Lecture Notes for

Math 135:
Algebraic Reasoning
for Teaching Mathematics

Steffen Lempp
Department of Mathematics, University of Wisconsin,
Madison, WI 53706-1388, USA
E-mail address: [email protected]
URL: http://www.math.wisc.edu/~lempp
Key words and phrases. algebra, equation, function, inequality, linear
function, quadratic function, exponential function, student
misconception, middle school mathematics

These are the lecture notes for the course Math 135:
“Algebraic Reasoning for Teaching Mathematics”
taught at UW-Madison in spring 2008.
The preparation of these lecture notes was partially supported
by a faculty development grant of the College of Letters and Science
and by summer support by the School of Education,
both of the University of Wisconsin-Madison.
Feedback, comments, and corrections are welcome!
This text is currently in a constant state of change,
so please excuse what I am sure will be numerous errors and typos!
Thanks to Dan McGinn and my Math 135 students in spring 2008
(Rachel Burgan, Erika Calhoon, Lindsay Feest, Amanda Greisch,
Celia Hagar, Brittany Hughes, and Shannon Osgar)
for helpful comments and corrections.
© 2008 Steffen Lempp, University of Wisconsin-Madison

This material may only be reproduced and distributed for cost of

reproduction and distribution for educational/non-profit purposes.
This book is currently available for free download from
http://www.math.wisc.edu/~lempp/teach/135.pdf.
Any dissemination of this material in modified form requires the
author’s written approval.
 Contents

Preface vi
Introduction to the Instructor viii
Introduction to the Student x

Chapter 1. Numbers and Equations: Review and Basics 1

1.1. Arithmetic in the Whole Numbers and Beyond:
A Review 1
1.1.1. Addition and Multiplication 1
1.1.2. Subtraction and Division 4
1.2. Letters and Algebraic Expressions 8
1.2.1. Letters in Algebra 8
1.2.1.1. Letters as Variables 9
1.2.1.2. Letters as Unknowns 10
1.2.1.3. Letters as Parameters 10
1.2.1.4. Letters as Constants: 11
1.2.2. Algebraic Expressions 12
1.3. Equations and Identities 13
1.4. Manipulating Algebraic Expressions 13
1.4.1. Using the definition of subtraction 14
1.4.2. “Combining Like Terms” 14
1.4.3. Removing Parentheses 15
1.5. Conventions in Algebra 15
1.5.1. Notation for Sets, and Some Special Sets of Numbers 15
1.5.2. Order of Operations and Use of Parentheses 17
1.6. Algebra as “Undoing Expressions” 19
Chapter 2. Linear Equations 21
2.1. Ratio and Proportional Reasoning 21
2.2. Velocity and Rate of Change 23
2.3. Linear Expressions 26
2.4. Line Equations 30
2.4.1. Graphing a Linear Equation 30
2.4.2. Different Forms of Equations for Lines 35
iii
iv CONTENTS

2.4.3. Varying the Parameters in a Line Equation 41

2.5. Solving a Linear Equation in One Variable 43
2.6. Solving Simultaneous Linear Equations in Two or More
Variables 52

Chapter 3. Order and Linear Inequalities 59

3.1. Ordering the Numbers 59
3.2. Order and the Arithmetical Operations 61
3.3. Linear Inequalities in One Unknown 64
3.4. Estimation and Approximation 69

Chapter 4. The Concept of a Function, and Functions Closely

Related to Linear Functions 72
4.1. The Concept of a Function 72
4.2. Range of a Function, Onto and One-to-One Functions,
and Inverse of a Function 77
4.2.1. Range of a Function, and Onto Functions 78
4.2.2. 1–1 Functions 78
4.2.3. Inverse of a Function 79
4.3. Some Functions Closely Related to Linear Functions 82
4.3.1. Step Functions 83
4.3.2. Piecewise Linear Functions 85
4.3.3. The Absolute Value Function 86
4.3.4. Direct and Inverse Variation 89

Chapter 5. Quadratic Functions, Equations and Inequalities 92

5.1. Introduction to Quadratic Functions 92
5.2. Solving Quadratic Equations 98
5.2.1. Reviewing the Binomial Laws 98
5.2.2. Solving Quadratic Equations by Factoring 99
5.2.3. Solving Quadratic Equations by Taking the Square Root100
5.2.4. Solving Quadratic Equations by Completing the
Square 102
5.2.5. Solving Quadratic Equations by the Quadratic Formula 104
5.2.6. The number of real solutions of a quadratic equation 105
5.3. A Digression into Square Roots
and the Complex Numbers 109
5.3.1. Square Roots 109
5.3.2. The Number i and the Complex Numbers 111
5.4. Graphing Quadratic Functions and
Solving Quadratic Equations Graphically 113
5.5. Quadratic Inequalities 121
 CONTENTS v

Chapter 6. Exponential and Logarithmic Functions, Equations

and Inequalities 124
6.1. Exponentiation: A Review of the Definition 124
6.2. Motivation for Exponentiation 128
6.3. The Exponential Functions 131
6.3.1. Properties of Exponential Functions 131
6.3.2. Another Application of Exponential Functions 135
6.4. The Logarithmic Functions and Solving Exponential
Equations 138
6.4.1. Definition and Properties of Logarithmic Functions 138
6.4.2. Applications of Logarithmic Functions 143
6.4.2.1. The slide rule 143
6.4.2.2. Radioactive Decay 144
6.4.2.3. Earthquakes and the Richter Scale 145
6.4.2.4. Sound Volume and Decibels 146
Bibliography 148
Index 149
 Preface

“The human mind has never invented

a labor-saving device equal to algebra.”
J. Willard Gibbs

The word “algebra” has several meanings in our society.

Historically, the word derives from the book “Al-Kitab al-Jabr wa-
l-Muqabala” (meaning “The Compendious Book on Calculation by
Completion and Balancing”) written by the Persian mathematician
Muhammad ibn Mūsā al-Khwārizmı̄ (approx. 780-850 C.E.), which was
the first book dealing systematically with solving linear and quadratic
equations, based on earlier work by Greek and Indian mathematicians.
(Curiously, the author’s last name is also the source for our word “al-
gorithm”, meaning a concise list of instructions, such as in a computer
program.)
In school mathematics, algebra (often called “elementary”, “inter-
mediate”, “high school”, or “college” algebra) follows the study of
arithmetic: Whereas arithmetic deals with numbers and operations,
algebra generalizes this from computing with “concrete” numbers to
reasoning with “unknown” numbers (“variables”, usually denoted by
letters) using equations, functions, etc.
In upper-level college algebra and beyond, algebra takes on yet
another meaning (which will not be the subject of this text): There,
algebra (now also often called “higher algebra”, “abstract algebra”, or
“modern algebra”) abstracts from the study of numbers to the study
of abstract objects which “behave like” numbers in a very broad sense;
the complex numbers are a very simple (pun intended!) example of
such a structure. It is the school mathematics meaning of the word
“algebra” which will be the topic of this text, and in which meaning
we will from now on exclusively use this word.
The history of algebra started with this meaning of the word, from
the Babylonians (around 1600 B.C.E.) to the Greeks, Indians and Chi-
nese (around 500 B.C.E. to 500 C.E.) to the Persians and Indians
(around 800 C.E. to 1100 C.E.) to the Italians and French (in the 16th
vi
 PREFACE vii

century). Until the Renaissance, most of the attention in algebra cen-

tered around solving equations in one variable. How to solve quadratic
equations by completing the square was already known to the Babyloni-
ans; the Italian mathematicians Scipione del Ferro and Niccolò Fontana
Tartaglia independently, and the Italian mathematician Lodovico Fer-
rari, gave the first general solution to the cubic and quartic equation,
respectively.1 (The Italian mathematician Gerolamo Cardano was the
first to publish the general solutions to cubic and quartic equations,
and the formula for the cubic equation bears his name to this day.)

The French mathematician François Viète (1540 - 1603, more com-

monly known by his Latinized name Vieta) is credited with the first
attempt at giving the modern notation for algebra we use today; before
him, very cumbersome notation was used. The 18th and 19th century
then saw the birth of modern algebra in the other sense mentioned
above, which also led to much more general techniques for solving equa-
tions, including the proof by the Norwegian mathematician Niels Hen-
rik Abel (in 1824) that there is no general “formula” to solve a quintic
equation. (For a brief history of algebra, see the Wikipedia entry on
algebra at http://en.wikipedia.org/wiki/Algebra#History.)

1Here, a quadratic equation is an equation involving numbers, x and x2 ; a cubic

equation is one also involving x3 ; a quartic equation is one additionally involving x4 ;
and a quintic equation (mentioned in the following paragraph) also involves x5 .
 Introduction to the Instructor

These lecture notes first review the laws of arithmetic and discuss
the role of letters in algebra, and then focus on linear, quadratic and ex-
ponential equations, inequalities and functions. But rather than simply
reviewing the algebra your students will have already learned in high
school, these notes go beyond and study in depth the concepts un-
derlying algebra, emphasizing the fact that there are very few basic
underlying idea in algebra, which “explain” everything else there is to
know about algebra: These ideas center around
• the rules of arithmetic (more precisely, the ordered field ax-
ioms), which carry over from numbers to general algebraic
expressions, and
• the rules for manipulating equations and inequalities.
These basic underlying facts are contained in the few “Propositions”
sprinkled throughout the lecture notes.
These notes, however, attempt to not just “cover” the material of
algebra, but to put it into the right context for teaching algebra, by
focusing on how real-life problems lead to algebraic problems, multi-
ple abstract representations of the same mathematical problem, and
typical student misconceptions and errors and their likely underlying
causes.
These notes just provide a bare-bones guide through an algebra
course for future middle school teachers. It is important to combine
them with a lot of in-class discussion of the topics, and especially with
actual algebra problems from school books in order to generate these
discussions. For these, we recommend using the following Singapore
math schoolbooks:
• Primary Math Textbook 5A (U.S. Edition),
• Primary Math Textbook 6A (U.S. Edition),
• New Elementary Math Textbook 1 (Syllabus D),
• New Elementary Math Textbook 2 (Syllabus D),
• New Elementary Math Textbook 3A (Syllabus D), and
• New Syllabus Additional Mathematics Textbook.
viii
 INTRODUCTION TO THE INSTRUCTOR ix

Only a few pages of the first and last of these books need to be used;
the other four books should be used more extensively. The course web
page http://www.math.wisc.edu/~lempp/teach/135.html gives an
idea of how to integrate these various components.
We finally refer to Milgram [2, Chapter 8] and Wu [6] for similar
treatments of algebra, and to Papick [3] and Szydlik/Koker [5] for
very different approaches. A very careful and thorough introduction to
algebra can also be found in Gross [1].
 Introduction to the Student

