Problems on Discrete

Mathematics1
Chung-Chih Li2
Kishan Mehrotra3
Syracuse University, New York
LATEX at January 11, 2007

(Part I)

No part of this book can be reproduced without permission from the authors.
[email protected]
3
[email protected]
2

Preface

This is not so much a Preface as it is an explanation of why these notes were

prepared in the first place. Year after year, students in CIS275 and CIS375
have commented that the text book was poor, useless, or difficult to read
etc. It is our impression that most of the students were lost in the details and
were unable to decide what are the important definitions and results and how
to apply these results to solve problems. In these notes we have attempted to
summarize the important definitions and results. Neither we have given detailed
explanations of the definitions nor we have proved all of the results; students
are strongly urged to read the recommended text book.
Our main emphasis is to provide the student a large number of problems and
their solutions. We expect that the students will attempt to solve the problems
on their own and look at a solution only if they are unable to solve a problem.
These problems are collections of home works, quizzes, and exams over the
past few years. Most of the problems are from Discrete Mathematics with applications by H. F. Mattson, Jr. (Wiley).
We hope that these notes will prepare a student to better understand basic
mathematics necessary of computer scientists.

Acknowledgment
Our most sincere thanks to Elaine Weinman for her her help in typing several
parts of these notes and, more importantly, for many editorial corrections.

Contents
Preface

Acknowledgment

iii

Basic Concepts

0 Preliminary

0.1

Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

0.2

Patterns of theorems and proof . . . . . . . . . . . . . . . . . . .

1 Sets
1.1

7
Definitions and Basic Theorems . . . . . . . . . . . . . . . . . . .

1.1.1

Definitions . . . . . . . . . . . . . . . . . . . . . . . . . .

1.1.2

Basic Theorems . . . . . . . . . . . . . . . . . . . . . . . .

11

1.2

Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13

1.3

Solutions

17

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

2 Logic

37

2.1

Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

39

2.1.1

Propositional Logic . . . . . . . . . . . . . . . . . . . . . .

39

2.1.2

Predicate Logic . . . . . . . . . . . . . . . . . . . . . . . .

43

2.1.3

Predicates and Sets

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

44

Logical Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

44

2.2.1

Laws of Logic . . . . . . . . . . . . . . . . . . . . . . . . .

45

2.2.2

Rules of Inference . . . . . . . . . . . . . . . . . . . . . .

47

2.2.3

Inference Rules for Quantified Predicates . . . . . . . . .

48

2.2

2.3

DNF and CNF . . . . . . . . . . . . . . . . . . . . . . . . . . . .

48

2.3.1

The DNF of a given wff . . . . . . . . . . . . . . . . . . .

49

2.3.2

The CNF of a given wff . . . . . . . . . . . . . . . . . . .

51

2.3.3

A shortcut to find the DNF and CNF . . . . . . . . . . .

53

2.4

Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

55

2.5

Solutions

65

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

3 Mathematical Induction
3.1

3.2

3.3

Concepts

101

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

3.1.1

Necessary Conditions of Using Mathematical Induction . 103

3.1.2

The Underlying Theory of Mathematical Induction . . . . 104

3.1.3

Mathematical Induction of the First Form (Weak Induction)105

3.1.4

Mathematical Induction of the Second Form (Strong Induction) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

Mathematical Induction and Recursive Definition . . . . . . . . . 107

3.2.1

Recursive Definitions for Functions . . . . . . . . . . . . . 107

3.2.2

Recursive Definitions for Sets and Structural Induction . 109

Nested Induction . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

3.3.1

The underlying logic of nested induction . . . . . . . . . . 112

3.4

Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

3.5

Solutions

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

4 Relations
4.1

155

Definitions, Theorems, and Comments . . . . . . . . . . . . . . . 157

4.1.1

Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . 157

4.1.2

Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . 161

4.2

Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

4.3

Solutions

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166

5 Functions
5.1

183

Definitions, Theorems, and Comments . . . . . . . . . . . . . . . 185

5.1.1

Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . 185

5.1.2

Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . 187

5.2

The Pigeonhole Principle . . . . . . . . . . . . . . . . . . . . . . 188

5.3

Asymptotic Notations . . . . . . . . . . . . . . . . . . . . . . . . 189

vi

II

5.4

Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194

5.5

Solutions

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196

Specific Topics

207

6 Integers

209

6.1

Floor and Ceiling Functions . . . . . . . . . . . . . . . . . . . . . 211

6.2

Divisibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212

6.3

Greatest Common Divisor . . . . . . . . . . . . . . . . . . . . . . 215

6.4

Congruence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

6.5

Solving Linear Congruence Equations . . . . . . . . . . . . . . . 228

6.6

Solving Linear Congruence Equations with multiple variables . . 231

6.7

Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
6.7.1

Chinese Remainder Theorem . . . . . . . . . . . . . . . . 233

6.7.2

Fermats Little Theorem and Eulers Theorem . . . . . . 236

6.7.3

RSA Cryptosystem . . . . . . . . . . . . . . . . . . . . . . 239

6.8

Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242

6.9

Solutions

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246

7 Binomial Theorem and Counting

269

7.1

The Binomial Theorem

. . . . . . . . . . . . . . . . . . . . . . . 271

7.2

Principles and Typical Problems for Counting . . . . . . . . . . . 274

7.2.1

Urns and Balls Model . . . . . . . . . . . . . . . . . . . . 276

7.2.2

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 282

7.3

Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285

7.4

Solutions

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292

8 Recurrence Relations and Generating Functions

8.1

Recurrence Relations . . . . . . . . . . . . . . . . . . . . . . . . . 331

8.1.1

8.2

329

Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . 333

Solving Recurrence Relations . . . . . . . . . . . . . . . . . . . . 334

8.2.1

Repeated Substitution Method . . . . . . . . . . . . . . . 334

8.2.2

Characteristic Root Method . . . . . . . . . . . . . . . . . 335

8.2.3

Generating Function Method . . . . . . . . . . . . . . . . 340

vii

8.2.4

An Example

. . . . . . . . . . . . . . . . . . . . . . . . . 342

8.3

Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346

8.4

Solutions

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349

9 Discrete Probability
9.1

369

Definitions and Terminologies

9.1.1

. . . . . . . . . . . . . . . . . . . 371

Examples and Discussion . . . . . . . . . . . . . . . . . . 374

9.2

Theorems of Probability . . . . . . . . . . . . . . . . . . . . . . . 376

9.3

Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378

9.4

Solutions

III

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382

Appendices

397

A Loop Invariance

399

B Sample Quizzes

405

viii

Part I

Basic Concepts

Chapter 0

Preliminary
Perspicuity is part of proof
Ludwig Wittgenstein

0.2. Conventions

0.1

Conventions
N : The set of natural numbers, i.e., {1, 2, 3, . . .}.
N0 : N {0}.
Z : The set of integers.
Q : The set of rational numbers.
R : The set of real numbers.

0.2

Patterns of theorems and proof

1. Implication:
Let X and Y be two mathematical statements.
If X then Y.

(1)

In logic1 we denote (1) as X Y. The meaning of (1) is: if the mathematical statement X is true, or the mathematical condition X holds, then
the mathematical statement Y is true.
To prove this kind of theorem, we first assume that the statement X is true.
Then, we have to prove that the statement Y is a logical (informally,
we say reasonable) consequence of X . If we are able to do so, then we
can claim that the theorem (statement) (1) is correct.
2. Equivalence:
Let X and Y be two mathematical statements.
X if and only if Y

(2)

The meaning of (2) is:

(if X then Y) and (if Y then X ).
In logic we denote (2) as X Y, which is equivalent to
(X Y) (Y X ).
To prove that X if and only if Y is true, we have to prove that if X
then Y and if X then Y are both true. Many textbooks used iff or
to denote if and only if.
1 Refer

to the logic chapter.

c Chung-Chih Li, Kishan Mehrotra

0. Preliminary
3. Disproving:
Given a mathematical theorem as (1), if we can find an example for X
and Y such that this example makes X to be true but Y to be false, then
we can claim that the theorem is incorrect. For example,
if x2 > 0 then x > 0
is incorrect. Because we can find 1 such that (1)2 > 0 is true but
1 < 0. Such examples are called counter examples.
4. Proving by contradiction:
This is an important technique for proving mathematical results. Suppose
a theorem is given as (1). The idea of proving by contradiction is: we first
assume that the theorem is wrong, i.e., we can have instances that make
X true and Y false. Then we argue that, if this is the case, we can lead to
a result showing that X is false. But the result contradicts our assumption
that X is true. Therefore, the assumption will never happen, and hence
the theorem is correct.
For example, consider
if x + 1 > 2 then x > 1.
To prove it by contradiction, we assume that there is an x such that,
x + 1 > 2 and x 1. From the assumption x 1, we have
x1x+11+1
x + 1 2.
The result x + 1 2 contradicts the assumption x + 1 > 2. Therefore, the
assumption will never happen, and hence the theorem is correct.
5. Proving by cases:
Suppose we are given a theorem with respect to the domain D. If possible,
we exhaust all of the examples in the domain to see if the theorem is
correct. But, in general, we are not able to do so because the domain is
usually an infinite set, and even worse, the domain can be uncountable,
e.g., real numbers. To overcome this problem, we divide the domain into
several categories and make sure that those categories cover the domain.
Then we exam each case to see if the theorem is valid. If the theorem holds
in every case, then the theorem is correct in the entire domain. Note that
we must use the idea of universal generalization2 in the proof of each case.

2 See

the logic chapter for more details.

c Chung-Chih Li, Kishan Mehrotra

Chapter 1

Sets
One can always make a theory, many theories,
to account known facts,
occasionally to predict new ones.
The test is aesthetic.
Gorge Thomson

1.1. Definitions and Basic Theorems

1.1

Definitions and Basic Theorems

Set theory is one of the most rigorous study in mathematics. In fact, the desire
to advance the modern set theory had been the desire of mathematicians who
wanted to pursue ultimate rigorousness in mathematics. Although the results
of securing our mathematical foundations turn out to be rather negative and we
are unfortunately in a loosing battle, the concept of sets and the notations used
in this battle are proven to be a indispensable tool in the study of mathematics
at any level.
Of course, we will not step into the dark side of the road in this book. In
stead, we will study some naive concepts of sets; most of them are intuitively
understandable from our daily-life experiences. For example, all students of
Syracuse University is a set; all students in the United State of America is a
superset of the set of students at Syracuse University. Since Dennis is a student
at Syracuse University, he is a member of the set of the students of Syracuse
university. Some students of Syracuse University are also students of Cornell
University, but none of them are students of Stanford University...., and so on.
Some of the sentences above are a bit awkward, but we live with them without
too many complains. However, when mathematicians come along, they have
to deal with mathematical objects that very often have to carry some relations
that are much more complicate than the descriptions we used in the sentences
above. Fortunately, the concepts and some immediate properties in the set
theory provide us a simple yet precise notation to simplify our works.
In addition to letting the reader be familiar with the basic terminologies and
properties of sets, another purpose of this chapter is to let the reader be used to
rigorous mathematical arguments by getting through the proofs step by step.

1.1.1

Definitions

Definition 1.1: A set S is a collection of distinct objects without regard to the

order of the objects given by any possible method of description. Usually,
we use a pair of braces, {}, to enclose the concerned collection.
Definition 1.2: The empty set is a null collection, denoted as or {}.
Definition 1.3: The universal set is the set that contains everything concerned,
usually denoted as U . In general, the context of the problem determines U .
Definition 1.4: The objects in a set S are called the members of S. Some
textbooks use elements instead.
Definition 1.5: Suppose a is a member of a set S. We denote this property as
a S. The property is known as the membership relation.
c Chung-Chih Li, Kishan Mehrotra

10

1. Sets

Definition 1.6: Let A, B be sets. A is a subset of B if and only if all members

of A are members of B. We use A B to denote that A is a subset of B.
If A 6= B, we say that A is a proper subset of B, denoted as A B.
Definition 1.7: Let A, B be sets. The intersection of A and B, denoted as
A B, is the set C such that every member of C is a member of both A and
B. In logic this set is defined as
A B = {x|x A & x B}.
Definition 1.8: Let A, B be sets. If A B = , we say that A and B are
disjoint.
Definition 1.9: Let A, B be sets. The union of A and B, denoted as A B, is
the set C such that every member of C is a member of either A or B. In
logic this set is defined as
A B = {x|x A or x B}.
Definition 1.10: Let A, B be sets. The set A B is defined as
A B = {x|x A & x 6 B}.
Definition 1.11: Let A, B be sets. The Cartesian product of A and B, denoted
as A B, is defined as
A B = {(a, b)|a A & b B}.
Definition 1.12: Let S be any set, the cardinality of S, denoted as |S|, is the
number of elements in S 1 .
Definition 1.13: Let U be the universal set and A any set in the universe.
Define
A = U A.
A is called the complement of A.
Definition 1.14: The Venn diagram consists of figures that show the relations
between sets. Example:

AB

Venn Diagram
1 This

naive definition is sufficient for our purpose set up in this book. And, it is intuitively
understandable for finite sets. A more serious mathematical setup is needed to understand
the cardinality of an infinite set, but that is far beyond the scope of this book.

c Chung-Chih Li, Kishan Mehrotra

1.1. Definitions and Basic Theorems

11

Definition 1.15: Let A be any set. The power set of A, denoted as Pr(A), is
the set of all possible subsets of A. In symbols,
Pr(A) = {S|S A}.

1.1.2

Basic Theorems

The following list contains the most important theorems that are used in many
mathematical proofs. We do not prove them here, but ask them as problems in
the problem section and prove them in the solution section.
Theorem 1.1: Let A be a set.
A = A;

A = .

Theorem 1.2: Let A be a set.

A A = A A = A.
Theorem 1.3: Let A, B be sets.
A = B if and only if (A B & B A).
Theorem 1.4: Let A, B, C be sets.
If (A B & B C), then A C.
Theorem 1.5: Let A, B be sets,
A A B,
A B A,

B A B,
A B B.

Theorem 1.6: The commutative rules: Let a, b be sets.

A B = B A,
A B = B A.
Theorem 1.7: The associative rules: Let A, B, C be sets.
A (B C) = (A B) C,
A (B C) = (A B) C.
Theorem 1.8: The distribution rules: Let A, B, C be sets.
A (B C) = (A B) (A C),
A (B C) = (A B) (A C).
c Chung-Chih Li, Kishan Mehrotra

12

1. Sets

Theorem 1.9: De Morgans laws: Let A, B be sets.

A B = A B,
A B = A B.
Theorem 1.10: Let A, B be two infinite sets. |A| = |B| if and only if there
exists a bijection2 f : A B.

2 For

the definition of bijection, please refer to chapter 5.

c Chung-Chih Li, Kishan Mehrotra

1.2. Problems

1.2

13

Problems

Problem 1: Determine
if each of the following objects is a member of Z;
{5}, {3, 1}, 7.12, 5, a = the 2,00th decimal digit in the base-10 expression for .
Problem 2: Let A be the set of digits in the base-10 expression of the
41
44
rational number 333
. Let B be the same for 333
. Prove that A = B.
Problem 3:
of

Prove or disprove: the set C of digits in the base-10 expression

40363 63637
3 33000 00000

equals the set A of the previous problem.

Problem 4:

Define sets
A = {1, {4}, {2}, 3, 4, 5},
B = {{{1, 4, 5, 3, 1}}},
C = {1, {3}, 2, 1},
D = {1, 1, 3},
E = {1, 4, {5}, {3}},
F = {1, 8, {1, 2, 3, 4}},

and,
a1 = 1, a2 = {2}, a3 = {2, 1}, a4 = {2, 1, 3, 4}, a5 = {3, 1, 5}.
For each of a1 , . . . , a5 , determine if it is a member of the sets A, . . . , F
respectively. Present your answer in the following table:
A

a1
a2
a3
a4
a5
Use an to stand for membership and a blank to stand for nonmembership.
Problem 5: Define A, B, C, D, E, and F as in Problem 4. Calculate the
following sets.
AC

BF

DC

C E

C (DF )

AE.

c Chung-Chih Li, Kishan Mehrotra

14

1. Sets

Problem 6:

Define

U = {3, 1, 3, 2},
V = {1, 3, {1, 3}, {1, 2, 3}}.

Is U V ? Is U V ?
Problem 7: Let A and B be sets. If A B, what does that tell you about
A B and A B?
Problem 8: Let A, B and S be sets. If A S and B S, what can you
say about A B?
Problem 9:

Find the cardinality of the set

S = {p/q | p, q N + , p, q, 10}.

Problem 10:

List all the subsets of

{1, 2, {b}}.

Problem 11:

List all the elements of

{b, c, d} {e, o}.

Problem 12:

Describe all sets that have no proper subsets.

Problem 13: Let A, B, C be sets. Suppose that A is a subset of B, and C

is a proper subset of B. Is A a proper subset of C?
Problem 14: Let A and B be any sets. From the definition of union it
follows that
if x A B, then x A or x B.
Consider now the analogue for inclusion. Can we say that for any set X,
if X (A B), then X A or X B?
Why or why not?
Problem 15:

Prove or disprove that

if (A B) (A B), then A = B.

Problem 16:

Let A, B, C be sets. Prove the distributive laws:

1. A (B C) = (A B) (A C).
2. A (B C) = (A B) (A C).
Problem 17:

Let A, B be sets. Prove the De Morgan laws:

1. A B = A B.
c Chung-Chih Li, Kishan Mehrotra

1.2. Problems

15

2. A B = A B.
Problem 18:

Let A, B, C be sets. Prove

A B = A B.

Problem 19:

Let A, B, C be sets. Prove

A (A B) = A B.

Problem 20:

Let A, B, C be sets. Prove

AB =

Problem 21:

if and only if

A B.

Let A B = . Show that

A B = A.

Problem 22:

Define the symmetric difference of sets A, B as

A4B := (A B) (B A).

Prove that symmetric difference is associative and commutative. Also

prove that for all sets A
A4A = .
Problem 23:

Let A, B be sets. Use Venn diagrams to show that

A B if and only if A B = A.

Problem 24:

Let A, B, C be any sets. Use

1. Venn diagrams,
2. set algebra, and
3. epsilon-argument,
respectively, to prove that
B C A if and only if A B C.
Problem 25:

Let A := {, 4} and B := {a, 3}. List the elements of A B.

Problem 26:

Let A, B, and C be any sets. Show that

A (B C) = (A B) (A C)

Problem 27:

Let A, B be sets. Prove or disprove that

if (A A) = (B B), then A = B.
c Chung-Chih Li, Kishan Mehrotra

16

1. Sets

Problem 28: Let A, B, U, and V be any sets such that A U and B V .

Is the following correct? Explain.
(A B) (U V ).
Problem 29:

Let A, B, C be sets. Prove

A (B C) = (A B) (A C).

Problem 30: Let A, B be sets, and C := A B. Use the result in Problem

29 to prove that
A B A C C C.
Problem 31:

Let A be a set. Define

diagonal of A A := {(a, a)|a A}.

Suppose A B. Prove or disprove the following identities.

1. (A B) (B A) = B B.
2. (the diagonal of A A) (A B) = the diagonal of B B.
Problem 32:

Let X = {1, 2, 3, a}. List the elements of P(X).

Problem 33:

Let X be the set {1, 2, 3, 4, 5, 6, {1}}. Find Y such that

Y = X (X P(X)).

Problem 34:
1.
2.
3.
4.
5.
6.

List the elements of the following sets.

P()
P({})
P(P())
{} P()
P()
P() P()

Problem 35:

List the elements of A P(A), where A = {a, 1}.

Problem 36:

Let A, B be sets. Prove

P(A B) = P(A) P(B).

Problem 37:

Prove or disprove that

if A P(B) & B P(A), then A = B.

Problem 38: Let A be the set of nonnegative integers and B the set of
nonnegative odd integers. Prove that the cardinality of A is equal to the
cardinality of B. [See Theorem 1.10 on page 12]
Problem 39:

Prove that for any set A, |A| < |Pr(A)|.

c Chung-Chih Li, Kishan Mehrotra

1.3. Solutions

1.3

17

Solutions

Solution 1:
1. {5} 6 Z; {5} is a set, although the set happens to contain an integer.
2. {3, 1} 6 Z; {3, 1} is a set not an integer.
3. 7.12 6 Z.
Note: One should not say that 7.12 is a real number, thus 7.12 6 Z. A
number that is real does not mean it cannot be an integer. Dont forget
that Z R. That means all integers are also real numbers.
4.

5 6 Z.

Note: One should not havea problem in pointing out that 5 6 Z. The
reason is that the value of 5 is not an integer, and that is. We cannot

conclude that the operator

will result in a number that is not an integer

on any input. For example,

4 Z. In some computer programming
languages, thedata type of 4 may be real causing the confusion. But in
mathematics, 4 is an integer; it does not matter how computers treat it.
5. Whatever the number a might be, we are sure that a is an integer, thus
a Z.

Solution 2:
41
44
= 0.123, and
= 0.132,
333
333
where a bar means the integers are repeated in that order forever. Thus,
A = {0, 1, 2, 3}, and B = {0, 1, 2, 3}.
Therefore, A = B.

2
c Chung-Chih Li, Kishan Mehrotra

18

1. Sets

Solution 3:

We patiently divide to get:

4036363637
= 0.12121212123.
33300000000

Thus, C = {0, 1, 2, 3}. Therefore, A = C.

Solution 4:

a1
a2
a3
a4
a5

The set A has six members: 1, {4}, {2}, 3, 4, and 5. Only a1 and a2 are
elements of A.
The set B has only one member, that is {{1, 3, 4, 5}}.
Lets take another example. F contains 3 elements: 1, 8 and {1, 2, 3, 4}.
Thus, a1 and a4 are elements in F . Note that 2 6 F although 2 {1, 2, 3, 4}
and {1, 2, 3, 4} F . This example shows that is not transitive3 .
2

Solution 5:
1. A C = {1}.
2. B F = .
3. D C = {1, 2, 3{3}}.
4. C E = {1, {3}}.
5. C (D F ) = {1, 2, {3}}.
6. A E = {1, 4}.

2
3 See

chapter 4 for the transitivity

c Chung-Chih Li, Kishan Mehrotra

1.3. Solutions
Solution 6:

19
Define
U = {3, 1, 3, 2}
V = {1, 3, {1, 3}, {1, 2, 3}}.

Since we do not consider multiset, thus U := {1, 2, 3}. Therefore, U is a member

of V . However, U is not a subset of V because 2 U , but 2 6 V .
2

Solution 7:

Since A B, we obtain the following Venn diagram.

It is clear that A B = A and A B = B.

Solution 8:
We can answer this problem by using Venn diagrams, but we
have to draw Venn diagrams that cover all possible cases of A and B to make
the proof complete.
Let us use another kind of argument for this problem. Two conditions are
given, i.e.,
A S,
B S.

(1.1)
(1.2)

Given x A B. From the definition of union, we have two cases: x is

either in A or in B.
Case 1: x A. Because A S, we know x S.
Case 2: x B. Because B S, we know x S.
In either case, we conclude that x S. Therefore, by the definition of , we
have A B S.
c Chung-Chih Li, Kishan Mehrotra

20

1. Sets

Can we claim that A B is a proper subset of S? The answer is that we

cannot. From (1.1) and (1.2), we know only that there is at least one element
x S such that x 6 A, and there is at least one element y S such that y 6 B,
but we dont know if x = y. To see this, consider

A = {1}, B = {2}, and S = {1, 2}.

It is clear that A S, B S, and A B = S.

In some textbooks the argument used above is called the -argument, because the above proof procedure is based on the properties of the membership
of elements to sets.
2

Solution 9:

Find the cardinality of the set,

S = {p/q | p, q N + , p, q, 10}.

The cardinality of a finite set is simply the number of distinct elements of the set;
we do not consider multisets. In other words, all identical elements are counted
as one. For example, 1/2, 2/4, 3/6, 4/8, and 5/10 are the same element.
In this example, we can list all p/q, remove extra identical elements, and
count the remaining. The cardinality of S is 63.
2

Solution 10:

The subsets of {1, 2, {b}} are

, {1}, {2}, {{b}}, {1, 2}, {1, {b}}, {2, {b}}, {1, 2, {b}}.

Remember that the empty set and {1, 2, {b}} itself are both subsets of the given
set.
2

c Chung-Chih Li, Kishan Mehrotra

1.3. Solutions

21

Solution 11:
{b, c, d} {e, o} = {(b, e), (c, e), (d, e), (b, o), (c, o), (d, o)} .

Solution 12:

The empty set is the only set that has no proper subset.

Solution 13:
The statement is wrong. The following Venn diagram disproves the statement.

AB
CB

A 6 C

The best way to disprove a theorem is to give a counter example. For

example, let
B = {1, 2, 3, 4, 5}, A = {1, 2, 3}, C = {3, 4, 5}.
It is easy to see that, A B, C B, but A 6 C.
The easiest counter example is when A = C = and B is any non-empty
set. In this case, A B, C B, but A is not a proper subset of C.
2

Solution 14:

The following statement is incorrect.

If X (A B), then X A or X B.

We can give a counter example to disprove the statement.

Let A = {1}, B = {2}, and X = {1, 2}. It is clear that X (A B), but
neither X A nor X B.
2

c Chung-Chih Li, Kishan Mehrotra

22

1. Sets

Solution 15:

For any two sets A and B,

if (A B) (A B), then A = B.

(1.3)

(A B) (A B).

(1.4)

We assume that
Then ask: Is A equal to B?
To prove A = B, we may prove that A B and B A.
1. To prove A B, assume x A.
xA xAB
xAB
xB

def. of union;
(1.4) and the def. of subset;
def. of intersection.

Therefore, A B.
2. Similarly, to prove B A, let x B.
xB xAB
xAB
xA

def. of union;
(1.4) and the def. of subset;
def. of intersection.

Therefore, B A.
From 1 and 2 above, we conclude that A = B.

Method 2:
We can use another way to prove A = B under the same condition given in
(1.4). Let us recall the following basic theorems: For any sets A, B and C,
1. A B A.
2. A A B.
3. A B C A C.
Thus,
(A B) A (A B).
And, since (A B) (A B) is given, we have
(A B) A (A B) (A B) A.
Therefore, A (AB) and (AB) A, hence A = AB. Similarly, B = AB.
Therefore, A = B.
2
c Chung-Chih Li, Kishan Mehrotra

1.3. Solutions

23

Method 3:
We can also prove (1.3) by way of contradiction, a very important technique
for mathematical reasoning. Mathematicians use this technique very often.
Suppose that
(A B) (A B), and A 6= B.
Without loss of generality, we assume that A 6 B 4 . That is, there is an element
a in A but not in B. We know that a A B and a 6 A B. From the
definition of subset, we have
A B 6 A B,
which contradicts the assumption that (A B) (A B). Hence, A B must
be true.
2

Solution 16:

Let A, B, C be sets.

1. A (B C) = (A B) (A C).
(a) To show that A(B C) (AB)(AC), suppose x A(B C).
We have two cases:
Case 1: x A. If x A, then x (A B) and x (A C). Thus,
x (A B) (A C).
Case 2: x (B C). If x (B C), then x B and x C. Thus,
x (A B) (A C).
Therefore, in either case we have A (B C) (A B) (A C).
(b) To show that A (B C) (A B) (A C), suppose x (A
B) (A C). Then x A B and x A C. We have two cases:
Case 1: x A. If x A, then x A (B C).
Case 2: x 6 A. If x 6 A, then x must be in B and C, because from
the assumption x A B and x A C. Thus, x B C, and
hence x A (B C).
In either case we have A (B C) (A B) (A C).
Therefore,
A (B C) = (A B) (A C).
2. A (B C) = (A B) (A C).
4 The other case is: A 6 B. For this case, the argument is symmetric to the one we
give, and there is no need to repeat it again. This is the reason we can say without loss of
generality

c Chung-Chih Li, Kishan Mehrotra

24

1. Sets
(a) To show that A(B C) (AB)(AC), suppose x A(B C).
x A (B C) x A and x (B C).
From the definition of union, x (B C) gives two cases:
Case 1: x B. If x B, then x (A B).
Case 2: x C. If x C, then x (A C).
Thus, x (A B) (A C). .
(b) To show that A (B C) (A B) (A C), suppose x (A
B) (A C). We have two cases.
Case 1: x A B. If x A B, then x A and x B. Because
x B, x must be in B C. Therefore, x A (B C).
Case 2: x A C. If x A C, then x A and x C. Because
x C, x must be in B C.
In either case, x A (B C).
Therefore, by putting (a) and (b) together, we have
A (B C) = (A B) (A C).

Solution 17:

Let A, B be sets. Prove

1. A B = A B.
(a) Let x A B.
x A B x 6 A B,
x 6 A and x 6 B,
x A and x B,
x A B.
(b) Let x 6 A B.
x 6 A B x A B,
x A or x B,
x 6 A or x 6 B,
x 6 A B.
Therefore, A B = A B.
2. A B = A B.
c Chung-Chih Li, Kishan Mehrotra

1.3. Solutions

25

(a) Let x A B.
x A B x 6 A B,
x 6 A or x 6 B,
x A or x B,
x A B.
(b) Let x 6 A B.
x 6 A B x A B,
x A and x B,
x 6 A and x
6 B,
x 6 A B.
Therefore, A B = A B.

Solution 18:

Let A, B, and C be sets. We want to prove

A B = A B.

From the definition, A B is the set of all elements in A but not in B. Thus,
if x A B, then x A and x 6 B, namely, x A and x B. From the
definition of intersection, x A B. Therefore,
A B A B.

(1.5)

For the other direction, if x A B, then x A and x B, namely x A

and x 6 B, thus x A B. Therefore,
A B A B.

(1.6)

From (1.5) and (1.6), we have

A B = A B.

(1.7)

Solution 19:

Let A, B, and C be sets. To prove

A (A B) = A B,
c Chung-Chih Li, Kishan Mehrotra

26

1. Sets

we will use the equality (1.7).

Let us start from the left-hand side,
A (A B) = A (A B),
= A (A B),
= A (A B),
= (A A) (A B),
= (A B),
= A B.

from (1.7)
from (1.7)
De Morgan law
distributive law

Therefore, the left-hand side is equal to the right-hand side.

Solution 20:
Let A, B, and C be sets. To prove that A B = if and only
if A B, we split our task into two subtasks.
1. The first subtask is to prove that A B = A B.
By way of contradiction, suppose
A B = and A 6 B.
If A 6 B, we can find an x, such that x A and x 6 B. But,
x A, x 6 B

x A & x B,
x A B,
A B 6= .

This leads to a contradiction.

2. Our second subtask is to prove that A B A B = .
Again, by way of contradiction, suppose
A B and A B 6= .
If A B 6= , there exists an x such that, x A B. But,
xAB

x A & x B,
xA&x
6 B,
A 6 B.

This leads to a contradiction.

From 1 and 2, we have proved
A B = A B.
c Chung-Chih Li, Kishan Mehrotra

1.3. Solutions

27

Solution 21:

Suppose A B = . Given any x, we have

xA

x 6 B
xB
x (A B)
A (A B)

because A B =

def. of subset.

From the property of intersection of two sets, we have (A B) A. Thus,

(A B) = A.
2

Solution 22:

Define
A4B := (A B) (B A).

Because is commutative, thus (AB)(BA) = (BA)(AB) = B4A.

Hence 4 is commutative.
Before proving the associativity of 4, let us prove an intermediate result:
A4B = (A B) (A B).

(1.8)

Recall equation (1.7), we know that A B = A B. Therefore, we can rewrite

A4B as
A4B = (A B) (A B).
(1.9)
A4B = (A B) (A B)
= (A B) (A B)
= (A B) (A B)
= [(A B) A] [(A B) B]
= [(A A) (B A)] [(A B) (B B)]
= (A B) (A B)

from (1.9),
De Morgan laws,
De Morgan laws,
distributive law,
distributive law,
commutative law.

Now, we prove the associativity of 4:

(A4B)4C = A4(B4C).
(A4B)4C
= [(A4B) C] [(A4B) C]
(1.9),
= [((A B) (A B)) C] [((A B) (A B)) C]
(1.9),(1.8),
= [(A B C) (A B C)] [(A B C) (A B C)] distributive,
= (A B C) (A B C) (A B C) (A B C).
c Chung-Chih Li, Kishan Mehrotra

28

1. Sets

Similarly, we can prove that

A4(B4C) = (A B C) (A B C) (A B C) (A B C).
Therefore, (A4B)4C = A4(B4C), i.e., 4 is associative.
Finally, A4A = (A A) (A A) = = .

Solution 23:

A B if and only if A B = A.

(i) A Venn diagram for A B = A. Note that A B.

(ii) A Venn diagram for A B 6= A. Note that A 6 B.

AB

Solution 24:
1. Using Venn diagrams:
(a) B C A A B C.
By way of contradiction, suppose B C A and A B 6 C. If
c Chung-Chih Li, Kishan Mehrotra

1.3. Solutions

29

AB 6 C, then we can find sets A, B, and C as shown in the following

Venn diagram, where the shaded area covers those elements that are
in A B but not in C. From the diagram we can see that B C 6 A,
because the shaded area is a subset of B C and is also a subset of
A.

