Chapter 2

Techniques of Proof
Note 3

Lecturer: Dr. Rajitha Ranasinghe Academic Year: 2021/2022

2.5 : MISTAKES IN PROOFS

There are many common errors made in constructing mathematical proofs. We will briefly describe
some of these here. Among the most common errors are mistakes in arithmetic and basic algebra.
Each step of a mathematical proof needs to be correct and the conclusion needs to follow logically from
the steps that precede it. Many mistakes result from the introduction of steps that do not logically
follow from those that precede it.

Example 1 What is wrong with this infamous “proof” of 1 = 2?

Proof : We use these steps, where a and b are two equal positive integers.

Step Reason

(1). a = b Given

(2). a2 = ab Multiply both sides of (1) by a

(3). a2 − b2 = ab − b2 Subtract b2 from both sides of (2)

(4). (a − b)(a + b) = b(a − b) Factor both sides of (3)

(5). a + b = b Divide both sides of (4) by a − b

(6). 2b = b Replace a by b in (5) because a = b and simplify

(7). 2 = 1 Divide both sides of (6) by b

Solution: Every step is valid except for one, step 5 where we divided both sides by a − b. The error is
that a − b equals zero; division of both sides of an equation by the same quantity is valid as long as
this quantity is not zero.

MAT 1023 1 Dr. P. G. R. S. Ranasinghe

Example 2 What is wrong with this “proof” of “If n2 is positive, then n is positive”.

Proof : Suppose that n2 is positive. Because the conditional statement “If n is positive, then n2 is
positive” is true, we can conclude that n is positive.


Solution :

• Let P (n) : n is positive and Q(n) : n2 is positive. Then our hypothesis is Q(n).

• The statement “If n is positive, then n2 is positive” is the statement ∀n(P (n) −→ Q(n)).

• From the hypothesis Q(n) and the statement ∀n(P (n) −→ Q(n)) we cannot conclude P (n),
because we are not using a valid rule of inference.

• Instead, this is an example of the fallacy of affirming the conclusion. A counterexample is

supplied by n = −1 for which n2 = 1 is positive, but n is negative.

Example 3 What is wrong with this “proof” of “If n is not positive, then n2 is not positive.”

Proof : Suppose that n is not positive. Because the conditional statement “If n is positive, then n2
is positive” is true, we can conclude that n2 is not positive.


Solution :

• Let P (n) and Q(n) be as in the solution of Example 2.

• Then our hypothesis is ¬P (n) and the statement “If n is positive, then n2 is positive” is the
statement ∀n (P (n) −→ Q(n)).

• From the hypothesis ¬P (n) and the statement ∀n (P (n) −→ Q(n)) we cannot conclude ¬Q(n),
because we are not using a valid rule of inference.

• Instead, this is an example of the fallacy of denying the hypothesis. A counterexample is supplied
by n = −1, as in Example 2.

MAT 1023 2 Dr. P. G. R. S. Ranasinghe

Example 4 Why Is the following argument correct? It supposedly shows that n is an even integer
whenever n2 is an even integer.

Suppose that n2 is even. Then n2 = 2k for some integer k. Let n = 2l for some integer l. This
shows that n is even.

Solution : This argument is incorrect. The statement “let n = 2l for some integer l” occurs in the
proof. No argument has been given to show that n can be written as 2l for some integer l. This is
circular reasoning because this statement is equivalent to the statement being proved, namely, “n
is even”. Of course, the result itself is correct; only the method of proof is wrong.

Remark : Making mistakes in proofs is part of the learning process. When you make a mistake that
someone else find, you should carefully analyze where you went wrong and make sure that you do not
make the same mistake again. Even professional mathematicians make mistakes in proofs. More than
a few incorrect proofs of important results have fooled people for many years before subtle errors in
them were found.

MAT 1023 3 Dr. P. G. R. S. Ranasinghe

2.6 : EXHAUSTIVE PROOF AND PROOF BY CASES
Sometimes we cannot prove a theorem using a single argument that holds for all possible cases.
We now introduce a method that can be used to prove a theorem, by considering different cases
separately. This method is based on a rule of inference that we will now introduce. To prove a
conditional statement of the form

(p1 ∨ p2 ∨ · · · ∨ pn ) −→ q
the tautology

[(p1 ∨ p2 ∨ · · · ∨ pn ) −→ q] ←→ [(p1 −→ q) ∧ (p2 −→ q) ∧ · · · (pn −→ q)]

can be used as a rule of inference. This shows that the original conditional statement with a hypothesis
made up of a disjunction of the propositions p1 , p2 , · · · , pn can be proved by proving each of the n
conditional statements pi −→ q, i = 1, 2, · · · , n, individually. Such an argument is called a proof by
cases. Sometimes to prove that a conditional statement p −→ q is true, it is convenient to use a
disjunction (p1 ∨ p2 ∨ · · · ∨ pn ) instead of p as the hypothesis of the conditional statement, where p
and (p1 ∨ p2 ∨ · · · ∨ pn ) are equivalent.

