CS/Math 240: Introduction to Discrete Mathematics

Reading 4 : Proofs
Author: Dieter van Melkebeek (updates by Beck Hasti and Gautam Prakriya)

Up until now, we have been introducing mathematical notation to capture concepts such as
propositions, implications, predicates, and sets. We need this machinery in order to be able to
argue properties of discrete structures in a rigorous manner. As we were introducing new concepts,
we stated various facts and gave proofs of some of them, but we were not explicit about what a
correct proof should look like. In this reading we start discussing what constitutes a valid proof of
a proposition and give some guidelines for writing proofs.

4.1 Proofs
We briefly mentioned what proofs were in the second reading. Let’s repeat some of this discussion,
and make our definition of a proof more precise.
Definition 4.1. A proof of a proposition P is a chain of logical deductions ending in P and starting
from some set of axioms.
Our definition of a proof mentions axioms and logical deductions, both of which require further
consideration. Let’s discuss them one by one.

4.1.1 Axioms
Axioms are statements we take for granted and do not prove. The set of axioms we use depends
on the area we work in. For geometry, we would use Euclid’s five axioms of geometry. Another set
of axioms are the ZFC axioms (the abbreviation stands for Zermelo, Fraenkel, and the axiom of
Choice) which form the basis of all set theory. However, both of these sets of axioms are small and
proving any substantial result starting just from those axioms requires a significant amount of work.
Thus, such sets of axioms are more suitable for a course on logic than a course on discrete structures.
In this course, we use a much larger set of axioms because our focus is on proof techniques and
their applications to discrete structures. Thus, we will consider any familiar fact from math at the
level of high school as an axiom. If you are unsure whether you can take something for granted on
an assignment, just ask.

4.1.2 Logical Deductions

Logical deductions, which are sometimes called inference rules, tell us how to construct proofs of
propositions out of axioms and other proofs. One example of an inference rule is modus ponens,
which says that if we have a proof of P and a proof of P ⇒ Q, then we also have a proof of Q.
We now define some terminology and notation for describing inference rules. An inference rule
consists of antecedents and a conclusion. Antecedents, also known as premises, are axioms or other
proofs. If all the antecedents are true, the inference rule says that we have a proof of the conclusion.
We sometimes also use the word consequence instead of the word conclusion. We illustrate this
terminology on the example of modus ponens.
Example 4.1: Modus ponens deals with two statements, P and Q. The antecedents are P and
P ⇒ Q, and the conclusion is Q. ⊠

1
4.2 Proof Techniques

To describe inference rules in a more compact way, we draw a horizontal line, place all an-
tecedents above the horizontal line (either on the same line or on multiple lines), and write the
conclusion below the horizontal line. In Figure 4.1a we show the notation for a general inference
rule with antecedents P1 , P2 , . . . , Pk and conclusion Q, and we give two ways of writing modus
ponens in Figures 4.1b and 4.1c.

P
P1 P2 . . . Pk P P ⇒Q P ⇒Q
Q Q Q
(a) General case (b) Modus ponens using (c) Modus ponens using
one line two lines

Figure 4.1: Notation for logical inference rules

4.2 Proof Techniques

In a proof, we apply logical deductions in order to reach the proposition we are proving from a set
of axioms. We will not attempt to turn every logical step into the form of Figure 4.1 in this course,
and will write proofs at a higher level.
We now present some proof techniques. We have seen examples of some of proofs already, but
some of the proofs were too complicated to serve as examples illustrating the proof techniques.
Below, we present proofs of simpler statements in order to highlight the proof techniques used.