A
A

C
B

(b) B C A A B C.
Again, by way of contradiction, suppose A B C and B C 6 A.
We can use the same Venn diagram. If B C 6 A, we can find the
shaded area contradicts our assumption that A B C.
2. Using set algebra;
B C A

(B C) A =
(B C) A =
(B C) A =
A (B C) =
(A B) C =
(A B) C =
(A B) C

definition of ,
problem 18,
problem 18,
commutative law,
associative law,
problem 18,
definition of .

3. Using epsilon-arguments:
(a) B C A A B C.
By way of contradiction, suppose B C A and A B 6 C. If
A B 6 C, then there exists an x such that x A B and x 6 C.
Because x A B, we know x B and x 6 A. And, because x B
and x 6 C, we know x B C. Therefore, B C 6 A.
(b) B C A A B C.
By way of contradiction, suppose A B C and B C 6 A. If
B C 6 A, there exists an x such that x B C and x 6 A. If so,
x B and x 6 C because x B C, and x A because x 6 A.
Therefore, A B 6 C because x A B and x 6 C.

c Chung-Chih Li, Kishan Mehrotra

30

1. Sets

Solution 25:
A B = {(, a), (, 3), (4, a), (4, 3)}.

Solution 26:
Let X and Y be two sets. There are two approaches to show
that X = Y . One approach is to show X Y and X Y , i.e, to show if a X
then a Y , and if a Y then a X. The other approach is to show if a X
then a Y , and if a 6 X then a 6 Y . Let us solve this problem by the second
approach.
1. Let a = (x, y), and suppose a A (B C).
(x, y) A (B C)
x A and y (B C)
x A and (y B or y C)
(x A and y B) or (x A and y C)
(x, y) A B or (x, y) A C
(x, y) (A B) (A C).
Therefore, a (A B) (A C).
2. Let a = (x, y), and suppose a 6 A (B C).
(x, y) 6 A (B C)
x 6 A or y 6 (B C)
x 6 A or (y 6 B and y 6 C)
(x 6 A or y 6 B) and (x 6 A or y 6 C)
(x, y) 6 A B and (x, y) 6 A C
(x, y) 6 (A B) (A C).
Therefore, a 6 (A B) (A C).
From (1) and (2), we have proved that
A (B C) = (A B) (A C).

Solution 27:
1. x A.

Suppose that (A A) = (B B).

x A (x, x) (A A)
(x, x) (B B)
x B.

c Chung-Chih Li, Kishan Mehrotra

1.3. Solutions

31

Therefore, A B.
2. x B.
x B (x, x) (B B)
(x, x) (A A)
x A.
Therefore, B A.

From the above, A = B.

Solution 28:
Then we have

Let A, B, U, and V be any sets such that A U and B V .

(A B) (U V ).

To prove this, let x = (a, b) and x A B. Thus, a A and b B. Since

A U and B V , we have a U and b V . Therefore, (a, b) U V, i.e.,
x U V.
2
We can prove this result by using the contradiction argument. Suppose the
result is incorrect, i.e., we can find sets A, B, U, and V such that
A U, B V and A B 6 U V.
If that is the case, then there exists (a, b) A B and (a, b) 6 U V. If
(a, b) 6 U V, then either a 6 U or b 6 V . Both cases contradict the assumption:
(A U and B V.)
2

Solution 29:

Let A, B, C be sets.

(a, b) A (B C)

a A and b (B C),
a A and (b B or b C ),
(a, b) (A B) or (a, b) (A C),
(a, b) (A B) (A C).

Therefore,
A (B C) = (A B) (A C).
Since we can use in every step of the proof above, we dont have to prove
the two directions of inclusion explicitly.
2

c Chung-Chih Li, Kishan Mehrotra

32

1. Sets

Solution 30:
i.e.,

To prove this problem, we will use the result in problem 29,

X (Y Z) = (X Y ) (X Z).

(1.10)

A similar equality we need is

(X Y ) Z = (X Z) (Y Z).

(1.11)

This can be proved in a similar manner discussed in problem 29.

1. To prove A B A C,
A C = A (A B)
= (A A) (A B)

because C = (A B),
(1.10).

Because (A B) ((A A) (A B)), we have (A B) (A C). 2

2. To prove A C C C,
C C = (A B) C
= (A C) (B C).

because C = (A B),
(1.11).

Because (A C) ((A C) (B C)), we have (A C) (C C).

Solution 31:

If A B, both (1) and (2) are incorrect.

We can give a counter example to disproves (1) and (2) of this problem. Let
A be the empty set and B be any non-empty set. Therefore, A B, B A, and
A A are empty sets, while B B is not.
2

Solution 32:

P(X) =

, {1}, {2}, {3}, {a}, {1, 2}, {1, 3}, {1, a}, {2, 3},

{2, a}, {3, a}, {1, 2, 3}, {1, 2, a}, {1, 3, a}, {2, 3, a}, {1, 2, 3, a}

}.

Note: Dont forget to include the empty set and the set X in P(X).

Solution 33:
We will prove the following equality first, so we do not have
to enumerate all elements in P(X). For any set X,
X = [X (X P(X))].
c Chung-Chih Li, Kishan Mehrotra

1.3. Solutions

33

From the definition of union, we know that

X [X (X P(X))].

(1.12)

From the definition of intersection, we know that

(X P(X)) X.

(1.13)

Union with X on both sides of (1.13), we get

[X (X P(X))] (X X).
And, since (X X) = X,
[X (X P(X))] X.

(1.14)

From (1.12) and (1.14), we get X (X P(X)) = X. For this problem,

X (X P(X)) = X = {1, 2, 3, 4, 5, 6, {1}}.

Solution 34:
1. P() = {}
2. P({}) = {, {}}
3. P(P()) = {, {}, {{}}, {, {}}}
4. {} P() = {(, )}
5. P() =
6. P() P() = {(, )}

Solution 35:
A = {a, 1},
P(A) = {, {a}, {1}, {a, 1}},

(a, ), (a, {a}), (a, {1}), (a, {a, 1}),

A P(A) =
.
(1, ), (1, {a}), (1, {1}), (1, {a, 1})
c Chung-Chih Li, Kishan Mehrotra

34

1. Sets

Solution 36:
x Pr(A B)

x (A B),
x A and x B,
x Pr(A) and x Pr(B),
x Pr(A) Pr(B).

Therefore, Pr(A B) = Pr(A) Pr(B).

Solution 37:
of power set,

Suppose that A Pr(B) and B Pr(A). From the definition

A Pr(B) A B,
B Pr(A) B A.

Therefore, A = B.

Solution 38:
Let A be the set of natural integers and B the set of odd
natural integers. Define function f : A B as
f (n) = 2n 1.
We want to prove that f is a bijective function from A to B.

1. Let a1 , a2 A, and a1 6= a2 . It is clear that 2a1 1 6= 2a2 1. Thus,

f (a1 ) 6= f (a2 ), and hence f is injective.
2. Let b B. Then
surjective.

b1
2

A, because it is a natural number. Therefore, f is

From the definition of cardinality, |A| = |B|.

c Chung-Chih Li, Kishan Mehrotra

1.3. Solutions
Solution 39:

35
We consider the following axiom:

Let A and B be any sets. |A| |B| if and only if there exists an injection g
from A to B.
To prove that |A| |Pr(A)|, we define g : A Pr(A) as
g(x) = {x}, for all x A.
g is injective because for all x, y A, if x 6= y, then {x} 6= {y}. Thus, by the
axiom, |A| |Pr(A)|.
Now, we have to prove that |A| 6= |Pr(A)|. By way of contradiction, suppose
|A| = |Pr(A)|. Then there is a bijective function f : A Pr(A). Thus, for any
a A, f (a) is a subset of A. Define the set S as
S = {x|x 6 f (x)}.
It is clear that S is a subset of A. Since f is surjective, there exists s A such
that f (s) = S. We have two cases: s S and s 6 S.
1. If s S, that means s f (s), then s 6 S.
2. If s 6 S, that means s 6 f (s), then s S.
Either case gives a contradiction. Therefore, there is no bijective function from
A to Pr(A), and hence |A| 6= |Pr(A)|.

From |A| |Pr(A)| and |A| 6= |Pr(A)|, we conclude that |A| < |Pr(A)|.

c Chung-Chih Li, Kishan Mehrotra

Chapter 2

Logic
It presupposes nothing but logic; that is,
logic is the only preceding theory.
Alfred Tarski

2.1. Definitions

2.1
2.1.1

39

Definitions
Propositional Logic

Definition 2.1: Propositions are mathematical statements such that their truth
or falsity can be told without ambiguity. A proposition is also called a primitive statement.
We use letters p, q, r, . . . to denote propositions. Thus, the values of p, q, r, . . .
are either T rue or F alse.
Example 2.1
p:
q:
r:
s:

2 + 2 = 5,
x2 2x + 1 has two identical roots,
{1, 2, },
there exists a 100-digit prime number,

are propositions. The value of p is F alse, and the values of q and r are
T rue. We will use T to denote T rue and F to denote F alse for the rest of
this book.
Comment: The statement s in the above example is a proposition, even
though we do not know for sure that there exists a 100-digit prime
number, but we are sure that s is either T or F . However, not every
mathematical statement has a truth value. The following two statements are not propositions.
u : This statement is false.
v : a S, where S is a set defined as S = {x|x 6 S}.
Definition 2.2: Suppose p, q, r, . . . are variables with values either T or F . We
call such variables propositional variables. We can assign T or F to any
propositional variable as we wish. propositional variables are also known as
atoms.
Definition 2.3: Let p, q be two propositions. We use the symbols:
, , , ,
to construct new propositions such as
p, p q, p q, p q, p q..
These symbols are called logical connectives; is read as not or negation,
as and or conjunction, as or or disjunction, p q is read as
if p then q or p implies q, and p q is read as p if and only if q.
Recall that all propositions have truth values either T or F . We decide the
values of the new propositions constructed above based on the values of p
and q and the rules of logical connectives defined below.
c Chung-Chih Li, Kishan Mehrotra

40

2. Logic

Definition 2.4:

1. The negation of p: p.
p
T
F

p
F
T

2. The conjunction of p and q: p q.

p
T
T
F
F

q
T
F
T
F

pq
T
F
F
F

3. The disjunction of p and q: p q.

p
T
T
F
F

q
T
F
T
F

pq
T
T
T
F

q
T
F
T
F

pq
T
F
T
T

4. Implication, if p then q: p q.
p
T
T
F
F

5. Equivalence, p if and only if q: p q.

p
T
T
F
F

q
T
F
T
F

pq
T
F
F
T

Definition 2.5: Given p q, we call q p the contrapositive, call q p

the converse, and call p q the inverse of p q.
Definition 2.6: Propositional formulas are recursively defined as follows:
1. T and F are propositional formulas.
2. All atoms are propositional formulas.
3. If f1 and f2 are propositional formulas, so are
f1 , f1 f2 , f1 f2 , f1 f2 , f1 f2 .
c Chung-Chih Li, Kishan Mehrotra

2.1. Definitions

41

Only the formulas generated by the rules above are propositional formulas.
Propositional formulas are also known as well-formed-formulas, or wffs in
short.
Example 2.2
pq, pq,
p q, p q,

(p q) r
(p q) r

are not wffs.

are wffs.

We are allowed to use parentheses with their natural purposes.

Definition 2.7: The operational priorities of logical connectives are defined below:
Precedence : {} > {, } > {, }.
In other words, is evaluated before and , and and
are evaluated before and . For example,
a b c a (b (c)).
The equivalence symbol above means: a bc is to be interpreted
as a (b (c)), and a (b (c)) can be abbreviated as a b c.
We can alternatively use one of them without introducing ambiguity.
Associativity : and are left associative; and are right associative.
For example,
a b c (a b) c,
a b c d a (b (c d)).
Definition 2.8: Let f be a wff in n variables. The truth table of f is a table that
contains all possible values of the variables in the rows (each row represents
one possibility), and the corresponding values of f are in the last column.
The tables in definition 2.4 are the truth tables of p, p q, p q, p
q, and p q, respectively. For complicated formulas, we can add some
intermediate columns to help us find the truth values of f .
Example 2.3

Let f = (p q) r. The truth table of f is

p
T
T
T
T
F
F
F
F

q
T
T
F
F
T
T
F
F

r
T
F
T
F
T
F
T
F

pq
T
T
F
F
F
F
F
F

(p q) r
T
F
T
T
T
T
T
T

We have added a column to present the truth values of p q. The 3rd row,
for example, indicates that if p = T, q = F, and r = T , then f is T .
c Chung-Chih Li, Kishan Mehrotra

42

2. Logic

Definition 2.9: Let f be a wff. f is called a tautology if and only if f is T

everywhere in the last column of its truth table.
Definition 2.10: Let f be a wff. f is called a contradiction if and only if f is
F everywhere in the last column of its truth table.
Definition 2.11: Let f1 and f2 be two wffs. Define
f1 f2 if and only if f1 f2 is a tautology.
We say f1 logically implies f2 .
Definition 2.12: Let f1 and f2 be two wffs. Define
f1 f2 if and only if f1 f2 is a tautology.
We say f1 and f2 are logically equivalent.
Definition 2.13: Let f be a wff. If the logical connectives that f contains are
and only, then the dual of f , denoted as f d , is a wff obtained from f
by the following rules: In f ,
1. replace T by F , and F by T ,
2. replace by ,
3. replace by .
Definition 2.14: Let f be a wff in n variables. The Disjunctive Normal Form,
DNF in short, of f is a logical equivalence of f , which is a disjunction of one
or more distinct (x1 x2 xn ), where xi , 1 i n, is a propositional
variable of f or its negation. Each xi is known as a literal.
Example 2.4

Let f = p q. The DNF of f is

(p q) (p q) (p q).

(2.1)

Because, (a) both f and the proposition (2.1) have the same truth values for
all possible values of p and q, and (b) the proposition (2.1) is a disjunction
of distinct formulas using the connective between literals.
Definition 2.15: Let f be a wff in n variables. The Conjunctive Normal Form,
CNF in short, of f is a logical equivalence of f , which is a conjunction of
one or more distinct (x1 x2 xn ), where xi , 1 i n, is a literal.
Example 2.5

Let f := p q. The CNF of f is

(p q).

c Chung-Chih Li, Kishan Mehrotra

2.1. Definitions

2.1.2

43

Predicate Logic

Definition 2.16: A predicate P (x1 , x2 , , xn ), n 0 is a mapping from the

concerned domains to {T, F }. In other words, if we replace x1 , , xn by
instances in their corresponding domains, we will get a proposition.
Example 2.6
P (x, y) :
Q(x, y) :

The following are all predicates.

x y 2 , Dx = Dy = N;
x is a y, Dx is the set of names,
Dy = {father, son, cat, dog, table,. . . }

We use Dx , Dy to denote the domains of x and y, respectively. Note that

Dx and Dy do not have to be the same. In the above example, P (3, 2)
is the proposition 3 22 with truth value F . Similarly, Q(Boo, dog) is a
proposition with truth value T if there is a dog named Boo.
Note: Any proposition is a predicate in zero variables.
Definition 2.17: Let P (x1 , x2 , , xn ), n 0 be a predicate. The set Dx1
Dx2 Dxn is called the domain of predicate P .
Definition 2.18: Let P (x1 , x2 , , xn ), n 0 be a predicate. The truth set of
P , denoted as TP , is a subset of Dx1 Dx2 Dxn such that for all
(a1 , a2 , . . . , an ) TP , P (a1 , a2 , . . . , an ) = T .
Definition 2.19: Let P (x1 , x2 , , xn ), n 0 be a predicate. The falsity set of
P , denoted as FP , is a subset of Dx1 Dx2 Dxn such that for all
(a1 , a2 , . . . , an ) FP , P (a1 , a2 , . . . , an ) = F .
Definition 2.20: Let P, Q be two predicates. We say that P and Q are logically
equivalent if and only if Tp = TQ .
Definition 2.21: Let P, Q be two predicates. We say that P logically implies
Q if and only if Tp TQ .
In addition to the connectives discussed in the previous section, we have two
special symbols called quantifiers in predicate logic. The two special symbols
are and . The quantifiers and are called the universal quantifier and the
existential quantifier and pronounced as for all and there exists respectively.
The meaning of and are defined below.
Definition 2.22: Let P (x) be a predicate in one variable.
1. Universal quantifier: x Dx P (x) is a proposition, and its value is T
if every instance a in Dx makes P (a) true.
2. Existential quantifier: x Dx P (x) is a proposition, and its value is
T if there exists some instance a in Dx that makes P (a) true.
c Chung-Chih Li, Kishan Mehrotra

44

2. Logic
If the domain Dx is clear from the context, we usually drop the domain
set and rewrite x Dx P (x) as xP (x) and x Dx P (x) as xP (x).
A predicate preceded with one or more quantifiers is called a quantified
predicate.

Definition 2.23: Let P (x1 , . . . , xi , . . . , xn ) be a predicate in n variables. Then,

xi P (x1 , . . . , xi , . . . , xn ) and xi P (x1 , . . . , xi , . . . , xn ) are predicates in n 1
variables, where xi is called a bounded variable and the rest of the variables
are called unbounded variables or free variables.
Example 2.7
Given the predicate P (x, y) in two variables x and y,
xP (x, y) and xP (x, y) are predicates in one variable y. The variable y
is a free variable, and x is a bounded variable in xP (x, y) and xP (x, y).

2.1.3

Predicates and Sets

Let P (x) and Q(x) be two predicates in one variable, and let Dx be the universe
of x. Then the associated truth and falsity sets satisfy the following properties.
1. TP FP = Dx .
2. TP FP = .
3. TP Q = TP TQ .
4. TP Q = TP TQ .
5. x[P (x) Q(x)] TP TQ .
6. x[P (x) Q(x)] (TP TQ ) 6= .

2.2

Logical Proof

Logical proof is a formal way of convincing that some statements are correct
based on given facts.
Let P and Q be two wffs, predicates, or quantified predicates. Starting from
P , if we can find a sequence of applications of laws of logic, rules of inference, or
tautologies to arrive at Q, step by step, we say there is a proof for the theorem
P = Q.
In theorem P Q, P is called the premise, and Q is called a logical conclusion
of P . The sequence of these steps is called a logical proof of P Q. A logical
proof is usually represented in a table like the following one.
c Chung-Chih Li, Kishan Mehrotra

2.2. Logical Proof

45

Premises:
Conclusion:

P
Q

1.
2.
3.
..
.

steps
P
q2
q3
..
.

reasons
Premises

..
.

i.
..
.

qi
..
.

..
.

n.

where are the names of the logical rules.

In short, a logical proof can be written as
P q2 q3 qi . . . Q.
P and Q are not necessarily two single statements; they can be a set of wffs,
predicates, and quantified predicates. For example, a theorem may look like
{P1 , P2 , . . . , Pn } = {Q1 , Q2 , . . . , Qm }.

(2.2)

On the other hand, a punctuation symbol , is considered as a . Therefore,

the theorem in (2.2) can be expressed as
(P1 P2 . . . Pn ) = (Q1 Q2 . . . Qm ).

2.2.1

Laws of Logic

Let f1 f2 be a tautology, i.e., f1 f2 . During the course of deriving the

sequence of a logical proof, if one of f1 and f2 appears, then we can attach the
other one to the sequence of the logical proof.
For example, if we know that f1 f2 and already have
P q2 q3 f1 ,
then the proof above can be extended to
P q2 q3 f1 f2 .
The following is a list of tautologies, where p, q, and r are wffs. One can use
the definitions of connectives to prove that they are tautologies. We call them
Laws of Logic. In a logical proof we are allowed to use the laws of logic directly
without further proof.
c Chung-Chih Li, Kishan Mehrotra

46

2. Logic
1. Law of double negation:
p p.
2. Absorption Laws:
p (p q) p,
p (p q) p.
3. Idempotent Laws:
p p p,
p p p.
4. Inverse Laws:

p p F,
p p T.

5. Identity Laws:
p T p,
p F p.
6. Domination Laws:

p F F,
p T T.

7. Commutative Laws:

(p q) (q p),
(p q) (q p).

8. Associative Laws:
(p (q r)) ((p q) r),
(p (q r)) ((p q) r).
9. Distributive Laws:
(p (q r)) ((p q) (p r)),
(p (q r)) ((p q) (p r)).
10. Contrapositive Law:
(p q) (q p)
11. De Morgans Laws:
(p q) (p q),
(p q) (p q).
12. No specific name is given, but this law is one of the most frequently used
laws in logical proof.
(p q) (p q).
c Chung-Chih Li, Kishan Mehrotra

2.2. Logical Proof

2.2.2

47

Rules of Inference

Let f1 f2 be a tautology, i.e., f1 f2 . During the course of deriving the

sequence of a logical proof, if f1 appears, then we can attach f2 to the sequence
of the logical proof.
For example, if we know that f1 f2 , and already have
P q2 q3 f1 ,
then the proof above can be extended to
P q2 q3 f1 f2 .
Let p, q, r, and s be wffs. The following tautologies are called Rules of
Inference.
1. Modus Ponens:
(p (p q)) = q.
2. Modus Tollens:
(q (p q)) = p.
3. Law of the Syllogism:
((p q) (q r)) = (p r).
4. Rule of Disjunctive Syllogism:
((p q) q) = p.
5. Rule of Contradiction:
(p F ) = p.
6. Rule of Conjunctive Simplification:
(p q) = p.
7. Rule of Disjunctive Amplification:
p = (p q).
8. Proof by cases:
((p r) (q r)) ((p q) r)
c Chung-Chih Li, Kishan Mehrotra

48

2. Logic

2.2.3

Inference Rules for Quantified Predicates

Let P (x) and Q(x, y) be two predicates in one and two variables respectively.
Following are some basic inference rules for quantified predicates.
1. Negation:
xP (x) xP (x),
xP (x) xp(x).
2. -Conversion (changing the name of the bounded variable):
xP (x) yP (y),
xP (x) yP (y).
3. Reordering quantifiers of the same kind:
xyQ(x, y) yxQ(x, y),
xyQ(x, y) yxQ(x, y).
4. Universal Specification:
xP (x) = P (a), any a Dx .
5. Existential Specification:
xP (x) = P (a), some a Dx .
6. Universal Generalization:
(any a Dx , P (a) = T ) = xP (x).
7. Existential Generalization:
(some a Dx , P (a) = T ) = xP (x).
Comment: We do not have a standard notation to distinguish a between any
a in 6 and some a in 7. One should find a way to make them clear in
the proof without confusion. See problem 50 of this chapter.

2.3

DNF and CNF

In this section we introduce a systematic way to find the DNF and CNF of any
given wff. The DNF and CNF are perhaps not mathematically dignified for
presenting any wff, but they are essentially the underlying presentation inside
c Chung-Chih Li, Kishan Mehrotra

2.3. DNF and CNF

49

modern digital computers. The DNF and CNF provide a convenient apparatus
in applications of Artificial Intelligence, Logical Programming, and many other
areas of research. Therefore, it behooves us to pay more attention to the DNF
and CNF.
We will pedagogically discuss wffs in three variables. One can easily extend
the method to a more general case.
Definition 2.24: Let f be a wff in variables x1 , x2 and x3 , and let i {1, 2, 3}.
1. Each term xi or its complement xi is called a literal.
2. A term of the form y1 y2 y3 , where yi = xi or yi = xi is called a
fundamental conjunction.
3. A term of the form y1 y2 y3 , where yi = xi or yi = xi is called a
fundamental disjunction.
4. A representation of f in a disjunction of fundamental conjunctions is
called a disjunctive normal form (DNF) or sum-of-product from.
5. A representation of f in a conjunction of fundamental disjunctions is
called a conjunctive normal form (CNF) or product-of-sum from.

2.3.1

The DNF of a given wff

DNF:

D1 D2 Dn .

(2.3)

Let f be a wff in three variables a, b, and c. There are eight possible fundamental conjunctions for a D in (2.3). These fundamental conjunctions are also
known as the building blocks for f in DNF. Let di be the ith building block
defined in the follows.

d1 : a b c

d2 : a b c

d3 : a b c

d4 : a b c
Building Blocks for DNF in three variables.
d5 : a b c

d6 : a b c

d7 : a b c

d8 : a b c
We use x to denote x due to space consideration.
Step 1: Evaluate the truth value of each building block. It is convenient to use
a truth table as shown below, where the truth value associated with each
building block is evaluated and followed by the truth value of f .
c Chung-Chih Li, Kishan Mehrotra

50

2. Logic

a
T
T
T
..
.
..
.

b
T
T
F
..
.
..
.

c
T
F
T
..
.
..
.

abc abc abc

T
F
F
F
T
F
F
F
T
..
..
..
.
.
.
..
..
..
.
.
.

..
.

..

f (a, b, c)
T
F
T
..
.
..
.

Step 2: We check the last column and mark the rows in which f s value is T .
For each marked row, we find and mark the building block with value T in
the row. Note that for each row, there is exactly one building block with
value T , and for each column of a building block, there is exactly one row
with value T .
Step 3: Finally, the DNF is the disjunction of the building blocks marked in
step 2. In the above partial example, a part of the DNF is
f (a, b, c) = (a b c) (a b c)

2
Example 2.8

Find the DNF of (a b) c.

Let f (a, b, c) = (a b)c, and d1 , d2 , . . . , d8 be the building blocks we just

defined. Follow the rules we just described step by step, we have the following
table.
a
T
T
T
T
F
F
F
F

b
T
T
F
F
T
T
F
F

c
T
F
T
F
T
F
T
F

d1

T
F
F
F
F
F
F
F

d2
F
T
F
F
F
F
F
F

d3
F
F
T
F
F
F
F
F

d4
F
F
F
T
F
F
F
F

d5
F
F
F
F
T
F
F
F

d6
F
F
F
F
F
T
F
F

d7
F
F
F
F
F
F
T
F

d8
F
F
F
F
F
F
F
T

f
T
T
T
F
T
T
T
T

According to the rule in step 3, the DNF of f is

d1 d2 d3 d5 d6 d7 d8 .
I.e.,
(a b) c

(a b c) (a b c) (a b c) (a b c)
(a b c) (a b c) (a b c).

2
c Chung-Chih Li, Kishan Mehrotra

2.3. DNF and CNF

2.3.2

51

The CNF of a given wff

CNF: C1 C2 Cn .

Let f be a wff in 3 variables a, b, and c. There are 8 building blocks for f in

CNF. Let ci be the ith building block defined in the follows.

c1
c2
c3
c4
c5
c6
c7
c8

:
:
:
:
:
:
:
:

a
a
a
a
a
a
a
a

b
b
b
b
b
b
b
b

c
c
c
c
c
c
c
c

Building Blocks for CNF in 3 variables.

Step 1: We construct the following truth table. As we did for finding DNF,
each building block must occupy one column, and the last column contains
the truth values of f .

a
T
T
T
..
.
..
.
F

b
T
T
F
..
.
..
.
F

c
T
F
T
..
.
..
.
F

abc abc
T
T
T
T
T
T
..
..
.
.
..
..
.
.
F
T

..
.

..

abc
T
F
T
..
.
..
.
T

..
.
..
.

f (a, b, c)
T

F
T
..
.
..
.

Step 2: Check the last column and mark the rows in which f s value is F . For
each marked row, we find and mark the building block with value F in
the row. Note that for each row, there is exactly one building block with
value F , and there is exactly one row with value F in each building blocks
column.
Step 3: Finally, the CNF is the conjunction of the building blocks marked in
step 2. In the above partial example, a part of the CNF is
f (a, b, c) = (a b c) (a b c)

2
c Chung-Chih Li, Kishan Mehrotra

52

2. Logic

Example 2.9

Find the CNF of (a b) c.

Let, f (a, b, c) = (a b) c, and c1 , c2 , . . . , c8 be the building blocks we

just defined.
a
T
T
T
T
F
F
F
F

b
T
T
F
F
T
T
F
F

c
T
F
T
F
T
F
T
F

c1
T
T
T
T
T
T
T
F

c2
T
T
T
T
T
T
F
T

c3
T
T
T
T
T
F
T
T

c4
T
T
T
T
F
T
T
T

c5
T
T
T
F
T
T
T
T

c6
T
T
F
T
T
T
T
T

c7
T
F
T
T
T
T
T
T

c8
F
T
T
T
T
T
T
T

f
T
T
T
F
T
T
T
T

Therefore,
(a b) c = (a b c).

2
Example 2.10

Find the DNF and CNF of a (b c).

Let f = a (b c), and d1 , d2 , . . . , d8 be the building blocks for DNF and

c1 , c2 , . . . , c8 the building blocks for CNF as defined.
For DNF:
a
T
T
T
T
F
F
F
F

b
T
T
F
F
T
T
F
F

c
T
F
T
F
T
F
T
F

d1

T
F
F
F
F
F
F
F

d2
F
T
F
F
F
F
F
F

d3
F
F
T
F
F
F
F
F

d4
F
F
F
T
F
F
F
F

d5
F
F
F
F
T
F
F
F

d6
F
F
F
F
F
T
F
F

Therefore, the DNF of f is d1 d4 , i.e.,

(a b c) (a b c).
c Chung-Chih Li, Kishan Mehrotra

d7
F
F
F
F
F
F
T
F

d8
F
F
F
F
F
F
F
T

f
T
F
F
T
F
F
F
F

2.3. DNF and CNF

53

For CNF:
a
T
T
T
T
F
F
F
F

b
T
T
F
F
T
T
F
F

c
T
F
T
F
T
F
T
F

c1
T
T
T
T
T
T
T
F

c2
T
T
T
T
T
T
F
T

c3
T
T
T
T
T
F
T
T

c4
T
T
T
T
F
T
T
T

c5
T
T
T
F
T
T
T
T

c6
T
T
F
T
T
T
T
T

c7
T
F
T
T
T
T
T
T

c8
F
T
T
T
T
T
T
T

f
T
F
F
T
F
F
F
F

Therefore, the CNF of f is c1 c2 c3 c4 c6 c7 , i.e.,

(a b c) (a b c) (a b c) (a b c) (a b c) (a b c).

2
2.3.3

A shortcut to find the DNF and CNF

Given any wff f , we can find the DNF and CNF directly from its truth set and
falsity set. Let us assume that f has three variables a, b, and c, and let the
format of the truth set Tf and the falsity set Ff be Da Db Dc .
Step 1: Find out the truth set Tf and the falsity set Ff of f .
Step 2: For DNF, we will use the truth set Tf . The building blocks that should
appear in the DNF of f are those having truth value T if we apply some
elements in Tf to them. In other words, for each (ta , tb , tc ) Tf , we choose
(xa xb xc ) according to the following rules.
1. xa = a if ta = T .
2. xa = a if ta = F .
3. xb and xc are decided by the same principle above.
For CNF, we will use the falsity set Ff . The building blocks that should
appear in the CNF of f are those having truth value F if we apply some
elements in Ff to them. In other words, for each (ta , tb , tc ) Ff , we
choose (xa xb xc ) according to the following rules.
1. xa = a if ta = T .
2. xa = a if ta = F .
3. xb and xc are decided by the same principle above.
c Chung-Chih Li, Kishan Mehrotra

54

2. Logic

Basically, the shortcut method and the truth table addressed with building
blocks are essentially the same. Lets consider the following example.
Example 2.11
Find the DNF and CNF of (a b) c from its truth set
and falsity set directly.
Let f = (a b) c.
table.
a
T
T
T
T
F
F
F
F

We first find Tf and Ff by using the following truth

b
T
T
F
F
T
T
F
F

c
T
F
T
F
T
F
T
F

ab
T
T
F
F
F
F
T
T

(a b) c
T
F
F
T
F
T
T
F

We have
Tf = {(T, T, T ), (T, F, F ), (F, T, F ), (F, F, T )},
Ff = {(T, T, F ), (T, F, T ), (F, T, T ), (F, F, F )}.
For DNF of f , consider Tf . We apply the rule described on all four elements of
Tf to get the associated blocks.
(T, T, T )
(T, F, F )
(F, T, F )
(F, F, T )

= (a b c),
= (a b c),
= (a b c),
= (a b c).

Thus, the DNF of f is

(a b c) (a b c) (a b c) (a b c).
In a similar manner, we use Ff of f to find CNF.
(T, T, F ) = (a b c),
(T, F, T ) = (a b c),
(F, T, T ) = (a b c),
(F, F, F ) = (a b c).
Thus, the CNF of f is
(a b c) (a b c) (a b c) (a b c).

2
c Chung-Chih Li, Kishan Mehrotra

2.4. Problem

2.4

55

Problem

Problem 1:

Define sets
A = {1, . . . , 10}, B = {3, 7, 11, 12}, C = {0, 1, . . . , 20}.

Which of the following are propositions?

(1)
(2)
(3)
(4)
(5)
(6)
(7)
Problem 2:

Find the possible wffs f and g in the following truth table.

a
T
T
F
F

Problem 3:

1+1=3
(A B) C
AB
(8 + 22)3 /102
7A
(B C) 9
C is an infinite set

b
T
F
T
F

f
F
T
T
T

g
T
T
F
T

If p q is false, what is the truth value of

((p) q) (p q)?

