nX
−2 nX
−1 nX
−2
2k + 2k (2n − 1) = 2k + (2n − 1) 2k
k=0 k=0 k=0 k=0
= (2n − 1) + (2n − 1)(2n−1 − 1) (by Equation (11.1))
= [1 + (2n−1 − 1)](2n − 1)
= 2n−1 (2n − 1)
= p (by Equation (11.2)).

This shows that the positive divisors of p that are less than p add up to
p. Therefore p is perfect, by definition of a perfect number. Thus p ∈ P , by
definition of P .
We have shown that p ∈ A implies p ∈ P , which means A ⊆ P . ■

Combined with the chart on the previous page, this theorem gives us
a new perfect number! The element p = 213−1 (213 − 1) = 33, 550, 336 in A is
perfect.
Observe also that every element of A is a multiple of a power of 2, and
therefore even. But this does not necessarily mean every perfect number
is even, because we’ve only shown A ⊆ P , not A = P . For all we know there
may be odd perfect numbers in P − A that are not in A .
Are there any odd perfect numbers? No one knows.
In over 2000 years, no one has ever found an odd perfect number, nor
has anyone been able to prove that there are none. But it is known that the
set A does contain every even perfect number. This fact was first proved by
Euler, and we duplicate his reasoning in the next theorem, which proves
that A = E , where E is the set of all even perfect numbers. It is a good
example of how to prove two sets are equal.
For convenience, we are going to use a slightly different definition of a
perfect number. A number p ∈ N is perfect if its positive divisors add up
to 2p. For example, the number 6 is perfect since the sum of its divisors is
1 + 2 + 3 + 6 = 2 · 6. This definition is simpler than the first one because we do
not have to stipulate that we are adding up the divisors that are less than
p. Instead we add in the last divisor p, and that has the effect of adding an
additional p, thereby doubling the answer.
278 Proofs Involving Sets

Theorem 11.2 If A = 2n−1 (2n − 1) : n ∈ N, and 2n − 1 is prime and E =

© ª

p ∈ N : p is perfect and even , then A = E .

© ª

Proof. To show that A = E , we need to show A ⊆ E and E ⊆ A .

First we will show that A ⊆ E . Suppose p ∈ A . This means p is even,
because the definition of A shows that every element of A is a multiple of a
power of 2. Also, p is a perfect number because Theorem 11.1 states that
every element of A is also an element of P , hence perfect. Thus p is an even
perfect number, so p ∈ E . Therefore A ⊆ E .
Next we show that E ⊆ A . Suppose p ∈ E . This means p is an even
perfect number. Write the prime factorization of p as p = 2k 3n1 5n2 7n2 . . .,
where some of the powers n1 , n2 , n3 . . . may be zero. But, as p is even, the
power k must be greater than zero. It follows p = 2k q for some positive
integer k and an odd integer q. Now, our aim is to show that p ∈ A , which
means we must show p has form p = 2n−1 (2n − 1). To get our current p = 2k q
closer to this form, let n = k + 1, so we now have

p = 2n−1 q. (11.3)

List the positive divisors of q as d1 , d2 , d3 , . . . , d m . (Where d1 = 1 and d m = q.)

Then the divisors of p are:

20 d 1 20 d 2 20 d 3 ... 20 d m
21 d 1 21 d 2 21 d 3 ... 21 d m
22 d 1 22 d 2 22 d 3 ... 22 d m
23 d 1 23 d 2 23 d 3 ... 23 d m
.. .. .. ..
. . . .
n−1 n−1 n−1 n−1
2 d1 2 d2 2 d3 ... 2 dm

Since p is perfect, these divisors add up to 2p. By Equation (11.3), their

sum is 2p = 2(2n−1 q) = 2n q. Adding the divisors column-by-column, we get
nX
−1 nX
−1 nX
−1 nX
−1
2k d 1 + 2k d 2 + 2k d 3 + · · · + 2k d m = 2n q.
k=0 k=0 k=0 k=0

Applying Equation (11.1), this becomes

(2n − 1)d 1 + (2n − 1)d 2 + (2n − 1)d 3 + · · · + (2n − 1)d m = 2n q

(2n − 1)(d 1 + d 2 + d 3 + · · · + d m ) = 2n q
2n q
d1 + d2 + d3 + · · · + d m = ,
2n − 1
Case Study: Perfect Numbers 279

so that
(2n − 1 + 1)q (2n − 1)q + q q
d1 + d2 + d3 + · · · + d m = = = q+ n .
2n − 1 2n − 1 2 −1

From this we see that 2nq−1 is an integer. It follows that both q and 2nq−1
are positive divisors of q. Since their sum equals the sum of all positive
divisors of q, it follows that q has only two positive divisors, q and 2nq−1 .
Since one of its divisors must be 1, it must be that 2nq−1 = 1, which means
q = 2n − 1. Now a number with just two positive divisors is prime, so q = 2n − 1
is prime. Plugging this into Equation (11.3) gives p = 2n−1 (2n − 1), where
2n − 1 is prime. This means p ∈ A , by definition of A . We have now shown
that p ∈ E implies p ∈ A , so E ⊆ A .
Since A ⊆ E and E ⊆ A , it follows that A = E . ■

Do not be alarmed if you feel that you wouldn’t have thought of this
proof. It took the genius of Euler to discover this approach.
We’ll conclude this chapter with some facts about perfect numbers.

