Solutions to selected exercises

 1a, page 4
Consider a vacuously true condition proposition, say p → q. By the definition of vacuously
true, the hypothesis p is false. In particular, it is not the case that p is true and q is false.
So the conditional proposition p → q is true by the definition of →.

 1b, page 4
In the notation of Definition 1.3.8, we have p = (a
∧ b) and q
= (c ∨ d) here. So the
converse is
(c ∨ d) → (a ∧
b), the inverse is ¬(a ∧ b) → ¬(c ∨ d) , and the contrapositive
is ¬(c ∨ d) → ¬(a ∧ b) .

 1c, page 5
If a proposition is true, then it cannot be false because by definition no proposition can be
both true and false.
If a proposition is not false, then it must be true because by definition no proposition can
be neither true nor false.

 1d, page 6
According to Convention 1.4.3, the compound expression ¬p → ¬q stands for (¬p) → (¬q).
• ¬p evaluates to F according to the definition of ¬.
• ¬q evaluates to T according to definition of ¬.
• ¬p → ¬q evaluates to T according to the definition of →.

In summary,
¬p → ¬q
T F
F T
T

 1e, page 6
(1) P and P evaluate to the same truth value under any substitution of propositions into
the propositional variables.
(2) If P and Q evaluate to the same truth value under any substitution of propositions into
the propositional variables, then so do Q and P .
(3) If P and Q evaluate to the same truth value under any substitution of propositions into
the propositional variables, and Q and R evaluate to the same truth value under any
substitution of propositions into the propositional variables, then so do P and R.

sol.1
 1f, page 8
No, as the following truth table shows.
p q r p∧q (p ∧ q) ∨ r q∨r p ∧ (q ∨ r)
T T T T T T T
T T F T T T T
T F T F T T T
T F F F F F F
F T T F T T F
F T F F F T F
F F T F T T F
F F F F F F F
Actually, to justify a “no” answer, it suffices to give one row of this truth table in which the
columns for (p ∧ q) ∨ r and p ∧ (q ∨ r) are different.

 1g, page 8
One can prove the Associativity, the Commutativity, the Identity, and the Annihilator parts
of Theorem 1.4.20 using the truth tables below.
p q r p∨q (p ∨ q) ∨ r q∨r p ∨ (q ∨ r)
T T T T T T T
T T F T T T T
T F T T T T T
T F F T T F T
F T T T T T T
F T F T T T T
F F T F T T T
F F F F F F F
p q r p∧q (p ∧ q) ∧ r q∧r p ∧ (q ∧ r)
T T T T T T T
T T F T F F F
T F T F F F F
T F F F F F F
F T T F F T F
F T F F F F F
F F T F F F F
F F F F F F F
p q p∨q q∨p p∧q q∧p p t c p∨c p∧t p∨t p∧c
T T T T T T T T F T T T F
T F T T F F F T F F F T F
F T T T F F
F F F F F F
Alternatively, one can prove the associativity of ∧ from that of ∨ as follows:
(p ∧ q) ∧ r ≡ (¬¬p ∧ ¬¬q) ∧ ¬¬r by the Double Negative Law;
≡ ¬(¬p ∨ ¬q) ∧ ¬¬r by De Morgan’s Laws;


≡ ¬ (¬p ∨ ¬q) ∨ ¬r by De Morgan’s Laws;


≡ ¬ ¬p ∨ (¬q ∨ ¬r) by the associativity of ∨;
≡ ¬¬p ∧ ¬(¬q ∨ ¬r) by De Morgan’s Laws;
≡ ¬¬p ∧ (¬¬q ∧ ¬¬r) by De Morgan’s Laws;
≡ p ∧ (q ∧ r) by the Double Negative Law.

sol.2
One can also prove the commutativity of ∧ from that of ∨ as follows:

p ∧ q ≡ ¬¬p ∧ ¬¬q by the Double Negative Law;

≡ ¬(¬p ∨ ¬q) by De Morgan’s Laws;
≡ ¬(¬q ∨ ¬p) by the commutativity of ∨;
≡ ¬¬q ∧ ¬¬p by De Morgan’s Laws;
≡q∧p by the Double Negative Law.

 2a, page 15
(1) False: note that ∀x ∈ Z ¬(x > 0 ∧ x < 1) is true, but ∀x ∈ R ¬(x > 0 ∧ x < 1) is false.
(2) True, because every integer is a real number.
(3) True, because every integer is a real number.
(4) False: note that ∃x ∈ R (x > 0 ∧ x < 1) is true, but ∃x ∈ R (x > 0 ∧ x < 1) is false.

There are many other counterexamples for (1) and (4).