Problem 4:

Construct the truth tables for the following:

1. (a T ) (F b)
2. (F a) (b F )
3. (a b) (a b)
Problem 5:

Which of the following is a tautology?

1. (a b) (a b),
2. (a b) (a b) (a b)
Problem 6:

Show that
(a b c) = (a b c),

but the converse (i.e., (a b c) (a b c)) is not true.

c Chung-Chih Li, Kishan Mehrotra

56

2. Logic

Problem 7: Let p, q, r denote the following statements about a triangle

ABC.
p : Triangle ABC is isosceles;
q : Triangle ABC is equilateral;
r : Triangle ABC is equiangular.
Translate each of the following into an English sentence.
1. q p
3. q r
5. r p

2. p q
4. p q

Problem 8: Let p, q, r denote primitive statements. Use truth tables to

prove the following logical equivalences.
1. p (q r) (p q) (p r)
2. [(p q) r] [(p r) (q r)]
Problem 9: Let p, q, r denote primitive statements. Use the laws of logic
to show that
[p (q r)] [(p q) r].
Problem 10: Let p, q, and r be primitive statements. Write the dual for
the following statements.
1. q p
2. p (q r)
3. p q
Problem 11:

Show that
((a b) c) ((a c) (b c)).

Problem 12: Work out the truth tables for modus ponens to show that it
is a logical implication but not an equivalence.
Problem 13:

Consider

Premises: If there was a ball game, then traveling was difficult. If they
arrived on time, then traveling was not difficult. They arrived on
time.
Conclusion: There was no ball game.
Determine whether the conclusion follows logically from the premises.
Explain by representing the statements symbolically and using rules of
inference.
c Chung-Chih Li, Kishan Mehrotra

2.4. Problem
Problem 14:

57
Consider

Premises: If Claghorn has wide support, then hell be asked to run for
the senate. If Claghorn yells Eureka in Iowa, he will not be asked
to run for the senate. Claghorn yells Eureka in Iowa.
Conclusion: Claghorn does not have wide support.
Determine whether the conclusion follows logically from the premises.
Explain by representing the statements symbolically and using rules of
inference.
Problem 15: Write the converse, inverse, contrapositive, and negation of
the following statement.
If Sandra finishes her work, she will go to the basketball game.
Problem 16:

Simplify
(p (r q q)) ((r t r) q).

Problem 17:

Simplify
(p (p q) (p q r)) ((p r t) t).

Problem 18:

The following is a logical proof for

(p (p q) (s r) (r q)) (s t).

Refer to the laws of logic and inference rule, and give reasons to justify
each step of the proof.
steps
1. p
2. p q
3. q
4. r q
5. q r
6. r
7. s r
8. s
9. s t
Problem 19:

reasons

The following is a logical proof for the inference:

Premises :

Conclusion :

(p q) r
r (s t)
s u
u t
p
c Chung-Chih Li, Kishan Mehrotra

58

2. Logic
Give a reason to justify each step of the proof.
steps
1. s u
2. u
3. u t
4. t
5. s
6. s t
7. r (s t)
8. (s t) r
9. (s t) r
10. t
11. (p q) r
12. r (p q)
13. r (p q)
14. p q
15. p

Problem 20:

reasons

The following is a logical proof for

((p q) (r s) (p r)) (q s).
steps
1. (q s)
2. q s
3. s
4. r s
5. r
6. p q
7. q
8. p
9. p r
10. r
11. r r
12. q s

reasons

1. Give a reason to justify each step of the proof. (Note: This is a

proof by contradiction.)
2. Give a direct proof.
Problem 21:

Prove that the following inference is valid.

Premises :

Conclusion :
c Chung-Chih Li, Kishan Mehrotra

p q
qr
r
p

2.4. Problem
Problem 22:

59
Show that the following premises are inconsistent.

1. If Jack misses many classes through illness, then he fails high school.
2. If Jack fails high school, then he is uneducated.
3. If Jack reads a lot of books, then he is not uneducated.
4. Jack misses many classes through illness and reads a lot of books.
Problem 23:

Let Dx := R. Which of the following are predicates?

(1)
(2)
(3)
(4)
(5)
(6)

Problem 24:

x2 + 1 < 0
x is odd
(x2 1)/(x + 1)
1+2=3
xN
sin2 x + cos2 x

Define
A = {x|x N, x 10};
B = {y; y N, y 15, y is even}.

Write two predicates, of which A B and B A are the truth sets

respectively.
Problem 25: Let P and Q be two predicates, and let their truth sets be
denoted as TP , TQ , and falsity sets be denoted as FP , FQ . Prove the
following identities.
1. TP TQ = TP Q
2. TP TQ = TP Q
3. FP FQ = FP Q
4. FP FQ = FP Q
Problem 26: What are the most general conditions on the truth sets TP
and TQ making
1. (P (x) Q(x)) = P (x) true
2. (P (x) Q(x)) = Q(x) true
Problem 27: Suppose the domain D for predicates P, Q, and S is {a, b, c}.
Express the following propositions without using quantifiers.
1. x P (x)
2. (x R(x)) (x S(x))
c Chung-Chih Li, Kishan Mehrotra

60

2. Logic

Problem 28:

Let Dx = Dy = {1, 2, 3, 4, 5}. Define the predicate P (x, y) as

P (x, y) := (y x) or (x + y > 6).

Find the truth sets of the following predicates:

1. P (x, y).
2. xP (x, y)
3. yP (x, y)
4. xP (x, y)
5. yP (x, y)
Problem 29:

Let V be the truth set of P (x, y). Thus, V Dx Dy .

1. Prove that the truth set of x Dx P (x, y) is the set of all second
coordinates of ordered pairs in V .
2. Prove that the truth set of x Dx P (x, y) is
{b; b Dy , Dx {b} V }.
Problem 30:

Let sets
Dx = {t; t R, 1 t 1} and
Dy = {r; r R, 0 r 1}

be the universes of x and y, respectively. Define

P (x, y) = (x + y 1) (y x 1);
Q(x, y) = x2 + y 2 1.
Prove that TP TQ , where TP is the truth set of P and TQ is the truth
set of Q.
Problem 31: Let A, B and S be sets. The following is a wrong proof to
claim that
S (A B) (S A or S B).
Wrong proof: If S (A B), then
x S x (A B),
x S (x A or x B),
(x S x A) or (x S x B),
S A or S B.

(1)
(2)
(3)
(4)

Point out which step is problematic and explain why.

[Hint: consider the definition of subset in term of quantified predicates.]
c Chung-Chih Li, Kishan Mehrotra

2.4. Problem
Problem 32:

61
Let P (x, y) be a predicate defined as
P (x, y) : (x y) z.

Express the negation of xy P (x, y) without in front of any quantifier.

Problem 33: Find the negations of the following two quantified predicates
without in front of any quantifier.
1. x y [(x > y) (x y > 0)].
2. x y [(x < y) z(x < z < y)].
Problem 34: Take P 1 through P 7 as premises. See what conclusion you
can logically derive. Explain.
P 1 : All the policemen on this beat eat with our cook.
P 2 : No man with long hair can fail to be a poet.
P 3 : Amos Judd has never been in prison.
P 4 : Our cooks cousins all love cold mutton.
P 5 : None but policemen on this beat are poets.
P 6 : None but her cousins ever eat with the cook.
P 7 : Men with short hair have all been to prison.
Problem 35: Consider No pigs have wings. Write this proposition as a
quantified predicate.
Problem 36:

Consider

Premises:
All soldiers can march.
Some babies are not soldiers.
Conclusion:
Some babies cannot march.
Determine whether the conclusion follows logically from the premises.
Explain.
Problem 37: Let Dx = N and Dy = N0 . Define P (x, y) as x divides y.
Find the truth values of the following quantified predicates.
1. yP (1, y)
2. xP (x, 0)
3. xP (x, x)
4. yxP (x, y)
5. yxP (x, y)
c Chung-Chih Li, Kishan Mehrotra

62

2. Logic
6. xy[(P (x, y) P (y, x)) (x = y)]
7. xyx[(P (x, y) P (y, x)) P (x, z)]

Problem 38: Let Dx and Dy denote the domains of x and y, respectively.

Consider the following quantified statement
x y [x + y = 17].
Determine the truth value of the quantified predicate in different domains.
1. Dx = Dy = the set of integers.
2. Dx = Dy = the set of positive integers.
3. Dx = the set of integers and Dy = the set of positive integers.
4. Dx = the set of positive integers and Dy = the set of integers.
Problem 39:

What is the DNF of

(a b) (a b)?

Problem 40:

What is the DNF of

(a b) (a b)?

Problem 41:

Find the CNFs of

1. a b.
2. (a b) c.
Problem 42:

Let the wff f be a (b c).

1. Use the shortcut method (use the falsity set) to find the CNF of f .
2. Use the propositional calculus to find the DNF of f .
Problem 43:
theory:

Consider the following mathematical statement in number

For every integer n bigger than 1, there is a prime strictly between n

and 2n.
1. Express the statement in terms of quantifiers, variable(s), predicates,
and the inequality symbols < or >.
2. Express the negation of the predicate found in 1 without using .
[Be careful to define the domain(s) of your variable(s)]
Problem 44: Let x and y range over all integers. Prove that for all x, y, if
xy is even, then at least one of x and y is even.
c Chung-Chih Li, Kishan Mehrotra

2.4. Problem
Problem 45:

63
Are the following arguments logically correct?

Premises:
All who are anxious to learn work hard.
Some of these boys work hard.
Conclusion:
Therefore some of these boys are anxious to learn.
Problem 46:

Are the following arguments logically correct?

Premises:
There are men who are soldiers.
All soldiers are strong.
All soldiers are brave.
Conclusion:
Therefore some strong men are brave.
Problem 47: Let the universe be a social club, and let x and y range over
the members of the club. Define the predicate P (x, y) as
P (x, y) := x loves y.
Translate the following quantified predicates into English sentences
1. xy P (x, y)
2. xy P (x, y)
3. xy P (x, y)
4. xy P (x, y)
Problem 48: Let the domain range over all real numbers. Find a possible
conclusion from the given premises.
Premises:
All integers are rational numbers.
The real number is not a rational number.
Problem 49: Let the domain range over all people in the USA. Find a
possible premise for the following inference.
Premises:
All librarians know the Library of Congress Classification System.
(unknown premise)
Conclusion:
Margaret knows the Library of Congress Classification System.
Problem 50:

Let P and Q be two predicates. Prove that

x[P (x) Q(x)] xP (x) xQ(x).
c Chung-Chih Li, Kishan Mehrotra

64

2. Logic

Problem 51:

Let P and Q be two predicates. Prove that

x[P (x) Q(x)] xP (x) xQ(x).

Problem 52:

Let P and Q be two predicates. Prove that

xP (x) xQ(x) x[P (x) Q(x)].

Problem 53:

Let P and Q be two predicates. Disprove that

x[P (x) Q(x)] xP (x) xQ(x).

Problem 54:

Prove or disprove the following statement.

There exist a and b, where a and b are irrational and ab is rational.

[Hint: Use the fact that 2 is irrational; a and b dont have to be distinct.]

c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

2.5

65

Solutions

Solution 1:

(1), (2), (5), and (7) are propositions.

(3), (4) and (6) are not propositions.

Note: (6) (B C) 9 is not a proposition. It is meaningless, but that does

not mean its truth value is F .

Solution 2:
There are infinitely many wffs that satisfy the given truth
table. The simplest two we can think of are

f = (a b), and

g = b a.

Solution 3:
Let p q be false. The only case in which p q is false is
when p = T and q = F . We can replace the occurrences of p and q by their
values and find the result step by step as the following:

((p) q) (p q)
= ((T ) F ) (T F )
= (F F ) T
= F T
= F

Refer to the laws of logic to justify each step above.

c Chung-Chih Li, Kishan Mehrotra

66

2. Logic

Solution 4:
1. Let f (a, b) = (a T ) (F b)
a
T
T
F
F

b
T
F
T
F

T
T
T
T
T

F
F
F
F
F

aT
T
T
T
T

F b
T
T
T
T

F a
T
T
F
F

bF
F
F
F
F

f (a, b)
T
T
T
T

2. Let f (a, b) = (F a) (b F )
a
T
T
F
F

b
T
F
T
F

T
T
T
T
T

F
F
F
F
F

f (a, b)
F
F
T
T

3. Let f (a, b) = (a b) (a b).

a
T
T
F
F

b
T
F
T
F

b a b
F
T
T
T
F
T
T
F

a b
T
T
F
T

f (a, b)
T
T
F
F

Solution 5:
We will check the truth tables to tell whether the statements
are tautologies or not.
1. Let f = (a b) (a b).
a
T
T
F
F

b
T
F
T
F

ab ab
T
T
F
F
F
F
T
F

f
T
T
T
F

The last column of the truth table contains an F in the last row. Thus, f
is not a tautology.
c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

67

2. Let g = (a b) (a b) (a b).
Note: g should be read as (a b) ((a b) (a b)), but not as
((a b) (a b)) (a b), because the precedence priority of
is higher than the precedence priority of .
s = (a b) (a b).
a
T
T
F
F

b
T
F
T
F

a b a b a b s
T
T
F
T
F
F
F
F
F
F
F
F
T
F
T
T

g
T
T
T
T

Therefore, g is a tautology because every truth value in the last column

of its truth table is true.

Solution 6:

Let
p
q
r
s
a
T
T
T
T
F
F
F
F

=
=
=
=
b
T
T
F
F
T
T
F
F

a b c
a b c
(a b c) (a b c)
(a b c) (a b c)
c
T
F
T
F
T
F
T
F

ab
T
T
F
F
F
F
F
F

ab p
T
T
T
F
T
T
T
F
T
T
T
F
F
T
F
T

q
T
F
T
T
T
T
T
T

r
T
T
T
T
T
T
T
T

s
T
T
T
F
T
F
T
T

From the truth table above, r is a tautology and s is not. Therefore,

(a b c) = (a b c),
but its converse is not true.

c Chung-Chih Li, Kishan Mehrotra

68

2. Logic

Solution 7:

Let

p: Triangle ABC is isosceles;

q: Triangle ABC is equilateral;
r: Triangle ABC is equiangular.

1. q p : If triangle ABC is equilateral, then it is isosceles.

2. p q : If triangle ABC is not isosceles, then it is not equilateral.
3. q r : Triangle ABC is equilateral if and only if it is equiangular.
4. p q : Triangle ABC is isosceles, but not equilateral.
5. r p : If triangle ABC is equiangular, then it is isosceles.

Solution 8:
1. To prove that p (q r) (p q) (p r), let
s = p (q r),
t = (p q) (p r),
u = (p (q r)) ((p q) (p r)).
p
T
T
T
T
F
F
F
F

q
T
T
F
F
T
T
F
F

r
T
F
T
F
T
F
T
F

qr
T
F
F
F
T
F
F
F

pq
T
T
F
F
T
T
T
T

pr
T
F
T
F
T
T
T
T

s
T
F
F
F
T
T
T
T

t
T
F
F
F
T
T
T
T

u
T
T
T
T
T
T
T
T

Since the last column contains truth value T for all possible values of p.q,
and r, therefore, p (q r) and (p q)(p r) are logically equivalent.

2. To prove that ((p q) r) ((p r) (q r)), let

s = (p q) r,
t = (p r) (q r),
u = ((p q) r) ((p r) (q r)),
c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

69

and construct the truth table:

p
T
T
T
T
F
F
F
F

q
T
T
F
F
T
T
F
F

r
T
F
T
F
T
F
T
F

pq
T
T
T
T
T
T
F
F

pr
T
F
T
F
T
T
T
T

qr
T
F
T
T
T
F
T
T

s
T
F
T
F
T
F
T
T

t
T
F
T
F
T
F
T
T

u
T
T
T
T
T
T
T
T

Clearly, (p q) r and (p r) (q r) are logically equivalent.

Solution 9:

Using the laws of logic we obtain:

p (q r)
p (q r)
(p q) r
(p q) r
(p q) r
(p q) r
(p q) r

Logical equivalence
Associative law
Double negation law
De Morgans law
Double negation law
Logical equivalence

Therefore, [p (q r)] [(p q) r].

Solution 10:
From the definition of duality it is not possible to give the
dual of a logical statement that contains or . We have to find its
logical equivalent statements that contain no logical connectives other than
and .
1. Since q p q p, hence the dual of q p is q p.
2. Since p (q r) p (q r) its dual is p (q r).
3. Reduction of p q to a formula that contains connectives only , , and
c Chung-Chih Li, Kishan Mehrotra

70

2. Logic
is given below.
pq

(p q) (q p)
(p q) (q p)
[(p q) q] [(p q) p]
(p q) (q q) (p p) (q p)
(p q) F F (q p)
(p q) (q p).

Thus, the dual of p q is (p q) (q p).

Solution 11:
Our goal is to show that the values in the last column of the
truth table are all true. Let,
s = (a b) c,
t = (a c) (b c),
u = ((a b) c) ((a c) (b c)).
a
T
T
T
T
F
F
F
F

b
T
T
F
F
T
T
F
F

c
T
F
T
F
T
F
T
F

ab ac bc s
T
T
T
T
T
F
F
F
F
T
T
T
F
F
T
T
F
T
T
T
F
T
F
T
F
T
T
T
F
T
T
T

t
T
F
T
T
T
T
T
T

Therefore, u is a tautology.

Solution 12:

u
T
T
T
T
T
T
T
T

Recall that the Modus Ponens rule is: (a (a b)) b. Let,

s = (a (a b)) b,
t = (a (a b)) b.
a
T
T
F
F

b
T
F
T
F

a b a (a b)
T
T
F
F
T
F
T
F

c Chung-Chih Li, Kishan Mehrotra

s
T
T
T
T

t
T
T
F
T

2.5. Solutions

71

From the truth table above, we know that s is a tautology. Therefore,

(a (a b)) b.
But t is not a tautology, and hence the Modus Ponens is not a logical equivalence.

Solution 13:

Let

BG:
T D:
AO:

Premises: BG T D,

there was a ball game,

traveling was difficult,
they arrived on time.
AO T D,

AO.

Conclusion: BG.
steps
1. BG T D
2. AO T D
3. AO
4. T D
5. T D BG
6. BG

reasons
Premises
Premises
Premises
2,3, Modus Ponens
1, Contrapositive
4,5, Modus Ponens

Therefore, the conclusion that there was no ball game is logically correct
based on the given premises.
2

Solution 14:

Let

CS:
RS:
CY :

Premises: CS RS,

Claghorn has wide support,

Claghorn is asked to run for the senate,
Claghorn yells Eureka
CY RS,

CY.

Conclusion: CS.
steps
1. CS RS
2. CY RS
3. CY
4. CY (CY RS)
5. RS
6. RS CS
7. RS (RS CS)
8. CS

reasons
Premises
Premises
premises
23
4, Modus Ponens
1, Contrapositive
56
7, Modus Ponens
c Chung-Chih Li, Kishan Mehrotra

72

2. Logic

Therefore, Claghorn doesnt have wide support is logically correct from

the given premises.
2

Solution 15:

Let p :
q:

Sandra finishes her work.

Sandra goes to the basketball game.

Implication: (p q)
If Sandra finishes her work, she will go to the basketball game.

Converse: (q p)
If Sandra goes to the basketball game, she will finish her work.

Inverse: (p q)
If Sandra does not finish her work, she will not go to the basketball game.

Contrapositive: (q p)
If Sandra does not go to the basketball game, she does not finish her work.

Negation: (p q)
Sandra finishes her work, and she does not go to the basketball game.

Solution 16:
(p (r q q)) ((r t r) q)
(p (r T )) ((t T ) q)
(p T ) (T q)
p q.

2
c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

73

Solution 17:
(p (p q) (p q r)) ((p r t) t)

(p (p q T ) (p q r)) ((p r t) (T t)
(p (((p q) T ) ((p q) r))) (((p r) t) (T t)
(p ((p q) (T r)) (((p r) T ) t)
(p ((p q) T ) (T t)
(p (p q)) t)
((p T ) (p q)) t
(p (T q)) t
(p T ) t
pt

(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)

Explanation: In (1) identity law is used to have (p q) (p q T ) and

t (T t). Associative law is use is (2). Distributive law is used in (3).
Domination law is used is (4). Identity law is used in (5) and (6) to drop two
T s and add one T , respectively. Distributive law is used in (7). Domination
law is used is (8). Finally, we use identity law in the last step.
2

Solution 18:
To prove (p (p q) (s r) (r q)) (s t), we will
take p, p q, s r, and r q as the given assumptions.
steps
1. p
2. p q
3. q
4. r q
5. q r
6. r
7. s r
8. s
9. s t

reasons
Assumption
Assumption
1,2, Modus Ponens
Assumption
4, Contrapositive
3,5, Modus Ponens
Assumption
6,7, Disjunctive Syllogism
8, Disjunctive Amplification

Solution 19:
Premises :
(p q) r, r (s t), s u, u t
Conclusion : p
c Chung-Chih Li, Kishan Mehrotra

74

2. Logic
steps
1. s u
2. u
3. u t
4. t
5. s
6. s t
7. r (s t)
8. (s t) r
9. (s t) r
10. r
11. (p q) r
12. r (p q)
13. r (p q)
14. p q
15. p

reasons
Premises
1, Conjunctive Simplification
Premises
2,3, Modus Ponens
1, Conjunctive Simplification
4,5, Conjunction
Premises
7, Contrapositive
8, De Morgans law
6,9, Modus Ponens
Premises
11, Contrapositive
12, De Morgans law
10,13, Modus Ponens
14, Conjunctive Simplification

Solution 20:
((p q) (r s) (p r)) (q s)

(2.4)

1. The following proof for (2.4) is a contradiction argument. By contradiction, we assume that the premises and the negation of the consequence
are both true, i.e., we have
Assumptions:

p q, r s, p r, (q s).

steps
reasons
1. (q s)
Assumption, by Contradiction)
2. q s
Negation of implication
3. s
2, Conjunctive Simplification
4. r s
Assumption
5. r
3,4, Disjunctive syllogism
6. p q
Assumption
7. q
2, Conjunctive Simplification
8. p
6,7, Modus Tollens
9. p r
Assumption
10. r
8,9, Disjunctive syllogism
11. r r
5,10, Conjunction
12. q s
Since 11 is a contradiction, 1 cant be true

2
2. A direct proof for (2.4).
Assumptions:
c Chung-Chih Li, Kishan Mehrotra

p q, r s, p r.

2.5. Solutions

75
steps
1. p q
2. q p
3. p r
4. p r
5. q r
6. r s
7. r s
8. q s

reasons
Assumption
Equivalence of 1
Assumption
Equivalence of 3
2,4, Syllogism
Assumption
Equivalence of 6
5,7, Syllogism

Solution 21:
Premises :
Conclusion :

steps
1. p q
2. p q
3. q p
4. q r
5. r q
6. r
7. q
8. q p
9. p

p q, q r, r
p

4,
5,6
2
7,8

reasons
Premises
From 1
From 1
Premises
Contrapositive
Premises
Modus Ponens
Contrapositive
Modus Ponens

Solution 22:
We say a system (a set of premises) is inconsistent if and only
if we can obtain some results from the system that contradict each other.
Let

M C:
F S:
U E:
RB:

Premises :

Jack
Jack
Jack
Jack

misses many classes.

fails school.
is uneducated.
reads a lot of books.

M C F S, F S U E, RB EU, M C RB.
c Chung-Chih Li, Kishan Mehrotra

76

2. Logic
steps
1. M C RB
2. M C
3. M C F S
4. F S
5. F S U E
6. U E
7. RB U E
8. RB
9. U E
10. U E U E

reasons
Premises
1, Conjunctive Simplification
Premises
2,3, Modus Ponens
Premises
4,5, Modus Ponens
Premises
1, Conjunctive Simplification
7,8, Modus Ponens
6,9, Conjunction

We have both U E and U E, and hence the premises are inconsistent.

Solution 23:
Recall the definition of predicates: A statement is a predicate
if we can replace every variable in the statement by any instance in its domain
to form a proposition.
(1), (2), (4) and (5) are predicates. (3) and (6) are not predicates.

Note: A predicate may not have any variable. Therefore, all propositions are
also predicates, and hence (4) is a predicate.

Solution 24:
as

Let Dx = N be the universe of x. We find A B and B A

A B = {1, 3, 5, 7, 9} and B A = {12, 14}.

Define predicates P (x) and Q(x) as

P (x) = (x A) (x 6 B) and Q(x) = (x 6 A) (x B).
Thus, TP = A B and TQ = B A.
Note: One does not give, for example, {x : x N, 10 < x 15, x is even } as
the answer because it is the true-set of Q, but not a predicate with the
truth set B A.

2
c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

77

Solution 25:
1. TP TQ = TP Q
x TP TQ

x TP and x TQ
P (x) = T and Q(x) = T
(P Q)(x) = T
x TP Q .

2. TP TQ = TP Q
x TP TQ

x TP or x TQ
P (x) = T or Q(x) = T
(P Q)(x) = T
x TP Q .

x F (P ) and x F (Q)
P (x) = F and Q(x) = F
(P Q)(x) = F
(why?)
x FP Q .

x F (P ) or x F (Q)
P (x) = F or Q(x) = F
(P Q)(x) = F
(why?)
x FP Q .

3. F (P ) F (Q) = FP Q
x F (P ) F (Q)

4. F (P ) F (Q) = FP Q
x F (P ) F (Q)

Solution 26:

Let TP Q denote the truth set of P (x) Q(x).

1. (P (x) Q(x)) P (x).

[(P (x) Q(x)) P (x)]

TP Q TP
(FP TQ ) TP
(TP TQ ) TP
TP TP and TQ TP .

c Chung-Chih Li, Kishan Mehrotra

78

2. Logic
The only possibility to make TP TP is when TP = , namely, TP = U .
If TP = U , then TQ TP is satisfied for any TQ . Therefore, TP = U is
the most general condition for (P (x) Q(x)) P (x) to be true.
2. (P (x) Q(x)) Q(x).
[(P (x) Q(x)) Q(x)]

TP Q TQ
(FP TQ ) TQ
FP TQ and TQ TQ
FP TQ .

Therefore, FP TQ is the most general condition for (P (x) Q(x))

Q(x) to be true.
Note: The most general condition means the least restricted or the weakest
condition. For example, consider 1 < a < 10 and 1 < a. 1 < a is weaker than
1 < a < 10, hence 1 < a is more general than 1 < a < 10.
2

Solution 27:

Let Dx = {a, b, c}.

1. xP (x) = P (a) P (b) P (c).

2. (xR(x)) (xS(x)) = (R(a) R(b) R(c)) (S(a) S(b) S(c)).

Solution 28:
Let Dx = Dy = {1, 2, 3, 4, 5}, and P (x, y) := (y x)(x+y >
6). Thus, the truth set of P (x, y) is a subset of Dx Dy .
1. TP (x,y) = Tyx Tx+y>6 .

Tyx

Tx+y>6

(1, 1),

(1, 2), (2, 2),

(1, 3), (2, 3), (3, 3),
=

(1, 4), (2, 4), (3, 4), (4, 4),

(1, 5), (2, 5), (3, 5), (4, 5), (5, 5)

(5, 2),

(4, 3), (5, 3),

=
(3, 4), (4, 4), (5, 4),

(2, 5), (3, 5), (4, 5), (5, 5)

c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

79

Tyx Tx+y>6

(1, 1),

(5, 2),
(1, 2), (2, 2),
(1, 3), (2, 3), (3, 3), (4, 3), (5, 3),
=

(1, 4), (2, 4), (3, 4), (4, 4), (5, 4),

(1, 5), (2, 5), (3, 5), (4, 5), (5, 5)

2. TxP (x,y) = TP (1,y) TP (2,y) TP (3,y) TP (4,y) TP (5,y) .

TP (1,y)
TP (2,y)
TP (3,y)
TP (4,y)
TP (5,y)

=
=
=
=
=

{y
{y
{y
{y
{y

: (y
: (y
: (y
: (y
: (y

1) (y
2) (y
3) (y
4) (y
5) (y

> 5)} = {1, 2, 3, 4, 5},

> 4)} = {2, 3, 4, 5},
> 3)} = {3, 4, 5},
> 2)} = {3, 4, 5},
> 1)} = {2, 3, 4, 5}.

Therefore, TxP (x,y) = {1, 2, 3, 4, 5}.

3. TyP (x,y) = TP (x,1) TP (x,2) TP (x,3) TP (x,4) TP (x,5) .
TP (x,1)
TP (x,2)
TP (x,3)
TP (x,4)
TP (x,5)

=
=
=
=
=

{x : (1 x) (x > 5)} = {1},

{x : (2 x) (x > 4)} = {1, 2, 5},
{x : (3 x) (x > 3)} = {1, 2, 3, 4, 5},
{x : (4 x) (x > 2)} = {1, 2, 3, 4, 5},
{x : (5 x) (x > 1)} = {2, 3, 4, 5}.

Therefore, TyP (x,y) = {1, 2, 3, 4, 5}.

4. TxP (x,y) = TP (1,y) TP (2,y) TP (3,y) TP (4,y) TP (5,y) = {3, 4, 5}.
5. TyP (x,y) = TP (x,1) TP (x,2) TP (x,3) TP (x,4) TP (x,5) = {1}.

Solution 29:

Let V = TP (x,y) .

1. The set of all second coordinates of ordered pairs in V can be expressed

as S = {y : x, (x, y) V }. Given any b,
b TxP (x,y) xP (x, b)
x, (x, b) V
b S.
Therefore, TxP (x,y) = S.
c Chung-Chih Li, Kishan Mehrotra

80

2. Logic
2. Given any k,
k TxP (x,y)

xP (x, k)
x Dx , (x, k) TP (x,y)
x Dx , (x, k) V
Dx {k} V
k {b : b Dy , Dx {b} V }.

Therefore, TxDx P (x,y) = {b : b Dy , Dx {b} V }.

Solution 30:
Recall that TP TQ is equivalent to P Q. Suppose
x Dx , y Dy , and (x + y 1) (y x 1). We prove P Q in the
following.
1. From x + y 1:
steps
(1) x + y 1
(2) 1 x
(3) 0 y
(4) 1 x + y
(5) 1 x + y 1
(6) (x + y)2 1
(7) x2 + 2xy + y 2 1

reasons
given
x Dx
y Dy
(2) + (3)
(1) + (4)
math (5)
math (6)

steps
(1) y x 1
(2) 0 y
(3) 1 x
(4) 1 x
(5) 1 y x 1
(6) (y x)2 1
(7) x2 2xy + y 2 1

reasons
given
y Dy
x Dx
(3) 1
(2) + (4)
math (5)
math (6)

2. y x 1

From x + y 1 and y x 1 we have derived

x2 + 2xy + y 2 1,
2

x 2xy + y
c Chung-Chih Li, Kishan Mehrotra

1.

(2.5)
(2.6)

2.5. Solutions

81

Take (2.5) + (2.6), we have

2x2 + 2y 2 2

x2 + y 2 1.

Therefore, we conclude that x2 + y 2 1, i.e.

((x + y 1) (y x 1)) (x2 + y 2 1).

Solution 31:
Recall the definition of subset. We say that S is a subset of X
if and only if for all x, if x is in S, then x is in X. We may rewrite the definition
as
SX

iff

(x S) (x X).

Please note that we are using in the above definition instead of , i.e.,
S is a subset of X if and only if (x S x X) is a tautology. Therefore, it
is legitimate not to express for all explicitly in the second definition.1
In order to easily find the error in this problem, let us rewrite the definition
of subset without omitting the universal quantifier.
SX

Therefore, S (A B)

iff

x(x S x X).

iff

x(x S x (A B))
x(x S (x A or x B))
x[(x S x A) or (x S x B)]
6 x(x S x A) or x(x S x B)
(S A or S B).

(1)
(2)
(3)
(4)

Step (3) in the original proof is incorrect.

We may consider an easy example to see why (2) 6 (3). Let S, A, and B be
sets defined as the following.
1 Please bear in mind that, in mathematics, if the universe is clear or of no importance in
the discourse, it is not uncommon to discard the universal quantifier if it is the outermost
quantifier in a predicate.

c Chung-Chih Li, Kishan Mehrotra

82

2. Logic
S = all children, A = all boys, B = all girls.

It is clear that (2) is true, which means for all x, if x is a child, then x is either
a boy or a girl. However, (3) is not true, which means either all children are
boys, or all children are girls.
2

Solution 32:
xy[(x y) z] xy[(x y) z]
xy[(x y) z]
xy[(x y) z].

Solution 33:
To find the negations of the given formulas, we use the basic
rules and De Morgans laws. Recall that xp(x) xp(x) and xp(x)
xp(x) for any predicate p(x). Thus,
1. x y [(x > y) (x y > 0)]

x y [(x > y) ((x y) > 0)]

x y [(x > y) ((x y) > 0)]
x y [(x > y) ((x y) > 0)]
x y [(x > y) ((x y) 0)]
x y [(x > y) (x y)].