4.2.1 Proofs “By Picture”

A common approach to constructing proofs is to capture a proposition using descriptive pictures
and then reason about the pictures. This is a very powerful technique as it allows us to use our
intuition. However, be warned that in some cases our intuition may lead us astray.
We now present a proof of the Pythagorean Theorem.

c
a

Figure 4.2: A right triangle with sides of length a, b, and c

Proposition 4.2 (Pythagorean Theorem). In a right triangle where the hypotenuse has length c
and the other two sides have lengths a and b, we have a2 + b2 = c2 .
Proof. Consider a right triangle like the one in Figure 4.2. We take four copies of the triangle and
arrange them in two different ways.
First, form a square with the hypotenuse as the side. Its area is c2 . We show this arrangement
in Figure 4.3a. The four copies of the right triangle are shaded in dark gray. The space shaded
light gray inside the square in Figure 4.3a is a square of side length b − a.
Now rearrange the five pieces differently, as shown in Figure 4.3b. The thick blue lines indicate
that we can view this arrangement as two squares of sides a and b placed next to each other. The
square on the right has length is b, so the area of that square is b2 . The square on the left has side
length a, so its area is a2 , and the total area is, therefore, a2 + b2 .

2
 4.2 Proof Techniques 4.2.2 Proving Implications

b a
c
a

b b−a
b b−a

a
a b−a a

b
(a) An arrangement of four copies (b) A rearrangement of Figure 4.3a.
of the right triangle from Figure
4.2. There is a square in the mid-
dle.

Figure 4.3: Two arrangements of four triangles and a square.

Since we obtained the second picture from the first one by rearranging, they have the same
area, which completes the proof that a2 + b2 = c2 .

The use of Venn diagrams to prove properties of sets is another example of a proof “by picture”.
While such proofs are often very appealing, they don’t constitute a valid proof in mathematics.
Pictures are typically used only to aid our intuition.

4.2.2 Proving Implications

We now consider statements of the form P ⇒ Q, and look at two approaches to constructing proofs
of such statements

4.2.2.1 Direct Proof

In order to prove the implication P ⇒ Q using a direct proof, we take the following three steps.

Step 1: Assume that P holds. We usually write this as the first sentence in the proof.

Step 2: Logically derive Q from P .

Step 3: Say that Q holds. This is usually the last sentence in the proof.

As an example, we prove the statement that if an integer is odd, then so is its square. We state
it as Theorem 4.3.

Theorem 4.3. (∀x ∈ Z) x is odd ⇒ x2 is odd.

The statements P and Q for the implication in Theorem 4.3 are P : “x is odd” and “Q: x2 is
odd”.
Note that the implication is universally quantified. Whenever we prove a universally quantified
statement, we have to prove it for every x in the domain. In our case, we have to prove the
implication P ⇒ Q for every integer x. It does not suffice to prove, say, that if 3 is odd, then 9 is
odd. We cannot make any assumption about x besides the fact that it’s odd.
For the purposes of presentation, we label where the three steps outlined earlier in the margin.

3
 4.2 Proof Techniques 4.2.2 Proving Implications

Proof of Theorem 4.3.

Step 1 Let x be an integer, and assume that x is odd.
Step 2
Since x is odd, we can write x as x = 2y + 1 for some y ∈ Z. In particular, y = x−1 2 . Then
x2 = (2y + 1)2 = 4y 2 + 4y + 1 = 2(2y 2 + 2y) + 1. Since y is an integer, so is 2y 2 + 2y, which means
that x2 = 2z + 1 for some z ∈ Z.
Step 3 Therefore, x2 is odd.

We usually highlight the end of the proof in some way. In the proof above, we used a square
in the lower right corner at the end of the last paragraph. Another common way to end a proof is
to write Q.E.D. This comes from Latin “quod erat demonstrandum”, which means “which is what
had to be shown”.