 2b, page 16
(3) Note that the following are true.


¬∀x ∈ D P (x)
↔ ¬∀x x ∈ D → P (x)
by Definition 2.2.1(5).
¬∀x x ∈ D → P (x)
↔ ∃x ¬ x ∈ D → P (x)
by part (1).
∃x ¬ x ∈ D → P (x)
↔ ∃x x ∈ D ∧ ¬P (x) by Example 1.4.23.
∃x x ∈ D ∧ ¬P (x) ↔ ∃x ∈ D ¬P (x) by Definition 2.2.4(5).

From these, we deduce that ¬∀x ∈ D P (x) is true if and only if ∃x ∈ D ¬P (x) is true.

(4) Note that the following are true.

¬∃x ∈ D P (x)
↔ ¬∃x x ∈ D ∧ P (x)
by Definition 2.2.4(5).
¬∃x x ∈ D ∧ P (x)
↔ ∀x ¬ x ∈ D ∧ P (x)
by part (2).
∀x ¬ x ∈ D ∧ P (x)
↔ ∀x ¬(x ∈ D) ∨ ¬P (x)

by De Morgan’s Laws.
∀x ¬(x ∈ D) ∨ ¬P (x)
↔ ∀x x ∈ D → ¬P (x) by the Implication part of Theorem 1.4.20.
∀x x ∈ D → ¬P (x) ↔ ∀x ∈ D ¬P (x) by Definition 2.2.1(5).

From these, we deduce that ¬∃x ∈ D P (x) is true if and only if ∀x ∈ D ¬P (x) is true.

 2c, page 19
(3) We have the following equivalences by Theorem 2.3.1.

¬∃x ∃y Q(x, y) ↔ ∀x ¬∃y Q(x, y).

∀x ¬∃y Q(x, y) ↔ ∀x ∀y ¬Q(x, y).

From these, we deduce that ¬∃x ∃y Q(x, y) is true if and only if ∀x ∀y ¬Q(x, y) is true.

(4) We have the following equivalences by Theorem 2.3.1.

¬∃x ∀y Q(x, y) ↔ ∀x ¬∀y Q(x, y).

∀x ¬∀y Q(x, y) ↔ ∀x ∃y ¬Q(x, y).

From these, we deduce that ¬∃x ∀y Q(x, y) is true if and only if ∀x ∃y ¬Q(x, y) is
true.

sol.3
 2d, page 20
(1) • This reads “there exists x in D such that, for every y in E, xy = 0”.
• Alternatively, one can express this as “there is an element x of D which, when
multiplied to any element y of E, gives a product of 0”.
• This is true because if we choose the element 0 of D, then no matter which element
y of E we multiply it to, we get a product of 0.
(2) • This reads “for every y in E, there is x in D such that xy = 0”.
• Alternatively, one can express this as “no matter which element y of E is given,
one can always multiply it to an element x of D to get 0”.
• This is true because no matter which element y of E is given, we can always
multiply it to the element 0 of D to get 0.
(3) • This reads “there exists x in D such that, for every y in E, xy < 0”.
• Alternatively, one can express this as “there is an element x of D which, when
multiplied to any element y of E, gives a negative product”.
• This is false because no element x of D, when multiplied to any element y of E,
always gives a negative product: let us consider all the elements of D one by one.
– For the element −1 of D, when we multiply it to the element −1 of E, we get
a non-negative product.
– For the elements 0 and 1 of D, when we multiply them to the element 1 of E,
we get non-negative products.
There are other counterexamples.
(4) • This reads “for every y in E, there exists x in D such that xy < 0”.
• Alternatively, one can express this as “one can make any element y of E negative
by multiplying it to some element x of D.”
• This is true, as one can verify exhaustively for each element of E.
– For the elements 1 and 2 of E, one can multiply them to the element −1 of D
to make a negative product.
– For the elements −1 and −2 of E, one can multiply them to the element 1
of D to make a negative product.
(5) • This reads “there exist x1 , x2 in D such that x1 + x2 = 2”.
• Alternatively, one can express this as “there are two (possibly equal) elements x1 ,
x2 of D whose sum is 2”.
• This is true because 1 is an element of D and 1 + 1 = 2.
(6) • In view of Theorem 2.4.9(1), its negation ¬∀y1 , y2 ∈ E y1 = y2 is equivalent to
∃y1 , y2 ∈ E y1 ̸= y2 .
• This negation reads “there exist y1 , y2 in E such that y1 ̸= y2 ”.
• Alternatively, one can express this negation as “there are two different elements
y1 , y2 of E”.
• This negation is true: for example, the numbers 1 and 2 are different elements
of E.
• So the given proposition is false, because its negation is true.