2. x y [(x < y) z(x < z < y)]

x y [(x < y) z (x < z < y)]

x y [(x < y) z (x < z < y)]
x y [(x < y) z (x < z < y)]
x y [(x < y) z (x < z < y)]
x y [(x < y) z ((z x) (y z))]
x y z [(x < y) ((z x) (y z))]

x y z [((x < y) (z x)) ((x < y) (y z))]

x y z [(z x < y) (x < y z)].

2
c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

83

Solution 34:
Let the domain Dx be the set of all people. Define the
following predicates.
P B(x) :
SC(x) :
M L(x):
P E(x) :
P R(x) :
CC(x) :
HC(x):
LC(x) :
a
:

x is a policeman in this beat.

x eats with our cook.
x is a man with long hair.
x is a poet.
x has been in prison.
x is our cooks cousin.
x is her cousin.
x loves cold mutton.
Amos Judd, who is a man.

Note: a is a constant in D. We dont think it is necessary to define another

predicate to test if x is a man. Thus, we simply assume Amos Judd is a man.
Now, we can rewrite the premises in term of the predicates above.

p1 :
p2 :
p3 :
p4 :
p5 :
p6 :
p7 :

x[P B(x) SC(x)].

x[M L(x) P E(x)]
P R(a).
x[CC(x) LC(x)].
x[P E(x) P B(x)].
x[SC(x) HC(x)].
x[M L(x) P R(x)].

And, we know that p2 is equivalent to the follows.

x[M L(x) P E(x)] x[M L(x) P E(x)]
x[M L(x) P E(x)]
x[M L(x) P E(x)].
Since a D and all predicates above except P 3 are quantified with universal
quantifiers, we can apply the universal specification rule and replace the variable
x by a to obtain propositions with truth values T as follows.
p1
p2
p3
p4
p5
p6
p7

:
:
:
:
:
:
:

P B(a) SC(a).
M L(a) P E(a).
P R(a).
CC(a) LC(a).
P E(a) P B(a).
SC(a) HC(a).
M L(a) P R(a).
c Chung-Chih Li, Kishan Mehrotra

84

2. Logic

We have

steps
1 P R(a) M L(a)
2 M L(a)
3 P E(a)
4 P B(a)
5 SC(a)
6 HC(a)

reasons
p7 , Contrapositive
1, p3 , Modus Ponens
2, p2 , Modus Ponens
3, p5 , Modus Ponens
4, p1 , Modus Ponens
5, p6 , Modus Ponens

Now, we can transfer the propositions back to English, which will tell us some
facts about Amos Judd.
M L(a):
P E(a) :
P B(a) :
SC(a) :
HC(a):

Amos
Amos
Amos
Amos
Amos

Judd
Judd
Judd
Judd
Judd

is a man with long hair.

is a poet.
is a policeman on this beat.
eats with our cook.
is her cousin.

We do not know if Amos Judd is our cooks cousin, and we do not know if
Amos Judd loves cold mutton.
2

Solution 35:

Let Dx be the set of all living things.

Define P (x) : x is a pig.

W (x) : x has wings.
The following predicates are equivalent.

x[P (x) W (x)]

x[P (x) W (x)]
x[P (x) W (x)]
x[P (x) W (x)].

We can choose any one of them as the answer.

Solution 36:

The following conclusion is logically incorrect.

Premises:

All soldiers can march.

Some babies are not soldiers.
Conclusion: Some babies cannot march.
To see this, let us define the following predicates,
S(x) : x is a soldier.
B(x) : x is a baby.
M (x) : x can march.
c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

85

We can restate our question as: Is the following inference valid?

x(S(x) M (x))
x(B(x) S(x))
x(B(x) M (x))

From the first premise and by the rule of universal specification we have, for
any a,
S(a) M (a),

but we cannot say

S(a) M (a)

because, unlike the contrapositive, the inverse is not an equivalence. Therefore,

although from the premises we know that there are some babies who are not
soldiers, we cannot use Modus Ponens to conclude that those babies cannot
march. In other words, if one is not a soldier, that doesnt mean one cannot
march. It is possible that all babies can march, while some of them are not
soldiers.
To see this more clearly, consider the following Venn diagram.

B
S
M

Where M is the set of all creatures that can march, B is the set of babies, and S
is the set of all soldiers. We interpret x S as S(x) is true, likewise for x M .
c Chung-Chih Li, Kishan Mehrotra

86

2. Logic

Recall S M x(x S x M ). In the Venn diagram S M ,

which gives the first premise: x(S(x) M (x)).
The diagram shows S B 6= . Therefore, there are some elements that are
in B but not in S, which is the second premise: x(B(x) S(x)).
However, the Venn diagram also shows that the conclusion is incorrect, because B is a subset of M , i.e., there is no baby that cant march.
2

Solution 37:
P as

Let Dx = N and Dy = N0 , and define two variable predicate

P (x, y) = x divides y.

1. yP (1, y) = T.
2. xP (x, 0) = T.
3. xP (x, x) = T.
4. yxP (x, y) = T.
Given any number y, there exists x, say x = 1, such that x divides y.
5. yxP (x, y) = T. Such a y is 0.
6. xy[(P (x, y) P (y, x)) (x = y)] = T.
Given any x and y, suppose that (P (x, y) P (y, x)) is true.
P (x, y) y = ax, a N0 ;
P (y, x) x = by, b N.
Thus, x = by = b(ax) = abx, and hence ab = 1. Because both a and b are
nonnegative integers, we know that a = b = 1. Therefore, x = y.
7. xyz[(P (x, y) P (y, z)) P (x, z)] = T.
Given any x, y, and z, suppose that (P (x, y) P (y, z)) is true.
P (x, y) y = ax, a N0 ;
P (y, z) z = by, b N0 .
Thus, z = by = b(ax) = abx. Since ab N0 , therefore, P (x, z) is true.

Solution 38:
Consider the quantified statement, x y [x + y = 17]. Let
Dx and Dy denote the universes of x and y, respectively.
c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

87

1. Dx = Dy = the set of integers.

x y [x + y = 17] = T rue.
In this case, given any integer x, we always can find one integer y = 17x,
such that x + y = 17.
2
2. Dx = Dy = the set of positive integers.
x y [x + y = 17] = F alse.
In this case, given any integer x > 17, we are not able to find another
positive integer y, such that x + y = 17.
2
3. Dx = the set of integers and Dy = the set of positive integers.
x y [x + y = 17] = F alse.
In this case, given any integer x > 17, we are not able to find another
positive integer y, such that x + y = 17.
2
4. Dx = the set of positive integers and Dy = the set of integers.
x y [x + y = 17] = T rue.
In this case, given any integer x, we always can find y = 17 x, such that
x + y = 17.

Solution 39:
of f first.

Let f (a, b) = (a b) (a b). Lets find the truth table

a
T
T
F
F

b
T
F
T
F

a b a b
T
T
F
T
T
F
T
T

f (a, b)
T
F
F
T

The building blocks for the DNF of a propositional formula with two variables
are (a b), (a b), (a b), and (a b).
a
T
T
F
F

b
T
F
T
F

ab ab

T
F
F
T
F
F
F
F

ab ab
F
F
F
F
T
F
F
T

f (a, b)
T
F
F
T

c Chung-Chih Li, Kishan Mehrotra

88

2. Logic

Therefore, f (a, b) = (a b) (a b).

Solution 40:

Let f (a, b) = (a b) (a b). The truth table of f :

a
T
T
F
F

b
T
F
T
F

ab
T
F
T
T

a b
F
T
T
T

f (a, b)
F
F
T
T

The associated truth table for the building blocks is:

a
T
T
F
F

b
T
F
T
F

ab ab
T
F
F
T
F
F
F
F

ab ab
F
F
F
F
T
F
F
T

f (a, b)
F
F
T
T

Therefore, f (a, b) = (a b) (a b).

Solution 41:
Remark: Since using building blocks and the truth tables to
find the DNF or CNF is pretty mechanical, it should not be difficult to find the
DNF and CNF in this way. Let us solve this problem by using propositional
calculus. Sometimes the propositional calculus method is easier than the truth
table approach, sometimes it isnt.
1. a b = a b = (a b).
2. For (a b) c:
(a b) c = (a c) (b c)
= ((a c) F ) (F (b c))
= ((a c) (b b)) ((a a) (b c))
= ((a c b) (a c b)) ((a b c) (a b c))
= (a b c) (a b c) (a b c).
Note: Why can we insert F without changing the value of the formula of the
problem? How about the case?
2

Solution 42:
Let f = a (b c). We will use the shortcut method to find
the CNF of f and propositional calculus to find the DNF.
c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

89

1. Shortcut Method: First, we find the truth table for f .

a
T
T
T
T
F
F
F
F

b
T
T
F
F
T
T
F
F

c
T
F
T
F
T
F
T
F

bc
T
F
F
T
T
F
F
T

a (b c)
T
F
F
T
F
F
F
F

Consequently, the falsity set of f is:

Ff = {(T, T, F ), (T, F, T ), (F, T, T ), (F, T, F ), (F, F, T ), (F, F, F )}.

Thus, the CNF of f is

(a, b, c) (a, b, c) (a, b, c) (a, b, c) (a, b, c) (a, b, c).

2. Propositional Calculus:

a (b c) = a ((b c) (b c))
= (a (b c)) (a (b c))
= (a b c) (a b c).

The last formula above is the DNF of f .

2
c Chung-Chih Li, Kishan Mehrotra

90

2. Logic

Solution 43:
In general, if . . . then . . . can be translated into a logical implication (), and there is or there are should use an existential quantifier
(). We need a predicate for a property described. And, or, and all used
in English sentences are equal to , , and , respectively, in the corresponding
mathematical expressions. With this in mind, lets examine the statement from
the number theory:
For every integer n bigger than 1, there is a prime strictly
between n and 2n.
Lets move one step forward:
For all integers, if the integer n is bigger than 1, then there is
a number which is a prime and strictly between n and 2n.
Let Dx be the set of all integers, and define predicate P (x) as
P (x) : x is prime.
1. We can obtain a logical sentence from the statement above:
n[(n > 1) x(P (x) (n < x < 2n))].

(2.7)

2. The negation of (2.7) is:

n((n > 1) x(P (x) (n < x < 2n)))

n((n > 1) x(P (x) (n < x < 2n)))

n((n > 1) x(P (x) (n < x < 2n)))
n((n > 1) x(P (x) (n < x < 2n)))
n((n > 1) x(P (x) (n < x < 2n)))
n((n > 1) x(P (x) (n < x < 2n)))
n((n > 1) x(P (x) (n < x < 2n)))
n((n > 1) x(P (x) ((n < x) (x < 2n))))
n((n > 1) x(P (x) ((n < x) (x < 2n))))
n((n > 1) x(P (x) ((x n) (x 2n))))

Solution 44:

We will use two different methods to prove this result. The

c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

91

first method uses the contradiction argument, and the second one will check all
the possible cases to show that the result is valid for all of them.
Let Dx be the set of integers. Define
P (x) : x is even.
We can restate the given result as the predicate,
P (xy) (P (x) P (y)).
Method 1:

(2.8)

By way of contradiction, assume that

P (xy) (P (x) P (y)).

(2.9)

By De Morgans law, (2.9) is equivalent to

P (xy) P (x) P (y).

(2.10)

If x and y are not even, then they are odd and can be expressed as
x = 2k + 1, y = 2q + 1
for some integers k and q. Thus,
xy = (2k + 1)(2q + 1)
= 4kq + 2k + 2q + 1
= 2(2kq + k + q) + 1
Because k and q are both integers, we know that 2kq + k + q is an integer too. Thus, xy is not even, i.e., P (xy) is false. This contradicts our
assumption that P (xy) is true. Therefore, (2.8) is correct.
2
Method 2: Given any two integers, we have three cases: 1. both are even, 2.
both are odd, 3. one is odd and the other one is even. We want to show
that, in each case, (2.8) is correct.
1. Both are even: Let x = 2k, y = 2q for some integers k and q.
P (xy)

(P (x) P (y))
P (4kq) (P (2k) P (2q))
T (T T )
T T
T

2. Both are odd: Let x = 2k + 1, y = 2q + 1 for some integers k and q.

P (xy)

(P (x) P (y))
P (4kq + 2k + 2q + 1) (P (2k + 1) P (2q + 1))
F (F F )
F F
T
c Chung-Chih Li, Kishan Mehrotra

92

2. Logic
3. One odd and one even: Let x = 2k + 1, y = 2q for some integers k
and q.
P (xy) (P (x) P (y))
P (4kq + 2q) (P (2k + 1) P (2q))
T (F T )
T T
T
We have seen that, in all cases, (2.8) is true. Therefore, it is a correct
statement for all integers.

Solution 45:
Let the domain Dx be the set of all people. Define the
following predicates with variables over Dx .
P (x): x is anxious to learn.
Q(x): x works hard.
R(x): x is one of those boys.
Now we can rewrite the premises and the conclusion in terms of the predicates defined above.
P 1 : x[P (x) Q(x)]
P 2 : x[R(x) Q(x)]
C : x[R(x) P (x)]
The conclusion C is not logically correct from the premises. Consider the
following Venn diagram for the truth sets of predicates P, Q, and R.

Dx
TP
TR

TQ

TP TQ that satisfies P 1. The shaded area, TR TQ , that is not empty

satisfies P 2. TR TP = indicates that there is no instance in the universe Dx
such that both R(x) and P (x) are true. Therefore, the conclusion C does not
follow from the premises.
2

c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

93

Solution 46:
Let the domain Dx be the set of all people. Define the
following predicates with variables over Dx .
P (x):
Q(x):
R(x):
S(x) :

x
x
x
x

is
is
is
is

a man.
a soldier.
strong.
brave.

The premises and the conclusion in terms of the predicates defined above
are:
P1 :
P2 :
P3 :
C:

x[P (x) Q(x)]

x[Q(x) R(x)]
x[Q(x) S(x)]
x[P (x) R(x) S(x)]

We use two different approaches to solve this problem.

Method 1: Let TP , TQ , TR , and TS be the truth sets of the predicates defined
above. From the premises we can have the following facts:
P1
P2
P3

TP TQ 6=
TQ TR
TQ TS

Based on the facts above, we want to prove the conclusion that is equivalent to
(TP TR TS ) 6= .
From P 1, we know there is a nonempty set such that TP TQ = .
steps
1. TQ
2. TQ TR
3. TR
4. TQ TS
5. TS
6. TP
7. (TP TR TS )
8. 6=
9. (TP TR TS ) 6=

reasons
Definition of
P2
1, 2, Definition of
P3
1, 4, Definition of
Definition of
3, 5, 6, Definition of
P1
steps 7, 8

Since (TP TR TS ) =
6 , there must be at least one element a in the
universe (domain Dx ) such that (P (a) R(a) S(a)) is true. Thus, we
can conclude that
x(P (x) R(x) S(x)) = T,
c Chung-Chih Li, Kishan Mehrotra

94

2. Logic
i.e., conclusion C is correct.

Method 2: Proof by using the laws of logic and inference rules for quantified
predicate calculus.

steps
reasons
1. x[P (x) Q(x)]
P1
2. P (a) Q(a)
1, Existential Specification
3. x[Q(x) R(x)]
P2
4. Q(a) R(a)
3, Universal Specification
5. Q(a)
2, Conjunctive Simplification
6. R(a)
4, 5, Modus Ponens
7. x[Q(x) S(x)]
P3
8. Q(a) S(a)
7, Universal Specification
9. S(a)
5, 8Modus Ponens
10. P (a)
2, Conjunctive Simplification
11. P (a) R(a) S(a)
6, 9, 10, conjunction
12. x(P (x) R(x) S(x)) 11, Existential Generalization

2
c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

95

Solution 47:
1. All members love each other.
2. There are some members who love some of the other members.
3. All members love some members.
4. There are some members who love all of the other members.

Solution 48:
There are many alternatives for the conclusion that can be
derived from the given premises. Let us just write down three of the most
obvious, but not trivial, conclusions and see how to derive them logically from
the premises.
We first define some needed predicates and translate the English sentences
into logical formulas:
P (x) :
Q(x) :
R(x) :
Premises:

P1 :
P2 :

x is an integer.
x is a rational number.
x is a real number.
x[P (x) Q(x)]
R() Q()

steps
reasons
1. x(P (x) Q(x))
P1
2. R() Q()
P2
3. P () Q()
1, Universal Specification
4. Q() P ()
3, Contrapositive
5. Q()
2, Conjunctive Simplification
6. P ()
4, 5, Modus Ponens
7. R()
2, Conjunctive Simplification
8. R() P ()
6, 7, Conjunction
9. x(R(x) P (x)) 8, Existential Generalization
10. P () R()
6, Disjunctive Amplification
11. [P () R()]
10, De Morgans Law
12. x(P (x) R(x)) 11, Existential Generalization
13. x(P (x) R(x))
12, Logical Equivalence
In step 6, is not an integer.
c Chung-Chih Li, Kishan Mehrotra

96

2. Logic
In step 9, there is a real but not rational number.

In step 13, not all numbers are both integer and real.

Solution 49:

Let m denote Margaret. Define

P (x) :
Q(x) :

x is a librarian.
x knows the system.

We want to use the rule of universal specification and Modus Ponens. If the
unknown premise is Margaret is a librarian, then we can have the following
inference.
P1 :
P2 :
C:

x[P (x) Q(x)]

P (m)
Q(m)

steps
1. x(P (x) Q(x))
2. P (m)
3. P (m) Q(m)
4. Q(m)

reasons
P1
P2
1, Universal Specification
2, 3, Modus Ponens

Where P (m) means: Margaret is a librarian.

Solution 50:

Prove

x[P (x) Q(x)] xP (x) xQ(x).

For direction, assume x[P (x) Q(x)].

steps
reasons
1. x[P (x) Q(x)]
Assumption
2. P (a) Q(a)
Existential Specification, 1
3. xP (x) Q(a) Existential Generalization, 2
4. xP (x) xQ(x)Existential Generalization, 3
c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

97

For direction, assume xP (x) xQ(x),

steps
reasons
1. xP (x) xQ(x)
Assumption
2. P (a) xQ(x)
1, Existential Specification
3. P (a) Q(b)
2, Existential Specification
4. [P (a) Q(b)] [P (b) Q(a)]
3, Disjunctive Amplification
5. [P (a) Q(a)] [P (b) Q(b)]
4, Associative, Commutative
6. x[P (x) Q(x)] x[P (x) Q(x)] 5, Existential Generalization
7. x[P (x) Q(x)]
AAA
This completes the proof.
Note: Be very careful in step 3 of the proof of direction. We have to use two
different symbols, a and b, because they are specified from two different
existential quantifiers, and they may or may not be equal.

Solution 51:

Prove

x[P (x) Q(x)] xP (x) xQ(x).

For direction, by way of contradiction, assume

x[P (x) Q(x)] [xP (x) xQ(x)].
steps
reasons
1. [xP (x) xQ(x)]
Assumption
2. [P (a) xQ(x)]
1, Universal Specification
3. [P (a) Q(a)]
2, Universal Specification
4. x[P (x) Q(x)]
3, Existential Specification
5. x[P (x) Q(x)]
4, Logical Equivalence
The conclusion in step 5 contradicts our assumption.
Note: We choose the same a in steps 2 and 3.
For direction, by way of contradiction, assume
[xP (x) xQ(x)] x[P (x) Q(x)]
steps
1. x[P (x) Q(x)]
2. x[P (x) Q(x)]
3. [P (a) Q(a)]
4. P (a) Q(a)
5. xP (x) xQ(x)
6. xP (x) xQ(x)
7. [xP (x) xQ(x)]

reasons
Assumption
1, Logical Equivalence
2, Existential Specification
3, De Morgans law
4, Existential Generalization
5, Logical Equivalence
6, De Morgans law
c Chung-Chih Li, Kishan Mehrotra

98

2. Logic
Step 7 gives a conclusion that contradicts our assumption:

xP (x) xQ(x).

Therefore, both directions hold. This completes the proof.

Solution 52:
To prove xP (x) xQ(x) x[P (x) Q(x)], by contradiction, assume that

[xP (x) xQ(x)] x[P (x) Q(x)].

steps
reasons
1. x[P (x) Q(x)]
Assumption
2. x[P (x) Q(x)]
1, Logical Equivalence
3. [P (a) Q(a)]
2, Existential Specification
4. P (a) Q(a)
3, De Morgans law
5. P (a)
4, Conjunctive Simplification
6. xP (x)
5, Existential Generalization
7. xP (x)
6, Logical Equivalence
8. Q(a)
4, Conjunctive Simplification
9. xQ(x)
8, Existential Generalization
10. xQ(x)
9, Logical Equivalence
11. xP (x) xQ(x)
7, 10, Conjunction
12. [xP (x) xQ(x)]
11, De Morgans law

Step 12 gives a conclusion that contradicts our assumption:

xP (x) xQ(x).

This proves the theorem.

c Chung-Chih Li, Kishan Mehrotra

2.5. Solutions

99

Solution 53:
To disprove x[P (x) Q(x)] xP (x) xQ(x), we will
construct a counter example.
Let the universe be {a, b}, and let p, q be two predicates with the truth
values defined in the following table.
a
T
F

P (x)
Q(x)
x[P (x) Q(x)] =
=
=
=

b
F
T

[P (a) Q(a)] [P (b) Q(b)

[T F ] [F T ]
T T
T

xP (x) xQ(x) = [P (a) P (b)] [Q(a) Q(b)]

= F F
= F
Therefore, x[P (x) Q(x)] 6 xP (x) xQ(x).

Solution 54:
We already know that 2is irrational. Now, consider the

2
2
number 2 . We dont know whether 2
is irrational or not. But it is
certain that there are only two cases.

2
Case 1:
2 is rational. Let a = 2 and b = 2. Then both a and b are
irrational and ab is rational.
2

2
Case 2:
2 is irrational. Let a = 2 and b = 2.
b

a =

2
2

2 = 2.

Therefore, both a and b are irrational and ab is rational.

In both cases we can claim that there are a and b, where a and b are irrational
and ab is rational.
2

c Chung-Chih Li, Kishan Mehrotra

Chapter 3

Mathematical Induction
To develop the skill of correct thinking is in the first place
to learn what you have to disregard.
In order to go on, you have to leave out;
this is the essence of effective thinking.
Kurt Godel

3.1. Concepts

3.1

103

Concepts

Mathematical induction is one of the most important and powerful techniques

for verifying mathematical statements. Many complicated mathematical theorems about integers can be proved easily by mathematical induction. It is easy
because the frame of the proof is unique and the underlying idea of mathematical
induction is intuitively understandable.
Think of infinitely many dominoes lined up. The proof by mathematical
induction is like the domino effect illustrated in the following figure:
......

If the first domino falls, and for any n if

the fall of the nth domino would result in
the knockdown of the (n + 1)st domino,
then any domino in the line will fall.

3.1.1

.......

......

......

......

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

.....
..

.....

.....
..

Necessary Conditions of Using Mathematical Induction

In the domino analogue above, to be able to (1) line up the dominoes, and (2)
identify the first domino in the line are two necessary conditions for knocking
down all of the dominoes. The two conditions are exactly the same necessary
conditions for being able to make use of mathematical induction.
Suppose a given mathematical statement is about a property over the domain
S. If all elements in S can be well ordered, i.e., (1) we can enumerate all elements
in S one by one, and (2) we can identify the first element in the enumeration,
then mathematical induction is very likely to be the approach for verifying the
statement.
In particular, any well-ordered set S can be represented as
S = {s0 , s1 , s2 , . . .}.
In this representation, (1) for any element a in S, there is n N such that
a = sn (all dominoes have been lined up properly), and (2) s0 is identified,
which is called the least element (s0 serves as the first domino in the line to
fall.)
Example 3.1

Sets such as natural numbers, integers, even numbers, odd

c Chung-Chih Li, Kishan Mehrotra

104

3. Mathematical Induction

numbers, multiples of 5, etc., are typical well-ordered sets that are often encountered in mathematics. They can be represented as:
Natural numbers
Integers
Even numbers
Odd numbers
Multiples of 5

:
:
:
:
:

{1, 2, 3, 4, . . .};
{0, 1, 1, 2, 2, 3, 3, . . .};
{0, 2, 2, 4, 4, . . .};
{1, 1, 3, 3, 5, 5, . . .};
{0, 5, 5, 10, 10, . . .}.

For more advanced mathematical inductive proof, the domain set may be
rational numbers, prime numbers, or any other well-ordered sets.

3.1.2

The Underlying Theory of Mathematical Induction

The underlying theory of mathematical induction is very simple. It is nothing

but repeated application of the logical rule, Modus Ponens. Suppose
S = {s0 , s1 , s2 , . . .},
and we have proved that
[P (s0 ) = T ] and [n NP (sn ) P (sn+1 )].
Given any i N, we can claim that P (si ) = T by the following inferences.
steps
1. P (s0 )
2. P (s0 ) P (s1 )
3. P (s1 )
4. P (s1 ) P (s2 )
5. P (s2 )
6. P (s2 ) P (s3 )
7. P (s3 )
..
.
P (si1 )
P (si1 ) P (si )
P (si )

reasons
proved
proved
1,2, Modus Ponens
proved
3,4, Modus Ponens
proved
5,6, Modus Ponens
..
.
Modus Ponens
proved
Modus Ponens

Therefore, we can claim that for all a in S, P (a) is a true statement.

Comment: In most problems, to prove P (s0 ) = T is trivial. The main task
is to prove that P (sn ) P (sn+1 ) for any n.
c Chung-Chih Li, Kishan Mehrotra

3.1. Concepts

3.1.3

105

Mathematical Induction of the First Form (Weak

Induction)

Suppose that the universal set (domain) is the set of non-negative integers, and
P (n) is any property of non-negative integers. We wish to prove that P (n) is
true for all n = 0, 1, 2, . . .; i.e., nP (n). Then the following procedure can be
applied.
Step 1: Prove that P (0) is true. This step is known as the basis step,
and the proved result, P (0) = T , is called the basis of induction.
Step 2: Let n be an arbitrary fixed integer, and assume that P (n) is
true. This assumption is called the weak inductive hypothesis.
Step 3: Use the assumption in step 2 to prove that P (n + 1) is true.
This step is known as the inductive step.
If we can prove the basis in step 1 and the implication in step 3, then we can
claim that P (n) is true for all n = 0, 1, 2, . . .. This method of proof is known as
the mathematical induction of the first form or the weak induction.
Comment: The above proof procedure is an application of the following rule
of inference,
1. P (0)
2. n[P (n) P (n + 1)]
3. nP (n),
where the first assertion, P (0), is proved in step 1, and the second assertion, n[P (n) P (n + 1)], is shown by picking an arbitrary value of n
for P (n) in step 2 and by the implication proved in step 3.
Comment: One should not be confused by the statement in step 2 let n be
an arbitrary fixed integer, and assume that P (n) = T and the goal for
all n P (n) = T that we want to prove. The letter n in step 2 denotes
an instance in the domain, and the letter n in the statement nP (n) is a
variable that ranges over the domain. After step 2, the instance n is fixed.
The following modification makes the difference explicit, but it also makes
the proof awkward, and hence, for simplicity, most textbooks do not use
it.
Step 2: For the inductive hypothesis, we assume that
P (n) = T when n = i for some i in the domain.
Step 3: In the inductive step, we prove that
P (i) P (i + 1).
c Chung-Chih Li, Kishan Mehrotra

106

3. Mathematical Induction
Then we conclude that for all n in the domain, P (n) = T .

Comment: In some instances, a weak inductive hypothesis cannot provide

sufficient ground to prove that P (n + 1) is true in step 3. We need a
stronger inductive hypothesis that is introduced in the next subsection.

3.1.4

Mathematical Induction of the Second Form (Strong

Induction)

Suppose that the universal set (domain) is the set of non-negative integers, and
P (n) is any property of non-negative integers. We wish to prove that P (n) is
true for all n = 0, 1, 2, . . .; i.e., nP (n). The following procedure can be applied.
Step 1: Prove that P (0) is true. This step is known as the basis step,
and the proved result P (0) = T is called the basis of induction.
Step 2: Let n be an arbitrary fixed integer in the domain, and assume
that P (0), P (1), . . . , and P (n) are true. This assumption is called
the strong inductive hypothesis.
Step 3: Use the assumption to prove that P (n + 1) is true. This step
is known as the inductive step.
If we can prove the basis in step 1 and the implication in step 3, then we can
claim that that P (n) is true for all n = 0, 1, 2, . . .. This method of proof is known
as the mathematical induction of the second form, or the strong induction.
Comment: The above proof procedure is an application of the following rule
of inference,
1. P (0)
2. n[(P (0) P (1) P (n)) P (n + 1)]
3. nP (n)
Comment: Sometimes we prove that P (n) is true for n S, where S =
{s0 , s1 , s2 , . . .}, and s0 may not be equal to 0, or in some cases S may not
be a set of numbers at all. If that is the case, the basis step changes to:
Step 1: Show that P (s0 ) is true.
Comment: In some problems, one may claim that P (n) is true for all n
{0, 1, 2, . . .}, but the basis is not P (0), but P (1) or P (2). It is important
to recognize this property; otherwise, incorrect proofs are obtained. See
problem 17 in the problem section of this chapter and its note on page
133.
c Chung-Chih Li, Kishan Mehrotra

3.2. Mathematical Induction and Recursive Definition

3.2

107

Mathematical Induction and Recursive Definition

Mathematical induction and recursive definitions are intimately related. They

should be viewed as two sides of the same coin. Recursion is a very useful
apparatus for defining sets, functions,1 and the programming syntax.2

3.2.1

Recursive Definitions for Functions

Theoretically, any computable function can be defined recursively. In the following we take the point of view that the argument of the function f takes values in
the set of non-negative integers. In general, recursively defined functions have
an infinitely large domain set.
A function f can be recursively defined as follows:
1: Define the value of the function at a few points. For example, f (0), f (1)
are specified. Such values are called the initial values.
2: Define the value of the function at n + 1 in terms of f (0), f (1), . . . , f (n),
and n itself.
3: Write a closing statement that in most cases reads Steps 1 and 2 are the
only two steps that define the function f .
Comment: The essence of recursive definition is its simplicity that helps us to
understand the function being defined, but not its efficiency when we are
asked to actually find the value of the function. We prefer to compute a
function by using its closed-form formula instead of by using its recursive
definition directly. Mathematical induction has nothing to do with finding
a closed-form formula for a given function,3 but it is a powerful technique
for verifying that a given closed-form formula is a correct one for the
recursively defined function.
Example 3.2 Let f be a function taking a non-negative integer as its argument, and be recursively defined as follows.
1: f (0) = 0.
1 Basic

concepts associated with functions are considered in Chapter 5.

definition 2.6 in Chapter 2 is an example of recursive definitions for the syntax of
well-formed formulas.
3 Chapter 8 will discuss details about solving recurrence relations.
2 The

c Chung-Chih Li, Kishan Mehrotra

108

3. Mathematical Induction
2: For all n 0, f (n + 1) = f (n) + (n + 1).

We note that

f (0) = 0,
f (1) = f (0) + 1 = 0 + 1 = 1,
f (2) = f (1) + 2 = 0 + 1 + 2 = 3,
f (3) = f (2) + 3 = 0 + 1 + 2 + 3 = 6,
..
.

and conjecture that f (n) is the summation of the first n natural numbers, i.e.,
the closed-form of f is given as
f (n) =

n(n + 1)
, for all n 0.
2

(3.1)

We can use mathematical induction to prove that equality (3.1) is correct.4

Inductive Basis: The initial value of f serves as the basis of the induction.
In particular, we prove that f (0) = 0(0+1)
. Its clear that the equality
2
(3.1) holds when n = 0, and hence the basis holds.
Note: f (0) = 0 is given as a part of the definition of f , and we have
verified it by using formula (3.1).
Inductive Hypothesis:
(3.1) holds, i.e.,

Assume that given any fixed n 0, the equality

f (n) =

n(n + 1)
.
2

Note: We removed the universal quantifier in the equality (3.1), because

n is fixed after this step.
Inductive Step:

In this step, we need to prove that

f (n + 1) =

(n + 1)(n + 2)
.
2

We first use the recursive definition of f to have

f (n + 1) = f (n) + (n + 1).

(3.2)

4 We do not explicitly define a predicate to be proved true in its domain for this example.
Implicitly, the predicate is:
n(n + 1)
P (n) : f (n) =
,
2
0
and Dn = N . In general, if the predicate is clear from the problem context, we can omit
it to improve the compactness of the proof. Please compare the solutions to Problems 6 and
7. On the other hand, if the subject is not clear, a well-stated predicate and its domain can
help us to move the first step (see Problem 32.) We will explicitly define a predicate in most
solutions in Section 3.5.

c Chung-Chih Li, Kishan Mehrotra

3.2. Mathematical Induction and Recursive Definition

109

Then, we use the hypothesis to replace f (n) in (3.2) by

n(n+1)
.
2

We have

f (n + 1) = f (n) + (n + 1)
n(n + 1)
+ (n + 1)
2
n(n + 1) + 2(n + 1)
=
2
(n + 1)(n + 2)
.
=
2

Therefore, we can claim that for all n 0, f (n) =

given closed-form for f is correct.