These notes are designed to make you a better teacher of mathe-

matics in middle school.
The crucial role of algebra as a stepping stone to higher mathe-
matics has been documented numerous times, and the mathematics
curricula of our nation’s schools have come to reflect this by placing
more and more emphasis on algebra. At the same time, it has be-
come clear that many of our nation’s middle school teachers lack much
of the content knowledge to be effective teachers of mathematics and
especially of algebra. (See the recent report of the National Mathemat-
ics Advisory Panel available at http://www.ed.gov/about/bdscomm/
list/mathpanel/report/final-report.pdf, especially its chapter 4
on algebra.)
After some review of the laws of arithmetic and the order of oper-
ations, and looking more closely at role of letters in algebra, we will
mainly study linear, quadratic and exponential equations, inequalities
and functions. But rather than simply reviewing the algebra you have
already learned in high school, we will go beyond and study in depth
the concepts underlying algebra, emphasizing the following topics:

• “problem-solving”: modeling real-life problems (“word prob-

lems”) as mathematical problems and then interpreting the
mathematical solution back into the real-world context;
• “proofs”: making mathematically grounded arguments about
mathematical statements and solutions;
• “analyzing student solutions”: examining the rationale behind
middle school and high school students’ mathematical work
and how it connects to prior mathematical understanding and
future mathematical concepts; analyzing the strengths and
weaknesses of a range of solution strategies (including standard
techniques); and recognizing and identifying common student
misconceptions;
• “modeling”: flexible use of multiple representations such as
graphs, tables, and equations (including different forms), and
x
 INTRODUCTION TO THE STUDENT xi

how to transition back and forth between them; and using

functions to model real-world phenomena;
• “symbolic proficiency”: solving equations and inequalities, sim-
plifying expressions, factoring, etc.
These lecture notes try to make the point that school algebra centers
around a few basic underlying concepts which “explain” all (or at least
most) of school algebra. These are
• the rules of arithmetic for numbers (including those for the
ordering of numbers), and how they naturally carry over from
numbers to general algebraic expressions; and
• the rules for manipulating equations and inequalities, and how
they allow one to solve equations and inequalities.
We will generally label these basic facts “Propositions” to highlight
their central importance. Many, if not most, of the student misconcep-
tions and errors you will encounter in your teaching career will come
from a lack of understanding of these basic rules, and how they natu-
rally “explain” algebra.
 CHAPTER 1

Numbers and Equations: Review and Basics

1.1. Arithmetic in the Whole Numbers and Beyond:

A Review
1.1.1. Addition and Multiplication. Very early on in learning
arithmetic, we realize that there are certain rules which make com-
puting simpler and vastly decrease the need for memorizing “number
facts”. For example, we see that
(1.1) 2+7=7+2
(1.2) 6+0=6
(1.3) 3×5=5×3
(1.4) 6×0=0
(1.5) 3×1=3
For a more elaborate example, we also see that
(1.6) 2 × (3 + 9) = 2 × 3 + 2 × 9
Is there something more general going on? Of course, you know
the answer is yes, and we could write down many more such examples.
However, this would be rather tiring, and we certainly cannot write
down all such examples since there are infinitely many; so we’d rather
find a better way to state these as general “rules”. In ancient and
medieval times, before modern algebraic notation was invented, such
rules were stated in words; e.g., the general rule for (1.1) would have
been stated as
“The sum of any two numbers is equal
to the sum of the two numbers in the reverse order.”
Obviously, this is a rather clumsy way to state such a simple rule,
and furthermore, when writing down rules this way, it is not always
easy to write out the rule unambiguously. (Pause for a moment and
think how you would write down the general rule for (1.6)!) And again,
you know, of course, how to write down these rules in general: We use
“letters” instead of numbers and think of the letters as “arbitrary”
1
2 1. NUMBERS AND EQUATIONS: REVIEW AND BASICS

numbers. So the above examples turn into the following general rules:
(1.7) a+b=b+a
(1.8) a+0=a
(1.9) a×b=b×a
(1.10) a×0=0
(1.11) a×1=a
(1.12) a × (b + c) = a × b + a × c
Two related issues arise now:
(1) For which numbers a, b, and c do these rules hold?
(2) How do we know these rules are true for all numbers a, b,
and c?
In elementary school, these rules can be “proved” visually for whole
numbers by coming up with models for addition and multiplication. A
common model for the addition of whole numbers a and b is to draw
a many “objects” (e.g., apples) in a row, followed by b many objects in
the same row, and then counting how many objects you have in total.
Similarly, a common model for the multiplication of whole numbers a
and b is to draw a rectangular “array” of a many rows of b many objects
(e.g., apples), and then counting how many objects you have in total.
All the above rules are now easily visualized. (E.g., see Figure 1.1 for
a visual proof of (1.1) and Figure 1.2 for a visual proof of (1.3).)

2 7

7 2
Figure 1.1. A visual proof that 2 + 7 = 7 + 2

An easy extension of these arguments works for fractions (and in-

deed for positive real numbers) if we go from counting objects to work-
ing with points on the positive number line: On the positive number
line (starting at a point identified with the number 0), mark off a “unit
interval” from 0 to a new point called 1. Repeatedly marking off more
unit intervals gives us points we identify with the whole numbers 2, 3,
4, etc. For a whole number n > 0, we can subdivide the unit interval
into n many subintervals of equal length (i.e., of length n1 ). Now again,
 1.1. ARITHMETIC IN THE WHOLE NUMBERS AND BEYOND: A REVIEW 3

3
5

Figure 1.2. A visual proof that 3 × 5 = 5 × 3

starting at 0, we repeatedly mark off intervals of this length and get

points we identify with the fractions n1 , n2 , n3 , n4 , etc. (We also identify
these points with the “directed arrows” from 0 to that point; so, e.g.,
1
2
is represented both by the point labeled in Figure 1.3, and by the
arrow from 0 to the point labeled 21 .)
We can now model addition of fractions (and indeed for positive real
numbers) a and b by marking off an interval of length a from 0, followed
by an interval of length b from the point marked a. The resulting point
can be identified with the sum a + b. Similarly, for multiplication, we
can draw two positive number lines, a horizontal one going to the right
and a vertical one going up, both starting at the same point. Now the
rectangle bounded below by the interval on the horizontal number line
from 0 to a, and bounded on the left by the interval on the vertical
number line from 0 to b, has area a × b and so can be identified with
this product. Once we have these models of addition and multiplication
of fractions (or, more generally, positive real numbers), we can again
easily verify the above rules (1.7)–(1.12). (Draw your own pictures!)
Two much more subtle rules about addition and multiplication (so
subtle that you may not even think about them any more) are the
following:
(1.13) (a + b) + c = a + (b + c)
(1.14) (a × b) × c = a × (b × c)
It is due to these rules that we can write a + b + c and a × b × c
unambiguously without giving it much thought! (Again draw a picture
4 1. NUMBERS AND EQUATIONS: REVIEW AND BASICS

for each to convince yourself that they are true. For (1.14), you will
have to go to a three-dimensional array.)
Exercise 1.1. Give pictures for each of (1.8) and (1.10)–(1.14) in
the whole numbers.
1.1.2. Subtraction and Division. At first, students are given
problems of the type 2 + 5 = or 3 × 4 = , asking them to fill in
the answer. A natural extension of this is to then assign problems of
the form 2 + = 7 or 3 × = 12, asking them to fill in the missing
summand or factor, respectively.
Let’s look at the problem of the “missing summand” first: Solving
2 + = 7 corresponds, e.g., to the following word problem:
“Ann has two apples. Her mother gives her some more
apples, and now she has seven apples. How many
apples did her mother give her?”
Students will quickly learn that this requires the operation of sub-
traction: Ann’s mother gave her 7−2, that is 5, apples. More generally,
the solution to a + = c is c − a. But then a problem arises: What
if a > c?1 Obviously, you now have to go from the “objects” model
to a more “abstract” model, e.g., the money model with debt: If you
have $2 and spend $5, then you are $3 “in debt”, i.e., you have −$3
dollars. (It is a good idea to emphasize the distinction in the use of the
symbol − in these two cases by reading 2 − 5 as “two minus five”, but
−3 as “negative three”.) So allowing subtraction for arbitrary numbers
forces us to allow a “new” kind of numbers, the negative numbers.
So we proceed from the whole numbers 0, 1, 2, . . . to the integers
0, 1, −1, 2, −2, . . . . You will need to allow your students to adjust to
this change, since now your answer to a question like “Can I subtract a
larger number from a smaller number?” will change, possibly confusing
your students. Of course, one should never answer “no” to this question
without some qualification; some students will know about the negative
numbers at an early age (especially in Wisconsin in the winter!). But
now the correct answer should be “it depends”, namely, it depends
on which number system we work in. At this point, you will want to
extend the number line to the left also (see Figure 1.3, it can also be
used to include fractions and negative fractions).
But this transition from the whole numbers to the integers imme-
diately raises three new questions about the arithmetical operations:
1Ican’t resist at this point to tell you my favorite German math teacher joke:
“What does a math teacher think if two students are in the class room and five
students leave? – If three students come back, the room is empty!”
 1.1. ARITHMETIC IN THE WHOLE NUMBERS AND BEYOND: A REVIEW 5

-3 -2 -1 0 ½ 1 2 3

Figure 1.3. The number line

(1) How should we define a + b and a − b when one or both of a

and b are negative? What about multiplication?
(2) How do we know that the rules (1.7)–(1.14) hold for addition
and multiplication of arbitrary (positive or negative) numbers?
(3) Do these rules also hold for subtraction and multiplication of
arbitrary (positive or negative) numbers?
Of course, you already know the answer to item (1). But how
would you justify it to your students? Maybe there is another way to
define addition with negative numbers? The answer is twofold: First
of all, one can justify the common definition of addition of integers
with a “real-life” model (such as the money-and-debt model) or with
the number line (which is really just an abstraction from a real-life
model). The justification for subtraction and multiplication is a bit
more delicate and may not appeal to your students as much. But one
can also justify the common definition of addition and multiplication
of integers by the fact that we want the rules (1.7)–(1.14) to be true for
addition and multiplication of integers also. So, in a sense, the rules in
item (2) above really are “rules” we impose rather than rules we can
“prove”, unless you are satisfied with the justification for item (1) by
the models for addition and multiplication.
So what about subtraction? We first define (−b) as the integer
which when added to b gives 0:
(1.15) b + (−b) = 0
This is now a new “rule” about addition, and (−b) is called the (ad-
ditive) inverse of b. Note that there is lots of room for confusion here:
(−b) (or simply −b for short) is sometimes called the “negative” of b,
a term I would avoid, since −b need not be negative. In fact, −b is
positive if b is negative; −b is 0 if b is 0; and −b is negative if b is
positive! (This will confuse quite a few of your students through high
school and even college, so be sure to emphasize this early on.)
We can then “define” subtraction via addition: a − b is defined as
a + (−b) for any integers a and b. Note that this works well with real-
life models of subtraction: Adding a capital of -$5 (i.e., a debt of $5)
to a capital of $2 is the same as subtracting $5 from $2.
6 1. NUMBERS AND EQUATIONS: REVIEW AND BASICS