 3a, page 24
No. In the case when p is a false proposition and q is a true proposition, we have both p → q
and ¬p true, but ¬q is false.

sol.4
 3b, page 27
(1) Suppose n is even. Use the definition of even integers to find an integer x such that
n = 2x. Then −n = −2x = 2(−x) where −x is an integer. So −n is even.
(2) Suppose n is odd. Use the definition of odd integers to find an integer x such that
n = 2x + 1. Then −n = −(2x + 1) = 2(−x − 1) + 1 where −x − 1 is an integer. So n is
odd.

 3c, page 28
(1) We deduced n is odd from the assumption that n is not even. This uses Proposi-
tion 3.2.17.
(2) We deduced n2 is not even from our knowledge that n2 is odd. This uses Proposi-
tion 3.2.21.

 3d, page 28
Proof. We show that ∀x ∈ Z 2x ̸= 1. Take any z ∈ Z. By trichotomy, we have z > 0 or
z = 0 or z < 0.
Case 1: suppose z > 0. Then z ∈ N. So

z⩾1 by the discreteness of N;

∴ 2z ⩾ 2 × 1 = 2 > 1
∴ 2z ̸= 1 by the irreflexivity of <.

Case 2: suppose z ⩽ 0. Then 2z ⩽ 2 × 0 = 0 < 1. So 2z ̸= 1 by the irreflexivity of <.

 3e, page 30
Let us read the two steps in Strong MI as follows.
(base step) show that P (i) is true for each integer i satisfying b ⩽ i ⩽ c;
(induction step) for each integer k ⩾ c, use the hypothesis that P (i) is true for each
integer i satisfying b ⩽ i ⩽ k to show that P (k + 1) is also true.
Consider the case when c = b − 1.
• For the base step, there is nothing to prove, because no integer i satisfies b ⩽ i ⩽ c =
b − 1.

• Consider the induction step when k = c. The hypothesis we can use is P (i) is true for
each integer i satisfying b ⩽ i ⩽ c = b − 1. However, there is no such i. So there is no
hypothesis one can use to show P (b). In other words, one simply needs to show that
P (b) is true here (without assuming anything).
• In the induction step when k = c + 1, one needs to show P (b) → P (b + 1).

• In the induction step when k = c + 2, one needs to show P (b) ∧ P (b + 1) → P (b + 2).

• Etc.
We deduce that P (b), P (b + 1), P (b + 2), . . . are all true by a series of modus ponens.

sol.5
 3f, page 31
Proof. Note that 2a1 −a2 b1 = b2 . As b2 is odd, it cannot be even by Proposition 3.2.17. So
a1 ⩽ a2 . Similarly, we know b1 = 2a2 −a1 b2 . As b2 is odd, it cannot be even by Proposi-
tion 3.2.17. So a2 ⩽ a1 . Combining the two inequalities, we have a1 = a2 . As 2a1 = 2a2 ̸= 0,
we deduce from 2a1 b1 = 2a2 b2 that b1 = b2 .

 4a, page 34
(1) We want to prove that E = Z+ , where E = {x + 1 : x ∈ Z⩾0 }.

Proof. (⇒) Let z ∈ E. Use the definition of E to find x ∈ Z⩾0 such that x + 1 = z.
Then x ∈ Z and x ⩾ 0 by the definition of Z⩾0 . As x ∈ Z, we know x + 1 ∈ Z because
Z is closed under +. As x ⩾ 0, we know x + 1 ⩾ 0 + 1 = 1 > 0. So z = x + 1 ∈ Z+ by
the definition of Z+ .
(⇐) Let z ∈ Z+ . Then z ∈ Z and z > 0. Define x = z − 1. As z ∈ Z, we know x ∈ Z
because Z is closed under −. As z > 0, we know x = z − 1 > 0 − 1 = −1, and thus
x ⩾ 0 as x ∈ Z. So x ∈ Z⩾0 by the definition of Z⩾0 . Hence the definition of E tells us
z = x + 1 ∈ E.

(2) We want to prove that F = Z, where F = {x − y : x, y ∈ Z⩾0 }.

Proof. (⇒) Let z ∈ F . Use the definition of F to find x, y ∈ Z⩾0 such that x − y = z.
Then x, y ∈ Z by the definition of Z⩾0 . So z = x − y ∈ Z as Z is closed under −.
(⇐) Let z ∈ Z.
• Case 1: suppose z ⩾ 0. Let x = z and y = 0. Then x, y ∈ Z⩾0 . So z = z − 0 =
x − y ∈ F by the definition of F .
• Case 2: suppose z < 0. Let x = 0 and y = −z. Then x, y ∈ Z⩾0 as z < 0. So
z = 0 − (−z) = x − y ∈ F by the definition of F .
So z ∈ F in all the cases.