3.2.2

n(n+1)
,
2

and hence the

Recursive Definitions for Sets and Structural Induction

We will introduce an important variant of mathematical induction called structural induction in this subsection. Let us first study how to define a set by using
recursive definitions. The idea is: (1) give a few initial elements for the set, and
(2) specify a few rules to construct new elements by using old elements that are
already in the set.
A set S can be recursively defined as follows:
1: Identify a few elements of the set S.
2: Explain how to obtain new elements of the set S from the old elements of
the set.
3: Write a closing statement that in most cases reads (1) and (2) are the
only two ways to generate elements of the desired set S.

Example 3.3
Let set S be a set of strings of 1s and 0s and be recursively
defined as follows:
1: 1 S, 100 S.
2: If s S, then 11s S.
3: If s S, then 00s S.
4: Nothing but strings generated according to rules 1, 2, and 3 are
elements in S.
The following table shows some elements in the set and the rules used to
c Chung-Chih Li, Kishan Mehrotra

110

3. Mathematical Induction

generate each element in the table, respectively.

s0
s1
s2
s3
s4
s5

elements
1
001
1100111
001100001
11001100001
00001100001

rules used
1
1, 3
1, 2, 3, 2
1, 3, 3, 2, 3
1, 3, 3, 2, 3, 2
1, 3, 3, 2, 3, 3

In the example above, given any element s S, according to the rules we

can construct two new elements from s, i.e., 11s (using rule 2) and 00s (using
rule 3). For example, if we know that s3 S, then we can apply rules 2 and 3 to
s3 to obtain 11s3 and 00s3 which are s4 and s5 , respectively, and they are also
to be included in S. On the other hand, 1111, for example, is not a member of
S because we cannot have 1111 by using the given rules.
Now, the question is, what is the relation between a recursively defined set
and mathematical induction? To answer this question, lets observe the elements
s0 , . . . , s5 of the set S obtained above. We find that every element has an even
number of 00 s and an odd number of 10 s. Is this observation correct in general.
How can we prove it? The best technique for proving this kind of problem is
mathematical induction. In general, if a set is recursively defined and we observe
that the recursive definition gives a common property shared by all elements in
the set, then we can use mathematical induction to prove the observation.
The difficulty is that the order of elements in the set S is no longer obvious
(although S is still a well-ordered set). To overcome this problem, we introduce
a variant of mathematical induction called structural induction. We do not
visually line up the elements in the set. Instead, we imagine that the elements in
the set are lined up according to the numbers of times the rules used to generate
them. For example, if s is somewhere in the line, then the next elements are
00s and 11s, because 00s and 11s are obtained from s by applying one of the
given rules one more time.
Therefore, if a given property P is claimed to be universal in the set S, then
we should be able to prove the implication:
s S [P (s) P (00s) P (11s)].
The steps of proof by structural induction are stated below. In general, let set
S be recursively defined as follows:
1. s0 , s1 , . . . , sk S.
2. If s S, then r0 (s), r1 (s), . . . , rl (2) S.
3. Only elements specified in 1 or generated by rules in 2 are elements
in S.
c Chung-Chih Li, Kishan Mehrotra

3.3. Nested Induction

111

We wish to prove that all elements in S have the property P , i.e.,

s S [P (s) = T ].
Then the following procedure can be applied.
Step 1: Prove that P (s0 ), P (s1 ), . . . , P (sk ) are all true. This step is
the basis step of structure induction.
Step 2: Let s be any arbitrary fixed element in the set S, and assume
that P (s) is true. This is the inductive hypothesis.
Step 3: Use the assumption to prove that P (r0 (s)), P (r1 (s)), . . . ,
P (rl (s)) are all true. This step is the inductive step.
If we can prove the basis in step 1 and the implication in step 3, then we
can claim that that P (s) is true for all s S.
This method is called structural induction, because we are examining the
structure of the elements in the set, where the structure is given by the rules.
Any non-initial element can be decomposed into a few small fragments, and its
property is the result of its fragments properties.
Example 3.4 Consider the set S defined in the example on page 109. Prove
by structural induction that every element in S has an even number of 00 s and
an odd number of 10 s.
Inductive Basis: It is clear that both strings 1 and 100 have an even
number of 00 s and an odd number of 10 s. Thus the inductive basis holds.
Inductive Hypothesis: Assume s S and s has an even number of 00 s
and an odd number of 10 s.
Inductive Step: It is clear that, if the assumption is true, then both 00s
and 11s have an even number of 00 s and an odd number of 10 s.
Therefore, the observation is correct.

3.3

Nested Induction

Nested induction is a special form of mathematical induction, by which we can

prove a mathematical statement that has more than one variables involved.
Nested induction is also known as double induction in case that the subject
mathematical statement has two variables. Consider the Ackerman function
c Chung-Chih Li, Kishan Mehrotra

112

3. Mathematical Induction

A : N N N defined as follows. For every m, n N,

(i) A(0, m) = m + 1;
(ii) A(n + 1, 0) = A(n, 1);
(iii) A(n + 1, m + 1) = A(n, A(n + 1, m)).
Theorem: For all m, n N, A(n, m) > m.
We will prove the theorem by nested induction directly before we give its
underlying theory to justify our steps. Let m and n range over N.
Basis: mA(0, m) > m. It plainly appears from definition (i).
Inductive Hypothesis: Fix n N, assume that
m[A(n, m) > m].
Inductive Steps: We want to prove that m[A(n + 1, m) > m].
For m = 0, we have A(n + 1, 0) = A(n, 1) by definition (ii), and by the
hypothesis, A(n, 1) > 1, Thus, A(n + 1, 0) > 1 > 0.
For m > 0, we have
A(n + 1, m) =
>

A(n, A(n + 1, m 1))

A(n + 1, m 1)

by def (iii)
by the hypothesis

A(n + 1, m)
=
>

A(n + 1, m 1) + 1
A(n, A(n + 1, m 2)) + 1
A(n + 1, m 2) + 1

by def (iii)
by the hypothesis

A(n + 1, m)
..
.

A(n + 1, m 2) + 2

A(n + 1, m)
=
=
>

A(n + 1, m m) + m
A(n + 1, 0) + m
A(n, 1) + m
1+m

by the hypothesis

Therefore, m[A(n + 1, m) > m].

3.3.1

The underlying logic of nested induction

We use the previous example, and define two-place and one-place predicates Q
and P , respectively, over natural numbers as follows.

Q(n, m) =
A(n, m) > m;

P (n)

= mQ(n, m).

c Chung-Chih Li, Kishan Mehrotra

3.3. Nested Induction

113

Thus, we can rewrite the theorem according the following logic equivalences.
nm[A(n, m) > m] nmQ(n, m) nP (n).
To prove nP (n) by mathematical induction, we apply the inference rule we
have been familiar with:
P (0)
n[P (n) P (n + 1)]
nP (n).
Use the definition of P , the inference above can be rewritten as:
mQ(0, m)
n[mQ(n, m) mQ(n + 1, m)]
nmQ(n, m).
Therefore, in the inductive proof we just shown, the inductive basis is mQ(0, m)
and the inductive hypothesis is mQ(n, m).

c Chung-Chih Li, Kishan Mehrotra

114

3. Mathematical Induction

3.4

Problems

Conventions for the rest of this chapter:

Unless we state otherwise, n ranges over N0 , i.e., n {0, 1, 2, 3, . . .}.
P
All indexing variables in
notation are integers.
TH stands for inductive hypothesis.
Problem 1:

The assertion
X

2k = 2n+1 ,

n 1

1kn

is incorrect. Find the mistake of the invalid proof in the following:

1. Assume that for a fixed n
X

2k = 2n+1 .

1kn

2. Add 2n+1 to both sides of the above equality. We have

X
2k + 2n+1 = 2n+1 + 2n+1
1kn

2k = 2 2n+1

1kn+1

2k = 2n+1+1 .

1kn+1

This proves the result.

Problem 2:

Consider the sequence a0 , a1 , a2 , . . ., defined as:

1: a0 = 2, and a1 = 3.
2: For n 2, define an = 2an1 an2 .
We are given the following assertion:
There is no element in the sequence equal to 1.
We present the following proof. What is wrong with the proof?
If we can prove that a0 6= 1 and the sequence is increasing, i.e., for all
n, an+1 > an , then we can claim that no element in the sequence can be
equal to 1.
c Chung-Chih Li, Kishan Mehrotra

3.4. Problems

115

Inductive Basis:

Its clear that a0 6= 1.

Inductive Hypothesis:
Inductive Step:
esis, we have

[The Basis Holds.]

Assume an+1 > an .

From the definition of the sequence and the hypothan+2 = 2an+1 an

> 2an+1 an+1 = an+1 .

Thus, an+2 > an+1 .

Problem 3:

[The Inductive Step Holds.]

Let n N, and consider

1 1! + 2 2! + 3 3! + + n n!.

1. Use the

(3.3)

notation to express the above expression.

2. Find the formula for it.

Problem 4:
tion.

Prove the formula obtained for (3.3) by mathematical induc-

Problem 5:

Let n N, and consider

1
1
1
1
+
+
+ +
.
1 4 4 7 7 10
(3n 2) (3n + 1)
P
notation to express the expression above.
1. Use the

(3.4)

2. Find the formula for it.

Problem 6:
tion.

Prove the formula obtained for (3.4) by mathematical induc-

Problem 7:

Prove that
X

k(k + 1) = n(n + 1)(n + 2)/3.

1kn

Problem 8:

Prove by mathematical induction that

X

3k =

0kn

Problem 9:

(3n+1 1)
.
2

Prove by mathematical induction that

X
1kn

k2 =

n(n + 1)(2n + 1)
.
6

(3.5)

c Chung-Chih Li, Kishan Mehrotra

116

3. Mathematical Induction

Problem 10:

Prove by mathematical induction that

X

k(k 1) =

1kn

Problem 11:

(3.6)

Prove by mathematical induction that

X

k(k 1)(k 2) =

1kn

Problem 12:
for

1
(n + 1)n(n 1).
3

1
(n + 1)n(n 1)(n 2).
4

(3.7)

Use the equalities (3.5), (3.6), and (3.7) to derive a formula

X
k3 .
(3.8)
1kn

Problem 13: Prove the closed form obtained for (3.8) by mathematical
induction .
Problem 14:

Prove by mathematical induction that for all positive odd n,

X

(2)k =

0kn

Problem 15:

Prove by mathematical induction that for all n 1,

X
1kn

Problem 16:

1
1
=1
.
k(k + 1)
n+1

Prove by mathematical induction that for all n 1,

X
1kn

Problem 17:

1
(1 2n+1 ).
3

2k + 1
1
=1
.
k 2 (k + 1)2
(n + 1)2

Prove the following by mathematical induction :

[
\
Ai =
Ai ,
1in

1in

where n is any integer and n 2, and A1 , . . . , An are any sets. This is a

generalized De Morgans law.
Problem 18: Let a and b be two integers. Prove by mathematical induction
that for all integers n 1, an bn is a multiple of a b.
Note that you are asked to prove this by mathematical induction, so you
cannot simply use the following fact:
an bn = (a b)(an1 + an2 b + an3 b2 + + abn2 + bn1 ).
c Chung-Chih Li, Kishan Mehrotra

3.4. Problems
Problem 19:

117
Let x be any real number and x > 1. Prove that for all n > 1
(1 + x)n > 1 + nx.

Problem 20:
that

Let x be any real number such that x > 0 and x 6= 1. Prove

x

x, xx , x(x ) , x(x

(xx )

,...

is an increasing sequence.
Problem 21: Prove by mathematical induction that if n is any positive odd
integer, then 1 + 3n is divisible by 4.
Problem 22: Prove that the sum of the cubes of any three consecutive
integers is divisible by 9.
Problem 23:

Prove by mathematical induction that for all n Z

1
if n is even,
(1)n =
1
if n is odd.

Z is the set of all integers.

Problem 24:

Let c(n) be a sequence of integers such that

c(0) = 1, c(1) = 1, c(2) = 3,

and for all n 1,

c(n + 2) = 3c(n + 1) 3c(n) + c(n 1).
Prove that for all n 0,
c(n) = n2 n + 1.
Problem 25:

Define b(n) as follows:

b(n + 2) + 2b(n + 1) + b(n) = 0, n 0,

and b(0) = 1 and b(1) = 1. Prove that

b(n) = (1 2n)(1)n , n 0.
Problem 26: Using the same definition of b(n) given in problem 25, prove
that if b(0) = 1 and b(1) = 3, then
b(n) = (1 + 2n)(1)n , n 0.
c Chung-Chih Li, Kishan Mehrotra

118

3. Mathematical Induction

Problem 27:

Let us consider a famous sequence called Fibonacci numbers:

f0 = 0,
f1 = 1,
fn = fn1 + fn2 , for n 2.

Prove by mathematical induction that, for all n 0,

!n
!n !
1
1+ 5
1 5
fn =

.
2
2
5
Problem 28: Let S be a set with n elements. Prove by mathematical induction that the total number of subsets of S having exactly two elements
is n(n 1)/2.
Problem 29: Let p(n) be the maximum number of intersection points of n
distinct lines in the plane. Prove by mathematical induction that for all
integers n 2, p(n) = n(n 1)/2.
q
Problem 30: Let x1 = 1 and xn+1 = x2n + x12 . Prove by mathematical
n
induction that for all n 1,

1 xn n.
Problem 31:

Define the harmonic numbers as

X 1
Hn =
, n 1.
i
1in

Prove by mathematical induction that

n
H2 n 1 +
,
2

n 0.

Problem 32: Let denote the empty string. Let A be any finite nonempty
set. A palindrome over A can be defined as a string that reads the same
forward as backward. For example, mom and dad are palindromes
over the set of English alphabets.
We define a set S as follows:
1. S
2. a A, a S
3. a A x S, axa S
4. All the elements in S must be generated by the rules above.
Prove by structural induction that S equals the set of all palindromes
over A.
c Chung-Chih Li, Kishan Mehrotra

3.4. Problems

119

Problem 33: Let set S be a set of strings of as and bs recursively defined

as follows:
1. a S, b S.
2. If S and S, then S.
3. Nothing but strings generated according to rules 1 and 2 are elements in S.
We also recursively define the reverse operation R on S as:
1. R(a) = a, and R(b) = b.
2. If S, then R(a) = R()a, and R(b) = R()b.
Prove by structural induction that for all , S,
R() = R()R().
Problem 34: Let S and R be as defined in the previous problem. Prove
that, for all S,
R(R()) = .
Problem 35: Let = {a, b} and be the null string. Define A by
the following rules.
1. A.
2. If A, then ab A.
3. If , A, then A.
4. Every A must come from a finite number of applications of rules
1, 2, or 3.
Prove by mathematical induction that, for every A, has equally
many as and bs.
Problem 36: Let A be defined as the previous problem. Prove by mathematical induction that, if A, then the number of as is equal to or
greater than the number of bs in every prefix of .
Note: A prefix of is or any initial segment of . For example, if
= aabbad, then , a, aa, aab, aabb, aabba, and aabbab are all possible
prefixes of .

c Chung-Chih Li, Kishan Mehrotra

120

3.5

3. Mathematical Induction

Solutions

Solution 1:
for n = 1,

The basis of the induction is not proved. It is incorrect, because

X

2k = 21 = 2,

1k1

which is not equal to 21+1 = 4.

Solution 2:
The proof is incorrect. The proof shows that a0 6= 1 and claims
that it is the basis of the induction. According to the hypothesis given in the
proof, the correct basis is a0 6= 1 and a1 > a0 .
To see the mistake more clearly, lets change the initial values of the sequence
to a0 = 3, and a1 = 2. We thus have a2 = 1. The inductive basis only proves
that a0 6= 1, but does not mention the problem that a0 > a1 .
2

Solution 3:
1.

1 1! + 2 2! + 3 3! + + n n! =

k k!.

1kn

2. In order to obtain a closed form for the sum above, we note that k =
(k + 1) 1 for any k. Thus,
1 1! + 2 2! + 3 3! + + n n!
= (2 1! 1!) + (3 2! 2!) + (4 3! 3!) + + ((n + 1) n! n!)
= (2! 1!) + (3! 2!) + (4! 3!) + + ((n + 1)! n!)
= (n + 1)! 1.
Therefore,

k k! = (n + 1)! 1.

(3.9)

1kn

c Chung-Chih Li, Kishan Mehrotra

3.5. Solutions

121

Solution 4:
Recall that our goal is to prove that equation (3.9) holds for
all values of n 1.
by mathematical induction:
Inductive Basis:

n = 1.
X

k k! = 1 1! = 1,

1k1

(1 + 1)! 1 = 1.
That means both sides of (3.9) are equal to 1.

[The Basis Holds.]

Inductive Hypothesis: Suppose

X
k k! = (n + 1)! 1.
1kn

Inductive Step:
X
k k! =
1kn+1

k k! + (n + 1) (n + 1)!

1kn

= (n + 1)! 1 + (n + 1) (n + 1)!
= (n + 2) (n + 1)! 1
= (n + 2)! 1.

[by IH]

[The Inductive Step Holds.]

Therefore, for all n N, (3.9) is correct.

Solution 5:
1.
1
1
1
1
+
+
+ +
=
1 4 4 7 7 10
(3n 2) (3n + 1)

X
1kn

1
.
(3k 2) (3k + 1)

2. In this problem we make use of the equality, for any k, 1 k n,

1
1
1 3k + 1 3k + 2
1

=
3 3k 2 3k + 1
3 (3k 2)(3k + 1)
1
=
.
(3k 2)(3k + 1)
c Chung-Chih Li, Kishan Mehrotra

122

3. Mathematical Induction
1
1
1
1
+
+
+ +
1 4 4 7 7 10
(3n 2) (3n + 1)
1 1 1
1 1 1
1 1
1
1
1
1
= ( ) + ( ) + ( ) + + (

)
3 1 4
3 4 7
3 7 10
3 3n 2 3n + 1

1 1 1 1 1 1
1
1
1
=
+ +
+ +

3 1 4 4 7 7 10
3n 2 3n + 1

1
1
1
.
=
3
3n + 1

Therefore, the closed form expression of the sum is

X
1kn

1
1
=
(3k 2) (3k + 1)
3

1
3n + 1

(3.10)

Solution 6:

Our goal is to prove (3.10) by mathematical induction.

Inductive Basis:

n = 1.
1
1
1

= ,
312 31+1
4

1
1
1
RHS =
1
= .
3
31+1
4
LHS =

[The Basis Holds.]

Inductive Hypothesis:
X
1kn

Suppose

1
1
=
(3k 2) (3k + 1)
3

c Chung-Chih Li, Kishan Mehrotra

1
3n + 1

3.5. Solutions

123

Inductive Step:
X
1kn+1

1
(3k 2) (3k + 1)

1
1
+
(3k 2) (3k + 1)
(3n + 1) (3n + 4)
1kn

1
1
1
=
1
+
[by IH]
3
3n + 1
(3n + 1) (3n + 4)

1
1
3
=
1
+
3
3n + 1 (3n + 1) (3n + 4)

1
3n + 4 3
=
1
3
(3n + 1) (3n + 4)

1
3n + 1
1
=
3
(3n + 1) (3n + 4)

1
1
=
1
3
3n + 4
=

[The Inductive Step Holds.]

Therefore, for all n N, (3.10) is correct.

Solution 7:

Let Dn = N, and 5
X
1
P (n)6 :
k(k + 1) = n(n + 1)(n + 2).
3
1kn

Inductive Basis:

n = 1.
12=

Therefore, P (1) = T.

1
(1 2 3).
3
[The Basis Holds.]

Inductive Hypothesis: Assume P (n) = T , i.e.

X
1
k(k + 1) = n(n + 1)(n + 2).
3
1kn

5 Beginning from this problem and onwards, we will explicitly define the predicate and its
domain.
6 Note: P (n) is a predicate; for any fixed value of n, its value is either T rue or F alse. It
does not represent the sum or its value on the other side of the equality, i.e.,
X
1
P (n) 6=
k(k + 1) and P (n) 6= n(n + 1)(n + 2).
3
1kn

c Chung-Chih Li, Kishan Mehrotra

124

3. Mathematical Induction
Inductive Step: We want to prove that P (n + 1) = T rue. In other
words, we want to prove that the following equality is correct:
X
1
k(k + 1) = (n + 1)(n + 2)(n + 3).
3
1kn+1

k(k + 1) =

1kn+1

k(k + 1) + (n + 1)((n + 1) + 1)

1kn

=
=
=
=
=
=

n(n + 1)(n + 2)
+ (n + 1)(n + 2)
3
n(n + 1)(n + 2)
+ (n + 1)(n + 2)
3
n(n + 1)(n + 2) + 3(n + 1)(n + 2)
3
(n + 1)(n + 2)(n + 3)
3
(n + 1)((n + 1) + 1)((n + 1) + 2)
3
1
(n + 1)(n + 2)(n + 3).
3

P (n + 1) = T.

[The Inductive Step Holds.]

Therefore, n Dn , P (n) is true.

Solution 8:

[by IH]

Let Dn = N, and
X

P (n) :

3k =

0kn

Inductive Basis:

n = 1.

(3n+1 1)
.
2

3k =30 + 31 = 4,

0k1

(31+1 1) 9 1
=
= 4.
2
2
Therefore, P (1) = T.

[The Basis Holds.]

Inductive Hypothesis:

Assume P (n) = T , i.e.,

X

0kn

c Chung-Chih Li, Kishan Mehrotra

3k =

(3n+1 1)
.
2

3.5. Solutions

125

Inductive Step:
to prove that

We want to prove that P (n + 1) is true, i.e., we want

X

3k =

0kn+1

(3n+2 1)
.
2

3k =

3k + 3n+1

0kn

0kn+1

n+1

+ 3n+1

[by IH]

3n+1 1 + 2 3n+1
2
3 3n+1 1
2
3n+1+1 1
2
3n+2 1
.
2

=
=
=
=
P (n + 1) = T.

[The Inductive Step Holds.]

Therefore, P (n) is true for all n in Dn .

Solution 9:

Let Dn = N, and
P (n) :

k2 =

1kn

Inductive Basis:

1
n(n + 1)(2n + 1).
6

n = 1.
12 =

1
123
6

Therefore, P (1) is true.

Inductive Hypothesis:
X
1ki

[The Basis Holds.]

Let i Dn , and assume P (i) = T , i.e.,
k2 =

1
i(i + 1)(2i + 1).
6
c Chung-Chih Li, Kishan Mehrotra

126

3. Mathematical Induction
Inductive Step:
X

Consider P (i + 1),

k 2 = 12 + 22 + + i2 + (i + 1)2

1ki+1

k 2 + (i + 1)2

1ki

=
=
=
=
=

1
i(i + 1)(2i + 1) + (i + 1)2
[by IH]
6
1
(i + 1)(i(2i + 1) + 6(i + 1))
6
1
(i + 1)(2i2 + 7i + 6)
6
1
(i + 1)(i + 2)(2i + 3))
6
1
(i + 1)((i + 1) + 1)(2(i + 1) + 1)).
6

P (i + 1) = T.

[The Inductive Step Holds.]

Therefore, n Dn , P (n) is T rue.

Solution 10:

Let Dn = N, and
P (n) :

k(k 1) =

1kn

Inductive Basis:

1
(n + 1)n(n 1).
3

n = 1.

1 (1 1) =

1
(1 + 1) 1 (1 1).
3

Therefore, P (1) = T rue.

Inductive Hypothesis:
X

[The Basis Holds.]

Let i Dn , and assume P (i) = T , i.e.,
k(k 1) =

1ki

c Chung-Chih Li, Kishan Mehrotra

1
(i + 1)i(i 1).
3

3.5. Solutions

127

Inductive Step:

Consider P (i + 1),

X
k(k 1) =
k(k 1) + (i + 1)((i + 1) 1)

X
1ki+1

1ki

1
(i + 1)i(i 1) + i(i + 1)
3
1
= (i + 1)i(i 1 + 3)
3
1
= (i + 1)i(i + 2).
3

P (i + 1) = T.

[by IH]

[The Inductive Step Holds.]

Therefore, n Dn , P (n) is T rue.

Solution 11:

Let Dn = N, and
X
1
P (n) :
k(k 1)(k 2) = (n + 1)n(n 1)(n 2).
4
1kn

Inductive Basis: For n = 1, the LHS of P (n) is 1(1 1)(1 2) = 0,

and the RHS is 41 (1 + 1)1(1 1)(1 2) = 0. Since both sides are equal,
P (1) = (0 = 0) = T.
[The Basis Holds.]
Inductive Hypothesis: Let i Dn , and assume P (i) = T , i.e.,
X
1
k(k 1)(k 2) = (i + 1)i(i 1)(i 2).
4
1ki

Inductive Step:
X

To prove that P (i + 1) is true, we need to show that

k(k 1)(k 2) =

1ki+1

1
(i + 2)(i + 1)i(i 1).
4

k(k1)(k 2)

1ki+1

k(k 1)(k 2) + (i + 1)i(i 1)

1ki

1
= (i + 1)i(i 1)(i 2) + (i + 1)i(i 1)
4
1
= (i + 1)i(i 1) ((i 2) + 4)
4
1
= (i + 1)i(i 1)(i + 2).
4

[by IH]

c Chung-Chih Li, Kishan Mehrotra

128

3. Mathematical Induction
P (i + 1) = T.

[The Inductive Step Holds.]

Therefore, n Dn , P (n) is T rue.

Solution 12:
Lets observe the following properties first. Let f (k) and g(k)
be two functions of k, and a be any constant. We have the following equalities.
X
X
X
(f (k) + g(k)) =
f (k) +
g(k).
1kn

1kn

1kn

af (k) = a

1kn

f (k)

1kn

One can verify that,

k 3 = k(k 1)(k 2) + 3k 2 2k.
= k(k 1)(k 2) + 2k(k 1) + k 2 .
Therefore,

k3 =

1kn

k(k 1)(k 2) + 2

1kn

k(k 1) +

1kn

k2 .

1kn

Then, we use the results from the previous problems to get,

X
1
2
1
k 3 = (n + 1)n(n 1)(n 2) + (n + 1)n(n 1) + n(n + 1)(2n + 1)
4
3
6
1kn

1
n(n + 1)[3(n 1)(n 2) + 8(n 1) + 2(2n + 1)]
12
1
=
n(n + 1)(3n2 9n + 6 + 8n 8 + 4n + 2)
12
1
=
n(n + 1)(3n + 3n2 )
12
1
= n2 (n + 1)2 .
4
=

Solution 13:

Let Dn = N, and
X
1
P (n) :
k 3 = n2 (n + 1)2 .
4
1kn

c Chung-Chih Li, Kishan Mehrotra

3.5. Solutions

129

Inductive Basis: n = 1. It is easy to verify that, for n = 1, both sides

of the equality are 1i, i.e.,
13 =

1 2
1 (1 + 1)2 .
4

Therefore, P (1) = T rue.

[The Basis Holds.]

Inductive Hypothesis:

Let i Dn , and assume P (i) = T , i.e.,

k3 =

1ki

Inductive Step:
X

1 2
i (i + 1)2 .
4

Consider P (i + 1),
X
k3 =
k 3 + (i + 1)3

1ki+1

1ki

=
=
=
=

1 2
i (i + 1)2 + (i + 1)3
4
1
(i + 1)2 (i2 + 4(i + 1))
4
1
(i + 1)2 (i2 + 4i + 4)
4
1
(i + 1)2 (i + 2)2 .
4

Thus, P (i + 1) = T.

[The Inductive Step Holds.]

Therefore, n Dn , P (n) = T.

Solution 14:

[by IH]

Let Dn = {1, 3, 5, 7, 9, . . .}, and define,

X

P (n) :

(2)k =

0kn

1
(1 2n+1 ).
3

Inductive Basis: n = 1. We choose 1 as the basis because 1 is the first

element of Dn in our enumeration, and note that the LHS of the equality
is
X
(2)k = (2)0 + (2)1 = 1,
0k1

and the RHS of the equality is

1
(1 22 ) = 1.
3
Therefore, P (1) = T.

[The Basis Holds.]

c Chung-Chih Li, Kishan Mehrotra

130

3. Mathematical Induction
Inductive Hypothesis: Let i Dn , and assume P (i) = T , i.e.,
X
1
(2)k = (1 2i+1 ).
3
0ki

Inductive Step: Let n = i + 2. We choose i + 2 because i + 2 is the

positive odd integer next to i.
X
X
X
(2)k =
(2)k +
(2)k
0ki+2

0ki

=
=
=
=
=
=
=
=

i+1ki+2

1
(1 2i+1 ) + (2)i+1 + (2)i+2
[by IH]
3
1
[1 2i+1 + 3(2)i+1 + 3(2)i+2 ]
3
1
[1 2i+1 + 3(2)i+1 + 3(2)(2)i+1 ]
3
1
[1 2i+1 + 3(2)i+1 6(2)i+1 ]
3
1
[1 (1 3 + 6) 2i+1 ]
3
1
(1 (2)2 (2)i+1 ) since i + 1 is even
3
1
(1 (2)i+3 )
3
1
(1 2i+3 ) since i + 3 is even.
3

P (i + 2) = T.

[The Inductive Step Holds.]

Therefore, n Dn , P (n) is true.

Solution 15:

Define P (n) for all n 1 as,

X
1
1
P (n) :
=1
.
k(k + 1)
n+1
1kn

Inductive Basis: For n = 1, it is easy to verify that both sides of the

equality are equal to 12 .
LHS =

X
1k1

RHS = 1

1
1
=
= 1/2.
k(k + 1)
12

1
= 1/2.
1+1

Therefore, P (1) is true.

c Chung-Chih Li, Kishan Mehrotra

[The Basis Holds.]

3.5. Solutions

131

Inductive Hypothesis:

Assume P (n) is true, i.e.,

X
1kn

Inductive Step:
X
1kn+1

1
1
=1
.
k(k + 1)
n+1

To prove P (n + 1) is true.

X
1
1
1
=
+
k(k + 1)
k(k + 1) (n + 1)(n + 2)
1kn

1
1
+
n + 1 (n + 1)(n + 2)
(n + 2) 1
=1
(n + 1)(n + 2)
n+1
=1
(n + 1)(n + 2)
1
=1
n+2
=1

Therefore, P (n + 1) is true.

[by IH]

[The Inductive Step Holds.]

Solution 16:

Define P (n) for all n 1 as,

P (n) :

X
1kn

Inductive Basis:

2k + 1
1
=1
.
k 2 (k + 1)2
(n + 1)2

n = 1.
X
1k1

2k + 1
2+1
= 2 2 = 3/4.
2
+ 1)
1 2

k 2 (k

1
= 3/4.
(1 + 1)2

Therefore, P (1) is true.

Inductive Hypothesis:
X
1kn

[The Basis Holds.]

Assume P (n) is true, i.e.
2k + 1
1
=1
.
2
+ 1)
(n + 1)2

k 2 (k

c Chung-Chih Li, Kishan Mehrotra

132

3. Mathematical Induction
Inductive Step:
X
1kn+1

To prove P (n + 1) is true.

X
2k + 1
2k + 1
2(n + 1) + 1
=
+
k 2 (k + 1)2
k 2 (k + 1)2
(n + 1)2 (n + 2)2
1kn

=1
=1
=1
=1
=1
=1

2(n + 1) + 1
1
+
2
(n + 1)
(n + 1)2 (n + 2)2
2
(n + 2) 2(n + 1) 1
(n + 1)2 (n + 2)2
2
n + 4n + 4 2n 2 1
(n + 1)2 (n + 2)2
2
n + 2n + 1
(n + 1)2 (n + 2)2
(n + 1)2
(n + 1)2 (n + 2)2
1
(n + 2)2

Therefore, P (n + 1) is true.

[by IH]

[The Inductive Step Holds.]

Solution 17:
Let Ai denote a set for any i N, and P (n) be the predicate,
for n {2, 3, 4, . . .},
\
[
P (n) :
Ai =
Ai .
1in

1in

Inductive Basis:

For n = 2, we need to prove that

A1 A2 = A1 A2 .

It is the classical De Morgans law in logic, and is known to be true. Hence,

the basis follows.
Inductive Hypothesis:

Assume n = k and P (k) = T, i.e.,

[
1ik

Inductive Step:

Ai =

Ai .

1ik

Let n = k + 1.
[

Ai = (

1ik+1

c Chung-Chih Li, Kishan Mehrotra

1ik

Ai ) Ak+1 .

3.5. Solutions
If we take (

133
S

Ai ) as A, and Ak+1 as B, we have

1ik

Ai ) Ak+1 = (

1ik

Ai ) Ak+1 .

1ik

Using the inductive hypothesis, we have

(

Ai ) Ak+1 = (

1ik

Ai ) Ak+1

1ik

Ai .

1ik+1

That proves P (k + 1) = T .