2.6.1 : EXHAUSTIVE PROOF

Some theorems can be proved by examining a relatively small number of examples. Such proofs are
called exhaustive proofs, or proofs by exhaustion because these proofs proceed by exhausting all
possibilities. An exhaustive proof is a special type of proof by cases where each case involves checking
a single example. We now provide some illustrations of exhaustive proofs.

Example 1 Prove that (n + 1)3 ≥ 3n if n is a positive integer with n ≤ 4.

Proof:

• We only need verify the inequality (n + 1)3 ≥ 3n when n = 1, 2, 3, and 4 .

• For n = 1, we have (n + 1)3 = 23 = 8 and 3n = 31 = 3 for n = 2, we have (n + 1)3 = 33 = 27

and 3n = 32 = 9; for n = 3, we have (n + 1)3 = 43 = 64 and 3n = 33 = 27 and for n = 4, we
have (n + 1)3 = 53 = 125 and 3n = 34 = 81.

MAT 1023 4 Dr. P. G. R. S. Ranasinghe

 • In each of these four cases, we see that (n + 1)3 ≥ 3n . We have used the method of exhaustion
to prove that (n + 1)3 ≥ 3n if n is a positive integer with n ≤ 4 .


Example 2 Prove that the only consecutive positive integers not exceeding 100 that are perfect
powers are 8 and 9. (An integer is a perfect power if it equals na , where a ∈ Z+ and a > 1.)

Proof:

• We use a proof by exhaustion. In particular, we can prove the assertion by examining positive
integers n not exceeding 100, first checking whether n is a perfect power, and if it is, checking
whether n + 1 is also a perfect power.

• A quicker way to do this is simply to look at all perfect powers not exceeding 100 and checking
whether the next largest integer is also a perfect power.

• The squares of positive integers not exceeding 100 are 1, 4, 9, 16, 25, 36, 49, 64, 81, and 100.

• The cubes of positive integers not exceeding 100 are 1, 8, 27, and 64 .

• The fourth powers of positive integers not exceeding 100 are 1, 16, and 81.

• The fifth powers of positive integers not exceeding 100 are 1 and 32.

• The sixth powers of positive integers not exceeding 100 are 1 and 64.

• There are no powers of positive integers higher than the sixth power not exceeding 100, other
than 1.

• Looking at this list of perfect powers not exceeding 100, we see that n = 8 is the only perfect
power n for which n + 1 is also a perfect power. That is, 23 = 8 and 32 = 9 are the only two
consecutive perfect powers not exceeding 100.


MAT 1023 5 Dr. P. G. R. S. Ranasinghe

2.6.2 : PROOF BY CASES
A proof by cases must cover all possible cases that arise in a theorem. We illustrate proof by cases
with a couple of examples. In each example, you should check that all possible cases are covered.

Example 3 Prove that if n is an integer, then n2 ≥ n

Proof.

• We can prove that n2 ≥ n for every integer by considering three cases, when n = 0, when n ≥ 1,
and when n ≤ −1

• We split the proof into three cases because it is straightforward to prove the result by considering
zero, positive integers, and negative integers separately.

• Case (i) : When n = 0, because 02 = 0, we see that 02 ≥ 0. It follows that n2 ≥ n is true in this
case.

• Case (ii) : When n ≥ 1, when we multiply both sides of the inequality n ≥ 1 by the positive
integer n, we obtain n · n ≥ n · 1. This implies that n2 ≥ n for n ≥ 1.

• Case (iii) : In this case n ≤ −1. However, n2 ≥ 0. It follows that n2 ≥ n.

• Because the inequality n2 ≥ n holds in all three cases, we can conclude that if n is an integer,
then n2 ≥ n


Example 4 Use a proof by cases to show that |xy| = |x||y|, where x and y are real numbers.

Proof:

• Because both |x| and |y| occur in our formula, we will need four cases: (i) x and y both non-
negative, (ii) x non-negative and y is negative, (iii) x negative and y non-negative, and (iv) x
negative and y negative.

