CSC/MAT A67 – Discrete Mathematics

M. Molloy and A. Tafliovich1

1
based on How To Prove It, by D. Velleman, and Discrete Mathematics, by
K. H. Rosen
 Proof Strategies
• Mathematicians use theorems to state answers to
mathematical questions.
• A theorem often states that if certain assumptions, called
hypotheses, are true, then some conclusion must also be true.
• Often hypotheses or conclusion contain free variables: we
mean the variable can stand for any element of the universe of
discourse.
• An assignment of values to these variables produces an
instance of the theorem.
• To prove a theorem, we must prove it for every instance.
• If there is even one instance for which the hypothesis is true,
but the conclusion is false, then the statement is incorrect.
• This instance is a counterexample.
 Proof Strategies: Counterexample
Example. Suppose a real number x > 3. Then x 2 > 10.
• universe of discourse: R
• hypothesis: x > 3
• conclusion: x 2 > 10
• logical form: ∀x, (x > 3) → (x 2 > 10)
• example instance: x = 4, hypothesis is true, conclusion is true
• example instance: x = 3.1, hypothesis is true, conclusion is
false (3.12 = 9.61 < 10)
• correct or incorrect? true or false? Incorrect. False.
• how do we prove it is false? We just did, by finding a
counterexample.
 Proof Strategies: Counterexample
Example. Suppose a real number x > 3. Then x 2 > 10.
• how do we prove it is false? We just did, by finding a
counterexample. Formally:

(1) Take c = 3.1

(2) (c > 3) ∧ ¬(c 2 > 10) 1 math
(3) ¬((c > 3) → (c 2 > 10)) 2 conditional, deMorgan
2
(4) ∃x, ¬((x > 3) → (x > 10)) 3 exist. generalisation
(5) ¬(∀x, (x > 3) → (x 2 > 10)) 4 quantifier negation

4
 Proof Strategies: Counterexample
In general, to prove that ∀x, P(x) → Q(x) is false, we can
provide a counterexample P(c) ∧ ¬Q(c) for c in the universe of
discourse.
Take c = . . . in the universe of discourse.
[Proof of P(c) ∧ ¬Q(c) goes here.]
Therefore ¬(∀x, P(x) → Q(x)).
Why does this work?

(1) Take c = . . .
(2) P(c) ∧ ¬Q(c)
(3) ¬(P(c) → Q(c)) 2 conditional, deMorgan
(4) ∃x, ¬(P(x) → Q(x)) 3 exist. generalisation
(5) ¬(∀x, P(x) → Q(x)) 4 quantifier negation

5
 Proof Strategies: Counterexample
Example. Suppose a real number x > 3. Then x 2 > 10.
• how do we prove it is false? By finding a counterexample. A
shorter version of the same proof:

(1) Take c = 3.1

(2) (c > 3) ∧ ¬(c 2 > 10) 1 math
(3) ¬(∀x, (x > 3) → (x 2 > 10)) 2 counterexample

6
 Shorthand notations and conventions
• As our proofs get more complex, we will gradually introduce
shorthand notations.
• It is important to understand that these are not new concepts,
just shorter ways of writing the same thing.

First batch of shorthand notations:

• ∀x, y , z . . . stands for ∀x∀y ∀z . . .
• ∃x, y , z . . . stands for ∃x∃y ∃z . . .
• “Take arbitrary x, y . . . ” stands for two lines:
• “Take arbitrary x . . . ”
• “Take arbitrary y . . . ”
 Shorthand notations and conventions
• So far, we always chose a different variable (e.g., c) as an
instance, when using Existential or Universal Generalisation.
• In mathematical proofs, we often use the same variable (e.g.,
x instead of c), understanding that the choice of variable does
not matter, since x is bounded in ∀x, . . . and ∃x, . . .:

(1) Take x = 3.1

(2) (x > 3) ∧ ¬(x 2 > 10) 1 math
(3) ¬(∀x, (x > 3) → (x 2 > 10)) 2 counterexample