[The Inductive Step Holds.]

Therefore, n 2, P (n) is T .
Note: If we want to claim that [n 1, P (n)] is also true, we have to prove the
special case, n = 1, and the inductive basis, n = 2, separately. It is trivial
to prove P (1) is true, but P (1) = T is not the basis of the induction. We
have to prove P (2) = T as the inductive basis, because P (1) does not
imply P (2).

Solution 18:

Let Dn = N, and
P (n) : (a b)|(an bn ),

where (ab)|(an bn ) means an bn is divisible by ab. In other words, we have

to prove that for any fixed n, there exists an integer k such that an bn = a(ab).
Note that k may be a function of a, b, and n.
Inductive Basis: For n = 1, an bn = a b and it is obvious that
(a b) divides (a b), i.e.,
(a b)|(a1 b1 ) is true.
Therefore, P (1) = T rue.
Inductive Hypothesis:

[The Basis Holds.]

Let i Dn , i > 1, and assume P (i) = T , i.e.,
(a b)|(ai bi ),
c Chung-Chih Li, Kishan Mehrotra

134

3. Mathematical Induction
Inductive Step:

We want to know if ai+i bi+1 is divisible by a b.

ai+1 bi+1 = aai bai + bai bbi = ai (a b) + b(ai bi ).

From the inductive hypothesis, we know that there exists an integer k
such that,
(ai bi ) = k(a b).
Therefore,
ai+1 bi+1 = ai (a b) + bk(a b)
= (a b)(ai + bk).
It is clear that ai + bk is an integer because a, b, i, and k are all integers.
Thus,
(a b)|(ai+1 bi+1 ).
P (i + 1) is true.

[The Inductive Step Holds.]

Therefore, for all n 1, an bn is divisible by a b.

Solution 19:

Suppose x > 0, Dn = {2, 3, 4, . . .}, and

P (n) : (1 + x)n > 1 + nx.

Inductive Basis:

n = 2.

(1 + x)2 = 1 + 2x + x2 > 1 + 2x, because n2 > 0.

[The Basis Holds.]
Inductive Hypothesis:

Let i Dn , and assume P (i) = T , i.e.,

(1 + x)i > 1 + ix.

Inductive Step:

We want to prove that P (i + 1) is true.

(1 + x)i+1 = (1 + x)i (1 + x)
> (1 + ix)(1 + x)
= 1 + (i + 1)x + ix

[by IH] and x > 0

2

> 1 + (i + 1)x by n2 > 0, i > 0.

[The Inductive Step Holds.]
c Chung-Chih Li, Kishan Mehrotra

3.5. Solutions

135

Note: The given condition x > 0 and the inductive hypothesis are both needed
to claim that
(1 + x)i (1 + x) > (1 + ix)(1 + x).
One can easily find a counter example if x < 0.

Solution 20:

We express the sequence

x

x, xx , x(x ) , x(x

(xx )

,...

as
a1 , a2 , a3 , . . . , . . .
. Thus,
a1 = x,
a2 = xx = xa1 ,
x

a3 = x(x

= x a2 ,

In general, an = xan1 for n 2. Let Dn = N, and

P (n) : an < an+1 .
We want to prove that, if x > 1, then for all n, P (n) is true.
Inductive Basis:
1.

n = 1. We discuss the two cases: x > 1 and 0 < x <

1. Suppose that 0 < x < 1. In this case, we know that log x < 0. Since
0 < x < 1, we have log x < x log x. It follows that x < xx .
2. If x > 1, then it is clear that x < xx if x > 1.
Thus, in both cases, we have
a1 < a2 ,

P (1) = T rue.
[The Basis Holds.]

Inductive Hypothesis:

Assume P (i) is true for some i in Dn , i.e.

ai < ai+1 .
c Chung-Chih Li, Kishan Mehrotra

136

3. Mathematical Induction
Inductive Step: We want to prove P (i + 1) is also true. Using the
hypothesis, we have
ai < ai+1 xai < xai+1
P (i + 1) = T.

because x > 1

[The Inductive Step Holds.]

Therefore, it is an increasing sequence.

Solution 21:
Let Dn be the set of all positive odd integers, i.e, Dn =
{1, 3, 5 . . .}, and define
P (n) : 1 + 3n is divisible by 4.
Inductive Basis: n = 1. We choose 1 as the base because 1 is the first
element in Dn . Apparently, 1 + 31 = 4, which is divisible by 4. Therefore,
P (1) = T.
[The Basis Holds.]
Inductive Hypothesis: Assume P (n) = T , i.e., 1 + 3n is divisible by
4. Note: n is a positive odd integer.
Inductive Step: We want to prove P (n+2) = T , i.e., we want to prove
that 1 + 3n+2 is divisible by 4.
Note: We choose n + 2 instead of n + 1 because n + 2 is the odd number
next to n fixed in the inductive hypothesis, whereas n + 1 is not an
element in Dn .
From the inductive hypothesis, we can assume that 1 + 3n = 4k, where k
is an integer. We have
1 + 3n+2 = 1 + 9 3n
= (1 + 3n ) + 8 3n
= 4k + 8 3n
= 4(k + 2 3n ).
Because k and n are integers, k + 2 3n must be an integer. Therefore,
1 + 3n+2 is divisible by 4, and hence P (n + 2) = T.
[The Inductive Step Holds.]
Therefore, P (n) is true for all n in Dn .

c Chung-Chih Li, Kishan Mehrotra

3.5. Solutions

137

Solution 22:
The domain of this problem includes both positive and negative integers. The easiest way is to split the domain into two parts: positive
integers and negative integers. Then, we prove the statement by mathematical
induction in the two sub-domain separately.
(1) Let Dn = {0, 1, 2, 3, . . .} Define:

P (n) : 9 n3 + (n + 1)3 + (n + 2)3 .

Inductive Basis:

n = 0.
03 + 13 + 23 = 9.

P (0) is true.

[The Basis Holds.]

Inductive Hypothesis:

Assume i Dn , and P (i) is true, i.e.,

i3 + (i + 1)3 + (i + 2)3 = 9k,

for some integer k.
Inductive Step:

Prove P (i + 1) is true.

(i + 1)3 + (i + 2)3 + (i + 3)3

= (i + 1)3 + (i + 2)3 + i3 + 9i2 + 27i + 27
= (i3 + (i + 1)3 + (i + 2)3 ) + 9i2 + 27i + 27
= 9k + 9(i2 + 3i + 3)

[by IH]

= 9(k + i + 3i + 3)
Because k + i2 + 3i + 3 is an integer, therefore

9 (i + 1)3 + (i + 2)3 + (i + 3)3 ,

P (i + 1) is true.

[The Inductive Step Holds.]

(2) Let Dn = {0, 1, 2, 3, . . .} Define:

P (n) : 9 n3 + (n 1)3 + (n 2)3

Inductive Basis:

n = 0.
03 + 13 + 23 = 9.

P (0) is true.

[The Basis Holds.]

c Chung-Chih Li, Kishan Mehrotra

138

3. Mathematical Induction
Inductive Hypothesis:

Assume i Dn , and P (i) is true, i.e.,

i3 + (i 1)3 + (i 2)3 = 9k,

for some integer k.
Inductive Step:

Prove P (i 1) is true.

(i 1)3 + (i 2)3 + (i 3)3

= (i 1)3 + (i 2)3 + i3 9i2 + 27i 27
= (i3 + (i 1)3 + (i 2)3 ) 9i2 + 27i 27
= 9k + 9(i2 + 3i 3)

[by IH]

= 9(k i + 3i 3)
Because k i2 + 3i 3 is an integer, therefore

9 (i 1)3 + (i 2)3 + (i 3)3 ,

P (i 1) is true.

[The Inductive Step Holds.]

Putting the two domains together, we have that the sum of the cubes of any
three consecutive integers is divisible by 9.
2

Solution 23:
We first prove that the statement is correct in the non-negative
part of the domain. Let Dn = {0} N, and define

P (n) : (1) =

1
1

if n is even,
if n is odd.

We will prove that P (n) = T for all n Dn .

Inductive Basis:

n = 0.

0 is even and (1)0 = 1. Thus P (0) = T .

Inductive Hypothesis:

n = i.

Suppose P (i) = T , i.e.,

(1)i =
c Chung-Chih Li, Kishan Mehrotra

1
1

if i is even,
if i is odd.

[The Basis Holds.]

3.5. Solutions

139

Inductive Step:

n = i + 1.
(1)i+1 = 1 (1)i

1 1
=
1 1

1
if
=
1
if

1
if
=
1
if

Therefore, P (i + 1) is true.

if i is even,
if i is odd.
i is even,
i is odd.
i + 1 is odd,
i + 1 is even.
[The Inductive Step Holds.]

We have proved that the predicate is true for all non-negative integers. We
can use the same technique shown in the previous problem, i.e., we can further
prove that the statement is correct for the other part of the domain (negative
integers). Or, we can use the result we just got and the following arguments.
From the proved result above, if n is a non-negative integer, we know that
2n is even and (1)2n = 1. Observe the following fact: if n N, then
(1)n =

1
(1)2n
=
= (1)n .
n
(1)
(1)n

Therefore, we can claim that for all i Z, the predicate P (i) is always true.
That proves the theorem.
2
The following is another proof for the above problem. It is a bit awkward,
but it shows how to reorder the domain so we can examine the entire domain
without missing any element of it. Lets consider the following ordered sequence,
Z : 0, 1, 1, 2, 2, 3, 3, 4, 4, . . .
It is clear that any number in Z must be in the sequence Z somewhere. Let zi
denote the ith number in Z, where i 1. Let Dn = N. Define

1
if zn is even,
zn
P (n) : (1) =
1
if zn is odd.
Thus, our original problem can be viewed as:
Prove that for all n Dn , P (n) is true.
Inductive Basis:

n = 1.

z1 = 0, therefore z1 is even and (1)0 = 1. Thus P (1) = T .

[The Basis Holds.]
c Chung-Chih Li, Kishan Mehrotra

140

3. Mathematical Induction
Inductive Hypothesis:

n = i.

Suppose P (i) = T , i.e.

(1)zi =

1
1

if zi is even,
if zi is odd.

Inductive Step: n = i + 1.
There are only two possible values for zi+1 in terms of zi , i.e., .zi+1 = zi
if zi is positive or zi+1 = zi + 1 if zi is negative. We will discuss this by
cases.
case 1: zi+1 = zi .
(1)zi+1 = (1)zi =

1
=
(1)zi

1/1
=1
1/ 1 = 1

if zi is even,
if zi is odd.

From Z we know that in this case if zi is even, then zi+1 is even, and
if zi is odd, then zi+1 is odd. Therefore,

1
if zi+1 is even,
(1)zi+1 =
1
if zi+1 is odd.
case 2: zi+1 = zi + 1.
(1)zi+1 = (1)zi +1 =

1
=
(1)zi

1/1 = 1 if zi is even,
1/ 1 = 1 if zi is odd.

From Z we know that in this case if zi is even, then zi+1 is odd, and
if zi is odd, then zi+1 is even. Therefore,

1
if zi+1 is even,
(1)zi+1 =
1
if zi+1 is odd.
In both cases, P (i + 1) is true.

[The Inductive Step Holds.]

This proves the problem.

Solution 24:

Let us rewrite the recurrent relation first:

n 1, c(n + 2) = 3c(n + 1) 3c(n) + c(n 1)

n 3, c(n) = 3c(n 1) 3c(n 2) + c(n 3).
Given c(0) = c(1) = 1, and c(2) = 3.
To prove that for all n 0, c(n) = n2 n + 1, let Dn = N0 , and define
P (n) : c(n) = n2 n + 1.
c Chung-Chih Li, Kishan Mehrotra

3.5. Solutions

141

Inductive Basis:

n = 0, 1, 2.
c(0) = 0 0 + 1 = 1.
c(1) = 1 1 + 1 = 1.
c(2) = 4 2 + 1 = 3.

Therefore, P (0) = P (1) = P (2) = T .

Inductive Hypothesis:
then P (i) = T , i.e.,

[The Basis Holds.]

Assume that, for a fixed n Dn , if 0 i n,

c(i) = i2 i + 1.

This is a strong hypothesis.

Inductive Step:

To prove that P (n + 1) = T .

To calculate the value of c(n+1), we can use the values of c(n), c(n1), and
c(n 2) given in the strong inductive hypothesis, because the arguments,
n, n 1, and n 2, are in the domain of the inductive hypothesis. We
have:
c(n) = n2 n + 1,
c(n 1) = (n 1)2 (n 1) + 1,
c(n 2) = (n 2)2 (n 2) + 1.
Therefore,
c(n + 1)
= 3c(n) 3c(n 1) + c(n 2)
= 3(n2 n + 1) 3((n 1)2 (n 1) + 1) + (n 2)2 (n 2) + 1
= 3n2 3n + 3 3n2 + 9n 9 + n2 5n + 7
= n2 + n + 1
= (n2 + 2n + 1) (n + 1) + 1
= (n + 1)2 (n + 1) + 1.
We have the result above by using the definition of c(n + 1) and the
hypothesis, and the result agrees with the given closed-form formula for
c(n + 1).
[The Inductive Step Holds.]
This completes the proof.

Solution 25:

Please be careful to adjust the domain of n as shown in the

c Chung-Chih Li, Kishan Mehrotra

142

3. Mathematical Induction

following.
n 0[b(n + 2) + 2b(n + 1) + b(n) = 0]
n 0[b(n + 2) = 2b(n + 1) b(n)]
n 2[b(n) = 2b(n 1) b(n 2)].
Thats why this problem has to state that b(n) is defined by the equation after
first two values. Please compare this problem to the previous one7 .
Define b(0) = b(1) = 1. Let Dn = N0 and
P (n) : b(n) = (1 2n)(1)n .
Inductive Basis:

n = 0, and n = 1.

We have to take P (0) and P (1) as our basis, because both b(1) and
b(2) are undefined, and hence b(0) and b(1) cannot be obtained by using
the equation b(n) = 2b(n 1) b(n 2) in the inductive step. [Same
reason for the next problem.]
b(0) = 1 = (1 0)(1)0 .
b(1) = 1 = (1 2)(1)1 .
P (0) = P (1) = T.

[The Basis Holds.]

Inductive Hypothesis: Let i Dn , and 1 i.

Suppose, if n i, then P (n) = T , i.e.,
n Dn , (n i) (b(n) = (1 2n)(1)n ).
Inductive Step:

Let n = i + 1.

b(n) = b(i + 1)
= 2b((i + 1) 1) b((i + 1) 2)
= 2b(i) b(i 1)
= 2(1 2i)(1)i (1 2(i 1))(1)i1
= 2(1 2i)(1)i (3 2i)(1)i1
= (1)i1 (2(1 2i)(1) (3 2i))
= (1)i1 (1 2i)
= (1)2 (1)i1 (1 2 2i)
= (1 2(i + 1))(1)i+1
= (1 2n)(1)n .
7 One

may ask why we dont define b(n) as

n 0[b(n) = b(n + 2) 2b(n + 1)]

by using the given equation directly without adjusting the domain of n. Can we define a
function based on its future values instead of its previous values? Theoretically, the answer is
yes, but thats beyond the scope of our interest of using mathematical induction.

c Chung-Chih Li, Kishan Mehrotra

3.5. Solutions

143

P (i + 1) = T.

[The Inductive Step Holds.]

Therefore, n Dn , P (n) is true.

Solution 26:

Define b(0) = 1, b(1) = 3. Let Dn = N0 , and

P (n) : b(n) = (1 + 2n)(1)n .

Inductive Basis:

n = 0, and n = 1.
b(0) = 1 = (1 + 0)(1)0 .
b(1) = 3 = (1 + 2)(1)1 .

P (0) = P (1) = T.

[The Basis Holds.]

Inductive Hypothesis: Let i Dn , and 1 i.

Suppose, if n i, then P (n) = T , i.e.
n Dn , (n i) (b(n) = (1 + 2n)(1)n ).
Inductive Step:

Let n = i + 1.

b(n) = b(i + 1)
= 2b((i + 1) 1) b((i + 1) 2)
= 2b(i) b(i 1)
= 2(1 + 2i)(1)i (1 + 2(i 1))(1)i1
= 2(1 + 2i)(1)i (2i 1)(1)i1
= (1)i1 (2(1 + 2i)(1) (2i 1))
= (1)i1 (3 + 2i)
= (1)2 (1)i1 (1 + 2 + 2i)
= (1 + 2(i + 1))(1)i+1
= (1 + 2n)(1)n .
P (i + 1) = T.

[The Inductive Step Holds.]

Therefore, n Dn , P (n) is true.

Solution 27:

Given
f0 = 0, f1 = 1,
fn = fn1 + fn2 , for n 2.
c Chung-Chih Li, Kishan Mehrotra

144

3. Mathematical Induction

We want to prove that, for all n 0,

!n
!n !
1
1+ 5
1 5
fn =

.
2
2
5
Inductive Basis:
1
5

n=0:

0
1+ 5
2

0
1 5
2

1
5

(1 1)

= 0 = f0 .
n=1:
1
5

1
1+ 5
2

1
1 5
2

1
5

1
5

1+ 5
2

1 5
2

1+ 51+ 5
2

= 1 = f1
[The Basis Holds.]
Inductive Hypothesis:

Assume that, for 1 k n,

!k
!k
1 5
1
1+ 5
.

fk =
2
2
5

Inductive Step: In the following simplification we combine the two

positive terms and the two negative terms, and use
2
2

3+ 5
1 5
3 5
1+ 5
=
,
and
=
2
2
2
2 .
fn+1 = fn + fn1
n n
1+ 5
= 15
12 5
+
2

1
5

1
5

1
5

1
5

n
1+ 5
2

n1
1+ 5
2
n1
1+ 5
2
n+1
1+ 5
2

n1
1+ 5
2

3+ 5
2

1
5

2
1+ 5
2

n
1 5
2

n1
1 5
2

n+1
1 5
2

n1
1+ 5
2

n1
1 5
2

n1
1 5
2

3 5
2

n1
1 5
2

2
1 5
2

[The Inductive Step Holds.]

c Chung-Chih Li, Kishan Mehrotra

3.5. Solutions

145

Solution 28:
Prove by induction that the total number of subsets having
exactly two elements in a set of n elements is n(n 1)/2.
It is clear that if A is a set of 2 elements, the only subset of A with two
elements is A itself. And since 1 = 2(2 1)/2, the basis holds.
Suppose we have a set A with elements, and let n 2. And suppose we
already know by the inductive hypothesis that there are n(n1)/2 many subsets
of A with 2 elements. Now, we add a new element a into A, and examine the
power set of the new A. What are those subsets with 2 elements? They are the
old subsets with 2 elements plus every singleton subset of the old A union with
{a}. Apparently, the old A has n-many singleton subsets. Therefore, the new
A has (n(n 1)/2 + n)-many subsets with 2 elements, where the new size of A
is n + 1. And
n(n 1)
(n + 1)n
+n=
.
2
2
This proves the inductive step and completes the proof of this problem.

One may also want to define a recurrent relation according to the discussion
above, and prove the result by induction more formally. The following is a proof.
Let t(n) be the number of subsets with 2 elements of a set with n elements.
Apparently, t(0) = 0. If n 1, we can recursively define t(n) as follows. For
n 1,
t(n) = t(n 1) + (n 1).
Now, lets prove the claim that for all n 0, t(n) = n(n 1)/2. Define
Dn = N0 , and
1
P (n) : t(n) = n(n 1).
2
Inductive Basis:

n = 0.
t(0) = 0 =

1
0 (0 1).
2

Therefore, P (0) = T.

[The Basis Holds.]

Inductive Hypothesis: Let i Dn .

Suppose, if n i, then P (n) = T , i.e.
t(n) =

1
n(n 1).
2
c Chung-Chih Li, Kishan Mehrotra

146

3. Mathematical Induction
Inductive Step:

Let n = i + 1.

t(n) = t(i + 1) = t(i) + i

1
1
= i(i 1) + i = i(i 1 + 2)
2
2
1
1
= (i + 1)((i + 1) 1) = n(n 1).
2
2
P (i + 1) = T.

[The Inductive Step Holds.]

Therefore, for any set with n elements, the set has n(n 1)/2-many subsets
having exactly 2 elements.
2

Solution 29:
Prove by induction that for all integers n 2 the set of all
points of intersection of n distinct lines in the plane has no more than n(n1)/2
elements. Give examples showing exactly that many, and also fewer.
Suppose we already have n lines on the plane with the maximum number
of intersection points. If we draw a new line on the plane, we cannot introduce
more than n new intersection points. Therefore, if p(n) is the maximum number
of the intersection points of n distinct lines, the recurrent relation will be
p(n) = p(n 1) + (n 1).
This is exactly the same as the recurrence relation in the previous problem. We
skip the proof here.
2

Solution 30:

Define P (n) for all n 1 as

P (n) : 1 xn n.

Inductive Basis:

n = 1. x1 = 1, and

1 x1 1.

Therefore, P (1) is true.

Inductive Hypothesis:

Inductive Step:

[The Basis Holds.]

Assume P (n) is true, i.e.,

1 xn n.

To prove P (n + 1) is true.

c Chung-Chih Li, Kishan Mehrotra

3.5. Solutions

147

From the hypothesis, we know that 1 xn n. Thus,

1 xn n 1 x2n n
1
1
1
1 + 2 x2n + 2 n + 2 .
xn
xn
xn
Because 1 x2n

1
x2n

1, we have

11+

1
x2n

and n +

1
n + 1.
x2n

Therefore,
1
1
1
x2n + 2 n + 2 n + 1
2
xn
xn
xn
1
1 x2n + 2 n + 1
xn

11+

1 x2n+1 n + 1

1 xn+1 n + 1.
P (n + 1) is true.
Method 2:

[The Inductive Step Holds.]

There is another way to prove the inductive step.

Inductive Step:

To prove P (n + 1) is true.

From the hypothesis we know 1 xn n, and

we know 0 1n , thus

1
n

1 x2n n
1
1
2 1
n
xn

1
xn

1. Moreover,

(3.11)
(3.12)

By adding (3.11) and (3.12), we get

1+

1
1
x2n + 2 n + 1.
n
xn

Because 1 1 + n1 , we have
1
n+1
x2n
1
1 x2n + 2 n + 1
xn

1 x2n +

1 x2n+1 n + 1

1 xn+1 n + 1.
P (n + 1) is true.

[The Inductive Step Holds.]

c Chung-Chih Li, Kishan Mehrotra

148

3. Mathematical Induction

Solution 31:
Inductive Basis:

n = 0.
H2 n = H1 =

X 1
1
0
= 1+ .
i
1
2

1i1

[The Basis Holds.]

Inductive Hypothesis:

Assume H2n 1 +

n
2

Inductive Step:
H2n+1 = H22n
1 1
1
1
1
1
= + + + n + n
+
+ + n
1 2
2
2 + 1 2n + 2
2 + 2n
1
1
1
+
+ + n
= H2 n + n
2 + 1 2n + 2
2 + 2n
n
1
1
1
1+ + n
+
+ + n
2
2 + 1 2n + 2
2 + 2n
2n

{
n
1
1
1
1+ + n
+ n
+ + n
2
2 + 2n
2 + 2n
2 + 2n
z

=1+

n
2n
+ n
2
2 + 2n

=1+

n+1
2

}|

[The Inductive Step Holds.]

Therefore, for all n 0, H2n 1 + n2 .

Solution 32:
Let A denote the set of all possible strings made from A,

where A . Let range over A , and let || denote the length of . We will
prove the theorem by mathematical induction on the length of strings. Let us
first define the domain and the predicate P (n).
Dn : {0, 1, 2, 3, . . .},
P (n) : A , [(|| = n) ( S is a palindrome)].
c Chung-Chih Li, Kishan Mehrotra

3.5. Solutions

149

Inductive Basis: n = 0 and n = 1.

By the definition of S, S, and by the definition of palindrome, is
also a palindrome. Therefore, P (0) = T.
If || = 1, then by rule 2 any single character from A is in P , and it is
also a palindrome over A. P (1) = T.
[The Basis Holds.]
Inductive Hypothesis: Let i Dn , and 1 i.
Suppose if n i, then P (n) = T , i.e.,
A , [(|| i) ( S is a palindrome)].
Inductive Step: Let n = i + 1. Because we assume that 1 i in the
hypothesis, we have 2 n. This will simplify our discussion because,
as you will see, we dont have to consider rules 1 and 2 in the course of
the inductive step. Please note that this is valid, because the cases when
n = 0 and n = 1 were proved in the basis step.
Let A and || = i + 1.
1. If S, then must satisfy rule 3. Rules 1 and 2 are ruled out
because 2 ||. Thus, must be a string like aa, where a A and
S. We also know that || = i 1, and by the strong inductive
hypothesis, S if and only if is a palindrome. Therefore, is a
palindrome over A, and aa is also a palindrome over A. Therefore,
S is a palindrome.
2. If is a palindrome over A and 2 ||, must be a string like
aa, where a A and is a palindrome over A. Since || = i 1,
and by the strong inductive hypothesis, S if and only if is a
palindrome. Therefore, S, and by rule 3, aa P . Therefore,
is a palindrome S.
P (i + 1) = T.

[The Inductive Step Holds.]

For any length n, P (n) is true, which means that S is the set of palindromes
over A.
Note: The proof given above uses the original form of mathematical induction.
We havent seen the power of structural induction yet. Please compare
the following proof with the previous one. We will see the full power of
structural induction in the last two problems of this chapter.
Method 2: Structural induction
Ds : A ,
P (s) : (s S s is a palindrome).
c Chung-Chih Li, Kishan Mehrotra

150

3. Mathematical Induction
Inductive Basis: By the definitions of S and palindrome, S and it
is a palindrome. Thus, P () = T.
[The Basis Holds.]
Inductive Hypothesis:

Let s A , and assume P (s) = T , i.e.,

s S s is a palindrome.
Inductive Step:

Our task is to prove

asa S asa is a palindrome.

We have two cases about s: (1) s S, and (2) s 6 S.

case 1: s S. By the definition of S, asa S. By the hypothesis, s is a
palindrome, and hence asa is also a palindrome. Thus,
[asa S asa is a palindrome] = T.
case 2: s 6 S. By the definition of S, asa 6 S. By the hypothesis, s is
not a palindrome, and hence asa is not a palindrome. Thus,
[asa 6 S asa is not a palindrome] = T.
By contrapositive, we have
[asa is a palindrome asa S] = T.
Together, we have asa S asa is a palindrome.
[The Inductive Step Holds.]
Therefore, for any string s, s S if and only if s is a palindrome.

Solution 33:
for all , S,

Let Ds = S, and define two variables predicate P (, ) as:

P (, ) : R() = R()R().

Inductive Basis:

= a, and = b.

For = a, let S. We have

R() = R(a) = R()a = R()R(a) = R()R().
Same as = b. Thus, for all S, P (a, ) = T and P (b, ) = T .
[The Basis Holds.]
c Chung-Chih Li, Kishan Mehrotra

3.5. Solutions

151

Inductive Hypothesis: Let , S. Assume that for all S,

P (, ) = P (, ) = T , i.e.,
R() = R()R() & R() = R()R().
Inductive Step: Our task is to prove that for all S, P (, ) is
true, i.e., R() = R()R(). Let S,
R() =
=
=
=
=

R(())
R()R()
R()R()R()
R()(R()R())
R()R()

[by IH]
[by IH]
[by IH]

Therefore, for all , S, R() = R()R().

Solution 34:

Let Ds = S, and define predicate P () as: for all S,

P () : R(R()) = .

We will use the result proved in the previous problem: for all , S,
R() = R()R().

(3.13)

Inductive Basis: = a, and = b. It is clear that P (a) = P (b) = T ,

because
R(R(a)) = R(a) = a, and R(R(b)) = R(b) = b.
[The Basis Holds.]
Inductive Hypothesis:

Let , S. Assume that

R(R()) = , and R(R()) = .

Inductive Step:

We want to prove that R(R()) = .

R(R()) = R(R()R())
= R(R())R(R())
=

by (3.13)
by (3.13)
[by IH]

[The Inductive Step Holds.]

Therefore, for all S, R(R()) = .

2
c Chung-Chih Li, Kishan Mehrotra

152

3. Mathematical Induction

Solution 35:
We will prove this problem by structural mathematical induction. For convenience, let () denote the number of as in , and () the number of bs in
. We shall prove that, for every A, () = ().
Inductive Basis:
Basis Holds.]

= . It is clear that () = () = 0.

Inductive Hypothesis:
() = () = n.

[The

Let , A, and () = () = m, and

Inductive Step:
By rule 2: = ab. Clearly, () = () = m + 1.
By rule 3: = . Thus, () = () = m + n and () = () =
m + n.
Thus, () = () in all cases.
Therefore, if A, then () = ().

[The Inductive Step Holds.]

Solution 36:
As with the previous problem, let () denote the number
of as in , and () the number of bs in . We shall prove that, if A, then
for any prefix of we have () ().
Inductive Basis: = . The only prefix of is itself. It is clear
that () = () = 0. Thus, the inductive basis holds.
[The Basis
Holds.]
Inductive Hypothesis: Fix , A, and assume that, for any , if
is a prefix of or , then () ().
Inductive Step:
By rule 2: We obtain a new string = ab. Given a prefix of , we
have the following cases.
1. = . In this case, it is clear that () ().
2. = a 0 , where 0 is a prefix of . Thus, () = (a 0 ) = ( 0 )+
1 and () = (a 0 ) = ( 0 ). By the inductive hypothesis,
( 0 ) ( 0 ). Thus, ( 0 ) + 1 ( 0 ), and hence () ().
3. = = ab. In this case, () = () + 1 and () = () + 1.
By the inductive hypothesis, () (). Thus, () ().
c Chung-Chih Li, Kishan Mehrotra

3.5. Solutions

153

By rule 3: We obtain a new string = . Given a prefix of , we

have the following cases.
1. is a prefix of . By the inductive hypothesis, () ().
2. = , where is a prefix of . By the inductive hypothesis,
we have () () and () (). Thus, () + ()
() + (). Since () = () + () and () = () + (),
it follow that () ().
[The Inductive Step Holds.]
This completes the proof.

c Chung-Chih Li, Kishan Mehrotra

Chapter 4

Relations
I wanted certainty in the kind of way in which
people want religious faith.
I thought that certainty is more likely to be found in
mathematics than elsewhere.
Bertrand Russell

4.1. Definitions, Theorems, and Comments

4.1

157

Definitions, Theorems, and Comments

The concept of relation is widely used in everyday life. For example, Sean and
Leon are brothers carries the concept of relation brother that Sean and Leon
have. Or, Dennis is the father of Sean carries the concept of relation father.
For some relations, we can exchange the subject and object in the English
sentences that represent some relations; while for some others we cannot. For
example, if Sean is a brother of Leon, then we also have the fact that Leon
is a brother of Sean (presumably that both Sean and Leon are boys), whereas
Dennis is the father of Sean shows that exchanging subject and object of the
sentence is not allowed.
In this chapter, we will study some widely used relations in mathematics
and give them mathematical characterizations.

4.1.1

Definitions

Definition 4.1: A relation R on sets S and T is a subset of the Cartesian

product S T . That is, R is a binary relation if R S T .
Comment: A relation R can be a subset of the Cartesian product of more
than two sets, i.e., R S1 S2 Sn , where n 2. In this chapter,
we are interested in binary relations.
Definition 4.2: If R S S, we call R a binary relation on S.
Definition 4.3: If R = S T , then R is called the complete relation.
Notations for relations
There are many methods to express a relation:
1. Ordered pairs notation: We have used this notation in the definitions
above. This notation directly comes from the definition of the Cartesian
product of sets.
Example 4.1

Let S = {a, b, c, d}, and

R = {(a, a), (b, a), (b, c), (c, d)}.

R is a relation on S. We say, b has relation R to a because (b, a) R, but

a does not have relation R to b because (a, b) 6 R
2. Infix notation: In this notation the property (x, y) R is written as
xRy. This is also the method of expressing a relation in everyday life.
c Chung-Chih Li, Kishan Mehrotra

158

4. Relations
For example, note the similarity in the infix notation and a typical everyday real-life relation x is a brother of y, or a typical mathematical
statement x is divisible by y.
Example 4.2
Consider the relation in Example 4.1. We have bRa
because (b, a) R, but we dont have aRb because (a, b) 6 R.

3. Matrix notation: When R is a finite relation, it can be written as a

matrix (also denoted by R) R = (ri,j ) where

1 if i, j S and (i, j) R
ri,j =
0 if i, j S and (i, j) 6 R
Example 4.3 Let S = {x, y, z}, T = {a, b, c, d}. We represent a relation
R S T in the following 3 4 matrix r.

R =

x
y
z

a b
1 0
0 1
1 0

c d
1 0
0 0
1 1

For examples, (x, c) R because rx,c = 1, and (z, b) 6 R because rz,b = 0.

Altogether, we have
R = {(x, a), (x, c), (y, b), (z, a), (z, c), (z, d)}.