This then also gives us an answer for item (3) above: We don’t have
to worry about new rules for subtraction, since we can use the old rules
for addition: It is fairly clear that
a × (b − c) = a × b − a × c
since this is simply
a × (b + (−c)) = a × b + (−(a × c))
(and since −(a × c) = a × (−c)).
On the other hand, in general a − b = b − a is false since it would
mean a + (−b) = (−a) + b. Similarly, a − (b − c) = (a − b) − c is false.
(How would that be rewritten?)
Let’s move on to division: Again, division first comes up when a
student transitions from solving 3 × 4 = to solving 3 × = 12, and
solving for the missing factor corresponds, e.g., to the following word
problem:
“Ann’s mother has twelve apples. If she wants to give
each of her three children the same number of apples,
how many will each child get?”
Again, solving this problem amounts to dividing: The solution to
3× = 12 is 12÷3, or 4. More generally, the solution to a× = c is c÷a.
But a similar problem as for subtraction arises: Here, a may not “evenly
divide” into c, e.g., in the above problem, when Ann’s mother has 13
apples. Here is where it becomes really confusing to your students:
There are two different ways in which school mathematics solves this
problem: division with remainder, and extending to fractions. In the
above example, this would mean that Ann’s mother could either give
four apples each to her three children and keep a “remainder” of one
apple, or that she could cut one apple into thirds and give each child
4 13 apples.2 Since both of these solutions are acceptable, you have
to make it clear to your students which solution you are aiming for.
(Generally, in the higher grades, you will want the second solution,
but division with remainder never disappears completely from your
students’ mathematics: In algebra and calculus, there is division of
polynomials with remainder, as we will see late in this course!)
In any case, we will for now focus on extending our numbers so
that we can always divide c ÷ a “evenly”. (Unless a = 0: Then it
2There is one other model of division: “Ann’s mother has 12 apples and wants
to give three apples to each of her children. How many children does she have?” If
you now increase the number of apples to 13, then only one kind of division makes
sense since cutting up children is generally frowned upon... So be careful before
you start a problem and want to vary it later!
 1.1. ARITHMETIC IN THE WHOLE NUMBERS AND BEYOND: A REVIEW 7

makes no sense since it would amount to solving the original problem

0 × = c; and there will either be no solution if c 6= 0, or any number
is a solution if c = 0; so c ÷ 0 never makes sense.) We proceed from
the whole numbers to fractions (which will always be positive or 0), or
more generally from the integers to the rational numbers (which may
also be negative). The number line from Figure 1.3 still serves as a
good visual model.
But this transition again raises a number of questions:
(1) How should we define the operations of addition, subtraction,
multiplication, and division for fractions, or, more generally,
for rational numbers?
(2) How do we know that the rules (1.7)–(1.15) still hold?
(3) Do these rules also hold for division?
In answering item (1), we need to be able to come up with real-
life examples to justify the “correct” way to perform the arithmetical
operations on fractions and more generally rational numbers. (See
Exercise 1.3 below.) But again one can also justify this by wanting the
rules (1.7)–(1.15) to still hold for these numbers, addressing item (2).
An elegant way to think about division is to first define the (multi-
plicative) inverse of a nonzero number a as that number, denoted by a1 ,
which when multiplied by a gives 1:

1
(1.16) a× =1
a

This is now a new rule for multiplication; and it works not only when a
is a positive whole number but indeed for any nonzero rational number.
And again, this removes the need for new rules for division since we
can reduce division to multiplication by defining a ÷ b as a × 1b . (See
Exercise 1.4.)
We summarize all the rules about the arithmetical operations of
number which we have discovered in the following

Proposition 1.2. The following rules hold for all real numbers a,
b and c:

(commutative law of addition) a+b=b+a

(associative law of addition) (a + b) + c = a + (b + c)
(law of the additive identity) a+0=a
(law of the additive inverse) a + (−a) = 0
8 1. NUMBERS AND EQUATIONS: REVIEW AND BASICS

(commutative law of multiplication) a×b=b×a

(associative law of multiplication) (a × b) × c = a × (b × c)
(law of the multiplicative identity) a×1=a
1
(law of the multiplicative inverse) a× =1 (when a 6= 0)
a
(distributive law) a × (b + c) = a × b + a × c

Exercise 1.3. Give several word problems for the division 12 ÷ 13 ,

both for partitive division (as in the apple example in the main text
above), and for measurement division (as in the footnote above).

Exercise 1.4. Explain which of the rules (1.9)–(1.12) and (1.14)

carry over to division; if they don’t carry over, are there corresponding
rules for division?

1.2. Letters and Algebraic Expressions

1.2.1. Letters in Algebra. Arithmetic only deals with numbers
and numerical expressions, i.e., expressions involving numbers and the
arithmetical operations (addition, subtraction, multiplication, division,
later on also roots, exponentiation, and logarithms). Each such ex-
pression has a definite value (or is undefined for some reason, such as
division by 0 or taking the square root of a negative number). Word
problems often can be reduced to evaluating numerical expressions,
e.g.,
“Ann has four apples. Her friend Mary gives her three
more apples each day for a (five-day) week. How many
apples does she have at the end of the week? ”
Clearly, this gives rise to the numerical expression 4 + (5 × 3), and
evaluating this numerical expression then gives the solution
“Ann has 19 apples at the end of the week.”
One extension of this problems is to leave some numbers, e.g., the
number of apples Ann has at the beginning, and the number of apples
she receives each day, unspecified. In this case, we typically use letters
to denote these unspecified quantities, and it is crucial that these letters
are clearly and unambiguously specified. So let’s set
a = number of apples Ann has at the beginning of the week
b = number of apples Ann receives from Mary each day
Now the word problem becomes
 1.2. LETTERS AND ALGEBRAIC EXPRESSIONS 9

“Ann has a apples. Her friend Mary gives her b more

apples each day for a (five-day) week. How many
apples does she have at the end of the week? ”
This gives rise to the (more general) expression a + (5 × b), and the
solution becomes
“Ann has a + (5 × b) apples at the end of the week.”
Of course, this expression can be rewritten, but we cannot simply
write it as a number any more since we don’t know a and b. We will
cover rewriting such expressions, one of the key features of algebra,
later in the course. For now, let’s concentrate on the use of letters in
algebra.
We can basically distinguish four different uses of letters:
1.2.1.1. Letters as Variables. This is how letters were used in the
example above: Since we didn’t know the number of apples Ann had
at the beginning or that she received each day from Mary, we used the
letters a and b for them, respectively. It doesn’t really matter which
letters we use, as long as we consistently use the same letter for the
same quantity, use different letters for different quantities, and clearly
state how each letter is used. It is often helpful to use mnemonic (i.e.,
easy-to-remember) letters. (E.g., in the above example, we could have
used b for the number of apples Ann has at the beginning of the week,
and d for the number of apples Ann receives d aily.) Later on, we will
distinguish between independent and dependent variables. In the above
example, a and b are independent (since neither depends on any other
quantity in the example), but if we set c = a + (5 × b) for the total
number of apples at the end of the week, then c is a dependent variable,
since it depends on a and b.
In the 5th grade Singapore math book “Primary Mathematics Text-
book 5A”, variables are introduced visually as bars in “bar diagrams”
rather than as letters, one of the best-known features of Singapore
mathematics. It is only in the next grade, in the Singapore math book
“Primary Mathematics Textbook 6A”, that variables are used as letters
instead, and it’s easy to see how one can translate the bar diagram pic-
ture into one or more equations involving variables, where the variables
correspond to bars of unknown “length” in the bar diagram.
For example, in exercise 1 on page 23 of the“Primary Mathematics
Textbook 5A”, Ryan and Juan share $410 between them, and Ryan
receives $100 more than Juan. The problem asks you to find Juan’s
share. (See Figure 1.4.) If we call Juan’s share x, we can easily translate
the bar diagram into an equation (x + 100) + x = 410.
10 1. NUMBERS AND EQUATIONS: REVIEW AND BASICS

Ryan

$410
Juan

? $100

Figure 1.4. What is Juan’s share?

1.2.1.2. Letters as Unknowns. This is probably the first way in

which letters are introduced in school mathematics and is very sim-
ilar to the previous use as variables. Now, however, we don’t think
of the letter as denoting an unspecified quantity, but as a specified
quantity which we don’t know (and usually would like to find). E.g.,
“Ann has four apples. Her friend Mary gives her some
more apples each day for a (five-day) week, the same
number each day. How many apples does Mary give
Ann each day of the week if Ann has 19 apples at the
end of the week?”
In this case, we can express the number of apples Ann has at the
end of the week in two distinct ways, as the numerical expression 19,
and as the expression 4 + (5 × b) where b is the number of apples
she receives each day. So, to solve the problem, we need to solve the
equation 4 + (5 × b) = 19, where the letter b denotes the unknown
quantity. (For a single unknown, the letter x is usually used, but this
is merely a convention.)
1.2.1.3. Letters as Parameters. This is probably the trickiest bit
about the use of letters in algebra since the role of a letter as a pa-
rameter depends on the context. Let’s look again at a variation of the
above word problem
“Ann has a apples. Her friend Mary gives her b more
apples each day for a (five-day) week. At the end of
the week, she has 19 apples.”
This is not really a word problem yet since I haven’t told you what
I am asking for! In fact, I could ask for two different quantities: I could
ask (1) for the number of apples Ann had at the beginning, or (2) for
the number of apples Mary gave her each day. Of course, as stated, you
can’t give me a numerical answer, but you can give me an expression
in each case:
 1.2. LETTERS AND ALGEBRAIC EXPRESSIONS 11