• The sixth perfect number is p = 217−1 (217 − 1) = 8589869056.

• The seventh perfect number is p = 219−1 (219 − 1) = 137438691328.
• The eighth perfect number is p = 231−1 (231 − 1) = 2305843008139952128.
• The twentieth perfect number is p = 24423−1 (24423 − 1). It has 2663 digits.
• The twenty-third perfect number p = 211213−1 (211213 − 1) has 6957 digits.

As mentioned earlier, no one knows whether or not there are any odd
perfect numbers. It is not even known whether there are finitely many or
infinitely many perfect numbers. It is known that the last digit of every
even perfect number is either a 6 or an 8. Perhaps this is something you’d
enjoy proving.
We’ve seen that perfect numbers are closely related to prime numbers
that have the form 2n − 1. Such prime numbers are called Mersenne
primes, after the French scholar Marin Mersenne (1588–1648), who popu-
larized them. The first several Mersenne primes are 22 − 1 = 3, 23 − 1 = 7, 25 −
1 = 31, 27 − 1 = 127 and 213 − 1 = 8191. To date, only 48 Mersenne primes are
known, the largest of which is 257,885,161 − 1. There is a substantial cash prize
for anyone who finds a 49th. (See http://www.mersenne.org/prime.htm.)
You will probably have better luck with the exercises.
 280 Proofs Involving Sets

Exercises for Chapter 11

Use the methods introduced in this chapter to prove the following statements.
1. Prove that 12 n : n ∈ Z ⊆ 2n : n ∈ Z ∩ 3n : n ∈ Z .
© ª © ª © ª

2. Prove that 6n : n ∈ Z = 2n : n ∈ Z ∩ 3 n : n ∈ Z .
© ª © ª © ª

3. If k ∈ Z, then n ∈ Z : n | k ⊆ n ∈ Z : n | k2 .
© ª © ª

4. If m, n ∈ Z, then x ∈ Z : mn | x ⊆ x ∈ Z : m | x ∩ x ∈ Z : n | x .
© ª © ª © ª

5. If p and q are positive integers, then pn : n ∈ N ∩ qn : n ∈ N 6= ;.

© ª © ª

6. Suppose A, B and C are sets. Prove that if A ⊆ B, then A − C ⊆ B − C .

7. Suppose A, B and C are sets. If B ⊆ C , then A × B ⊆ A × C .
8. If A, B and C are sets, then A ∪ (B ∩ C ) = ( A ∪ B) ∩ ( A ∪ C ).
9. If A, B and C are sets, then A ∩ (B ∪ C ) = ( A ∩ B) ∪ ( A ∩ C ).
10. If A and B are sets in a universal set U , then A ∩ B = A ∪ B.
11. If A and B are sets in a universal set U , then A ∪ B = A ∩ B.
12. If A, B and C are sets, then A − (B ∩ C ) = ( A − B) ∪ ( A − C ).
13. If A, B and C are sets, then A − (B ∪ C ) = ( A − B) ∩ ( A − C ).
14. If A, B and C are sets, then ( A ∪ B) − C = ( A − C ) ∪ (B − C ).
15. If A, B and C are sets, then ( A ∩ B) − C = ( A − C ) ∩ (B − C ).
16. If A, B and C are sets, then A × (B ∪ C ) = ( A × B) ∪ ( A × C ).
17. If A, B and C are sets, then A × (B ∩ C ) = ( A × B) ∩ ( A × C ).
18. If A, B and C are sets, then A × (B − C ) = ( A × B) − ( A × C ).
19. Prove that 9n : n ∈ Z ⊆ 3n : n ∈ Z , but 9n : n ∈ Z 6= 3n : n ∈ Z
© ª © ª © ª © ª

20. Prove that 9n : n ∈ Q = 3n : n ∈ Q .

© ª © ª

21. For each a ∈ R, let A a = ( x, a( x2 − 1)) ∈ R2 : x ∈ R . Prove that

© ª \ © ª
A a = (−1, 0), (1, 0) .
a∈R
[3 − x2 , 5 + x2 ] = [3, 5].
\
22. Prove that
x∈R
23. Suppose A, B, C and D are sets. Prove that ( A × B) ∪ (C × D ) ⊆ ( A ∪ C ) × (B ∪ D ).
24. Prove 4k + 5 : k ∈ Z = 4k + 1 : k ∈ Z .
© ª © ª

25. Prove 12a + 4b : a, b ∈ Z = 4 c : c ∈ Z .

© ª © ª

26. Prove 12a + 25 b : a, b ∈ Z = Z.

© ª

27. Suppose B 6= ; and A × B ⊆ B × C . Prove A ⊆ C .

28. Prove that (Z × N) ∩ (N × Z) = N × N.