4.2.2.2 Indirect Proof

In an indirect proof of the implication P ⇒ Q, we prove the contrapositive implication ¬Q ⇒ ¬P .
Since the contrapositive of an implication is logically equivalent to the original implication, proving
the contrapositive also proves the original implication. We outline the steps in an indirect proof
below.
Step 1: Say that we use an indirect proof, and state the contrapositive of the implication we are
proving.
Step 2: Use a proof technique of our choice to prove the contrapositive.
Step 3: Conclude that the original implication is proved.
The inference rule for an indirect proof is
¬Q ⇒ ¬P
.
P ⇒Q
We often use an indirect proof if the implication we want to prove contains negations. For such
implications, an indirect proof is a natural way to go because taking the contrapositive removes
negations from the statement of the implication. We demonstrate this technique on the proof of the
statement which says that if a positive real number is not rational, its square root is not rational
either. Notice that we use a direct proof to prove the contrapositive in this example.
√
Theorem 4.4. (∀x ∈ R+ )x ∈ /Q⇒ x∈ / Q.
Again, we highlight the steps we mentioned earlier in the margin.

Proof.
√
Step 1 We prove the contrapositive of our statement, that is, we show that if x is rational, then so
is x.
Step 2 We use a direct proof to prove the contrapositive.
√
Let x be a positive real number, and assume that x ∈ Q. Then
√ a
x= for some a ∈ Z, b ∈ N, b ̸= 0. (4.1)
b
√
Taking the square of both sides of (4.1) yields x = (a/b)2 . Now ( x)2 = x and (a/b)2 = a2 /b2 , so
x = a2 /b2 . Since a is an integer, so is a2 , and since b is a positive integer, so is b2 , which means
that x is a rational number.
√
Step 3 It follows that if x ∈
/ Q, then x ∈ / Q.

4
 4.2 Proof Techniques 4.2.3 Proving Equivalences

4.2.3 Proving Equivalences

Recall that an equivalence is a statement of the form P ⇐⇒ Q.

4.2.3.1 Proving Two Implications

One way to rewrite an equivalence P ⇐⇒ Q is (P ⇒ Q) ∧ (Q ⇒ P ). Thus, if both implications
hold, then so does the equivalence. Thus, to prove an equivalence is true, we can take the following
steps

Step 1: State that we prove two equivalences

Step 2: Prove the implication P ⇒ Q using a proof technique of your choice. Make sure you state
your choice of proof technique.

Step 3: Prove the implication Q ⇒ P using a proof technique of your choice.

Step 4: In the conclusion, state what you proved.

The inference rule for this kind of proof is

P ⇒Q Q⇒P
.
P ⇐⇒ Q
As an example, we prove another relationship between an integer and its square. As will be apparent
from the proof, Theorem 4.5 is a strengthening of Theorem 4.3.

Theorem 4.5. (∀x ∈ Z) x is even ⇐⇒ x2 is even.

Proof.
Step 1 Let x ∈ Z. We prove the equivalence “x is even ⇐⇒ x2 is even” by proving the implications
“x is even ⇒ x2 is even” and “x2 is even ⇒ x is even”.
Step 2 We first prove that if x is even, then so is x2 . We do so by a direct proof.
If x is even, we can write x = 2y for some y ∈ Z. By squaring both sides, we get x2 = (2y)2 =
4y = 2(2y 2 ). Now 2y 2 is an integer, which means that x2 is even.
2

Step 3 Next, we prove that if x2 is even, then so is x. We prove this using an indirect proof.
Observe that the statement “if x2 is even, then so is x” is the contrapositive of Theorem 4.3.
Recall that Theorem 4.3 says that if x is odd, then so is x2 . Its contrapositive says that if x2 is
not odd, then x is not odd. But this is equivalent to saying that if x2 is even, then so is x, which
is what we wanted to show.
Step 4 It follows that x is even if and only if x2 is even.

The proof of Theorem 4.5 uses a theorem we proved earlier. Using theorems to prove other
theorems is common practice. The programming analog of this is calling a function instead of
writing all the code for that function again from scratch. It makes no sense to reprove a fact that
has already been proved, just like it makes no sense to rewrite a function that is provided to us.
Of course, there are exceptions to this rule, such as if we want to prove a slightly different fact or if
we want to override some function in a subclass, but in general, we should attempt to use existing
work before we “reinvent the wheel”.