For (1), you could tell me that Ann had 19 − (5 × b) many apples at
the beginning, where b is the number of apples Mary gives her each day.
So in this case, we treat a as an unknown and b as a parameter, and we
express the solution (i.e., in our case, a) in terms of the parameter b.
Why on earth would I want to know this? Well, I can now solve the
problem as soon as you tell me b, and I can solve it very quickly if you
want to know a for different values of b. So, at the beginning, Ann has
four apples if Mary gives her three each day, but Ann has nine apples
if Mary gives her only two each day.
On the other hand, for (2), you could tell me that Mary gives Ann
19−a
5
many apples each day, where a is the number of apples Ann has at
the beginning. So in this case, the roles of a and b are reversed: Now a
is the parameter, and b is the unknown. And the solution b is given in
terms of a.
It is this switch in the roles of letters (as unknowns or parameters)
which is very confusing to students, so when you reach this stage, you
need to make this very clear and explicit.
Note that the distinction between a letter as a variable and a letter
as a parameter is somewhat fluid, but roughly the following: A variable
is a number we don’t know and want to leave unspecified. A parameter
is a specific number which we know but which we have not yet specified.
So, in item (1) above, we can think of b as a specific number of apples
Mary gives Ann every day. (Then b is a parameter.3) Or we can
think of b as an unspecified number for which we want to study how
a = 19 − (5 × b) (as a dependent variable) depends on the independent
variable b. Thus the distinction of the role of a letter as a parameter
or as a variable very much depends on the context.
1.2.1.4. Letters as Constants: This last use of letters is rather ob-
vious in general: There are certain letters which are used for fixed
numbers. The most common one is π, defined as the ratio of the cir-
cumference of a circle divided by its diameter. This is an irrational
number and thus a non-repeating decimal, and there is no way to write
it down “easily” using integers, fractions, or even roots. (The values
3.14, 227
, or 3.14159 are only approximations to its true value.) Another
well-known constant in calculus is the number e, the infinite sum
1 1 1
1+ + + + ...
1 1·2 1·2·3
which is again an irrational number and equals approximately 2.718.
It is very useful in calculus but not really before. A third well-known
3E.g., b could be some really complicated number which, if plugged in now,
would really make our computation very complicated.
12 1. NUMBERS AND EQUATIONS: REVIEW AND BASICS

number of the complex number i, the square root of −1. (This number
is used to allow us to take the square root of negative numbers and
is thus not a “real number”, i.e., a number which we can write as a
decimal or which we can identify with a point on the number line.) In
physics, etc., there are other constants, such as g (for the gravitational
constant). The only thing to remember about these constants is that
one should not use the same letters for variables, unknowns or param-
eters when these constants are present in the context of a problem.
(And the letter π should probably never ever be used in any meaning
other than the usual one.)

1.2.2. Algebraic Expressions. Above, we defined a numerical

expression as an expression involving numbers and the arithmetical op-
erations (addition, subtraction, multiplication, division, later on also
roots, exponentiation, and logarithms). An algebraic expression is an
expression involving both numbers and letters as well the arithmetical
operations. We think of the algebraic expression as a numerical expres-
sion where not all numbers involved are specified and thus expressed
as letters.
First, a short aside: Constants take sort of an intermediate role:
They should really be viewed as actual numbers, but since they are
generally not expressible as rational numbers (or repeating decimals),
we often leave them as letters; so the circumference of a circle of radius
2 units is normally given as 4π (the exact answer) rather than 12.57,
say (which is only an approximate answer).
Algebraic expressions can be evaluated by replacing each letter by
a number (so that the same letter is replaced by the same number
everywhere in the expression, whereas different letters may, but don’t
have to be, replaced by different numbers). E.g., in the example of the
last subsection, we evaluated 19 − (5 × b) when b = 3 to get a value
of 4, and when b = 2 to get a value of 9.
More importantly, algebraic expressions can be rewritten as alge-
braic expressions which take the same value no matter what numbers
replace the letters in them. For this, we usually use the rules given
in Proposition 1.2, which hold for all numbers and thus also for all
algebraic expressions. For example, 19−a 5
can be rewritten as 19 5
− 51 a
using the distributive law and the definition of subtraction and divi-
sion. Learning how to rewrite algebraic expressions is one of the most
important topics of algebra and requires a lot of practice.
 1.4. MANIPULATING ALGEBRAIC EXPRESSIONS 13

1.3. Equations and Identities

Equality is one of the basic notions of algebra, expressing the iden-
tity of two expressions. It is first introduced in the role of “Evaluate
2 + 5 = ”. In this role, it corresponds to the way a pocket calculator
would understand it, namely, “Add 2 and 5 and give the answer” rather
than the role in which it is normally used in algebra, namely, simply
as a statement that two expressions are equal. This can easily lead to
problems: Solving 2 + = 7 does not involve addition but subtraction,
or, at first, possibly “guessing” the right answer and then adding 2 to
check if the sum is 7.
What is often confusing is that there are two types of equations.
Identities are equations which are always true no matter by what num-
bers we replace the letters, as long as the expressions on both sides are
still defined. The rules in Proposition 1.2 are examples of identities,
and many more can be derived from them, e.g., a − (b + c) = a − b − c,
etc.
Any equation which is not an identity is thus an equation which is
false when each letter is replaced by some number and both sides are
still defined(!). So, e.g., xx = 1 is an identity since for all nonzero x,
this equation is true, and for x = 0, the left-hand side is undefined. On
the other hand, x − 1 = 2x − 3 is not an identity since, e.g., when we
replace x by 1, it becomes the equation 1 − 1 = 2 − 3, which is false.
Of course, if we had replaced x by 2, then we would have gotten the
true equation 2 − 1 = 4 − 3, but that is irrelevant here: The equation
x − 1 = 2x − 3 is only an identity if it is true for all “values” of x, i.e.,
for all possible replacements of x making both sides defined.

1.4. Manipulating Algebraic Expressions

In early algebra, students are often asked to “simplify” complicated
algebraic expressions, i.e., to give an algebraic expression identical to
the original one using identities in each step. (Here, we call two al-
gebraic expressions A and B identical if, no matter which numbers
replace the letters in them, A and B evaluate to the same number.)
Usually, such tasks appeal the identities from Proposition 1.2 as well as
the definition of subtraction and division. But the exact way in which
these steps can be justified can be obscure at times, so we will go over
the two most error-prone steps in these simplifications here and justify
them by our identities and definitions. Of course, I’m not trying to
advocate that manipulating algebraic expressions should be taught the
way I present this on the next page or so, but for you as a teacher, it
should be valuable to know why one can perform these simplifications,
14 1. NUMBERS AND EQUATIONS: REVIEW AND BASICS

and that they are merely consequences of our identities in arithmetic

and our definition of subtraction!
But first, let’s prove a little fact about subtraction.

1.4.1. Using the definition of subtraction. In subsection 1.1.2,

we defined subtraction by addition, namely, a − b = a + (−b), where
(−b) was the number such that b + (−b) = 0. (Note that this definition
can only be used after you have introduced negative numbers! Before
then, you have to model subtraction with “real-life” situations, like
taking away a part from a whole: “Mary had five apples and gave two
apples to Ann. How many did she have left?”)
In this subsection, we want to show that for any number a,
(1.17) − a = (−1)a.
To show this, we add both −a and (−1)a to a:
(1.18) a + (−a) = 0 (by definition)
(1.19) a + (−1)a = 1a + (−1)a (by (1.11))
= (1 − 1)a (by (1.12))
= 0a (by arithmetic)
=0 (by (1.10))
Now we have to appeal to intuition (but we could also show formally
using our identities!) that there is exactly one number b which, when
added to a, gives 0, and so that number b equals both −a and (−1)a
by definition and (1.18). This gives the desired identity (1.17).

1.4.2. “Combining Like Terms”. Let’s now start on the first of

the two common steps in simplifying algebraic expressions,“combining
like terms”. Here is an example:
(1.20) 4x + 5y − 3x − 6y = x − y
Sounds pretty straightforward? It is, but when the expressions become
more complicated, it can be confusing. And, most importantly, how
can we justify it?
We first appeal to our identities (1.7) and (1.13):
4x + 5y − 3x − 6y = 4x − 3x + 5y − 6y.
Now we have the “like terms” adjacent and can use the distributive
law:
(1.21) 4x − 3x + 5y − 6y = (4 − 3)x + (5 − 6)y = 1x + (−1)y = x − y
 1.5. CONVENTIONS IN ALGEBRA 15

Of course, the last step uses our identities (1.8) and (1.17). But the
main step in combining like terms consists in the use of the distributive
law (1.12) in the first equality sign of (1.21)!
1.4.3. Removing Parentheses. The other common step in sim-
plifying algebraic expressions is “removing” parentheses when adding
or subtracting several terms. Here are two typical examples:
(1.22) (a − b) + (c − d) = a − b + c − d
(1.23) (a − b) − (c − d) = a − b − c + d
The first identity (1.22) is pretty simple to justify using the defini-
tion of subtraction and our identities (1.7) and (1.13):
(a−b)+(c−d) = (a+(−b))+(c+(−d)) = a+(−b)+c+(−d) = a−b+c−d
The second identity (1.23) is one of the most common sources of errors
in algebra; it requires lots of practice, and a good justification. Here it
is:
(a − b) − (c − d)
= (a − b) + [−(c + (−d))] (by definition of subtraction)
= (a − b) + [(−1)(c + (−1)d)] (by (1.17))
= (a − b) + [(−1)c + (−1)2 d] (by the distributive law)
= (a − b) + ((−c) + 1d) (by (1.17) and arithmetic)
=a−b−c+d (by (1.7), (1.13) and (1.8))
As you can see, this justification, even for a “small” expression like
(a − b) − (c − d) , can become quite tricky, and so it is no wonder that
removing parentheses after a minus sign is so hard for students!