4. Directed graph representation: In this representation the relation R

is a directed graph with nodes as all elements of the set S, and for x, y S
the graph contains a directed edge from x to y if and only if (x, y) R.
Example 4.4

Let S = {a, b, c, d}, and

R = {(a, b), (a, c), (a, d), (b, b), (c, a), (d.b), (d, c)}.
The following figure shows two different representations of the same relation R.

a
b
c
d

a
0
0
1
0

b
1
1
0
1

c d
1 1
0 0
0 0
1 0

c
6
?
a

?
- b
6

Matrix
c Chung-Chih Li, Kishan Mehrotra

Directed Graph

4.1. Definitions, Theorems, and Comments

159

Definition 4.4: Let S be a non-empty set and R S S a relation.

1. R is called a reflexive relation if and only if
for all x in S, we have (x, x) R.
2. R is called a symmetric relation if and only if
for all x, y S, if (x, y) R, then (y, x) R.
3. R is called a transitive relation if and only if
for all x, y, z, S, if (x, y) R and (y, z) R, then (x, z) R.
Comment: The above definition can be extended to include the case S =
. In this special case, the only possible relation on S would be . The
empty relation, , satisfies all three properties stated above.
Definition 4.5: A relation R on S is called an equivalence relation if it satisfies
the following three properties:
1. R is a reflexive relation,
2. R is a symmetric relation, and
3. R is a transitive relation.
Example 4.5
Let S = {v0 , v1 , v2 , . . . , vn } be the set of variables in a
certain computer program. If vi = vj means the values of vi and vj are the
same, then at any moment during the course of running the program =
is an equivalence relation on S.
Definition 4.6: Let S be a non-empty set. A partition P of S is a set of subsets
of S,
P = {A1 , A2 , . . . , Ak },
such that,
1.

k
S
i=1

Ai = S, and

2. for all i, 1 i k,

(a) Ai S,
(b) Ai 6= ,
(c) Ai Aj = , where 1 j k and i 6= j.

The elements of the set P , A1 , A2 , . . . , Ak are called cells.

Definition 4.7: Let P = {A1 , A2 , . . . , Ak } and let Q = {B1 , B2 , . . . , Bl } be two
partitions of a nonempty set S. The partition Q is called a refinement of P
if and only if for any Bj Q there exists one and only one Ai P such that
Bj Ai .
c Chung-Chih Li, Kishan Mehrotra

160

4. Relations

Definition 4.8: Suppose that R S S is an equivalence relation, and x S is

a fixed element. The set [x]R = {y : xRy, y S} is known as the equivalence
class of x.
Definition 4.9: Let R S S be an equivalence relation. It can be seen
that the set of equivalence classes produced by R is a partition of S, i.e.,
{[x]R : x S} is a partition of S. This set is also known as the quotient set
and is often denoted as S/R.
Example 4.6
Let S = {1, 2, 3, 4, 5, 6}. Consider the relation R given in
the following matrix.
1
2
3
4
5
6

R =

1
1
1
0
0
0
1

2
1
1
0
0
0
1

3
0
0
1
0
0
0

4
0
0
0
1
1
0

5
0
0
0
1
1
0

6
1
1
0
0
0
1

We have the following facts:

1. R is an equivalence relation on S because it is reflexive, symmetric, and
transitive.
2. Let
P = {{1, 2, 6}, {3}, {4, 5}}.
Then P is a partition of S. In fact, P is the quotient set S/R given by
the relation R because
[1]R = [2]R = [6]R = {1, 2, 6},
[3]R = {3},
[4]R = [5]R = {4, 5}.
3. Let
Q = {{1, 2}, {3}, {4, 5}, {6}}.
Then Q is a refinement of P .
Definition 4.10: A relation R on S is called an irreflexive relation if and only
if for all x S, (x, x) 6 R.
Definition 4.11: A relation R on S is called an asymmetric relation if and only
if for all x, y S, if (x, y) R then (y, x)
/ R.
Definition 4.12: A relation R on S is called an antisymmetric relation if and
only if for all x, y S, if (x, y) R, then either x = y or (y, x) 6 R. In
another words, if both (x, y) and (y, x) are in R, then x = y.
c Chung-Chih Li, Kishan Mehrotra

4.1. Definitions, Theorems, and Comments

161

Definition 4.13: A relation R on S is called a partial order relation if it satisfies

the following three properties:
1. R is a reflexive relation,
2. R is an antisymmetric relation, and
3. R is a transitive relation.
Comment: In some occasions, we allow a partial order relation to be
irreflexive. The reflexivity or irreflexivity of a partial order relation
will not affect some major properties of the partial order relation we
are interested in.
Example 4.7
Let S = {0, 1, 2, 3 . . .}. The relation is a partial order
relation on S. In fact, if S is any collection of real numbers, then is a
partial order relation on S.

4.1.2

Theorems

There is an intimate relation between an equivalence relation R S S and a

partition P of S. This relation is precisely stated below:
Theorem 4.1 Let R S S be an equivalence relation. Then R generates
a unique partition of PR = {A1 , A2 , . . . , Ak } of S such that, for all x, y
i, 1 i k, (x, y) Ai if and only if (x, y) R.
Conversely given a partition P = {A1 , A2 , . . . , Ak } of S, a unique equivalence relation RP is defined as follows:
(x, y) RP if and only if i, 1 i k, (x, y) Ai .
Theorem 4.2 Let R S S be an equivalence relation and x, y S. Then
the following property holds: Either [x]R = [y]R or [x]R [y]R = .

Comment: Given an equivalence relation on S, the above theorem shows that

the collection of all possible equivalence classes give a partition of S.

c Chung-Chih Li, Kishan Mehrotra

Should we push the transitive closure here?

162

4.2

4. Relations

Problems

Problem 1: Let S be a set, and R be a relation on S. Rewrite the definitions

of reflexive, symmetric, and transitive relations into quantified predicates.
Problem 2: Same as above, rewrite the definitions of irreflexive, asymmetric, and antisymmetric relations into quantified predicates.
Problem 3:
1. Write the inclusion relation on the power set P({a, b, c}) in the
matrix notation.
2. Is an equivalence relation on P({a, b, c})?
3. Is a partial order relation on P({a, b, c})?
Problem 4:

Define the relation R on the power set P(A) of a set A as,

a, b P(A)[(a, b) R a b 6= ].

Is R reflexive, symmetric, or transitive? Why?

Problem 5: Let A be the set of all lines in the plane, and R1 , R2 be two
relations on A defined as follows:
1. Definition of R1 : For all L, L0 A,
(L, L0 ) R1 iff L is perpendicular to L0 .
2. Definition of R2 : For all L, L0 A,
(L, L0 ) R2 iff L is perpendicular or parallel to L0 .
Mark the following table properly to indicate the properties of the relations R1 and R2 have.
reflexive
symmetric
antisymmetric
irreflexive
asymmetric
transitive
R1
R2
Problem 6: Let N be the set all positive integers, and R1 , R2 be two relations on N defined as follows:
1. Definition of R1 : For all a, b N,
(a, b) R1 iff a 6= b.
2. Definition of R2 : For all a, b N,
(a, b) R2 iff
c Chung-Chih Li, Kishan Mehrotra

a
= 2i for some integer i 0.
b

4.2. Problems

163

Mark the following table properly to indicate the properties of the relations R1 and R2 have.
reflexive
symmetric
antisymmetric
irreflexive
asymmetric
transitive
R1
R2
Problem 7:

Define the relation R on N as

c, d N

(c, d) R iff c + d is even.

1. Prove that R is an equivalence relation.

2. How many equivalence classes does R have?
Problem 8: Let S be the set of all students at a school. Define the relation
R as: For all students x and y in S,
xRy iff x and y are taking the same class.
Is R an equivalence relation?
Problem 9: Let R be a reflexive and transitive relation on a set S. Define
a relation Y on S as follows:
a, b S [(a, b) Y ((a, b) R (b, a) R.)]
Prove that Y is an equivalence relation.
Problem 10:
1. Give an example of a relation that is symmetric and
transitive, but not reflexive.
2. What is wrong with the following proof that every symmetric and
transitive relation R is reflexive? If (a, b) R then (b, a) R by
symmetry. By transitivity, (a, a) R. Therefore R is reflexive.
Problem 11: Let S be a set, P (x) and Q(x) be two predicates in one variable x S, and L be a binary equivalence relation on S. Suppose
x S[P (x) Q(x)].
Define a binary relation R on S as follows:
(x, y) R [((x, y) L) P (x) Q(y)].
Is the relation R reflexive, symmetric, or transitive? Why?
Problem 12: An isolated point for a relation R on a set A is an element a,
where a A and (a, x) 6 R and (x, a) 6 R for any x A.
c Chung-Chih Li, Kishan Mehrotra

164

4. Relations
1. Given R and A as above, write a predicate to define the isolated
point.
2. Suppose R is a binary relation on A such that, R is symmetric
and transitive, and R has no isolated points. Prove that R is an
equivalence relation.

Problem 13: Let A be a set having a total of n elements, where n 1.

Let R be any equivalence relation on A. Prove that the total number of
ordered pairs in R is odd if n is odd, even if n is even.
[Hint: Consider the matrix notation for relations.]
Problem 14:

Let R and S be two relations both on A.

1. If R and S are both reflexive, is R S also reflexive?

2. If R and S are both symmetric, is R S also symmetric?
3. If R and S are both transitive, is R S also transitive?
Problem 15:

Let R and S be two relations both on A.

1. If R and S are both reflexive, is R S also reflexive?

2. If R and S are both symmetric, is R S also symmetric?
3. If R and S are both transitive, is R S also transitive?
Problem 16:

Let A, B, C be any three sets. And let

D =(A (B C)),
E =((A C) B),
F =(A C).

Suppose each of D, E and F is nonempty. Prove that {D, E, F } is a

partition of A.
[Hint: Use an epsilon argument, but first draw a Venn diagram to help
yourself see whats going on.]
Problem 17:

Define R to be the following relation on N N:

For all a, b, x, y in N,
(a, b)R(x, y) iff ay = bx.
Prove that R is an equivalence relation.
Problem 18: Let A = {1, 2, 3, 4}. For each of the following three relations
on A, prove or disprove that it is an equivalence relation and, if it is one,
write down its equivalence classes.
R1 = {(1, 1), (2, 2), (3, 4), (3, 3), (4, 4)}

(1, 1), (2, 2), (3, 4), (4, 4), (1, 2), (2, 1),
R2 =
(3, 3), (4, 3), (1, 3), (1, 4), (3, 1), (4, 1)
R3 = {(1, 1), (2, 2), (3, 4), (3, 3), (4, 4), (1, 2), (2, 1), (4, 2), (2, 3)}.
c Chung-Chih Li, Kishan Mehrotra

4.2. Problems

165

Problem 19:

List all the partitions of {a, b, c}.

Problem 20:

Given any set S, is {S} a partition of S?

Problem 21:
follows:

Let Z be the set of all integers. Define relation R on N as

a, b N, (a, b) R iff i Z

a
= 2i .
b

1. Prove that R is an equivalence relation.

2. List the cells of the partition given by R.
Problem 22:

How many equivalence relations are there on A = {1, 2, 3}?

Problem 23: Let R1 and R2 be equivalence relations on a set S. Prove or

disprove that R1 R2 is also an equivalence relation on S.
Problem 24: Let R1 and R2 be partial order relations on a set S. Prove
or disprove that R1 R2 is also a partial order relation on S.
Problem 25: Let P(N) be the power set of natural numbers. Prove that
the subset relation is a partial order relation on P(N).
Problem 26:

Let S = {1, 2, . . . , 10}. Define the following four sets as:

P1
P2
P3
P4

={{1, 3, 8}, {2, 4, 6}, {5, 7, 10}, {9}}.

={{7, 4, 3, 8}, {1, 5, 10, 3}, {2, 6}}.
={{1, 5, 9}, {2, 10, 4, 7}, {8, 3, 6}}.
={{4, 2}, {3, 8}, {6}, {10, 7}, {1}, {5}, {9}}.

1. Which of the sets above is a partition of S?

2. Which of the partitions is a refinement of which other partitions?
Problem 27:

The set {1, 2, 3, 4, 5, 6} is partitioned as

P = {{1, 2, 3}, {4, 5}, {6}}.

How many refinements does P have?

Problem 28:

Let S be any non-empty set and suppose that

P1 = {C1 , . . . , Cm } and P2 = {D1 , . . . , Dn }

are two partitions of S. Define set Q as:

Q = {Ci Dj ; 1 i m, 1 j n} {}
1. Prove that Q is a is a partition of S.
2. Prove that Q is a refinement of both P1 and P2 .

c Chung-Chih Li, Kishan Mehrotra

166

4.3

4. Relations

Solutions

Solution 1:

Let S be a set, and R be a relation on S.

R is reflexive

iff

a S, (a, a) R.

R is symmetric iff

a, b S[(a, b) R (b, a) R].

R is transitive iff

a, b, c S, [(a, b) R (b, c) R (a, c) R].

Note: For the definition of symmetry, we can also rewrite as:

R is symmetric iff a, b S[(a, b) R (b, a) R].

Solution 2:

Let S be a set, and R be a relation on S.

R is irreflexive

iff

a S, (a, a) 6 R.

R is asymmetric

iff

a, b S[(a, b) R (b, a) 6 R].

R is antisymmetric iff

a, b S, [(a, b) R (b, a) R a = b].

Solution 3:
1. Relation matrix of on the power set P({a, b, c}):

{a}
{b}
{c}
{a, b}
{b, c}
{a, c}
{a, b, c}

{a}
1
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0

{b} {c} {a, b}

1
1
1
0
0
1
1
0
1
0
1
0
0
0
1
0
0
0
0
0
0
0
0
0

{b, c}
1
0
1
1
0
1
0
0

{a, c}
1
1
0
1
0
0
1
0

{a, b, c}
1
1
1
1
1
1
1
1

2. is not an equivalence relation because it is not symmetric. For example,

{a} {a, b}, whereas {a, b} 6 {a}.
c Chung-Chih Li, Kishan Mehrotra

4.3. Solutions

167

3. is a partial order relation. From the relation matrix, it is easy to

verify that is reflexive and antisymmetric. In general, the transitivity
is not obvious to be seen from the relation matrix. But in this particular
problem, we can see the property directly from the basic theorems in set.
That is, for any sets A, B, C P({a, b, c}), if A B and B C, then
A C. Therefore, is transitive.

Solution 4:
defined as,

Given any set A, the relation R on the power set P(A) is

a, b P(A) [(a, b) R a b 6= ].

1. R is not reflexive because is an element of the power set of any set A,

and = . By the definition, (, ) 6 R.
2. R is symmetric because the intersection operator is commutative, thus,
if a b 6= , then b a 6= .
3. R is not transitive, because if a b 6= and b c =
6 doesnt mean that
a c 6= . Here is a counter example: let a = {1, 2}, b = {2, 3}, and
c = {3, 4}.

Solution 5:

R1
R2

Answers:

reflexive
symmetric
antisymmetric
irreflexive
asymmetric
transitive

Explanation: Let l1 l2 denote line l1 is perpendicular to line l2 , and l1 k l2

denote l1 is parallel to l2 . Consider the following figure.

l2

90

90

l1

l3
c Chung-Chih Li, Kishan Mehrotra

168

4. Relations

R1 :

1. It is clear that is not reflexive because any line is not perpendicular

to itself. In another words, is irreflexive.
2. is symmetric because
l1 l2 l2 l1 .
3. is not asymmetric because
l1 l2 and l2 l1 .
4. is not antisymmetric because
(l1 l2 l2 l1 ) 6 l1 = l2 .
5. is not transitive because
(l1 l2 l2 l3 ) 6 l1 l3 .

R2 :

1. R2 is reflexive because for all l, we know that l k l, thus

l k l (l k l) (l l) (l, l) R2 .
2. R2 is symmetric because,
(l1 , l2 ) R2 (l1 k l2 ) (l1 l2 )
(l2 k l1 ) (l2 l1 )
(l2 , l1 ) R2 .
3. R2 is not asymmetric. As shown in the example above, both (l1 , l2 )
and (l2 , l1 ) are in R2 .
4. R2 is not antisymmetric. As shown in the example above both (l1 , l2 )
and (l2 , l1 ) are in R2 , and l1 6= l2 .
5. R2 is transitive, because, for all l1 , l2 and l3 , if (l1 , l2 ) R2 and
(l2 , l3 ) R2 , we have four cases:
case
case
case
case

1:
2:
3:
4:

l1
l1
l1
If

k l2 and l2 k l3 , l1 k l3 (l1 , l3 ) R2 .
k l2 and l2 l3 , l1 l3 (l1 , l3 ) R2 .
l2 and l2 k l3 , l1 l3 (l1 , l3 ) R2 .
l1 l2 and l2 l3 , l1 k l3 (l1 , l3 ) R2 .

In each cases, we have (l1 , l3 ) R2 .

c Chung-Chih Li, Kishan Mehrotra

4.3. Solutions
Solution 6:

R1
R2

169
Answers:

reflexive
symmetric
antisymmetric
irreflexive
asymmetric
transitive

Explanation:
R1 :

1. R1 is not reflexive because for all a N, a = a.

2. R1 is irreflexive because for all a N, a = a, thus (a, a) 6 R1 .
3. R1 is symmetric because for all a, b N, if a 6= b then b 6= a, i.e., if
(a, b) R1 , then (b, a) R1 .
4. R1 is not asymmetric because there exist a, b N, a 6= b and b 6= a.
Note: We must not say that R1 is not asymmetric because R1 is
symmetric. Consider the empty relation, it is symmetric, asymmetric, and antisymmetric!!
5. R1 is not antisymmetric because there exist different a and b in N
such that, a 6= b and b 6= a.
6. R1 is not transitive because, for example, (1, 2), (2, 1) R1 but
(1, 1) 6 R1 .

R2 :

1. R2 is reflexive because for all a N,

a
a

= 1 = 20 , this (a, a) R4 .

2. R2 is not symmetric because if (a, b) R2 , then ab = 2i , where i 0,

but ab = 2i , where i 0. Therefore, (b, a) 6 R2 .
3. R2 is not asymmetric. Because, if we let a = b, we can have both
(a, b) and (b, a) in R2 .
4. R2 is antisymmetric. If (a, b) R2 and (b, a) R2 , we have
and ab = 2j , where i, j 0. Then

a
b

= 2i

a
b

= 1,

a
b
= 1 = 2i+j .
b
a
Thus, i + j = 0. Since i, j 0, we have i = j = 0. Therefore,
and hence a = b.
5. R2 is transitive because, if (a, b), (b, c) R2 , then
where i, j 0.

a
b

= 2i and

b
c

= 2j ,

a
a b
= = 2i+j , where i + j 0.
c
b
c
Therefore, (a, c) R2 .
c Chung-Chih Li, Kishan Mehrotra

170

4. Relations

Solution 7:

If c+d is even, then we can find an integer i such that, c+d = 2i.

1. R is an equivalence relation because,

Reflexive: For all a N, a+a = 2a, which is even. Therefore, (a, a) R.
Symmetric: If a+b is even, then b+a is even. Therefore, for all a, b N,
if (a, b) R, then (b, a) R.
Transitive: Let (a, b) R and (b, c) R. Then,
a + b = 2i
b + c = 2j

(4.1)
(4.2)

where, i and j are integers. Take (1)+(2), we have a+2b+c = 2i+2j.

Thus, a + c = 2(i + j b). Since i + j b is an integer, we know that
a + c is even. Therefore, (a, c) R.
2. There are two equivalence classes of R:
E :={x : x N, x is even}.
O :={x : x N, x is odd}.
Because, if x E and (x, y) R, that means x + y is even. If x and
x + y are even, y must be even. Therefore, y E. Likewise, if x O and
(x, y) R, that means x + y is even. Therefore, y must be odd, hence
y O. And since E O = N, no other equivalence class is possible.

Solution 8:
R is not an equivalence relation due to the lack of transitivity.
Let x, y, z B. Suppose x and y taking discrete mathematics, y and z taking
programming languages. That means (x, y) R and (y, z) R. But, that does
not mean x takes programming languages, nor z takes discrete mathematics.
Therefore, (x, z) R is not necessarily true.
2

Solution 9:
Reflexivity:

Given a reflexive and transitive relation R on S.

For all a S, (a, a) R, because R is reflexive.
(a, a) R (a, a) R and (a, a) R
(a, a) Y.

c Chung-Chih Li, Kishan Mehrotra

4.3. Solutions
Symmetry:

171
For all a, b S, from the definition of Y , we have
(a, b) Y (a, b) R and (b, a) R
(b, a) R and (a, b) R
(b, a) Y.

Transitivity:

For all a, b, c S, we have

(a, b), (b, c) Y

[(a, b) R (b, a) R] [(b, c) R (c, b) R]

[(a, b) R (b, c) R] [(c, b) R (b, a) R]
(a, c) R (c, a) R because R is transitive.
(a, c) Y.

Therefore, Y is an equivalence relations.

Solution 10:
1. Let S = {1, 2, 3}, and define R on S as:
R = {(1, 1), (2, 2), (1, 2), (2, 1)}.
This is an example of a relation that is symmetric and transitive, but not
reflexive.
2. If R is reflexive, we require that for every element a A, (a, a) has to be
in R. But, if R is symmetric and transitive, given any a in A, (a, b) may
not be in R. Therefore, we cant assure that (b, a) R, and apply the
transitivity to conclude that (a, a) is in R. In the above example, if a = 3,
we dont have any relation (3,b) in R. Thus, in no way we can apply
symmetric and transitive properties of R to conclude that (3, 3) R.

Solution 11:

Let P (x) and Q(x) be two predicates such that,

x S[P (x) Q(x)],

(4.3)

and L be an equivalence relation on S. Define the relation R on S as follows:

(x, y) R [((x, y) L) P (x) Q(y)].
c Chung-Chih Li, Kishan Mehrotra

172

4. Relations

1. R is not reflexive, because for all x in S,

P (x) Q(x) P (x) Q(x)
(P (x) Q(x)).
That means (P (x) Q(x)) is false, hence, (x, x) 6 R.
2. R is not symmetric. Assume (x, y) R. From the definition of R, we
know that
((x, y) L) P (x) Q(y),
P (x) is true. From (4.3), Q(x) is false. Thus, ((y, x) L) P (y) Q(x),
is false. Therefore, (y, x) 6 R.
3. R is transitive. Suppose (x, y) R and (y, z) R. From the definition of
R, we know that
[(x, y) L) P (x) Q(y)] [(y, z) L P (y) Q(z)]
[(x, y) L (y, z) L] P (x) Q(z)
(x, z) L P (x) Q(z) because L is transitive.
(x, z) R.

Solution 12:

Let R be a binary relation on A.

1. The predicate P that defines the isolated points is: For all a A,
P (a) = (a A) [x A((a, x) 6 R (x, a) 6 R)].
a is an isolated point of R if and only if p(a) is true.

2. We want to prove that if a symmetric and transitive relation without any

isolated points, then it is also reflexive.
Given such a relation R on A. For any a A, because a is not an isolated
point, there must exist b A such that, (a, b) R or (b, a) R. If
(a, b) R, we know that (b, a) R because R is symmetric. Likewise, if
(b, a) R, we know that (a, b) R. In either case, by the transitivity of
R, we have (a, a) R. Therefore, R is reflexive.

c Chung-Chih Li, Kishan Mehrotra

4.3. Solutions

173

Solution 13:
Consider the relation matrix of a equivalence relation on a set
A with n elements.
1 2 3 n
1 1
2 1

3
1

.. ..
..
..
.
. .
.
.. ..
.
..
. ..
. .
n 1
Because R is symmetric, the number of 1s in the upper right corner must be
equal to the number of 1s in the lower left corner. Therefore, the total number
of 1s in both corners is an even number. Let it be 2k.
Because R is reflexive, all entries on the diagonal must be 1, and the total
number is n. Thus, the total number of ordered pairs in R is 2k + n.
Therefore, if n is even, then 2k + n is even, and if n is odd, then 2k + n is
odd.
2

Solution 14:

Given two binary relations R and S on A.

1. Suppose both R and S are reflexive.

Given any a A, we know that
(a, a) R (a, a) S (a, a) (R S).
Therefore, if both R and S are reflexive, so is R S.
2. Suppose both R and S are symmetric.
Assume (x, y) (R S).
(x, y) (R S) (x, y) R (x, y) S
(y, x) R (y, x) S
(y, x) (R S).
Therefore, if both R and S are symmetric, so is R S.
3. Suppose both R and S are transitive.
Assume (x, y) (R S) and (y, z) (R S).
(x, y) (R S) (y, z) (R S)
(x, y) R (x, y) S (y, z) R (y, z) S
(x, z) R (x, z) S
(x, z) (R S).
Therefore, if both R and S are transitive, so is R S.
c Chung-Chih Li, Kishan Mehrotra

174

4. Relations

Solution 15:

Given two binary relations R and S on A.

1. Suppose both R and S are reflexive.

Given any a A, we know that
(a, a) R (a, a) S (a, a) (R S).
Therefore, if both R and S are reflexive, so is R S.
2. Suppose both R and S are symmetric.
Assume (x, y) (R S).
(x, y) (R S) (x, y) R (x, y) S
(y, x) R (y, x) S
(y, x) (R S).
Therefore, if both R and S are symmetric, so is R S.
3. If both R and S are transitive, R S is not necessary to be a transitive
relation. We can see this in the following example. Let A = {a, b}, and
define relations R and S on A as:
R = {(a, b)}; S = {(b, a)}.
Thus, we have R S = {(a, b), (b, a)}, and it is not transitive. Because
(a, b), (b, a) R, but (a, a) 6 R.
Note: Make sure you know why R and S are transitive.

Solution 16:
If we want to claim that a collection of sets is a partition of
set A, we have to make sure 3 things.
1. None of the sets in the collection is empty.
2. The union of the sets in the collection is equal to A.
3. The sets in the collection are mutually disjointed, i.e., they are disjointed
to each other.
For this problem, lets first draw a Venn diagram. Its not the only case, but it
will help us to build up the feeling.
c Chung-Chih Li, Kishan Mehrotra

4.3. Solutions

175

C
:= (A C) B
:= A (B C)
:= A C

1. A (B C), (A C) B, and A C are given to be nonempty. Therefore,

we only have to prove that 2. and 3. are satisfied.
2. To prove that (A (B C)) ((A C) B) (A C) = A:
Let a (A (B C)) ((A C) B) (A C). There are 3 cases:
(a) a (A (B C))
(a A) (a 6 (B C))
a A.
(b) a ((A C) B)
(a (A C)) (a B)
a (A C)
(a A) (a 6 C)
a A.
(c) a (A C)
(a A) (a C)
a A.
Therefore,
((A (B C)) ((A C) B) (A C)) A
Let a A. Consider the following two cases:
(a) a C. If so, a (A C).
(b) a 6 C. There are two sub-cases:
i. a B.
(a A) (a 6 C) (a (A C)).
Therefore, a ((A C) B).
c Chung-Chih Li, Kishan Mehrotra

176

4. Relations
ii. a 6 B.
(a 6 B) (a 6 C) (a 6 (B C)).
Therefore, a (A (B C)).
From above, and the definition of union, we have
a ((A (B C)) ((A C) B) (A C)).
Therefore,
A ((A (B C)) ((A C) B) (A C)).
That prove 2.

3. To prove that (A(B C)), ((AC)B), (AC) are mutually disjointed:

Be careful, X Y Z = does not imply X, Y, Z are mutually disjointed.
Therefore, it is not sufficient if we simply prove that
(A (B C)) ((A C) B) (A C) = .
(a) a (A (B C))
a A a 6 (B C)
a A a 6 B a 6 C
a 6 ((A C) B) a 6 (A C).
(b) a ((A C) B)
a (A C) a B
a A a 6 C a B
a A a (B C) a 6 (A C)
a 6 (A (B C)) a 6 (A C).
(c) a (A C)
aAaC
a A a (B C) a 6 (A C)
a 6 (A (B C)) a 6 ((A C) B).
That proves the claim.

Solution 17:
c Chung-Chih Li, Kishan Mehrotra

4.3. Solutions

177

Note: Do not be confused by the notation (a, b)R(c, d). (a, b)R(c, d) means
((a, b), (c, d)) R, in another words, (a, b) and (c, d) have R relation. And,
another common mistake is to consider (a, b) as an instance of R. (a, b) is
only an object in N N, and this object may or may not have R relation
to other objects in N N. Therefore, if (a, b) N N, the statement
(a, b) R is nonsense.
Reflexivity: Given any (a, b) N N. ((a, b), (a, b)) R since ab = ba.
Thus, R is reflexive.
Symmetry:

Suppose ((a, b), (c, d)) R,

ad = bc
cb = da
((c, d), (a, b)) R.

Therefore, R is symmetric.
Transitivity:

Suppose ((a, b), (c, d)) R and ((c, d), (e, f )) R,

ad = bc cf = de
adcf = bcde
(af )(dc) = (be)(dc)
af = be
((a, b), (e, f )) R.

R is transitive.

Therefore, R is an equivalence relation.

Solution 18:
R1
R2
R3

reflexive symmetric

transitive

None of them is an equivalence relation.

R1 : Not symmetric. (3, 4) R but (4, 3) 6 R. Note that R1 is transitive.
Think why.
R2 : Not transitive. (2, 1) R2 , (1, 3) R2 but (2, 3) 6 R2 .
R3 : Not symmetric. (3, 4) R3 but (4, 3) 6 R3 . R3 is not transitive either,
because (1, 2) R3 , (2, 3) R3 but (1, 3) 6 R3 .
c Chung-Chih Li, Kishan Mehrotra

178

4. Relations

Solution 19:

The possible partitions of {a, b, c}:

{{a, b, c}}, {{a}, {b, c}}, {{a, b}, {c}}, {{b}, {a, c}}, {{a}, {b}, {c}}.

Solution 20:
If S is a non-empty set, then {S} is a partition of S. If S = ,
the partition of S is not {}. The partition of is .
2

Solution 21:
1. Reflexivity: For a N,
R is reflexive.
Symmetry:

a
a

= 1 = 20 . Therefore, (a, a) R, and hence

Suppose (a, b) R.
a
= 2i
b
b
i Z, = 2i
a
(b, a) R because i Z.

(a, b) R i Z,

Transitive:

Suppose (a, b), (b, c) R.

a
= 2i
b
b
(b, c) R j Z, = 2i
c

(a, b) R i Z,

a b
a
= = 2i+j .
c
b
c
Since i + j Z, therefore, (a, c) R.
Therefore, R is an equivalence relation.
2. Given any natural number x, there exist k, i N such that, k is not
divisible by 2 and x = k2i . For any natural number y = k2j , we have
x
ij
, and i j is an integer. Thus (x, y) R, namely, x and y are in
y =2
c Chung-Chih Li, Kishan Mehrotra

4.3. Solutions

179

the same cell. Therefore, the cells of R are as follows:

{1 2i | i = 0, 1, 2, . . .},
{3 2i | i = 0, 1, 2, . . .},
{5 2i | i = 0, 1, 2, . . .},
{7 2i | i = 0, 1, 2, . . .},
..
.

Solution 22:
By theorem 4.1, each partition determines an equivalence
relation on A. Therefore, this problem is exactly same problem 19. We just
have to list all possible partitions of A. They are:
{{1, 2, 3}}, {{1}, {2, 3}}, {{1, 2}, {3}}, {{2}, {1, 3}}, {{1}, {2}, {3}}.

Solution 23:

Let S = {a, b, c}, and

R1 = {(a, a), (b, b), (c, c), (a, b), (b, a)}
R2 = {(a, a), (b, b), (c, c), (a, c), (c, a)}.

Both R1 and R2 are equivalence relations, but R1 R2 is not. Because, R1 R2

is not transitive, where (b, a), (a, c) (R1 R2 ), but (b, c) 6 (R1 R2 ).
Note: The results in problem 15 do not directly give us the answer to this
problem. The two relations given in problem 15 are not equivalence relations.

Solution 24:
Let R1 and R2 be two partial order relation on S. The union
R1 R2 may not be a partial order relation, because the union operation does
not preserve antisymmetry and transitivity. Consider the following example.
Let S = {a, b, c}, and
R1 = {(a, a), (b, b), (c, c), (a, b)}
R2 = {(a, a), (b, b), (c, c), (b, a), (b, c)}.
c Chung-Chih Li, Kishan Mehrotra

180

4. Relations

Both R1 and R2 are partial order relations, but R1 R2 is not. Because both
(a, b) and (b, a) are both in R1 R2 but a 6= b, thus R1 R2 is not antisymmetric.
(a, b) and (b, c) are both in R1 R2 but (a, c) 6 R1 R2 , thus R1 R2 is not
transitive.
Note: Although we have shown that both antisymmetry and transitivity cannot be preserved by the union operation, but one is enough for us to claim
that partial order relations are not closed under union.

Solution 25:
that:
Reflexivity:

From the basic theorem for sets in chapter 1, it is easy to see

For all A P(N), A S.

Antisymmetry:
Transitivity:

For all A, B P(N), if A B and B A, then A = B.