5
4.2 Proof Techniques 4.2.4 Proof by Contradiction

4.2.3.2 Chains of Equivalences

Another way to prove that the equivalence P ⇐⇒ Q is true is to demonstrate a sequence of
statements P1 , P2 , . . . , Pk such that P is P1 , P1 ⇐⇒ P2 , . . . , Pk−1 ⇐⇒ Pk , and Pk is Q.
Formally, the inference rule is
P1 ⇐⇒ P2 P2 ⇐⇒ P3 · · · Pk−1 ⇐⇒ Pk
.
P1 ⇐⇒ Pk
When we prove an equivalence by proving two implications, we argue P ⇒ Q and Q ⇒ P separately.
But if all steps in the proof are equivalences and not just implications, we prove P ⇒ Q and Q ⇒ P
at the same time.
As an example of this proof technique, we prove a relationship between even and odd numbers.

Theorem 4.6. (∀x ∈ Z) x2 is even ⇐⇒ (x + 1)2 is odd.

Proof. Pick x ∈ Z. We prove the equivalence “x2 is even ⇐⇒ (x + 1)2 is odd” by constructing a
chain of equivalences.
We know by Theorem 4.5 than x2 is even if and only if x is even. The latter is true if and only
if x + 1 is odd. Note that Theorem 4.5 also tells us that x2 is odd if and only if x is odd. Therefore,
using x + 1 instead of x in our alternative formulation of Theorem 4.5, we see that x + 1 is odd if
and only if (x + 1)2 is odd.
Thus, we have shown that x2 is even if and only if (x + 1)2 is odd.

4.2.4 Proof by Contradiction

The last two sections discussed proof techniques that apply to proving specific kinds of statements.
With this section, we move towards more general proof techniques.
In order to prove a statement P by contradiction, we assume that P is false, and show that this
assumption leads to a false statement. The inference rule for a proof by contradiction is
¬P ⇒ false
.
P
At a first sight, an argument by contradiction looks strange because the reasoning happens in
an “absurd world” where we assume a false statement. Therefore, such proofs are less intuitive, and
you should avoid them as much as possible. It is often possible to rewrite a proof by contradiction
as a direct proof or an indirect proof. In some cases, however, proof by contradiction is the only
possibility.
We saw an example of a proof by contradiction in the reading on sets where we showed that the
power set of the natural numbers is not countable. In this reading we prove a simpler statement.
We also use the next proof to show how to label equations. Labeling some of our equations
with numbers or other symbols makes it easy to refer to them later without writing long English
sentences.
√
Theorem 4.7. 2 ∈ / Q.

Proof. We give a√ proof by contradiction. √

Assume that 2 is rational. Then we can write 2 = a/b for some a ∈ Z, b ∈ N, b ̸= 0.
We first rewrite a/b so that the numerator and the denominator share no common factors. Let
gcd(a, b) be the greatest common divisor of a and b, and define a′ = a/ gcd(a, b) and b′ = b/ gcd(a, b).

6
4.2 Proof Techniques 4.2.5 Proof by Cases

Note that a/b = a′ /b′ because dividing the numerator and the denominator of a fraction by the
same number doesn’t change the value of the fraction. We now have
√ a′
2= ′ where gcd(a′ , b′ ) = 1. (4.2)
b
By squaring (4.2), we get 2 = (a′ /b′ )2 = (a′ )2 /(b′ )2 , so

2(b′ )2 = (a′ )2 . (4.3)

Therefore, (a′ )2 is even, which implies that a′ is even by Theorem 4.5. Hence, we can write

a′ = 2c, c ∈ Z. (4.4)

Substituting (4.4) into (4.3) tells us that 2(b′ )2 = (2c)2 = 4c2 , so (b′ )2 = 2c2 . It follows that (b′ )2
is even. By Theorem 4.5, this means that b′ is even.
We have shown that a′ and b′ are both even, which means that 2 is a common divisor of a′ and
b . This is a contradiction with the fact that gcd(a′ , b′ ) = 1, so the statement that 2 is√a common
′

divisor of a′ and b′ is false. What led to this contradiction was the assumption that 2 √ ∈ Q, so
this assumption is false. Therefore, the opposite of our assumption is true, and we have 2 ∈ / Q,
which is what we wanted to show.