8
 Proof Strategies: Direct Proof
Example. Suppose x > 3 and y < 2 (real numbers). Then
x 2 − 2y > 5.
• universe of discourse: R
• hypotheses: x > 3 and y < 2
• conclusion: x 2 − 2y > 5
• logical form: ∀x∀y , ((x > 3) ∧ (y < 2)) → (x 2 − 2y > 5)
• example instance: x = 4 and y = 1, hypotheses are true,
conclusion is true
• example instance: x = 2 and y = 1, one hypothesis is false,
conclusion is false
• correct or incorrect? true or false? Correct. True.
• how do we prove it?
 Proof Strategies: Direct Proof
Example. Suppose x > 3 and y < 2 (real numbers). Then
x 2 − 2y > 5.
Ideas / scratch work:

Givens: Goal:
x >3 x 2 − 2y > 5
y <2

• Since x > 3, then x is positive, and so x ∗ x > 3 ∗ x, and so

x 2 > 3x.
• Since x > 3, then 3 ∗ x > 3 ∗ 3, and so 3x > 9.
• Thus, x 2 > 3x > 9.

• Since y < 2, then −2 ∗ y > −2 ∗ 2, and so −2y > −4.

• Then x 2 − 2y > 9 − 4, and so x 2 − 2y > 5.

 Proof Strategies: Direct Proof
To prove a goal of the form P → Q:
Suppose P.
[Proof of Q goes here.]
Therefore P → Q.

Why does this work? It is impossible for P to be true and Q to be

false. Thus, P → Q.

Rule of Inference Name

Suppose P
Q Implication
∴ P→Q

11
 Proof Strategies: Direct Proof
Example: If a, b, c ∈ R, a > b, and c > 0, then ac > bc.
Logical form: ((a > b) ∧ (c > 0)) → (ac > bc)

(1) Suppose (a > b) ∧ (c > 0)

(2) a>b 1 simplification
(3) c >0 1 simplification
(4) ac > bc 2,3
(5) ((a > b) ∧ (c > 0)) → (ac > bc) 1,4 implication

12
 Proof Strategies: Direct Proof
Example: If a, b, c ∈ R, a > b, and c > 0, then ac > bc.
Equivalent Logical form: (a > b) → ((c > 0) → (ac > bc))
(Prove it!)

(1) Suppose a > b

(2) Suppose c > 0
(3) ac > bc 1,2
(4) (c > 0) → (ac > bc) 2,3 implication
(5) (a > b) → ((c > 0) → (ac > bc)) 1,4 implication

13
 Proof Strategies: Direct Proof
To prove a goal of the form (P1 ∧ . . . ∧ Pn ) → Q:
Suppose P1 .
...
Suppose Pn .
[Proof of Q goes here.]
Therefore (P1 ∧ . . . ∧ Pn ) → Q.

Example: ((a > b) ∧ (c > 0)) → (ac > bc)

(1) Suppose a > b

(2) Suppose c > 0
(3) ac > bc 1,2
(4) ((a > b) ∧ (c > 0)) → (ac > bc) 1,2,3 implication
 Proof Strategies: Direct Proof
Example. Suppose x > 3 and y < 2 (real numbers). Then
x 2 − 2y > 5.
Logical form: ∀x, y , ((x > 3) ∧ (y < 2)) → (x 2 − 2y > 5)

(1) Take arbitrary x, y

(2) Suppose x > 3
(3) Suppose y < 2
(4) x2 = x · x > 3 · x > 3 · 3 = 9 2: mult. by pos.
(5) − 2y > −4 3: multiply by neg.
2
(6) x − 2y > 9 − 4 = 5 4,5: add
(7) ((x > 3) ∧ (y < 2)) → (x 2 − 2y > 5) 2,3,6 impl.
2
(8) ∀x∀y , (x > 3) ∧ (y < 2) → x − 2y > 5 1,7 univ.gen.
 Proof Strategies: Direct Proof
• Definition: an even integer is an integer that can be written
as 2k for some integer k.
• n is even: ∃k, n = 2k
• Definition: an odd integer is a integer that can be written as
2k + 1 for some integer k.
• n is odd: ∃k, n = 2k + 1
Exercise: Use inference rules to prove each of the following
universally quantified statements.
1. The square of an odd integer is odd.
• For any integer n, if n is odd, then n2 is odd.
• ∀n, (∃k, n = 2k + 1) → (∃k, n2 = 2k + 1)
2. The sum of two odd integers is even.
• For any integers n and m, if n and m are odd, then n + m is
even.
• ∀n∀m, (∃k, n = 2k + 1) ∧ (∃k, m = 2k + 1) → (∃k, n + m = 2k)