1.5. Conventions in Algebra

There are a number of conventions about sets and operations which
will make algebra simpler once we have mastered them, since they
allow us to write statements more precisely, more succinctly, and with
less clutter. (All notation and conventions introduced in this section
should be introduced in school only gradually and will require repeated
practice.)
1.5.1. Notation for Sets, and Some Special Sets of Num-
bers. One of the fundamental notion of mathematics is that of a set.
Loosely speaking, a set is a “well-defined collection of objects”. (Note
that I am not actually carefully defining what a set is here, and for
good reason: The notion of a set cannot be defined precisely, we can
16 1. NUMBERS AND EQUATIONS: REVIEW AND BASICS

only state “properties” of sets.) In any case, let’s fix some notation
about sets. (We will restrict ourselves to sets of numbers in the follow-
ing. Not all of the notation introduced here will be used right away; so
consider this subsection more of a reference for later on!)
The simplest example of a set is a finite set, and we can simply list
all its elements. So the set containing exactly the numbers 0, 2, and 5
would be written as {0, 2, 5}. A special case of a finite set is the empty
set containing no elements at all, usually written as {} or ∅. We write
n ∈ S to denote that a number n is a member, or an element, of a
set S.
Infinite or very large sets cannot be written this way. (For very large
sets, this is at least not practical.) In that case, we will resort to using
“dots” and write, e.g., {0, 1, . . . , 100} for the set of whole numbers less
than or equal to 100. For the infinite set of all whole numbers, we can
use notation like {0, 1, 2, . . . }, etc.
Some sets of numbers are so common that one fixes standard nota-
tion: We denote
• by N the set of all whole numbers 0, 1, 2, . . . ;
• by Z the set of all integers 0, 1, −1, 2, −2, . . . ;
• by Q the set of all rational numbers (i.e., all numbers of the
form ab where a, b ∈ Z and b 6= 0), this set is also the set of all
real numbers which can be written as repeating decimals;
• by R the set of all real numbers (i.e., all numbers which can
be written as decimals, possibly non-repeating); and
• by C the set of all complex numbers. (We will introduce this
set in detail later on.)
(Here, N stands for “natural” number, which is what mathematicians
usually call a whole number; Z stands for the German word “Zahl”,
meaning “number”; and Q stands for the word “quotient”.)
Now that we have all this, we can use the “set-builder” notation
{. . . | . . .} to define new sets. E.g., the set of all positive integers can
be written as {n ∈ Z | n > 0}; and the set of all real numbers less
than 2 can be written as {r ∈ R | r < 2}.
Furthermore, later on, it will be very convenient to have notation
for intervals of real numbers, i.e., sets of real numbers less than, greater
than, or between fixed real numbers. Given real numbers a and b, we
use
• [a, b] for the set {r ∈ R | a ≤ r ≤ b};
• (a, b) for the set {r ∈ R | a < r < b};
• [a, ∞) for the set {r ∈ R | a ≤ r}; and
• (−∞, b] for the set {r ∈ R | r ≤ b}.
 1.5. CONVENTIONS IN ALGEBRA 17

The sets [a, b), (a, b], (a, ∞), and (−∞, b) are defined similarly. (Note
that the symbols “∞” and “−∞” do not denote real numbers but are
merely used for convenience. Therefore, there can be no sets of the
form (a, ∞], [a, ∞], [−∞, b), or [−∞, b].)
Finally, we will sometimes use convenient notation to combine sets.
Given two sets A and B of numbers, we let
• A ∪ B = {n | n ∈ A or n ∈ B} be the union of the sets A
and B;
• A ∩ B = {n | n ∈ A and n ∈ B} be the intersection of the
sets A and B; and
• A−B = {n | n ∈ A and n ∈ / B} be the difference of the sets A
and B.
For example, if A = {0, 1, 2} and B = {0, 2, 4}, then
A ∪ B = {0, 1, 2, 4}
A ∩ B = {0, 2}
A − B = {1}
1.5.2. Order of Operations and Use of Parentheses. While
the last subsection may have been rough going for you, very quickly
defining a lot of notation which may be unfamiliar to you, this sub-
section will be fairly smooth sailing: We need to define, for numerical
or algebraic expressions, in what order the operations are to be per-
formed; and most of this will already be familiar to you from school,
even though you may never have thought about it this carefully and
explicitly. Note that everything we define in this subsection is only
“convention”, i.e., we arbitrarily define the order of operations a cer-
tain way to make things more convenient; we could have defined it
completely differently, but it’s certainly more convenient to have con-
ventions everyone agrees with!
So suppose you want to write an expression denoting the sum of
the difference of a and the product of 4 and b, and the difference of 2
and the quotient of 5 and c. (Note that it’s hard for anyone to even
comprehend what I just wrote here, so we desperately need a better
way to write expressions!) In any case, we could simply write this as
(1.24) (a − (4 × b)) + (2 − (5 ÷ c))
since the first rule on the order of operations is certainly that anything
inside parentheses must be evaluated first; and if parentheses are nested
one inside the other, then we start from the inside out.
But this expression is ridiculously redundant, of course, as you have
surely noticed; we have literally cluttered it with many unnecessary
18 1. NUMBERS AND EQUATIONS: REVIEW AND BASICS

parentheses. So we use the second rule on the order of operations: All

multiplications and divisions are performed before any additions and
subtractions. This allows us to rewrite the above expression much more
simply as
(1.25) (a − 4 × b) + (2 − 5 ÷ c)
We now come to the third rule on the order of operations: Additions
and subtractions without parentheses are performed from left to right;
and similarly multiplications and divisions are performed from left to
right. This allows us to rewrite our expression even more simply as
(1.26) a−4×b+2−5÷c
Pause for a second here and convince yourself that we have actually
used one of our rules on arithmetic to save parentheses, specifically, we
have used the identity
(1.27) (a − 4 × b) + (2 − 5 ÷ c) = ((a − 4 × b) + 2) − 5 ÷ c
Now we introduce two conventions:
• In order to avoid confusion between the multiplication sym-
bol × and the letter x (especially when hand-written!), we
drop the symbol × unless we multiply two numbers; in that
latter case, we use the symbol · instead. So, e.g, 4 × b is
rewritten as 4b, and 4 × 5 is rewritten as 4 · 5.
• Instead of the division symbol ÷, we use the fraction notation.
So, e.g, 5 ÷ b is rewritten as 5b , and (5 − a) ÷ (2 + b) is rewrit-
ten as 5−a
2+b
. Note here that the fraction notation also saves on
parentheses! If A and B are two complicated algebraic expres-
A
sions, then B is an abbreviated form of (A) ÷ (B), i.e., the
operations inside the expressions A and B are performed first,
and then we divide the value of the former by the value of the
latter.
Using these conventions, we arrive at the final simplified form of the
expression (1.24) as
5
(1.28) a − 4b + 2 −
c
without any parentheses at all!
Later on, when we introduce roots and exponentiation, we also have
to agree on the order of operations there. For roots, we simply indicate
the order
√ of operations by the length of the horizontal
√ bar in the root
1
symbol n and think of it as ( ) n . So, e.g., 2+3√ means that we first
take the square root of 2 and then add 3, whereas 2 + 3 means that
 1.6. ALGEBRA AS “UNDOING EXPRESSIONS” 19

we first add 2 and 3 and then take the square root. For exponentiation
(a fancy word for “taking the power of”), we agree that the power
is taken before any other operation is performed (unless parentheses
are involved). So, e.g., a + b2 means that we first square b and then
add a, whereas (a + b)2 means that we first add a and b and then
square. (A word on how to read these aloud: The first is usually read
as “a plus b squared”, whereas the latter is usually read as “a plus b
quantity squared”.)
We summarize the rules on the order of operations as follows:
(1) Any operation inside parentheses is performed before any op-
eration outside them. If nested parentheses are involved, we
A
perform the operations from the inside out. Any fraction B
(for
√ expressions A and B) is taken to be (A) ÷ (B). Any root
n 1
A (for an expression A) is taken to be (A) n .
(2) If parentheses do not determine the order between operations
then we first take powers, then perform any multiplications
and divisions in order from left to right, and then any additions
and subtractions in order from left to right.
Clearly, these conventions require quite a bit of practice before a
student becomes proficient at them!

1.6. Algebra as “Undoing Expressions”

We have now set up some of the foundations of algebra: We have
discussed some important properties of arithmetical operations which
hold for numbers and thus must carry over to arbitrary algebraic ex-
pressions. We have isolated some of the different ways in which letters
are used in algebra. We have talked about algebraic expressions and
equations. We have gone over the rules which govern the evaluation of
numerical expressions, and therefore also algebraic expressions. And
we have set up some notation for sets of numbers.
Earlier, we talked about algebra as “arithmetic with letters”. There
is one more way in which one can look at algebra, namely, as the “art
of undoing expressions”. Here is an example:
“Ann has four apples. Her friend Mary gives her three
more apples each day for a (five-day) week. How many
apples does she have at the end of the week?”
This is not an algebra but an arithmetic problem: The answer is
simply given by the numerical expression 4 + (5 · 3) apples, i.e., 19
apples. The corresponding pure arithmetic problem is 4 + (5 · 3) = .
Now let’s turn things around. Instead of asking for the “answer” 19,
20 1. NUMBERS AND EQUATIONS: REVIEW AND BASICS

let’s give the answer and ask for one of the things in the question; let’s
solve 4 + (5 · ) = 19. The corresponding word problem now becomes
“Ann has four apples. Her friend Mary gives her the
same number of apples each day for a (five-day) week.
At the end of the week, she has 19 apples. How many
apples did Mary give Ann each day?”
Why do I call this “undoing an expression”? Let b be the number
of apples Mary gives Ann each day, and c the number of apples Ann
has at the end. Then we have the equation 4 + (5 · b) = c. The first
problem gives b and asks for c, which amounts to simply evaluating
the expression 4 + (5 · b) when b = 3. The second problem gives c and
asks for b. This is very different: It amounts to “solving” the equation
4 + (5 · b) = 19, i.e., “undoing” (rather than evaluating) the expression
4 + (5 · b) to find out for which b it evaluates to 19. In this case, the
“undoing” goes in two steps:
4 + (5 · b) = 19 turns into 5 · b = 19 − 4
5 · b = 19 − 4 turns into b = (19 − 4) ÷ 5
Now we have “undone” the expression 4+(5·b) and reduced the solution
to an arithmetic problem, namely, evaluating the numerical expression
(19 − 4) ÷ 5.
Many algebra problems we will encounter later on will be of this
nature. And, of course, we will encounter situations later on where it is
much harder to “undo” an expression, where there may be no solution
at all, or where there are several or possibly infinitely many solutions.
So let’s get started.
 CHAPTER 2