 4b, page 35
• {1} ∈ C but {1} ̸⊆ C;

• {2} ̸∈ C but {2} ⊆ C;

• {3} ∈ C and {3} ⊆ C; and
• {4} ̸∈ C and {4} ̸⊆ C.

 4c, page 38
Let z ∈ A. In particular, we know z ∈ A or z ∈ B. So z ∈ A ∪ B by the definition of ∪.

 4d, page 39
Assume A ⊆ B and A ⊆ C. Take any z ∈ A. Then z ∈ B and z ∈ C as A ⊆ B and A ⊆ C
by assumption. So z ∈ B ∩ C by the definition of ∩. As the choice of z in A was arbitrary,
this shows A ⊆ B ∩ C.

sol.6
 4e, page 39

B B

A A
C C

(A \ B) ∪ (B \ C) A\C

No. For a counterexample, let A = C = ∅ and B = {1}. Then

(A \ B) ∪ (B \ C) = ∅ ∪ {1} = {1} =
̸ ∅ = A \ C.

 4f, page 39
Ideas. (1) The set of all sets?
( )
n o
(2) {. . . . . . } ?

 4g, page 39
Maybe, but is it better?

Proof. Take any set R. Split into two cases.

• Case 1: assume R ∈ R. Then ¬(R ̸∈ R). So ¬(R ∈ R ⇒ R ̸∈ R). Hence

∃x ¬(x ∈ R ⇔ x ̸∈ x).

• Case 2: assume R ̸∈ R. Then ¬(R ∈ R). So ¬(R ̸∈ R ⇒ R ∈ R). Hence

∃x ¬(x ∈ R ⇔ x ̸∈ x).

In either case, we showed ¬∀x (x ∈ R ⇔ x ̸∈ x). So the proof is finished.

 5a, page 44
Let P (n) be the predicate

For all ordered n-tuples (x1 , x2 , . . . , xn ) and (y1 , y2 , . . . , yn ),

(x1 , x2 , . . . , xn ) = (y1 , y2 , . . . , yn ) ⇔ x1 = y1 and x2 = y2 and . . . and xn = yn .

over Z⩾2 .

(Base step) P (2) is true by the definition of ordered pairs.

(Induction step) Let k ∈ Z⩾2 such that P (k) is true, i.e., all ordered k-tuples (x1 , x2 , . . . , xk )
and (y1 , y2 , . . . , yk ),

(x1 , x2 , . . . , xk ) = (y1 , y2 , . . . , yk ) ⇔ x1 = y1 and x2 = y2 and . . . and xk = yk .

sol.7
 For all (k + 1)-tuples (x1 , x2 , . . . , xk+1 ) and (y1 , y2 , . . . , yk+1 ),

(x1 , x2 , . . . , xk , xk+1 ) = (y1 , y2 , . . . , yk , yk+1 )

⇔ (x1 , x2 , . . . , xk ), xk+1 = (y1 , y2 , . . . , yk ), yk+1 by Definition 5.1.10(2);
⇔ (x1 , x2 , . . . , xk ) = (y1 , y2 , . . . , yk ) and xk+1 = yk+1 by Definition 5.1.1;
⇔ x1 = y1 and x2 = y2 and . . . and xk = yk and xk+1 = yk+1
by the induction hypothesis.

This shows P (k + 1) is true.

Hence ∀n ∈ Z⩾2 P (n) is true by MI.

 5b, page 46
2 2 2 2 2 2

1 1 1 1 1 1

0 0 0 0 0 0
S R R S
2 2 2 2

1 1 1 1

0 0 0 0
R◦S S◦R

No. For instance, we have (2, 2) ∈ R ◦ S because (2, 0) ∈ S and (0, 2) ∈ R, but (2, 2) ̸∈
S ◦ R because no y ∈ A makes (2, y) ∈ R and (y, 2) ∈ S. (There are exactly three other
counterexamples.)

 6a, page 52
• R is reflexive as x ⩽ x for all x ∈ Q;
• R is not symmetric as 0 ⩽ 1 but 1 ̸⩽ 0, for example;
• R is transitive as x ⩽ y and y ⩽ z imply x ⩽ z for all x, y, z ∈ Q.

 6b, page 52
• R′ is not reflexive as 0 ̸< 0, for example;
• R′ is not symmetric as 0 < 1 but 1 ̸< 0, for example;
• R′ is transitive as x < y and y < z imply x < z for all x, y, z ∈ Q.