16
 Proof Strategies: Direct Proof
Solutions: ∀n, (∃k, n = 2k + 1) → (∃k, n2 = 2k + 1)

(1) Take arbitrary integer n

(2) Suppose ∃k, n = 2k + 1
(3) n = 2j + 1 for some integer j 2 exist. inst.
2 2
(4) n = 2(2j + 2j) + 1 3
(5) 2j 2 + 2j is an integer j is an integer
2
(6) ∃k, n = 2k + 1 4,5 exist. gen.
(7) (∃k, n = 2k + 1) → (∃k, n2 = 2k + 1) 2,6 implication
2
(8) ∀n, (∃k, n = 2k + 1) → (∃k, n = 2k + 1) 1,7 univ. gen.

17
 Proof Strategies: Direct Proof
Solutions:
∀n, m, (∃k, n = 2k + 1) ∧ (∃k, m = 2k + 1) → (∃k, n + m = 2k)

(1)Take arbitrary integers n, m

(2) Suppose ∃k, n = 2k + 1
(3) Suppose ∃k, m = 2k + 1
(4) n = 2i + 1 for some integer i 2 exist. inst.
(5) m = 2j + 1 for some integer j 3 exist. inst.
(6) n + m = 2(i + j + 1) 4,5
(7) i + j + 1 is an integer 4,5: i, j integers
(8) ∃k, n + m = 2k 6,7 exist. gen.
(9) (∃k, n = 2k + 1) ∧ (∃k, m = 2k + 1) → (∃k, n + m = 2k) 2,3,8 impl.
(10)∀n∀m, (∃k, n = 2k + 1) ∧ (∃k, m = 2k + 1) → (∃k, n + m = 2k) 1,9 u.g.

18
 Proof Strategies: Contrapositive
Contrapositive Law:

P → Q is equivalent to ¬Q → ¬P

To prove a goal of the form P → Q:

Suppose Q is false.
[Proof of ¬P goes here.]
Therefore P → Q.

Rule of Inference Name

Suppose ¬Q
¬P Contrapositive
∴ P→Q

19
 Proof Strategies: Contrapositive
Example. Suppose a, b, and c are real numbers and a > b. Prove
that if ac ≤ bc, then c ≤ 0.
Ideas / scratch work:

Givens: Goal:
a, b, c are real ac ≤ bc → c ≤ 0
a>b

Notice an easier way: prove the contrapositive:

Givens: Goal:
a, b, c are real c > 0 → ac > bc
a>b

20
 Proof Strategies: Contrapositive
Solutions: ∀a, b, c, (a > b) → ((ac ≤ bc) → (c ≤ 0))

(1) Take arbitrary a, b, c

(2) Suppose a > b
(3) Suppose c > 0
(4) ac > bc 2,3
(5) (c > 0) → (ac > bc) 3,4 implication
(6) (ac ≤ bc) → (c ≤ 0) 5 contrapositive
(7) (a > b) → ((ac ≤ bc) → (c ≤ 0)) 2,5 implication
(8) ∀a, b, c, (a > b) → ((ac ≤ bc) → (c ≤ 0)) 1,7 univ.gen.