Linear Equations

2.1. Ratio and Proportional Reasoning

Probably the first time a student encounters linearity is via the
concept of ratio, in a problem such as this one:
“John has twelve apples, which are either green or
red. For every green apple, he has three red apples.
How many red apples does he have?”
So the problem states explicitly that the ratio of the number of
green apples to the number of red apples is 1 : 3. And one can then
solve the problem by finding out that John has nine red apples, so
in unsimplified form, the ratio of the number of green apples to the
number of red apples is 3 : 9.
Ratio can be a confusing concept, since it can be expressed in a
number of different (and “different-looking”) ways:
(1) For every green apple, John has three red apples.
(2) The ratio of the number of green apples to the number of red
apples is 1 : 3.
(3) Out of every four apples, three apples are red.
(4) The ratio of the number of red apples to the number of all
apples is 3 : 4.
(5) Three quarters of John’s apples are red.
(6) 75% of John’s apples are red.
Here, statements (1) and (2) are basically the same, just written in
different notation. However, statements (1) and (3) “look” quite differ-
ent, since the former expresses a “ratio” between parts (“green apples”
and “red apples”), whereas the latter expresses a ratio between a part
and the whole (“red apples” and “all apples”). Statements (4), (5),
and (6) formalize the ratio in statement (3) more and more abstractly,
statement (4) in terms of ratio similar to statement (2); statement (5)
in terms of fractions (still basically the same as (3)); and statement (6)
in terms of percentage, which now takes the whole (the “unit”) not as
four as in all previous statements, but as 100. Figure 2.1 illustrates
this ratio graphically.
21
22 2. LINEAR EQUATIONS

green apples red apples

Figure 2.1. Ratio of the number of green apples to the

number of red apples

Note that the ratio notation can be taken further by relating not
just two quantities but several. E.g., let’s modify the ratio problem to
read:
“John has 18 apples, which are either green, red, or
yellow. For every green apple, he has three red apples
and two yellow apples. How many red apples does he
have?”
Now the ratio of the number of green apples to the number of red
apples to the number of yellow apples is given as 1 : 3 : 2 (and we can
still solve for the number of red apples as being 9).
The concept of ratio directly leads to the concept of proportional
reasoning. Let’s look at the following table for the original version of
our problem:

number of green apples 1 2 3 4 5 n

number of red apples 3 6 9 12 15 3n
number of all apples 4 8 12 16 20 4n

Table 2.1. The number of green and red apples vs. all apples

Each column contains numbers which are in ratio 1 : 3 : 4. (The last

column states this fact for a general number n of green apples.) Given
any entry of the table, we can easily compute the other two entries of
that column. Also note that, going from one column to the next, the
numbers in each row increase by a fixed amount for each row, namely,
by 1, 3, and 4, respectively. On the other hand, the ratio is usually
changed when we add the same number to the entries of a column:
Suppose, in our original word problem, John receives two more green
apples and two more red apples, so that he now has five green apples
 2.2. VELOCITY AND RATE OF CHANGE 23

and eleven red apples. Then the new ratio of the number of green
apples to the number of red apples is 5 : 11, which is different from the
old ratio 1 : 3.
Similar examples can be found in many other contexts, e.g., in
recipes, which call for the ingredients to be in a certain ratio (such as
two cups of flour to one cup of sugar, etc.).
Percentage (from the Latin “per centum”, i.e., per one hundred) is
a special way to express a ratio, where one thinks of one quantity in
a ratio as “the whole” and sets that number equal to 100; the other
quantities are then expressed in proportion to that number. E.g., in
our first word problem, the ratio of the number of green apples to the
number of red apples to the number of all apples is 25 : 75 : 100, and
so we can say that there are 25% green apples and 75% red apples.
Usually, the “whole” is the largest of the numbers (and the others are
considered “parts” of it), but this need not always be the case.
Exercise 2.1. Think of several everyday examples of ratio and
percentage where the largest number is not considered as 100%.
2.2. Velocity and Rate of Change
Another common “linear” phenomenon is velocity:
“A train moves at a constant velocity of 120 km/h.
How far will the train travel in half an hour? In two
hours? In 3 12 hours?”
Again we can make up a table:
1
time traveled (in hours) 0 2
1 2 3 3 12
distance traveled (in km) 0 60 120 240 360 420

Table 2.2. Time and distance traveled by a train of

constant velocity 120 km/h

First a brief comment: Why didn’t I just call 120 km/h the speed
of the train rather than its velocity? There is a small but important
difference between these, and we’ll get to it in more detail later on. For
now, let’s just remember that velocity can be negative (when the train
moves backward), whereas speed is always positive (or zero, when the
train is not moving). So velocity takes into account the direction (in
one dimension, say, along a rail line), whereas speed does not.
In some sense, the above velocity problem is again a ratio problem,
since time and distance traveled are in ratio 1 : 120. But that is not how
24 2. LINEAR EQUATIONS

we normally think of this (even though, mathematically speaking, it is

exactly the same!). The reason for this is twofold: For one, time and
distance are measured in different units. But that alone doesn’t really
make it different, since, e.g., in a recipe, we might use the ratio of one
cup of sugar to one teaspoon of salt. What makes time and distance
“really” different is that they measure different physical quantities. So
instead of considering this a ratio, we normally think of it as a rate, or
more precisely, a rate of change, and express this rate of change as a
quotient:

distance
(2.1) velocity =
time

Now Table 2.2 tells us that this quotient is fixed for any time.
(However, since time and distance are measured not only in whole
numbers, we don’t have the property of Table 2.1 any more that as we
move along a row, the numbers change by a fixed amount, since we
have deliberately added some half-hour intervals.)
A quotient as in (2.1) is just one example of a rate of change. And,
of course, a rate of change need not be constant at all: The train might
speed up or slow down. For now, however, we will concentrate on the
case of a constant rate of change. Examples of a constant rate of change
abound in everyday situations:
(1) the cost of (loose) fruit and vegetables at the grocery store is
computed by the weight (say, by the kilogram), so there is a
fixed cost per kilogram;
(2) the cost of electricity you use is computed by the kWh (kilo-
watt hour), so there is a fixed cost per hour;
(3) the water volume of a swimming pool usually increases at a
constant rate when it is filled (letting the time start when the
pool starts filling up); etc.
In each case, we can identify a constant quotient. Let’s make this
relationship more formal: Usually, there is one independent quantity
(often called x) and one dependent quantity (often called y) depending
on the independent quantity. In the train example, x is the time trav-
eled and y is the distance traveled. In the above examples (1)–(3), x is
the weight, the electricity used (more precisely, the electrical energy in
kWh), and the time since the pool starts filling up, respectively. And y
is the cost (in both (1) and (2)) and the water volume, respectively.
Since we are assuming that the ratio (or quotient) between these
two quantities x and y is constant, we’ll denote this ratio by k for now;
 2.2. VELOCITY AND RATE OF CHANGE 25

we thus have
y
(2.2) k=
x
for any value of x and any “corresponding” y. In the train example, k is
the velocity; in the other three examples, it is the cost per kilogram,
the cost per kWh, and the increase in water volume (per time unit,
say, per minute), respectively.
We can now rewrite (2.2) as
(2.3) y =k·x
to make the dependence of y on x more explicit. For our train example,
switching to the more common notation of t for time, v for velocity,
and s for distance, we arrive at
(2.4) s=v·t
The equation (2.4) is just one way to represent this situation. An-
other representation is via a table as in Table 2.2. A third representa-
tion is visual via a graph as in Figure 2.2.

Figure 2.2. Time t and distance s = v · t traveled by a

train of constant velocity v = 120 km/h

What are the advantages and disadvantages of these three repre-

sentations?
• An equation lets us quickly compute the exact value of the dis-
tance s for any time t. As we will see later, an equation is also
26 2. LINEAR EQUATIONS

handy when you need a description of how the distance varies

with time inside another, more complicated computation. On
the other hand, an equation is not so easy to “visualize” with-
out some practice.
• A table lets us quickly look up the distance s for the particular
values of t listed in the table. But it has us “guessing” the
distance for other values of t. In fact, strictly speaking, we
can’t really tell the distance for other values of t at all; we
are really just “guessing” them since we can’t tell from a table
how the distance “really” varies as time varies.
• A graph is a great visual aid and lets us see “at a glance” how
the distance varies with time. The drawbacks are twofold: It’s
not really possible to read from a graph the precise distance at
a given time t (and this is impossible for values of t “off” the
graph). And it’s hard to work with the graph when you have
to use these values inside another, more complicated computa-
tion. (Graphs become a bit more complicated when they can
no longer be drawn “without lifting the pencil” off the page,
i.e., when the value of s suddenly “jumps” at some value of t.
We’ll talk about such graphs in more detail later.)
Since tables can only tell us the distance for finitely many values
of t, they are generally not used as much in algebra. There are excep-
tions, however; e.g., you may know that the values of the trigonometric
functions at specific multiples of π are handy to have around, and even
to have memorized. (On the other hand, it’s generally not hard but
only tedious at worst to generate a table from an equation, say.)
From now on, we will mostly concentrate on equations and graphs.
One of the main tasks in algebra will be to transfer back and forth
between these two representations, a task which requires lots and lots
of practice. A graphing calculator can be of great help in visualizing
the graphs of more complicated equations; but for simpler equations,
you, and hopefully later on your students, should be able to visualize
their graphs “mentally” and sketch them quickly.

2.3. Linear Expressions

We now slightly generalize from the above examples of ratio and
proportion. Here is an example:
“A pay phone charges 50¢ for the first minute and 10¢
for each additional minute of a long-distance call.”
Let’s look at a table of how much a phone call costs. There are four
things to note right away in Table 2.3:
 2.3. LINEAR EXPRESSIONS 27

number of minutes 1 2 3 4 5 6 7 8 9 10
cost in cents 50 60 70 80 90 100 110 120 130 140

Table 2.3. Cost of a pay phone call

(1) We still have the property as in Table 2.1 that each row in-
creases by a fixed amount (by 1 and 10, respectively).
(2) We no longer have the properties from Tables 2.1 and 2.2 that
each column has numbers in the same ratio: E.g., the ratios 1 :
50 and 2 : 60 from the second and third column are different.
(3) As in Table 2.1, but unlike in Table 2.2, it doesn’t make “real-
world” sense to look at the cost of a phone call lasting 0 min-
utes.
(4) As in Table 2.1, but unlike in Table 2.2, it doesn’t make sense
to look at the cost of a phone call lasting a number of minutes
which is not a whole number (such as half a minute). We’re
assuming here that the phone company charges the same for
a “fractional” minute (i.e., a time interval lasting less than a
minute) as for a whole minute. So, e.g., a phone call lasting 2
minutes and 1 second, and another call lasting 2 minutes and
59 seconds, will both be billed as a 3-minute call by the phone
company.
As you can see, there are a whole lot of new issues coming up, and
we need to separate them out one by one. We’ll defer the issue of
whole-number minutes in item (4) until Section 4.3.1; we’ll assume for
now that we just “round up” and so consider only whole numbers.
The issue in item (3) is fairly easy to resolve if we just look for
the “pattern” in Table 2.3: We could simply assume that the phone
company charges 40¢ for a 0-minute phone call. This is, of course, not
how this works out in the real world, but it makes the table follow the
pattern noted in item (1).
So we are down to considering the issues from items (1) and (2).
Let’s look at the graph for the cost of the phone call (see Figure 2.3).
Note that in light of item (3) above, I didn’t draw this graph as a line
but only used dots for positive whole-number values of the number of
minutes.
What we would like, of course, is an equation describing this graph.
Let’s denote by t the number of minutes of the phone call, and by c the
cost of the phone call in cents. Using the information from the original
problem, we notice that t − 1 is the number of additional minutes, and
28 2. LINEAR EQUATIONS

Figure 2.3. Cost of a pay phone call

so the cost of a phone call can be calculated by the equation

(2.5) c = 50 + 10(t − 1)
where 50¢ is the cost of the first minute and 10¢ is the cost for each
additional minute.
Alternatively, using the “trick” from above and letting 40¢ be the
“cost” of a “0-minute” phone call, we notice that if we subtract this
amount from all the other possible costs, then we do indeed get a fixed
ratio, namely, 1 : 10, for all the columns in Table 2.3. This reasoning
quickly leads to the equation
(2.6) c = 10t + 40
where 40¢ is the “cost” of a 0-minute call and 10¢ is the cost per
“additional” minute. (It is also not hard to see that equations (2.5)
and (2.6) give the same cost using our rules of arithmetic.)
It should be clear that manipulating expressions such as 10t + 40
or 50 + 10(t − 1) is very important to algebra. So let’s look at letters
and algebraic expressions in more detail.
Here is the way Singapore math books (in Primary Mathematics
6A) first introduce letters (as variables): Think of the letter m, say, as
the number of marbles in a bag (where we assume that each bag con-
tains the same, but unknown, number of marbles). Now it makes sense
 2.3. LINEAR EXPRESSIONS 29

to write five bags of marbles together with three additional individual

marbles as 5m + 3 (see Figure 2.4).