 6c, page 52
The arrow diagram below represents the relation R in the exercise:

sol.8
This relation is not reflexive because 2 ̸R 2. It is not symmetric because 3 R 2 but 2 ̸R 3. It
is not transitive because 2 R 1 and 1 R 2 but 2 ̸R 2.

 6d, page 52
Proof. (⇒) Assume R is transitive. Let (x, z) ∈ R ◦ R. Use the definition of R ◦ R to find
y ∈ A such that (x, y) ∈ R and (y, z) ∈ R. This means x R y and y R z. So x R z by the
transitivity of R. Hence (x, z) ∈ R.
(⇐) Assume R◦R ⊆ R. Let x, y, z ∈ A such that x R y and y R z. This means (x, y) ∈ R
and (y, z) ∈ R. So (x, z) ∈ R ◦ R by the definition of R ◦ R. Our assumption then implies
(x, z) ∈ R. Hence x R z.

 6e, page 54
Yes, as shown below.
Proof. We know x ∈ [x] by Lemma 6.3.5(1). So x ∈ S ∩ [x] by the hypothesis. This implies
S ∩ [x] ̸= ∅. Hence S = [x] by Lemma 6.3.6.

 6f, page 55
The divisibility relation on Z is not antisymmetric because 1 | −1 and −1 | 1, but 1 ̸= −1.

 6g, page 56
The divisibility relation on Z+ is antisymmetric, as shown below. So it is a partial order by
Example 6.1.8. It is not total because 2 ∤ 3 and 3 ∤ 2.
Proof of antisymmetry. Let us first show that if a, b ∈ Z+ such that a | b, then a ⩽ b.
Let a, b ∈ Z+ such that a | b. Then the definition of divisibility gives k ∈ Z such that b = ak.
Note k = b/a > 0 as both a and b are positive. Since k ∈ Z, this implies k ⩾ 1. Thus
b = ak ⩾ a × 1 = a, as required.
Now let a, b ∈ Z+ such that a | b and b | a. Then the previous paragraph tells us a ⩽ b
and b ⩽ a. So a = b.

 6h, page 56
We use the version of Strong MI with c = b − 1 given by Exercise 3.2.25.
Proof. Let P (n) be the predicate “n ̸∈ S” over Z⩾b .
(Induction step) Let k ∈ Z⩾b−1 such that P (b), P (b + 1), . . . , P (k) are true, i.e., that
b, b + 1, . . . , k ̸∈ S. If k + 1 ∈ S, then k + 1 is the smallest element of S because
S ⊆ Z⩾b , which contradicts our assumption. So k + 1 ̸∈ S. This means P (k + 1) is
true.
Hence ∀n ∈ Z⩾b P (n) is true by Strong MI.

 7a, page 61
Only (b), (c), (d) and (f) represent functions.

 7b, page 63
Proposition 7.3.1 implies
• idB ◦ f is a function A → B; and
• (idB ◦ f )(x) = idB (f (x)) = f (x) for all x ∈ A.
So idB ◦ f = f by Proposition 7.2.7.

sol.9
 7c, page 63
(1) Yes, because (f ◦ f )(x) = f (f (x)) = f (1231) = 1231 = f (x) for all x ∈ Z in this case.
(2) Yes, because (f ◦ f )(x) = f (f (x)) = f (x) for all x ∈ Z in this case.
(3) No, because (f ◦ f )(1) = f (f (1)) = f (−1) = 1 ̸= −1 = f (1) in this case.
(4) No, because (f ◦ f )(0) = f (f (0)) = f (1) = 4 ̸= 1 = f (0) in this case.

(5) No, because (f ◦ f )(2) = f (f (2)) = f (4) = 16 ̸= 4 = f (2) in this case.

 7d, page 64
(b) is a surjection but not an injection; (c) is neither a surjection nor an injection; (d) is an
injection but not a surjection; and (f) is both an injection and a surjection. Thus only the
last one is a bijection.

 7e, page 65
Proof. Let y ∈ B. Define x = f −1 (y). Then

(f ◦ f −1 )(y) = f (f −1 (y)) by Proposition 7.3.1;

= f (x) by the definition of x;
=y by Proposition 7.4.11, as f −1 (y) = x;
= idB (y) by the definition of idB .

So f ◦ f −1 = idB by Proposition 7.2.7.

 8a, page 69
Proof of the converse to the Pigeonhole Principle. Suppose n ⩽ m. Define f : A →
B by setting f (xi ) = yi for each i ∈ {1, 2, . . . , n}. This definition is unambiguous because the
x’s are different. To show injectivity, suppose i, j ∈ {1, 2, . . . , n} such that f (xi ) = f (xj ).
The definition of f tells us f (xi ) = yi and f (xj ) = yj . Then yi = f (xi ) = f (xj ) = yj . So
i = j because the y’s are different. This implies xi = xj .