21
 Proof Strategies: Contrapositive
√
Example: If n = ab, where a, b are positive integers, then a ≤ n
√
or b ≤ n.
√ √
Logical form: ∀a, b, n, (n = ab) → ((a ≤ n) ∨ (b ≤ n))
universe of discourse is positive integers
Contrapositive:
√ √
¬((a ≤ n) ∨ (b ≤ n)) → ¬(n = ab)
√ √
eqv (a > n) ∧ (b > n) → (n ̸= ab) de Morgan

22
 Proof Strategies: Contrapositive
√ √
Solutions: ∀a, b, n, (n = ab) → ((a ≤ n) ∨ (b ≤ n)) universe of
discourse is positive integers

(1) Take arbitrary a, b, n

√
(2) Suppose a > n
√
(3) Suppose b > n
√ √
(4) ab > ( n)( n) = n 2,3 all positive
(5) ab ≠ n 4
√ √
(6) (a > n) ∧ (b > n) → (n ̸= ab) 2,3,5 implication
√ √
(7) (n = ab) → ((a ≤ n) ∨ (b ≤ n)) 6 contrapositive
√ √
(8) ∀a, b, n, (n = ab) → ((a ≤ n) ∨ (b ≤ n)) 1,7 univ.gen

23
 Proof Strategies: Biconditional
Biconditional Law:

P ↔ Q is equivalent to (P → Q) ∧ (Q → P)

To prove a goal of the form P ↔ Q:

Prove P → Q.
Prove Q → P.
Therefore P ↔ Q.

Rule of Inference Name

Suppose P
Q
Suppose Q Biconditional
P
∴ P↔Q

24
 Proof Strategies: Biconditional
Example: Prove x is even iff x 2 is even, for x ∈ Z.

(1) Take arbitrary integer x

(2) Suppose x is even
(3) ∃k, x = 2k 2 def. of even
(4) x = 2i for some integer i 3 exist. inst.
2 2 2
(5) x = (2i) = 2(2i ) 4
(6) 2i 2 is an integer
(7) ∃k, x 2 = 2k 5,6 exist. gen.
2
(8) x is even 7 def. of even
2
(9) x is even → x is even 2,8 implication
 Proof Strategies: Biconditional
[continuing]

(10) Suppose x is odd

(11) ∃k, x = 2k + 1 10 def. of odd
(12) x = 2j + 1 for some integer j 11 exist.inst.
2 2 2
(13) x = (2j + 1) = 2(2j + 2j) + 1 12
2
(14) 2j + 2j is an integer
(15) ∃k, x 2 = 2k + 1 13,14 exist. gen.
2
(16) x is odd 15 def. of odd
(17) x is odd → x 2 is odd 10,16 implication
2
(18) x is even → x is even 17 contrapositive
(19) x is even ↔ x 2 is even 9,18 biconditional

Question: What assumption did we make?

26
Proof Strategies: Exhaustive Proof / Proof by Cases
To prove a goal of the form (P1 ∨ . . . ∨ Pn ) → Q:
Prove P1 → Q.
...
Prove Pn → Q.
Therefore (P1 ∨ . . . ∨ Pn ) → Q.

Why does this work?

((P1 ∨ . . . ∨ Pn ) → Q) ↔ ((P1 → Q) ∧ . . . (Pn → Q))

Rule of Inference Name

Suppose P
Q
Suppose R Cases
Q
∴ (P ∨ R) → Q

27
Proof Strategies: Exhaustive Proof / Proof by Cases
Sometimes P1 ∨ . . . ∨ Pn is a tautology: it describes all possible
situations in the universe of discourse.
Prove P1 → Q.
...
Prove Pn → Q.
Therefore Q.

Why does this work?

((a tautology) → Q) ↔ Q
Rule of Inference Name
Suppose P
Q
Suppose ¬P Cases
Q
∴ Q
28
Proof Strategies: Exhaustive Proof / Proof by Cases
Example: If n is an integer, then n2 ≥ n.

(1) Take arbitrary integer n

(2) Suppose n ≥ 1 Case #1
2
(3) n =n·n ≥n·1=n 2 mult. by positive
(4) Suppose n = 0 Case #2
2
(5) n =0=n≥n 4
(6) Suppose n ≤ −1 Case #1
2
(7) n ≥ 0 > −1 ≥ n square is ≥ 0
2
(8) n ≥n 2-7 cases
2
(9) ∀n, n ≥ n 1,8 univ.gen.