5m + 3

Figure 2.4. Representing five bags of marbles and

three individual marbles

This way allows students to explore the concept of an “unknown”

variable m (which may stand for any fixed number, namely, in this
case, the number of marbles in each bag) while still treating m as just
a “concrete” number which they can add and subtract. (Multiplication
and division of variables is a bit harder to model this way and probably
would require a different model.)
An expression of the form kx + b (where k and b are numbers and x
is a variable) is called a linear expression. (The name obviously derives
from the fact, as we will study later and as you probably already know,
that the graph of y = kx + b is a line.) We have already seen some
examples of linear expressions above. Many word problems lead to
linear expressions. Here is a so-called “mixture problem”:
“A ball consisting of an iron-nickel alloy has a volume
of 100cm3 . Determine its mass, given that iron has
a specific density of 7.9g/cm3 and nickel a specific
density of 8.9g/cm3 .”
Obviously, we can’t solve this problem numerically since we are not
given how much iron the alloy contains. So let’s denote the volume of
iron (in cm3 ) in the ball by V . Then

(2.7) 7.9V + 8.9(100 − V )

gives the mass of the ball in grams, since the volume of the nickel must
be 100 − V in cm3 , and so, by proportional reasoning, the mass of the
iron and the nickel in the alloy is 7.9V and 8.9(100 − V ) in grams,
respectively.1

1Actually, the fact that iron and nickel form an alloy here changes the specific
density slightly, but not enough to affect the computation enough to matter.
30 2. LINEAR EQUATIONS

2.4. Line Equations

The last section should have convinced you that equations of the
form
(2.8) y = mx + b
and their graphs are very important in algebra; so we will study them
mostly in the abstract in this section, with some asides about appli-
cations. (We have changed notation slightly here to conform with the
usual conventions on these equations: We have called the independent
variable x, the dependent variable y, and we have denoted the two pa-
rameters by m and b. As usual for parameters, we think of them as
fixed numbers.) An equation as in (2.8) is often called a linear equa-
tion, but it is not the most general form of a linear equation as we will
see later on.
2.4.1. Graphing a Linear Equation. Before we can formally
define the graph of a linear equation, we need to formally define the
(two-dimensional) coordinate system, which connects algebra with ge-
ometry (or, more precisely, coordinate geometry). A coordinate system
consists of a horizontal number line (usually called the x-axis if x is
the independent variable) and a vertical number line (usually called
the y-axis if y is the dependent variable). (So both axes are two-way
infinite lines, called the coordinate axes. Of course, if the variables are
denoted by different letters, we also use these letters to denote the cor-
responding axes.) The point where the two axes intersect is called the
origin and usually denoted by the letter O. (See Figure 2.5.)
We also introduce the set of (ordered) pairs of reals (x, y) by let-
ting x and y vary over the real numbers, and by defining two pairs of
reals (x1 , y1 ) and (x2 , y2 ) to be equal if both x1 = x2 and y1 = y2 . We
now identify each pair of reals (x, y) with the point P in the coordinate
plane obtained by moving from the origin x units to the right (or, if
x < 0, −x units to the left) and y units up (or, if y < 0, −y units
down). We observe that for each pair of reals, there is exactly one
such point in the coordinate plane. Similarly, for each point P in the
coordinate plane, there is exactly one pair of reals found as follows:
Draw a horizontal and a vertical line through the point P ; let x be
the number on the x-axis (viewed as a number line) at which the ver-
tical line intersects the x-axis, and let y be the number on the y-axis
(viewed as a number line) at which the horizontal line intersects the
y-axis. Then the point P corresponds to the pair of reals (x, y). We
call x the x-coordinate of P , and y the y-coordinate of P . Figure 2.5
illustrates this for the point P = (2, 3): The vertical line through P
 2.4. LINE EQUATIONS 31

Figure 2.5. The coordinate system with two points labeled

intersects the x-axis at 2, and the horizontal line through P intersects

the y-axis at 3.
Now the graph of the equation y = mx + b from (2.8) consists of the
set of all points (x, y) in the real plane satisfying this equation. (For
an example, see Figure 2.6 for the graph of the equation y = 12 x + 21 .)
We want to argue that the equation (2.8) always describes a line in
the plane. For this, we first compute the ratio of the change of y to the
change of x: Fix any two distinct points A = (x1 , y1 ) and B = (x2 , y2 )
on the graph. Since A and B satisfy the equation (2.8), we have

(2.9) y1 = mx1 + b and y2 = mx2 + b.

If x1 = x2 then, by (2.9), y1 = y2 and so A = B, contrary to our

assumption. Thus x1 6= x2 (which will allow us below to divide by
x2 − x1 ).
32 2. LINEAR EQUATIONS

Figure 2.6. The line y = 12 x + 1

2

We now compute the ratio (i.e., the quotient) of the y-change over
the x-change as
y2 − y1 (mx2 + b) − (mx1 + b)
=
x2 − x1 x2 − x1
mx2 − mx1
=
x2 − x1
m(x2 − x1 )
=
x2 − x1
=m
Therefore, this ratio does not depend on the choice of the points A
and B on the graph of the equation (2.8) and in fact always equals the
parameter m.
Now let’s pick three distinct points A, B, and B 0 on the graph
(and assume they are arranged in this order from left to right, and
that m > 0 so that B 0 and B both lie “above” A in the coordinate
plane). In order to show that the graph of our equation really does
form a line, we need to show that these three points are collinear. (For
an example, see Figure 2.7 for the graph of the equation y = 12 x + 12
with the points A, B, and B 0 , on the graph. Don’t be fooled by the
picture, however! We haven’t shown yet that the three points A, B
and B 0 lie on a line! So far, we don’t know that yet!) If we let the
point C (C 0 , respectively) be the point where the vertical line through B
 2.4. LINE EQUATIONS 33

Figure 2.7. The slope of the line y = 21 x + 1

2
as mea-
sured with two different triangles

(B 0 , respectively) and the horizontal line through A intersect, then we

can form two triangles 4ABC and 4AB 0 C 0 . (Here, the picture in
Figure 2.7 is correct: The points A, C, and C 0 really do lie on one line
since the sides AC and AC 0 are both horizontal (i.e., parallel to the
x-axis).) Now the previous calculation yields (for the lengths of the
sides of the triangles) that
BC B0C 0
m= =
CA C 0A
0
Furthermore, the angles ∠C and ∠C are both right angles, and thus
the triangles 4ABC and 4AB 0 C 0 are similar. Since the sides AC
and AC 0 are parallel (and, by similarity, the angles ∠BAC and ∠B 0 AC 0
coincide), the sides AB and AB 0 are parallel as well. Thus the points A,
B, and B 0 on the graph of the equation (2.8) do indeed lie on a line,
justifying our picture in Figure 2.7. We call the quotient m the slope
of the line given by the equation y = mx + b.
Exercise 2.2. How does the picture in Figures 2.6 and 2.7 change
when m is negative? What happens when m = 0? Give similar proofs
that the graphs form a line in each case.
Now that we have found a visual interpretation of the parameter m,
what about the parameter b? Setting x = 0, we obtain from the equa-
tion (2.8) that the point (0, b) is on our line. This is where the line
34 2. LINEAR EQUATIONS

crosses the y-axis. Therefore, b is called the y-intercept of the line, and
the equation
y = mx + b
is called the slope-intercept form of the line.

Figure 2.8. The slope of the line y = 21 x + 1

2
as mea-
sured with two different triangles

We also note that the converse of the statement we just proved

above is true: A (non-vertical) line in the coordinate plane is given by
an equation of the form y = mx + b. (We’ll talk about the special case
of vertical lines a bit later.) This is because we can run the above proof
“backwards”: Fix a non-vertical line. It must intersect the y-axis at
exactly one point A = (0, b) for some real number b. Fix any other
two points B = (x1 , y1 ) and B 0 = (x2 , y2 ) on the line. Then the line
segments AB and AB 0 are parallel. For now, let’s assume that both B
and B 0 lie to the right of A on the line, and that the y-coordinates of
both B and B 0 are > b so that the points B and B 0 lie “above” A in the
coordinate plane. So we can draw points C and C 0 (see Figure 2.8) on
the intersection of the horizontal line through A, and the vertical lines
through B and B 0 , respectively. Since all three sides of the triangles
4ABC and 4AB 0 C 0 are pairwise parallel, these triangles are similar,
and so
BC B0C 0
=
CA C 0A
for the lengths of their sides. This implies that the quotient BC
CA
does
not depend on the choice of the point B, and so we denote this quotient
 2.4. LINE EQUATIONS 35

by m. If we now let B = (x, y) be any point on the line to the right

of A, then
y−b
m=
x−0
which can be rewritten as
y − b = mx
and so
y = mx + b
We have thus shown the following
Theorem 2.3. (1) The graph of any equation y = mx + b is
a non-vertical line with slope m intersecting the y-axis at the
point (0, b).
(2) Any non-vertical line ` is the graph of an equation of the form
y = mx + b where b is determined by the y-coordinate of the
point where the line ` intersects the y-axis, and m is the quo-
tient
y2 − y1
x2 − x1
for any two distinct points B = (x1 , y1 ) and B 0 = (x2 , y2 ) on
the line `.
Exercise 2.4. Strictly speaking, we have shown part (2) of the
above theorem only for the case when the line is “increasing”, and
then only in the case where the points B and B 0 lie to the right of the
point A = (0, b). Finish the proof by analyzing the cases where the line
is “decreasing” or horizontal, and where one or both of the points B
and B 0 lie to the left of the point A = (0, b).
2.4.2. Different Forms of Equations for Lines. We have al-
ready seen one form of an equation for a (non-vertical) line, namely,
the slope-intercept form
y = mx + b
2
where m is called the slope and b the y-intercept, i.e., the y-coordinate
of the point at which the line intersects the y-axis. So we can easily
write down the equation of a (non-vertical) line if we are given its slope
and y-intercept.
In reality, however, we are much more likely to be given two points
on the line, both in geometry or in real-life applications such as the
following:
2The Singapore math books call the slope of a line its gradient. In U.S. text-
books, this term is normally used in a different (but similar) meaning in calculus.
36 2. LINEAR EQUATIONS