Proof of a partial converse to the Dual Pigeonhole Principle. Suppose n ⩾ m > 0.

Define f : A → B by setting, for each i ∈ {1, 2, . . . , n},
(
yi , if i ⩽ m;
f (xi ) =
ym , otherwise.

This definition is unambiguous because the x’s are different. It is surjective because given
any yi ∈ B, we have xi ∈ A such that f (xi ) = yi .

Proof. (1) Suppose f and g are surjective. Let z ∈ C. Use the surjectivity of g to find
y ∈ B such that z = g(y). Then use the surjectivity of f to find x ∈ A such that
y = f (x). Now z = g(y) = g(f (x)) = (g ◦ f )(x) by Proposition 7.3.1, as required.
(2) Suppose f and g are injective. Let x1 , x2 ∈ A such that (g ◦ f )(x1 ) = (g ◦ f )(x2 ).
Then g(f (x1 )) = g(f (x2 )) by Proposition 7.3.1. The injectivity of g then implies
f (x1 ) = f (x2 ). So the injectivity of f tells us x1 = x2 , as required.

sol.10
 8b, page 70
Proof. To show that A has the same cardinality as range(f ), it suffices to find a bijection
A → range(f ). We use the function f0 : A → range(f ) defined by setting f0 (x) = f (x) for
each x ∈ A.

(Well-defined) Let us first check that the definition of f0 given indeed assigns every element
of A exactly one element of range(f ), as required by the definition of functions. Take
any x0 ∈ A. As x0 ∈ A and f is a function with domain A, there is exactly one object
that is equal to f (x0 ). So we only need to check that this object is indeed an element
of the codomain of f0 , i.e., range(f ). Recall range(f ) = {f (x) : x ∈ A} by definition.
As x0 ∈ A, we see that f (x0 ) ∈ range(f ), as required.
(Surjective) Take any y ∈ range(f ). Use the definition of range(f ) to find x ∈ A such that
y = f (x). Then y = f (x) = f0 (x) by the definition of f0 .
(Injective) Let x1 , x2 ∈ A such that f0 (x1 ) = f0 (x2 ). Then the definition of f0 tells us
f (x1 ) = f (x2 ). So x1 = x2 as f is injective.

Additional comment. Sometimes when one defines a function, it is not immediately clear
that the definition given indeed defines a function, i.e., that it defines an object satisfying
the definition of functions. In such cases, a proof should be provided. In many other cases,
it is clear that the definition given really defines a function, and so no additional explanation
is needed.

 8c, page 70
Proof. Let n ∈ N and f : N → {1, 2, . . . , n}. Define fn : {1, 2, . . . , n + 1} → {1, 2, . . . , n} by
setting fn (x) = f (x) for all x ∈ {1, 2, . . . , n + 1}. Then fn is not injective by the Pigeonhole
Principle. Let x1 , x2 ∈ {1, 2, . . . , n + 1} such that x1 ̸= x2 but fn (x1 ) = fn (x2 ). Then
f (x1 ) = f (x2 ) by the definition of fn . This show f is not injective.
The paragraph above shows, in particular, that there is no bijection N → {1, 2, . . . , n},
for any n ∈ N. So N is infinite.

 8d, page 70
(1) True. This follows directly from Definition 8.2.1.

(2) False. The f : N → N satisfying f (0) = 0 and f (x+1) = x for each x ∈ N is a surjection
that is not an injection, but N has the same cardinality as N.
(3) False. The f : N → N satisfying f (x) = x + 1 for each x ∈ N is an injection that is not
a surjection, but N has the same cardinality as N.

(4) False. The function f : {0, 1} → {0, 1} satisfying f (0) = 0 = f (1) is neither a surjection
nor an injection, but {0, 1} has the same cardinality as {0, 1}.

 9a, page 75
Proof. By symmetry, it suffices to show only one direction. Suppose A is countable. If A is
finite, then B is also finite by Lemma 8.2.8; thus B is countable and we are done. So suppose
A is infinite. Then A has the same cardinality as N by the definition of countability. Then
the symmetry and the transitivity of same-cardinality tell us B has the same cardinality of N.
So B is countable.

sol.11
 9b, page 76
Proof sketch. Suppose B is finite. Use the finiteness of B to find n ∈ N and a bijection
f : {1, 2, . . . , n} → B. Run the following procedure.
1. Initialize i = 0.
2. While A \ {g1 , g2 , . . . , gi } =
̸ ∅ do:
2.1. Let mi+1 be the smallest element in

m ∈ {1, 2, . . . , n} : f (m) ∈ A \ {g1 , g2 , . . . , gi } .