What is the tautology we used? (n ≥ 1) ∨ (n = 0) ∨ (n ≤ −1)

29
 Proof Strategies
Example: If p ∈ Z and 2p − 1 is prime, then so is p.
Definition: An integer p is prime if p > 1 and it is not a product of
two smaller positive integers.
Contrapositive: If p is not prime, then 2p − 1 is not prime.
An integer p is not prime when:
(p ≤ 1) ∨ (∃a, b, (0 < a < p) ∧ (0 < b < p) ∧ (p = ab))
Looks like we need a proof by cases.
First part of the proof:
(1) Take arbitrary p
(2) Suppose p is not prime
(3) (p ≤ 1) ∨ (∃a, b, (0 < a < p) ∧ (0 < b < p) ∧ (p = ab)) def.
(4) Suppose p ≤ 1 case #1
p
(5) 2 −1≤1 4
(6) 2p − 1 is not prime def.
30
 Proof Strategies
(7) Suppose ∃a, b, (0 < a < p) ∧ (0 < b < p) ∧ (p = ab) case #2
(8) (0 < a < p) ∧ (0 < b < p) ∧ (p = ab) some a, b 7 exist.gen
(9) 0<a<p 8 simplification
(10) 0<b<p 8 simplification
(11) p = ab 8 simplification
(12) Take x = 2b − 1, y = 1 + 2b + 22b + · · · + 2(a−1)b
(13) xy = 2p − 1 * next page
(14) x = 2b − 1 < 2p − 1 10
(15) b>1 8,10 a < p = ab
(16) x = 2b − 1 > 21 − 1 = 1 12,15
p
(17) 0 < y < xy = 2 − 1 16,12 x > 1
p p p
(18) (0 < x < 2 − 1) ∧ (0 < y < 2 − 1) ∧ (xy = 2 − 1)
(19) ∃x, y , (0 < x < 2p − 1) ∧ (0 < y < 2p − 1) ∧ (xy = 2p − 1)
(20) 2p − 1 is not prime
31
 Proof Strategies
[continuing]
(21) 2p − 1 is not prime 4,6,7,20 cases
p
(22) p is not prime → 2 − 1 is not prime 2,21 implication
p
(23) 2 − 1 is prime → p is prime 22 contrapositive
p
(24) ∀p, 2 − 1 is prime → p is prime 23 univ.gen.
(*) Details from previous page:
x · y =(2b − 1) · (1 + 2b + 22b + · · · + 2(a−1)b )
=2b · (1 + 2b + 22b + · · · + 2(a−1)b ) − (1 + 2b + 22b + · · · + 2(a−1)b )
=(2b + 22b + · · · + 2ab ) − (1 + 2b + 22b + · · · + 2(a−1)b )
=2ab − 1
=2p − 1
What strategies did we use? Contrapositive, Proof by Cases,
Implication.
32
 Proof Strategies
What would this proof look like in a typical math textbook?
Take arbitrary integer p, and suppose p is not prime. Then, by definition
of a prime number, either p ≤ 1 or it is a product of two smaller positive
integers. If p ≤ 1, then 2p − 1 ≤ 1, and so 2p − 1 is not prime. If p is a
product of two smaller positive integers a and b, we have 0 < a < p,
0 < b < p, and p = ab. Take x = 2b − 1 and
y = 1 + 2b + 22b + · · · + 2(a−1)b . Then (Insert proof that xy = 2p − 1
from previous slide). Since b < p, we have x = 2b − 1 < 2p − 1. Since
p = ab and a < p, we have a < ab, and so b > 1, and so
x = 2b − 1 > 1. Since x > 1 and xy = 2p − 1, we have y < 2p − 1. Also
y > 0, since y = 1 + 2b + · · · + 2(a−1)b . Thus, we have 2p − 1 is a
product of two smaller positive integers x and y , and so it is not prime.
Math textbooks take a long time to read! We need to carefully
examine the overall stucture, as well as each step, of each proof.
 Proof Strategies: Contradiction
Adam: Charlie entered the warehouse yesterday.
• Betty thinks Adam is lying.
Betty: There is only one way in: the front door. Then Charlie
entered through the front door.
Betty: It was sunny, so Charlie was easy to see.
Betty: There is a 24hr recording of the front door, so it recorded
Charlie entering.
• Betty plays the video that does not have Charlie entering.
• Conclusion: Adam was lying.