For all A, B, C P(N), if A B and B C, then A C.

Therefore, is a partial order relation on P(N).

Solution 26:
1. P2 is the only one that is not a partition of S, because, in which {7, 4, 3, 8}
{1, 5, 10, 3} 6= .
2. P4 is a refinement of both P1 and P3 , because itself is a partition of S and
every element of P4 is a subset of one of the elements in P1 and P3 .

Solution 27:
The number of refinements of a partition P is the number of
the ways to further partition cells in P . The cell {1, 2, 3} has 5 ways, {4, 5} has
2 ways, and {6} has one way. Therefore, the total number of refinements of P
is 5 2 1 = 10.
2

c Chung-Chih Li, Kishan Mehrotra

4.3. Solutions

181

Solution 28: S Since has been removed from Q by definition, we have to

prove only that Q = A and, for any p Q and q Q, if p 6= q, then p q = .
S
1. To prove Q = A, consider
[
a
Q i, j [a (Ci Dj )]

by the definition of Q

i, j [a Ci and a Dj ]
by the definition of
a A since P1 and P2 are two partiontions on A.
2. To prove, for all p, q Q, if p 6= q then p q = , let p = (Ci Dj ) and
q = (Ck Dl ). Suppose p 6= q. If so, it must be the case that, i 6= k
or j 6= l. Since P1 and P2 are two partitions, we have if i 6= k then
Ci Ck = ; if j 6= l then Dj Dl = . Thus,
p q = (Ci Dj ) (Ck Dl )
= (Ci Ck ) (Dj Dl )
= ( (Dj Dl )) or ((Ci Ck ) )
= .
From 1. and 2., Q is a partition of A.
Moreover, given any (Ci Dj ) Q, since (Ci Dj ) Ci , it follows that Q
is a refinement of P1 . Likewise, (Ci Dj ) Dj , and hence Q is a refinement of
P2 .
2

c Chung-Chih Li, Kishan Mehrotra

Chapter 5

Functions
A new problem ... is like a young rice plant,
which can only thrive and bear fruit when it is carefully
grafted onto an old stem according to
the rules of art of the gardener,
the stem here being the secure treasury of mathematical knowledge.
David Hilbert

5.1. Definitions, Theorems, and Comments

5.1

185

Definitions, Theorems, and Comments

Function is one of the most frequent used terms in mathematics. For example,
f (x) = x2 , g(x, y) = x2 + y 2 , ... are typically functions we have studied since
middle school. In general, f and g are the names of the functions, which are
used to identify the functions. x2 and x2 + y 2 are the bodies of the functions
f and g respectively, which show the meanings of the functions. x and y are
called arguments.
In this chapter, we will introduce important formal terminologies and definitions for functions, and study some typical properties of functions.

5.1.1

Definitions

Definition 5.1: Let S and T be two sets. We formally write f : S T to

say that f is a mapping of the elements of the set S to elements of the set
T . The set S is known as the domain set, and the set T is known as the
codomain set.
Let x S. f (x) = y means that if we evaluate the function f at x, the
value y will be obtained.
Definition 5.2: f : S T is a function stands for a mapping which takes
each element of the set S and associates with it one and only one element
of the set T . For x S the value f (x) T is also known as the image of x.
Comment: Most of the time, function, mapping, correspondence,
and transformation are different words that carry the same the same
concept. We choose to use function in our discussion.
Definition 5.3: Let f : S T be a function and x S. If there is an y T
such that f (x) = y, we say that the function, f , is defined at x.
Definition 5.4: Let f : S T be a function. If f is defined at every x S,
then such a function is called total. If there is an x S such that, for every
y T , f (x) 6= y, in another words, f is undefined at x, then such a function
is called partial.
Comment: We only consider total functions in this chapter. Therefore,
all functions in the following discussion refer to total functions.
Comment: Function is a special case of relation. It is also possible to
define the concept of function in terms of relation. A function f is a
relation f S T that satisfies a very special property, namely, if
(x, y) f and (x, z) f , then y = z. This condition says that there is
a unique value associated with each argument x S. One can observe
that this condition is equivalent to Definition 2
c Chung-Chih Li, Kishan Mehrotra

186

5. Functions

Definition 5.5: Let f : S T be a function. As stated earlier, S is known

as the domain of f , and T is known as the codomain of f . The set
{y : x S such that f (x) = y and y T }
is known as the range.
Definition 5.6: Let f : S T be a function. If f satisfies the property,
whenever f (x1 ) = y and f (x2 ) = y, then x1 = x2 ,
then f is known as a one-to-one function. In logics,
x1 , x2 S [(f (x1 ) = f (x2 )) (x1 = x2 )].
A one-to-one function is also known as an injection.
Definition 5.7: Let f : S T be a function. If the range of f = T , then f is
known as an onto function. In logic,
y T x S [f (x) = y].
An onto function is also known as a surjection.
Definition 5.8: If f : S T is a function that is one-to-one and onto, then
the function is also known as a bijection.
Definition 5.9: If f : S S is a bijection and S is finite, then such function
f is also known as a permutation of S.
Definition 5.10: Consider a function f : S S. (Note that the domain and
codomain sets are the same.) The function defined such that f (x) = x for
all values of x S is known as the identity function. It is customary to
denote the identity function as i or iS if we wish to remember the set on
which the identity function is defined.
Definition 5.11: Suppose there are two functions f : S T and g : T U ,
(Note that the codomain of f and the domain of g are the same.) Then we
denote the composition of these two functions as g f : S U and define
it as follows:
x S, (g f )(x) = g(f (x)).
Definition 5.12: Suppose f : S T is a function. The inverse function of f
is a function g : T S such that the following two properties are satisfied:
f g = iS and g f = iT .
It is customary to denote the inverse function of f by f 1 .
c Chung-Chih Li, Kishan Mehrotra

5.2. Definitions, Theorems, and Comments

187

Definition 5.13: Suppose that f : S T is a function. A new function

defined using the properties of f is known as an induced function (induced
by f ). We define two such functions below. Let P(S) and P(S) denote the
power sets of S and T , respectively.
1. The function fb : P(S) P(T ) is induced by f and is defined as: For
all A S,
fb(A) = {y : y = f (x) for x A}.
2. The function fb1 : P(T ) P(S) is induced by f and is defined as:
For all B T ,
fb1 (B) = {x : x S such that f (x) B}.
Definition 5.14: Let S be a set, A S, and f a function defined as follows:

f (x) =

1
0

if x A
if x
6 A.

Then f is known as the characteristic function of the set A. Typical notation

for a characteristic function is A .

5.1.2

Theorems

Theorem 5.1: The inverse of a function f : S T exists if and only if f

is a bijection.
Theorem 5.2: If f : S S is a permutation, then f 1 : S S is also a
permutation.
Theorem 5.3: If f : S S and g : S S are permutations, then f g
and g f are both permutations.
Theorem 5.4: Let f : X Y , A X and B Y. We have the following
properties.
1. fb1 (fb(A)) A.
2. fb(fb1 (B)) B.
3. fb1 (fb(A)) = A, when f is injective.
4. fb(fb1 (B)) = B, when f is surjective.
c Chung-Chih Li, Kishan Mehrotra

188

5.2

5. Functions

The Pigeonhole Principle

The Pigeonhole Principle is an easy yet powerful tool in the study of combinatorial mathematics. For example, many difficult proofs of the theorem in
Ramsey Theory are based on the principle. Consider the following intuitively
understandable statement:

The Pigeonhole Principle: If we wish to distribute k pigeons to n pigeonholes and k > n, then there must exist a pigeonhole with at least two
pigeons.

This principle is based on an important observation of the one-to-one functions. Let S and T be two finite sets and |S| < |T |, i.e., the number of elements
in S is less than the number in T . Then it is impossible to define a one to one
function from s to T .
Theorem 5.5 Let S = {1, 2, . . . , 9}. Any six-element-subset of S must have
at least two elements with sum equal 10.
Proof: The set S = {1, 2, . . . , 9} can be partitioned into five cells as follows.
{1, 9}, {2, 8}, {3, 7}, {4, 6}, {5}.
By the Pigeonhole Principle, if 6 different numbers are picked from S, then at
least one cell must contribute 2 numbers. Clearly, the sum of the two numbers
is 10 according to our partition.
2
Theorem 5.6 Let x0 , x1 , x2 , . . . , xn2 be any sequence of n2 + 1 distinct numbers. Then, there must exist a subsequence of n + 1 numbers that is either
increasing or decreasing. That is, there exists a sequence xi0 , xi1 , . . . , xin
such that, i0 < i1 < < iin and either xi0 < xi1 < . . . < xin or
xi0 > xi1 > . . . > xin .
Proof: For each xi in the sequence, let (ai , bi ) be the pair of numbers such that,
ai is the length of the longest increasing subsequence ending at xi and bi is the
length of the longest decreasing subsequence ending at xi . We observe that, if
0 i, j n and i 6= j, then (ai , bi ) 6= (aj , bj ). Since we have n2 + 1 numbers
in the sequence, it is impossible to associate every number in the sequence a
pair of numbers each is less than n. (Note: the number of possible pairs of two
numbers from {1, 2, . . . n} is n2 .) Therefore, there must be an xi that is the end
of an increasing or decreasing subsequence of length at least n + 1.
2
c Chung-Chih Li, Kishan Mehrotra

5.3. Asymptotic Notations

5.3

189

Asymptotic Notations

This is section, we introduce an important notation in computer science. It

is about the set of functions that are considered equivalent in the context of
computational complexity analysis.
We use the following conventions throughout this section. We use f, g, or h
to denote a total function from N to N. P and Q are one-place predicates over
N. m, n, x, y, and these with subscripts are variables range over N. Let R+ be
the set of all positive real numbers, and a, b, c, range over R+ . In addition to
the standard quantifier in predicate logic introduced in Chapter 2, we bring up
the following notations for convenience.

Definition 5.15: xP (x)

Definition 5.16: xP (x)

nx[x n P (x)].

nx[(x n) P (x)].

Theorem 5.7

x[P (x) Q(x)] [ xP (x) xQ(x)].

Proof: For () direction:

x[P (x) Q(x)]

nx (x n) [P (x) Q(x)]

nx (x n) [P (x) Q(x)]

nx [(x n) P (x)] [(x n) Q(x)]

n x[(x n) P (x)] x[(x n) Q(x)]

= nx[(x n) P (x)] nx[(x n) Q(x)]
nx[(x n) P (x)] nx[(x n) Q(x)]

xP (x) xQ(x).

For () direction: We will use the following implication:

[(A C) (A B) (C D)] = [A (B D)].

xP (x) xQ(x)
nx[(x n) P (x)] nx[(x n) Q(x)]
x[(x n0 ) P (x)] x[(x n1 ) Q(x)], where n0 , n1 N

x [(x n0 ) P (x)] [(x n1 ) Q(x)] .

c Chung-Chih Li, Kishan Mehrotra

190

5. Functions

We have two cases: n0 n1 and n0 < n1 . We now consider the first case.
Suppose n0 n1 . In this case, we have (x > n0 ) (x > n1 ). Thus,

x [(x n0 ) P (x)] [(x n1 ) Q(x)]

x [(x > n0 ) (x > n1 )] [(x n0 ) P (x)] [(x n1 ) Q(x)]

= x[(x > n0 ) (P (x) Q(x))]
nx[(x > n) (P (x) Q(x))]

x[P (x) Q(x)].

For the other case, n0 < n1 , the proof is similar.

However, the distributive law of over does not hold. Consider the
following properties. We leave the proofs to the reader as exercises.
Exercise:

1. Prove that xP (x) xP (x).

2. Prove that x[P (x) Q(x)] = [ xP (x) xQ(x)].

3. Disprove that x[P (x) Q(x)] = [ xP (x) xQ(x)].

4. Prove that x[P (x) Q(x)] [ xP (x) xQ(x)].

5. Prove that x[P (x) Q(x)] = [ xP (x) xQ(x)].

6. Disprove that x[P (x) Q(x)] = [ xP (x) xQ(x)].

Theoretically, we can modify a function at finitely many places in an efficient
way. This viewpoint leads to the use of asymptotic notations where we neglect
finitely many points at which the functions may violate some nice properties
concerned. O(g), (g), and (g) are sets of functions from N to N defined in
the following.

Definition 5.17: f O(g) c n[f (n) cg(n)].

Definition 5.18: f (g) c n[cg(n) f (n)].

Definition 5.19: f (g) c1 c2 n[c1 g(n) f (n) c2 g(n)].

Theorem 5.8
f (g) [f O(g) f (g)].
c Chung-Chih Li, Kishan Mehrotra

5.3. Asymptotic Notations

191

Proof:
f (g)

c1 c2 x[c1 g(x) f (x) c2 g(x)]

c1 c2 x [c1 g(x) f (x)] [f (x) c2 g(x)]
h
i

c1 c2 x[c1 g(x) f (x)] x[f (x) c2 g(x)]

c1 c2 x[c1 g(x) f (x)] c1 c2 x[f (x) c2 g(x)]

c x[cg(x) f (x)] c x[f (x) cg(x)]

f (g) f O(g).

2
Theorem 5.9
[f1 O(g) f2 O(g)] (f1 + f2 ) O(g).
Proof: For () direction:
f1 O(g) f2 O(g)

c x[f1 (x) cg(x)] c x[f2 (x) cg(x)]

x[f1 (x) c1 g(x)] x[f2 (x) c2 g(x)]

x[f1 (x) c1 g(x) f2 (x) c2 g(x)]

x[(f1 (x) + f2 (x) (c1 + c2 )g(x)]

c x[(f1 (x) + f2 (x) cg(x)]

(f1 + f2 ) O(g).

For () direction:
(f1 + f2 ) O(g)

c x[(f1 (x) + f2 (x) cg(x)]

c x[f1 (x) cg(x) f2 (x) cg(x)]

h
i

c x[f1 (x) cg(x)] x[f2 (x) cg(x)]

= c x[f1 (x) cg(x)] c x[f2 (x) cg(x)]

f1 O(g) f2 O(g).

2
Theorem 5.10
[f1 (g) f2 (g)] = (f1 + f2 ) (g).
c Chung-Chih Li, Kishan Mehrotra

192

5. Functions

Proof:
f1 (g) f2 (g)

c1 c2 x[c1 g(x) f1 (x) c2 g(x)] c1 c2 x[c1 g(x) f2 (x) c2 g(x)]

x[a1 g(x) f1 (x) b1 g(x)] x[a2 g(x) f2 (x) b2 g(x)]

x[a1 g(x) f1 (x) b1 g(x) a2 g(x) f2 (x) b2 g(x)]

x[(a1 + a2 )g(x) (f1 (x) + f2 (x)) (b1 + b2 )g(x)]

c1 c2 x[c1 g(x) (f1 (x) + f2 (x)) c2 g(x)]

(f1 + f2 ) (g).

2
Theorem 5.11
f (g) g (f ).

f (g)

x c1 g(x) f (x) c2 g(x)

x [c1 g(x) f (x)] [f (x) c2 g(x)]

1
1
x [g(x) f (x)] [ f (x) g(x)]
c1
c2

1
1
f (x) g(x) f (x)
x
c2
c1

g (f ).

2
Exercise 1 Prove:
[f (g) g (h)] = f (h).

(5.1)

Theorem 5.12
f (g) (f ) = (g).
Proof: The proof of () direction is trivial.
For () direction, suppose f (g). We shall prove (f ) (g) and
(f ) (g).
c Chung-Chih Li, Kishan Mehrotra

5.3. Asymptotic Notations

193

1. Let h (f ). Since h (f ) and f (g), by (5.1), we have h (g).

Thus, (f ) (g).
2. Let h (g). Since f (g), by Theorem 11, we have g (f ). Then,
by (5.1), h (g) and g (f ) imply h (f ). Thus, (g) (f ).

Therefore, (g) = (f ).
Let a, b R+ . Then, an (bn ) a = b.

Theorem 5.13

Proof: The proof of () direction is trivial.

For () direction, suppose an (bn ).
an (bn )

=
=
=
=

If x , then

log c1
n

n
n
n
x [c1 b a c2 b ]

n
n
n
x [log(c1 b ) log a log(c2 b )]

x [(log c1 + n log b) n log a (log c2 + n log b)]

log c1
log c2
+ log b log a
+ log b .
x
n
n

0 and

log c2
n

0. Thus, when x , we have

log b log a log b.

Therefore, a = b.

c Chung-Chih Li, Kishan Mehrotra

194

5.4

5. Functions

Problems

Problem 1: List all total and partial functions with domain {1, 2} and
codomain {a, b}.
Problem 2:

Which of the following functions are: injective, surjective?

1. g : R R. g(x) := 2x + 1.
2. h : N N. h(x) := x2 + 2.
Problem 3: Let D = {1, 2, 3, 4} be the domain and N be the codomain of
the functions f1 and f2 defined as follows:
1. For all b in D, f1 (b) := b2 .
2. For all b in D, f2 (b) := the smallest prime that divides 3b + 1.
Express f1 and f2 in two relations.
Problem 4: Let A be any set. Consider relations on A and functions from
A to A as sets of ordered pairs. Find all equivalence relations on A which
are also functions from A to A. Explain.
Problem 5: Let f and g be injective functions, f : X Y and g : W
X, for sets W, X, Y . Prove that the composition f g is injective.
Problem 6: Give an example of two functions f : D Y and g : Y W
such that D, Y, and W are finite sets and g f is bijective, but neither
f nor g is bijective.
Problem 7:
that

Let f be a function f : X Y , and A X, B X. Prove

f(A B) f(A) f(B).

Is the reverse inclusion true?

Problem 8: Let A and B be subsets of X such that A B. Prove that
f(A) f(B).
Problem 9:

Let B Y and A = f1 (B). Prove that

f(A) B

and that in some cases equality does not hold.

Problem 10: Let f be a function and let A be a subset of the domain of f .
Assume that f1 (f(A)) A. Prove that A = f 1 (f(A)).
Problem 11:

Let f : X Y be a function.

1. Prove that if f is injective and A X, then f1 (f(A)) = A. Explain

where you use the injectivity.
c Chung-Chih Li, Kishan Mehrotra

5.4. Problems

195

2. Prove that if f is surjective and B Y , then f(f1 (B)) = B.

Explain where you use the surjectivity.
i

Let the function composition f f f be denoted as f i . If

|
{z
}
i = 0, then f i is the identity function. Let f be a permutation of a finite
set X. For each a X define S(a) := {f i (a); i = 0, 1, 2, . . .}. Define the
relation R on X by the rule: for all a, b X, aRb iff b S(a). Prove that
R is an equivalence relation on X.

Problem 12:

Problem 13: Suppose there are n people lined up in some order. Show that
you can achieve any other ordering of them by successively having some
two adjacent people trade places.
Problem 14: Prove that in a group of 700 people, there must be 2 people
who have the same first and last initials.
Problem 15: Prove that at any party there must be two people who have
shaken hands with the same number of others present.

c Chung-Chih Li, Kishan Mehrotra

196

5.5

5. Functions

Solutions

Solution 1:

f0
f1
f2
f3
f4

= ,
= {(1, a)}
= {(1, b)}
= {(2, a)}
= {(2, b)}

f5
f6
f7
f8

= {(1, a), (2, b)}

= {(1, b), (2, a)}
= {(1, a), (2, a)}
= {(1, b), (2, b)}

Note that f0 , f1 , f2 , f3 , and f4 are partial functions, and f5 , f6 , f7 and f8 are

total functions. Dont forget that is a partial function.
2

Solution 2:
1. g : R R. x R, g(x) := 2x + 1.
(a) g is injective. Because, for all x1 , x2 R,
x1 6= x2 2x1 + 1 6= 2x2 + 2
g(x1 ) 6= g(x2 ).
(b) g is surjective. Because, for all y R,
x R, x = (y 1)/2,
x R, f (x) = y.
2. h : N N. x N, h(x) := x2 + 2.
(a) h is injective. Because, for all x1 , x2 0,
x1 6= x2 x21 + 2 6= x22 + 2
h(x1 ) 6= h(x2 ).
and, x1 , x2 N x1 , x2 0.
(b) h is not surjective. Because, given any y N, we cannot always find
x N such that, f (x) = y. For example, if y = 1, there is no x N
such that f (x) = x2 + 2 = 1.

Solution 3:
c Chung-Chih Li, Kishan Mehrotra

5.5. Solutions

197

1. f1 = {(1, 1), (2, 4), (3, 9), (4, 16)}.

b
b2

1
1

2
4

3
9

4
16

2. f2 = {(1, 2), (2, 7), (3, 2), (4, 13)}.

b
3b + 1
f2

1
4
2

2
7
7

3
10
2

4
13
13

Note: For many applications in computer science and mathematics, by

convention, we consider 2 as the smallest prime number. But, if you
examine the definition of prime numbers: p is a prime number iff p is a
positive integer and p is divisible by 1 and p only, you may consider 1 as
the smallest prime number. Then,
f2 = {(1, 1), (2, 1), (3, 1), (4, 1)}.

Solution 4:
Let A be any set {a, b, c, . . .}. The only equivalence relation R
on A which is also a function from A to A is,
{(a, a), (b, b), (c, c), . . .}.
Why? The restriction of a relation being a function is that the relation has to
be single valued, i.e., if (x, y) and (x, z) in the relation, then y = z. And, since
R is an equivalence relation, we know that for each element a A, (a, a) R.
Therefore, (a, x) 6 R unless a = x.
2

Solution 5:
Let f and g be injective functions, f : X Y and g : W X,
for sets W, X, Y . We first prove that f g : W Y is well defined. The only
thing we have to do is to prove that f g as a relation on W Y is single valued.
Given w1 , w2 W . If w1 = w2 and g(w1 ), g(w2 ) are defined, then
g(w1 ) = g(w2 ),
because g is a well defined function. And, if g(w1 ) = g(w2 ) and f (g(w1 )),
f (g(w2 )) are defined, then
f (g(w1 )) = f (g(w2 )),
c Chung-Chih Li, Kishan Mehrotra

198

5. Functions

because f is a well defined function. Therefore, if w1 = w2 and f g(w1 ),

f g(w2 ) are defined, then
f g(w1 ) = f g(w2 ).
That means f g is a well defined function from W to Y .
To prove that f g is injective, let x, y W , and suppose that x 6= y and
f g(x), f g(y) are defined.
x 6= y g(x) 6= g(y)
f (g(x)) 6= f (g(y))

g is injective
f is injective.

Therefore, f g is injective.

Solution 6:

Let D = {1, 2}, Y = {a, b, c}, W = {, }. Define

f : D Y

as

1
x
f (x) a

2
b

g : Y W

as

x
a
g(x)

Wh have
g f : D W

as

1
x
(g f )(x)

f is not surjective, and g is not injective, but g f is bijective.

Solution 7:

Let f : X Y , and A X, B X.

y f (A B)

x (A B), f (x) = y
x, x A, x B, f (x) = y
x A, f (x) = y and x B, f (x) = y
y f (A) and y f (B)
y (f (A) f (B))

Therefore,
f (A B) f (A) f (B).
However, the the reverse inclusion is not true. Here is a counter example.
Define f : Z Z, f (x) = x2 . Let, A = {2}, B = {2}. We have
f (A) = {4}, f (B) = {4}, f (A) f (B) = {4}, f (A B) = .
c Chung-Chih Li, Kishan Mehrotra

5.5. Solutions

199

Therefore, in general, f (A B) 6 f (A) f (B)

Solution 8:

Let y f(A).
y f(A)

x A, f (x) = y
x B, f (x) = y
y f(B).

because A B

Therefore, f(A) f(B).

Solution 9:
The only thing having to do with this problem is the definitions
of the sets f(A) and f1 (B). We carefully write down the definitions as follows.
Let f : X Y , A X and B Y.
f( A) := {b | a A, f (a) = b}.

(5.2)

f1 (B) := {a | a X, f (a) B}.

(5.3)

Given A = f1 (B). To prove that f(A) B, let b f(A).

b f(A) a A, f (a) = b. by def. (5.2)
Fix this a. Because A = f1 (B), we know that a f1 (B). From the definition
in (5.3),
a f1 (B) f (a) B.
Therefore, b B.
If f is not surjective, then f(A) B is not necessary to be true. For example,
define
f : {0, 1} {0, 1},
f (0) = 0,
f (1) = 0.
Let B = {0, 1}, A = f1 (B). We have
A = {0, 1},
f(A) = {0}.
Therefore, B 6 f(A).

2
c Chung-Chih Li, Kishan Mehrotra

200

5. Functions

Solution 10:

Let f : X Y , A X. We first prove that

A f1 (f(A)).
a A f (a) f(A)
a f1 (f(A))

(5.4)

by def. (5.3).

Therefore, A f1 (f(A)).
If f1 (f(A)) A is given, then, with the fact (5.4) we just proved, we can
conclude that A = f 1 (f(A)).
2

Solution 11:

Let f : X Y.

1. With the fact (5.4) that we have proven in the previous problem, this
problem is in fact asking to prove that if f is injective, and A is any
subset of X, then
f1 (f(A)) A.
Suppose f is injective.
a f1 (f(A)) f (a) f(A)
x0 A, f (a) = f (x0 )
x0 A, a = x0
f is injective
a A.
Therefore, f1 (f(A)) A, plus (5.4), gives
f1 (f(A)) = A.
2. Lets first prove a result similar to (5.4).
f(f1 (B)) B.
b f(f1 (B)) a f1 (B), f (a) = b
a X, f (a) B, f (a) = b
b B.
That proves (5.5). Now, suppose f is surjective.
b B a X, f (a) = b.
c Chung-Chih Li, Kishan Mehrotra

(5.5)

5.5. Solutions

201

Because f is surjective, given any b in the codomain, we are able to find

such a in the domain and f (a) = b. Fix this a. From the definition in
(5.2), we know that
a f1 (B) f (a) = b

f (a) f(f1 (B)) f (a) = b

b f(f1 (B)).

Therefore, B f(f1 (B)), plus (5.5) and f is surjective, gives

f(f1 (B)) = B.

Solution 12:
Reflexivity:

By the definition, f 0 = iX . Thus, for all a X.

a = f 0 (a), a S(a).

Therefore, (a, a) R.
Symmetry: Given any a, b X, and (a, b) R, i.e., exists i 0, b = f i (a).
Consider the following sequence,
f 0 (a), f 1 (a), . . . , f i (a), . . . , f j (a), . . . , f j+k (a), . . . ,
where f 0 (a) = a and f i (a) = b. Because f is a permutation on X, we are
able to find j, k such that
i j, 0 k, f j (a) = f j+k (a) = c,
where c X.1 Now, we want to prove that a is in the cycle
f j (a), f j+1 (a), . . . , f j+k (a).
For any two sequences, let
a0 , a1 , . . . , an b0 , b1 , . . . , bn iff a0 = b0 , a1 = b1 , . . . , and an = bn .
By way of contradiction, suppose that
a 6 {f j (a), f j+1 (a), . . . , f j+k (a)}.
There are two cases:
1 In

fact, i = j and b = c, but suppose we dont know this fact.

c Chung-Chih Li, Kishan Mehrotra

202

5. Functions
Case 1: j k. Because f is injective, f (a) = f (b) implies a = b. By
the assumption that f j (a) = f j+k (a) = c, we have
f j (a) = f j+k (a) f (f j1 (a)) = f (f k+j1 (a))
f j1 (a) = f k+j1 (a)
f j2 (a) = f k+j2 (a)
..
.
f 1 (a) = f k+1 (a)
f 0 (a) = f k (a).
Therefore, f k (a) = f 0 (a) = a, and
f 0 (a), f 1 (a), . . . , f j (a) f k (a), f k+1 (a), . . . f j+k (a).
From the assumption in this case that j k j + k, we know
a {f j (a), f j+1 (a), . . . , f j+k (a)}.
That leads a contradiction.
Case 2: k < j. With the same reason that f is injective, we can have
the following deduction,
f j (a) = f j+k (a) = c f j1 (a) = f j+k1 (a)
f j2 (a) = f j+k2 (a)
..
.
f jk (a) = f j (a).
Therefore, we have
f jk (a), f jk+1 (a), . . . , f j (a) f j (a), f j+1 (a), . . . f j+k (a),
where f jk = f j (a) = c and 0 < j k. From the assumption,
a 6 {f j (a), f j+1 (a), . . . f j+k (a)}
a
/ {f jk (a), f jk+1 (a), . . . , f j (a)}.
Now, if j k is still greater than k, we repeat the above arguments,
until, after n times, we have
f jnk (a), f jnk+1 (a), . . . , f jnk+k (a)
f jnk+k (a), f jnk+1 (a), . . . f jnk+2k (a),

c Chung-Chih Li, Kishan Mehrotra

5.5. Solutions

203

where j nk k. We also know that

f jnk (a), f jnk+1 (a), . . . , f jnk+k (a)
f jnk+k (a), f jnk+k+1 (a), . . . f jnk+2k (a)


f j (a), f j+1 (a), . . . f j+k (a),
and,

a 6 {f jnk (a), f jnk+1 (a), . . . , f jnk+k (a)}.

Then, we use the similar argument in case 1. We will have

f 0 (a), f 1 (a), . . . , f jnk (a) f k (a), f k+1 (a), . . . f jnk+k (a).
a {f jnk (a), f jnk+1 (a), . . . , f jnk+k (a)}
a {f j (a), f j+1 (a), . . . f j+k (a)}.
That leads a contradiction.
In both cases, the assumption will lead a contradiction. Therefore, we can
conclude that
a {f j (a), . . . , f j+k (a)}.
(5.6)
Let a = f l (a), where i j l j + k and
a = f l (a) = f li (f i (a)) = f li (b).
Since l i 0, thus a S(b), and (b, a) R. Therefore, R is symmetric.
Transitivity:

For all a, b, c X,
(a, b), (b, c) R

b S(a) and c S(b)

i, j 0, b = f i (a) and c = f j (b)
c = f j (f i (a))
c = f i+j (a)
(a, c) R.

R is transitive.
Therefore, R is an equivalence relations.

Solution 13:
If we do some experiment, we will find that it is not so difficult
to see that we are able to reorder a line of people by successively switching two
adjacent people. The problem is, how to prove it formally. It is indeed what we
try to learn in this class. As to conclude the homework solutions for this class,
let me prove it by mathematical induction on the number of people lined up.
c Chung-Chih Li, Kishan Mehrotra

204

5. Functions
At first, lets define a class of functions as the following. For any x, y,

y if a = x,
x if a = y,
fx,y (a) =
(5.7)

a otherwise.

Apparently, if we restrict the domain X to a finite one that includes x and y,

then fx,y is a permutation on X, which only exchanges x and y. Therefore, if
we can prove that any order can be achieved from any other order by using the
functions in this class, then we would have proved this problem.
Inductive Basis:

n = 2. It is clear that f1,2 can do the job.

[The Basis Holds.]

Inductive Hypothesis: If there are n people lined up, then any order of the
line can be achieved from any other order by using the functions in the
class defined in (5.7).
Inductive Step:
order.

Suppose there are n + 1 people lined up in the following

a1 , a2 , . . . , an , an+1 .

(5.8)

And, given any another order,

b1 , b2 , . . . , bn , bn+1 .

(5.9)

Suppose b1 = ak , where 1 k n + 1. We can apply

f1,2 f2,3 . . . fk2,k1 fk1,k

(5.10)

to (5.8), and result in

ak , a1 , . . . , ak1 , ak+1 , . . . , an+1 .

(5.11)

Now, we have to permute (5.11) into (5.9). Since ak is already in the first
position required by the target order (5.9), we only have to permute
a1 , . . . , ak1 , ak+1 , . . . , an+1

(5.12)

b2 , . . . , bn , bn+1 .

(5.13)

into
From the inductive hypothesis, we can make a composition function g of
the functions in (5.7) to do the job. Let f 0 denote the composition function
(5.10). We have g f 0 , which will perform the permutation from (5.8) into
(5.9), and g f 0 is a composition of functions from the class defined in
(5.7).
[The Inductive Step Holds.]
c Chung-Chih Li, Kishan Mehrotra

5.5. Solutions

205

That completes the proof.

Solution 14:
initials are

The number of all possible combinations of the first and last

26 26 = 676.

Thus, by the Pigeonhole Principle, assigning 700 people to 676 different initials
must introduce at least 2 people who have the same initials.
2

Solution 15:
Suppose there is a party with n people. We may consider the
following two cases.
Case 1: Every body has shaken hands. Then, the possible numbers of people
for each person who has shaken hands with are 1, 2, 3, . . . , n 1. Every
body got a number from those possible numbers. We have n people,
and n 1 numbers. By the Pigeonhole Principle, at least one number is
assigned to two people.
Case 2: There is one person who does not shake hands with any others. In
this case, the possible numbers become 0, 1, 2, . . . , n 2. Please note,
n 1 is not allowed, otherwise, there is one person has shaken hands with
everybody in the party, and that contradicts the assumption. Just like
case 1, we have n people and n 1 numbers. By the Pigeonhole Principle,
at least one number is assigned to two people.

c Chung-Chih Li, Kishan Mehrotra