2.2. Set gi+1 = f (mi+1 ).

2.3. Increment i to i + 1.
On the one hand, suppose we are at line 2.2 when this procedure is run. Then the loop
condition tells us A \ {g1 , g2 , . . . , gi } =
̸ ∅. If ai ∈ A \ {g1 , g2 , . . . , gi }, then ai = f (m)
for some m ∈ {1, 2, . . . , n} because f is a surjection {1, 2, . . . , n} → B and A ⊆ B. This
says {m ∈ {1, 2, . . . , n} : f (m) ∈ A \ {g1 , g2 , . . . , gi }} =
̸ ∅. So mi+1 must exist by the
Well-Ordering Principle.
On the other hand, by the choices of mi+1 and gi+1 on lines 2.2 and 2.1,

each gi+1 ∈ A \ {g1 , g2 , . . . , gi }. (∗)

Case 1: this procedure stops after finitely many steps. Then a run results in

m1 , m2 , . . . , mℓ and g1 , g2 , . . . , gℓ

where ℓ ∈ N. Define g : {1, 2, . . . , ℓ} → A by setting g(i) = gi for all i ∈ {1, 2, . . . , ℓ}.

Notice A\{g1 , g2 , . . . , gℓ } = ∅ as the stopping condition is reached. This says any element
of A is equal to some gi , thus some g(i). So g is surjective. We know g is injective by (∗).
As g is a bijection {1, 2, . . . , ℓ} → A, we deduce that A is finite.

Case 2: this procedure does not stop. Then a run results in

m1 , m2 , m3 , . . . and g1 , g2 , g3 , . . . .

Define g : N → A by setting g(i) = gi+1 for all i ∈ N. As one can verify, this g is a bijection
N → A. Recall that f is a bijection {1, 2, . . . , n} → B. So Proposition 7.4.3 tells us f −1 is a
bijection B → {1, 2, . . . , n}. Define h : N → {1, 2, . . . , n} by setting h(i) = f −1 (g(i)) for all
i ∈ N. This h is injective because, if i, j ∈ N such that h(i) = h(j), then

f −1 (g(i)) = f −1 (g(j)) by the definition of h.

∴ g(i) = g(j) as f −1 is injective.
∴ i=j as g is injective.

This contradicts Exercise 8.2.7. Therefore, this case is not possible.

 9c, page 77
Proof. As f is an injection A → B, Exercise 8.2.5 implies that A has the same cardinality
as range(f ). Since A is infinite, Lemma 8.2.8 tells us range(f ) is also infinite. Recall from
Remark 7.2.3(2) that range(f ) ⊆ B. Hence B is infinite too by Proposition 9.2.6(1).

sol.12
 9d, page 77
Given any function f : A → P(A), we will produce an element of P(A) that is not equal to
f (x) for any x ∈ A. This will show that there is no surjection f : A → P(A), and thus A
cannot have the same cardinality as P(A).
Let f : A → P(A). Define R = {x ∈ A : x ̸∈ f (x)}. Then R ∈ P(A) by the definition of
power sets. We claim that R ̸= f (x) for any x ∈ A. We prove this by contradiction. Suppose
we have a ∈ A such that R = f (a). From the definition of R,


∀x ∈ A x ∈ R ⇔ x ̸∈ f (x) . (†)
As R = f (a), applying (†) to x = a gives
a ∈ f (a) ⇔ a ̸∈ f (a). (‡)
Split into two cases.
• Case 1: assume a ∈ f (a). Then a ̸∈ f (a) by the ⇒ part of (‡). This contradicts our
assumption that a ∈ f (a).
• Case 2: assume a ̸∈ f (a). Then a ∈ f (a) by the ⇐ part of (‡). This contradicts our
assumption that a ̸∈ f (a).
In either case, we get a contradiction. This completes the proof of the claim and thus of the
remark.

 9e, page 78
According to the definition of countability, we need to show that P(A) is infinite, and that
P(A) does not have the same cardinality as N.
Let f : A → P(A) defined by setting f (a) = {a} for each a ∈ A. Then f is injective
because if a1 , a2 ∈ A such that f (a1 ) = f (a2 ), then {a1 } = {a2 }, and thus a1 = a2 by the
definition of set equality. As A is infinite, Corollary 9.2.7(1) tells us that P(A) is infinite too.
As A is countable and infinite, it must have the same cardinality as N by the definition of
countability. However, Theorem 9.3.1 tells us A does not have the same cardinality as P(A).
Thus P(A) cannot have the same cardinality as N by Proposition 8.2.4.