Is this “good reasoning”? Why or why not?

 Proof Strategies: Contradiction
To prove a goal of the form P:
Suppose ¬P.
(a contradiction)
Therefore P.

Why does this work?

(P → (a contradiction)) ↔ ¬P

Rule of Inference Name

Suppose ¬P
(a contradiction) Contradiction
∴ P

35
 Proof Strategies: Contradiction
Example: If a is a natural number, and a2 is even, then a is even.
• universe of discourse: natural numbers
• hypothesis: a2 is even
• conclusion: a is even
• definitions: n is even: ∃k, n = 2k. n is odd: ∃k, n = 2k + 1.
• logical form: ∀a, (a2 is even) → (a is even)
• fact: (a is even) ↔ ¬(a is odd)
Proof plan:

Suppose a2 is even
Suppose a is odd for contr.
...
a contradiction

36
 Proof Strategies: Contradiction
Solutions: ∀a, (a2 is even) → (a is even)
(1) Take arbitrary a
(2) Suppose a2 is even
(3) Suppose a is odd for contr.
(4) ∃k, a = 2k + 1 3 def.
(5) a = 2i + 1 4 exist.inst.
(6) a2 = (2i + 1)2 = 2(2i 2 + 2i) + 1 5
2
(7) 2i + 2i is an integer
(8) ∃k, a2 = 2k + 1 6,7 exist.gen.
2
(9) a is odd 8 def.
(10) a contradiction 2,9
(11) a is even 3,10 contradiction
2
(12) (a is even) → (a is even) 2,11 implication
2
(13) ∀a, (a is even) → (a is even) 1,12 univ.gen.
37
 Proof Strategies
Anther useful short form:
Rule of Inference Name
∀x, P(x) → Q(x)
P(c) Universal Modus Ponens
∴ Q(c)

Why does this work?

(1) ∀x, P(x) → Q(x)

(2) P(c)
(3) P(c) → Q(c) universal instantiation
(4) Q(c) Modus Ponens

38
 Proof Strategies: Contradiction
√
Example: 2 is irrational.
• Definition: x is rational iff there are integers a, b such that
x = ba .
• We say ba is “in lowest terms” if a and b have no factors in
common (other than 1).

Proof plan:
√
Suppose 2 is rational for contr.
√ a
∃a, b ∈ N, 2 = in lowest terms
b
...
a contradiction

39
 Proof Strategies: Contradiction
Result from our previous proof: ∀a, (a2 is even) → (a is even) (*)
Universe of discourse: natural numbers.
√
(1) Suppose 2 is rational for contr.
√ a
(2) ∃a, b, ( 2 = ) ∧ (a, b have no common factors ̸= 1) 1 def.
b
√ x
(3) ( 2 = ) ∧ (x, y have no common factors ̸= 1) 2 exist.inst.
y
2 2
(4) x = 2y 3 simpl., mult. by pos.
2
(5) y is an integer
2
(6) ∃k, x = 2k 4,5 exist.gen.
2
(7) x is even 6 def.
(8) x is even (*),7 univ. modus ponens
(9) x = 2i 8 def., exist.inst.
2 2 2 2
(10) y = x /2 = (2i) /2 = 2i 4,9
2 2
(11) y is even 10, i ∈ Z, exist.gen., def.
(12) y is even (*),11 univ. modus ponens
(13) y = 2j 12 def., exist.inst.
(14) 2 is a common factor of x, y 9,13
(15) x, y have no common factors ̸= 1 3 simpl.
(16) a contradiction 14,15
√
(17) 2 is irrational 1,16 contradiction

40