MAT 1023 6 Dr. P. G. R. S. Ranasinghe

 • We denote by p1 , p2 , p3 , and p4 , the proposition stating the assumption for each of these four
cases, respectively.

• Case (i): We see that p1 −→ q because xy ≥ 0 when x ≥ 0 and y ≥ 0, so that |xy| = xy = |x||y|.

• Case (ii) : To see that p2 −→ q, note that if x ≥ 0 and y < 0, then xy ≤ 0, so that |xy| =
−xy = x(−y) = |x||y|

• Case (iii) : To see that p3 −→ q, we follow the same reasoning as the previous case with the
roles of x and y reversed.

• Case (iv) : To see that p4 −→ q, note that when x < 0 and y < 0, it follows that xy > 0. Hence,
|xy| = xy = (−x)(−y) = |x||y|

• Because |xy| = |x||y| holds in each of the four cases and these cases exhaust all possibilities, we
can conclude that |xy| = |x||y|, whenever x and y are real numbers.


Sometimes we can eliminate all but a few examples in a proof by cases, as Example 5 illustrates.

Example 5 Show that there are no solutions in integers x and y of x2 + 3y 2 = 8.

Proof:

• We can quickly reduce a proof to checking just a few simple cases because x2 > 8 when |x| ≥ 3
and 3y 2 > 8 when |y| ≥ 2

• This leaves the cases when x equals −2, −1, 0, 1, or 2 and y equals −1, 0, or 1. We can finish
using an exhaustive proof.

• To dispense with the remaining cases, we note that possible values for x2 are 0, 1, and 4, and
possible values for 3y 2 are 0 and 3, and the largest sum of possible values for x2 and 3y 2 is 7.

• Consequently, it is impossible for x2 + 3y 2 = 8 to hold when x and y are integers.



MAT 1023 7 Dr. P. G. R. S. Ranasinghe

2.7 : Without Loss Of Generality (WLOG)
In the proof in Example 4, we dismissed Case (iii), where x < 0 and y ≥ 0, because it is the
same as Case (ii), where x ≥ 0 and y < 0, with the roles of x and y reversed. To shorten the proof,
we could have proved Cases (ii) and (iii) together by assuming, without loss of generality, that x ≥ 0
and y < 0. Implicit in this statement is that we can complete the case with x < 0 and y ≥ 0 using
the same argument as we used for the case with x ≥ 0 and y < 0, but with the obvious changes.
In general, when the phrase “without loss of generality” is used in a proof, we assert that by proving
one case of a theorem, no additional argument is required to prove other specified cases. That is, other
cases follow by making straightforward changes to the argument, or by filling in some straightforward
initial step.

Example 6 Show that if x and y are integers and both xy and x + y are even, then both x and y
are even.
Proof:

• We will use proof by contraposition, the notion of without loss of generality, and proof by cases.

• First, suppose that x and y are not both even. That is, assume that x is odd or that y is odd
(or both).

• Without loss of generality, we assume that x is odd, so that x = 2m + 1 for some integer k.

• We need to show that xy is odd or x + y is odd.

• Consider two cases: (i) y even, and (ii) y odd.

• In (i), y = 2n for some integer n, so that

x + y = (2m + 1) + 2n = 2(m + n) + 1
is odd.

• In (ii), y = 2n + 1 for some integer n, so that

xy = (2m + 1)(2n + 1) = 2(2mn + m + n) + 1

is odd. 
MAT 1023 8 Dr. P. G. R. S. Ranasinghe
Note : Our use of without loss of generality within the proof is justified because the proof when y is
odd can be obtained by simply interchanging the roles of x and y in the proof we have given.

Remarks : A common error of reasoning is to draw incorrect conclusions from examples. No matter
how many separate examples are considered, a theorem is not proved by considering examples unless
every possible case is covered. The problem of proving a theorem is analogous to showing that a
computer program always produces the output desired. No matter how many input values are tested,
unless all input values are tested, we cannot conclude that the program always produces the correct
output.

2.8 : UNIQUENESS PROOFS

Some theorems assert the existence of a unique element with a particular property. In other words,
these theorems assert that there is exactly one element with this property. To prove a statement of
this type we need to show that an element with this property exists and that no other element has
this property. The two parts of a uniqueness proof are:

• Existence : We show that an element x with the desired property exists.

• Uniqueness : We show that if y 6= x, then y does not have the desired property.