 10a, page 82
(⊆) Let x ∈ A \ B. Then x ∈ A and x ̸∈ B by the definition of \. The latter implies x ̸∈ A
or x ̸∈ B. So it is not the case that x is both in A and in B by De Morgan’s Laws. Thus the
definition of ∩ tells us x ̸∈ A ∩ B. As x ∈ A, we deduce that x ∈ A \ (A ∩ B) according to
the definition of \.
(⊇) Let x ∈ A \ (A ∩ B). Then x ∈ A and x ̸∈ A ∩ B by the definition of \. In view of
the definition of ∩, the latter implies that it is not the case that x is both in A and in B. So
either x ̸∈ A or x ̸∈ B by De Morgan’s Laws. As x ∈ A, we deduce that x ̸∈ B. So x ∈ A \ B
by the definition of \.

 10b, page 83
Let C be the set of all countable sets and I be the set of all infinite sets here. Note that
all sets are either countable or infinite because finite sets are by definition countable. So
|C ∪ I| = 40. Also, by the Inclusion–Exclusion Rule,
|C ∪ I| = |C| + |I| − |C ∩ I|
= 12 + 31 − |C ∩ I|.
Hence, the number of sets that are both countable and infinite here is
|C ∩ I| = 12 + 31 − |C ∪ I| = 12 + 31 − 40 = 3.

sol.13
 10c, page 89
Define Γ = {B, I1 , C, O1 , N1 , D, I2 , T, I3 , O2 , N2 , A, L}. In view of the General Multiplication
Rule,
     
    number of number of number of
number of number of ways to

 ways to
 
 ways to 
  
permutations = permutations of  × arrange  × arrange  × arrange  .
     
of Γ BICONDITIONAL I1 , I2 , I3 into O1 , O2 into N1 , N2 into
3 positions 2 positions 2 positions
 
number of
∴ 13! = permutations of  × 3! × 2! × 2! by Corollary 10.3.7.
BICONDITIONAL
13!
From this, we see that the number of permutations of BICONDITIONAL is 3! 2! 2! = 13!/24.

 11a, page 94
Note that, in G, there is no edge between the left vertices, and there is no edge between the
right vertices. So vertices in a path in G must alternate between the left vertices and the
right vertices. As 1 and 3 are both on the left, the length of the path between 1 and 3 must
be even, and so be at least 2. So such a path must have length 2 or 4 because there are only
6 vertices in the graph, and paths cannot “use” the same vertex twice or more.

Case 1: consider those paths between 1 and 3 of length 2. There are 3 choices for
the remaining vertex from {a, b, c}. So there are exactly 3 such paths.
Case 2: consider those paths between 1 and 3 of length 4. Such a path contains one
more left vertex and two more right vertices. This left vertex must be 2 because this
is the only remaining vertex on the left. For the two right vertices, they can be any
two distinct ones from {a, b, c}, and order matters for these two vertices. So there are
P (3, 2) = 3 × 2 = 6 such paths.
In total, there are exactly 3 + 6 = 9 paths between 1 and 3 in G.

 11b, page 97
As one can verify exhaustively, (a) and (b) are drawings of connected graphs. It is clear that
(d) is a drawing of graph that is not connected. That (c) is a drawing of a graph that is not
connected can be more easily seen by redrawing the same graph as follows:

 12a, page 103

• By splitting into a ⇒ part and a ⇐ part, we used Technique 3.2.9 to prove a bicondi-
tional proposition.

• We used Technique 3.2.3 to prove the ⇒ part.

• The ⇐ part is proved using contraposition, i.e., Technique 3.2.7.
• In the ⇐ part, to prove that there is an edge e in G such that (V(G), E(G) \ {e})
remains connected, we produced a witness e; this is Technique 3.2.1.

sol.14
 • In the ⇐ part, we split into two cases: one when x1 xk ̸∈ E(P ), the other when x1 xk ∈
E(P ); this uses Technique 3.2.11.
• Etc.

 12b, page 104

• We used Strong MI, i.e., Technique 3.2.23.

• In the base step, we used a proof by contradiction, i.e., Technique 3.2.16, to show that
G does not have a loop.
• Etc.

 12c, page 105

• We used an algorithmic proof, i.e., Technique 9.2.3, to produce a subgraph Hk of G
|V(G)| = |V(Hk )| = |E(Hk )|.
• Etc.

 12d, page 108

• We used a counting argument here, i.e., Technique 10.3.16.
• Etc.

 12e, page 108

• We used a counting argument here, i.e., Technique 10.3.16.
• Etc.

sol.15