4.2.5 Proof by Cases

Another technique that is generally applicable is proving a statement by cases. We have actually
used this technique already, but did not state it explicitly. Proving logical equivalence of proposi-
tional formulas by truth tables is a proof by cases in which every row of the truth table corresponds
to one case.
In a proof of a proposition P by cases, we come up with a set of conditions C1 , C2 , . . . , Ck such
that every one of them implies P , and where we are guaranteed in any situation that at least one
of these conditions is true. Formally, we have the following inference rule:

(C1 ∨ C2 ∨ · · · ∨ Ck ) C1 ⇒ P C2 ⇒ P · · · Ck ⇒ P
.
P
We give an example of this proof technique by proving an elementary fact about graphs.

Theorem 4.8. Every group of 6 people contains a subgroup of 3 people who are mutual acquain-
tances or a subgroup of 3 people who are mutual strangers.

Proof. Let Alice be one of the people. Consider two cases.

Case 1. Suppose that Alice knows at least three people from the group. We have two subcases
in this situation.
Case 1.1. Suppose that among Alice’s acquaintances, there are at least two people who know
each other. Then Alice and those two people are all mutual acquaintances.
Case 1.2. Now suppose that no acquaintance of Alice knows anybody else who knows Alice. In
that case, any group of 3 of Alice’s acquaintances is a group of mutual strangers.
Since either there is a pair of Alice’s acquaintances who know each other (case 1.1), or there
isn’t one (case 1.2), we have shown that if Alice has at least three acquaintances, then the group
contains a subgroup of three mutual acquaintances or a subgroup of three mutual strangers. This
completes the proof for case 1.

7
4.2 Proof Techniques 4.2.5 Proof by Cases

Case 2. Now suppose that Alice knows at most two people from the group. We have two cases
in this situation.
Case 2.1. Suppose that among people Alice doesn’t know, there are at least two people who
don’t know each other. Then Alice and those two people are all mutual strangers.
Case 2.2. Now suppose that all people not known by Alice know each other. In that case, any
group of 3 such people is a group of mutual acquaintances.
Since either there is a pair of people not known by Alice who also don’t know each other (case
2.1), or all people not known by Alice know each other (case 2.2), we have shown that if Alice has
at most two acquaintances, then the group contains a subgroup of three mutual acquaintances or
a subgroup of three mutual strangers. This completes the proof for case 2.
Finally, since Alice either has at least three (Case 1) or at most two (Case 2) acquaintances,
we have shown that the group contains a subgroup of three mutual acquaintances or a subgroup of
three mutual strangers.

Observe that the proof for Case 2 is the same as the proof for Case 1, except the roles of the
words “acquaintance” and “stranger” are switched.
We mentioned that Theorem 4.8 was a statement about graphs, yet we did not use any graph
terminology in the theorem or its proof. Here is the connection. Represent the six people as vertices
in a graph. There is an edge connecting two vertices if the people those vertices correspond to know
each other. Now Alice is one vertex in the graph, and is connected to some number of vertices by
edges. In Case 1, there area at least 3 edges connecting Alice to other vertices. Furthermore, in
Case 1.1, there is at least one edge between vertices representing people who know Alice, and in
Case 1.2 there are no such edges. We show these two cases in Figures 4.4a and 4.4b.

Alice Eve Alice Eve

Bob Bob

Tom Tom

(a) Case 1.1: Some pair of Alice’s ac- (b) Case 1.2: No pair of Alice’s ac-
quaintances know each other. The three quaintances know each other. There
thick lines show that Alice, Bob and are no edges between the vertices
Paul are mutual acquaintances. labeled Eve, Bob and Tom, which
means that those three people are
mutual strangers.

Figure 4.4: Graphs representing Case 1 in the proof of Theorem 4.8.