Equivalently, we can show that if x and y both have the desired property, then x = y.

Remark : Showing that there is a unique element x such that P (x) is the same as proving the
statement ∃x(P (x) ∧ ∀y(y 6= x −→ ¬P (y))).

Example 7 Show that if a and b are real numbers and a 6= 0, then there is a unique real number
r such that ar + b = 0.

Proof:

• First, note that the real number r = −b/a is a solution of ar + b = 0 because

 
b
a − + b = −b + b = 0.
a

MAT 1023 9 Dr. P. G. R. S. Ranasinghe

 • Consequently, a real number r exists for which ar + b = 0. This is the existence part of the
proof.

• Second, suppose that s ∈ R such that as + b = 0. Then ar + b = as + b, where r = −b/a.

• Subtracting b from both sides, we find that ar = a.

• Dividing both sides of this last equation by a(6= 0), we see that r = s.

• This means that if s 6= r, then as + b 6= 0. This establishes the uniqueness part of the proof.


2.9 : FORWARD AND BACKWARD REASONING

To begin a direct proof of a conditional statement, you start with the premises. Using these
premises, together with axioms and known theorems, you can construct a proof using a sequence
of steps that leads to the conclusion. This type of reasoning called forward reasoning, is the most
common type of reasoning used to prove relatively simple results. Similarly, with indirect reasoning,
you can start with the negation of the conclusion and, using a sequence of steps, obtain the negation
of the premises.
Unfortunately, forward reasoning is often difficult to use to prove more complicated results, be-
cause the reasoning needed to reach the desired conclusion may be far from obvious. In such cases,
it may be helpful to use backward reasoning. To reason backward to prove a statement q, we find a
statement p that we can prove with the property that p −→ q.

Example 8 Given two positive real numbers x and y, their arithmetic mean is (x + y)/2 and their
√
geometric mean is xy. When we compare the arithmetic and geometric means of pairs of distinct
positive real numbers, we find that the arithmetic mean is always greater than the geometric mean.
Can we prove that this inequality is always true?

√
Solution : To prove that (x + y)/2 > xy when x and y are distinct positive real numbers, we can
work backward. We construct a sequence of equivalent inequalities. The equivalent inequalities are

MAT 1023 10 Dr. P. G. R. S. Ranasinghe

 (x + y) √
> xy
2
(x + y)2
> xy
4
(x + y) √
> xy
2
(x + y)2
> xy
4
(x + y)2 > 4xy

x2 + 2xy + y 2 > 4xy

x2 − 2xy + y 2 > 0

(x − y)2 > 0.

Because (x − y)2 > 0 when x 6= y, it follows that the final inequality is true. Because all these
√
inequalities are equivalent, it follows that (x + y)/2 > xy when x 6= y. Once we have carried out
this backward reasoning, we can easily reverse the steps to construct a proof using forward reasoning.

Proof:

• Suppose that x and y are distinct positive real numbers. Then (x − y)2 > 0.

• This implies that x2 − 2xy + y 2 > 0.

• Adding 4xy to both sides of the above inequality and arranging terms yields

(x + y)2
> xy
4

which upon taking the square root yields

(x + y)2 √
> xy.
2 

MAT 1023 11 Dr. P. G. R. S. Ranasinghe

2.10 : FINAL REMARKS
In this chapter, we introduced the basic methods used in proofs. We also described how to leverage
these methods to prove a variety of results. We will use these proof methods in all subsequent chapters.
Following are some tips of writing good proofs.

• Often you’ll need to do some rough work to figure out how your proof should go. Some of the
time, you will be able to get a good idea of how a proof should go by noticing that it can be
similar to the proof of something else.

• Understand every line that you write, and do not make bogus claims.

• Understanding the proofs in the lectures/ lecture notes as a way of understanding of the type
of reasoning that is involved in a proof.

• How you lay out a proof is at least as important as the content (it’s not ok if ‘it’s all in there
somewhere’).

• Try to make your proofs easier to follow by including brief phrases where appropriate. For
example, rather than writing (statement A) implies (statement B) it may be more enlightening
to write (statement B) follows from (statement A) because. . .

• Try to write out many of them, understanding every step. Ask others about unclear points. Try
to follow proofs in class or in books, and ask about unclear points.

“The importance of proofs goes well beyond a university degree. It is eventually about using reason
in everyday life. This could contribute to solving major and global problems.”

.............................................................

MAT 1023 12 Dr. P. G. R. S. Ranasinghe