“A car travels at constant speed. At 8:30 a.m., it

crosses a state line (where the mile markers start at 0).
At 10:10 a.m., it passes mile marker 100.”
So letting t be the number of hours since midnight and s the dis-
tance from the state line (to be taken as negative before the car crosses
the state line), and converting times to fractions, the problem tells us
that the graph of the car’s motion (which is a line since the car travels
at constant speed) contains the two points (8 12 , 0) and (10 16 , 100).
So how do we obtain an equation for a (non-vertical) line if we only
know two (distinct) points on it? The answer is, of course, that the
two points tell us the slope. First remember the “mantra”
rise
slope =
run
which really just says that the “y-change” (“rise”) is proportional, i.e.,
in a fixed ratio, to the “x-change” (“run”). The word “rise” may be
somewhat confusing since it may be negative, i.e., y may decrease as x
increases; but then the slope is simply negative. (As a third case, y
may just be constant as x changes, in which case the slope is 0, and
the graph of the equation is a horizontal line.)
Let’s call the two given points on the line P0 and P1 , and assume
that their coordinates are (x0 , y0 ) and (x1 , y1 ). Then the slope can be
computed as
y1 − y0
(2.10) m=
x 1 − x0
Similarly, given any other point P = (x, y) on this line, we have
y − y0
(2.11) m=
x − x0
The equation (2.11) is called the point-slope form of a line equation
and can, with after little bit of algebra, also be written as
y − y0 = m(x − x0 )
or
y = mx + (−mx0 + y0 )
Of course, in this last form, it is basically the same as the slope-
intercept form, except that the parameter b is now more complicated.
But the point-slope form (which is useful if we are given the slope
and just some point P0 = (x0 , y0 ) on a line) isn’t really what we were
after at first, when we asked for a line equation given two (distinct)
points on a (non-vertical) line. However, it is now easy to combine
 2.4. LINE EQUATIONS 37

equations (2.10) and (2.11) to obtain the two-point form of a line equa-
tion:
y − y0 y1 − y0
(2.12) =
x − x0 x1 − x0
Again, with a little bit of algebra, this can be rewritten as
y1 − y0
y − y0 = (x − x0 )
x1 − x0
or, in slope-intercept form, as
y1 − y0 y1 − y0
y= x + (− x 0 + y0 )
x1 − x 0 x1 − x0
And again, of course, in this last form, this is basically the same as
the slope-intercept form, except that now both parameters m and b
are very complicated. (Note that in equations (2.10)–(2.12), not only
the letter m, but also the letters x0 , x1 , y0 , and y1 are parameters, and
only the letters x and y are variables, something which many students
will find highly mystifying!)
The above three forms of the line equation all apply exactly to all
non-vertical lines. There is a fourth and more general form of the line
equation, which also applies to vertical lines and, not surprisingly, is
called the general form of the line equation:

(2.13) Ax + By + C = 0

Here A, B, and C are parameters, i.e., fixed numbers, and x and y are
variables, namely, the x- and y-coordinates of points in the coordinate
plane.
We will tacitly assume that at least one of A and B is nonzero,
since otherwise the equation (2.13) degenerates to the trivial equation
C = 0, which is either true or false, independent of the values of the
variables x and y, and thus does not describe a line. (The equation
C = 0 either describes the whole plane, if the parameter C is zero,
or the empty set otherwise.3) On the other hand, if at least one of A
and B is nonzero, then the equation (2.13) really does describe a line,
as we will now see.
We consider four cases for the general line equation (2.13) sepa-
rately. (Here, the third case will not exclude the first two cases!)

3Note that in this equation C = 0, the letter C is the parameter and not the
unknown! The unknowns, or variables, in C = 0 are x and y!
38 2. LINEAR EQUATIONS

Case A = 0: By our tacit assumption, we must then have B 6= 0,

and so the equation reads By + C = 0 and can be rewritten as
C
(2.14) y=−
B
This is just a special case of the slope-intercept form of a line, with
slope 0, and so describes a horizontal line, which crosses the y-axis at
C
the point (0, − B ). Note that equation (2.14) does not depend on the
value of the variable x, as is to be expected of a horizontal line; the
C
equation simply fixes a particular value of the variable y, namely, − B .

Case B = 0: By our tacit assumption, we must then have A 6= 0,

and so the equation reads Ax + C = 0 and can be rewritten as
C
(2.15) x=−
A
This equation does not depend on the value of the variable y and
thus describes a vertical line, which crosses the x-axis at the point
(− CA , 0); the equation simply fixes a particular value of the variable x,
namely, − CA . Note that equation (2.15) cannot be rewritten in slope-
intercept form or either of the other two forms introduced earlier in
this section.
Case C = 0: First, note that in this case, one of A or B may also
equal 0, so this case does not exclude either of the first two special cases.
However, since the equation now reads Ax + By = 0, we immediately
see that the origin (0, 0) is a point on the line. If A or B also equals 0,
then, by the previous cases, the equation describes the x-axis or the
y-axis as a line, respectively. If neither A nor B equals 0, then the
equation can be rewritten as
A
(2.16) y=− x
B
Since this line crosses both the x-axis and the y-axis at the origin, both
the y-intercept (defined before as the y-coordinate of the point where
a line crosses the y-axis) and the x-intercept (defined now as the x-
coordinate of the point where a line crosses the x-axis) equals 0. And
A
the slope of the line clearly equals m = − B .
Case A, B, C 6= 0: In that case, the equation (2.13) can be rewritten
as
A C
y =− x−
B B
A
and so has nonzero slope m = − B . Also, since the parameter C does
not equal 0, the line does not cross the axes at the origin. Rather, by
 2.4. LINE EQUATIONS 39

setting x = 0 and and solving for y, and then setting y = 0 and solving
for x, respectively, it is not hard to see that the line contains the two
(distinct) points (− CA , 0) and (0, − B
C
) on the x- and y-axis, respectively.
So the x-intercept of the line is − B , and its y-intercept is − CA . Since
C

we can multiply the equation (2.13) by any nonzero number without

affecting its truth, we may assume (having multiplied equation (2.13)
by the appropriate number) that the parameter C (after “renaming”
of parameters) equals −1. Then equation (2.13) reads (with the new
values for our parameters A, B, and C)
Ax + By − 1 = 0
or
Ax + By = 1
Now the x- and y-intercepts are A1 and B1 , respectively. Defining new
parameters a and b to equal A1 and B1 , respectively, we can thus rewrite
equation (2.13) (still assuming that the parameter C equals −1) as
x y
(2.17) + =1
a b
This is another “normal form” of an equation of a line (not containing
the origin). It is sometimes called the intercept form of such a line.
(See Figure 2.9 for the example −2x
+ y1 = 1.)

x y
Figure 2.9. The line −2 + 1
= 1 with x-intercept a =
−2 and y-intercept b = 1
Finally, let’s look at the five different forms of a line equation in
terms of a “real-life” example. In each case, a car will cross a state
line (where the mile markers start at 0) and travel at a constant speed.
We use the independent variable t for the time elapsed in hours since
40 2. LINEAR EQUATIONS

noon, and the dependent variable s for the distance traveled in miles
starting at the state line (taking s to be negative before the car crosses
the state line). In all versions of this example, we will get an equation
for the same line, see Figure 2.10.

Figure 2.10. The line s = 60t+90 describing the travel

of a car

Slope-intercept form:
“A car travels at a constant speed of 60mph. At noon,
it has traveled 90 miles past the state line.”
Slope m = 60, s-intercept b = 90, equation
s = 60t + 90
Point-slope form:
“A car travels at a constant speed of 60mph. At
1 p.m., it has traveled 150 miles past the state line.”
Slope m = 60, point P0 = (1, 150), equation
s − 150
60 = or s − 150 = 60(t − 1)
t−1
Two-point form:
“A car travels at a constant speed. At 1 p.m., it has
traveled 150 miles past the state line; at 2:30 p.m., it
 2.4. LINE EQUATIONS 41

has traveled 240 miles past the state line.”

Points P0 = (1, 150) and P1 = (2.5, 240), equation
s − 150 240 − 150 240 − 150
= or s − 150 = (t − 1)
t−1 2.5 − 1 2.5 − 1
Intercept form and general form:
“A car travels at a constant speed. At 10:30 a.m., it
crosses the state line; at noon, it has traveled 90 miles
past the state line.”
t-intercept a = −1.5, s-intercept b = 90, equation in
intercept form
t s
+ =1
−1.5 90
and (after clearing denominators by multiplying by
−90) in general form
60t − s + 90 = 0
Although the different phrasings of this problem may seem some-
what contrived, they show that each form of a line equation can be
modeled by a “real-life” example, which is what I was trying to con-
vince you of.
Exercise 2.5. Find a “very different” real-life example for the
equation y = −2x − 2 and explain how this problem would be phrased
in terms of the four different forms of the line equation (slope-intercept,
point-slope, two-point, and intercept form).
2.4.3. Varying the Parameters in a Line Equation. Let’s
now see what happens when we vary the parameters. Most of this
should be clear from the “visual” or “physical” interpretation of the
parameters in the different forms of the line equation covered in the
previous subsection.
In the slope-intercept form of a line equation
(2.18) y = mx + b
what happens if we vary the slope m and the y-intercept b? (See
Figure 2.11 for examples.)
Here are some fairly obvious remarks about the slope m:
• If m is positive, then the line described by (2.18) “goes up” or
“is increasing”, i.e., y increases as x increases.
• If m is negative, then the line described by (2.18) “goes down”
or “is decreasing”, i.e., y decreases as x increases.
42 2. LINEAR EQUATIONS

Figure 2.11. Examples of varying slope and y-intercept

of a line

• If m is 0, then the line described by (2.18) is horizontal, i.e.,

y remains constant as x varies.
• As m increases (and assuming it is always positive), the line
described by (2.18) becomes “steeper”.
• As m decreases (and assuming it is always positive), the line
described by (2.18) becomes “flatter” or “less steep”.
Similarly, here are some remarks about the y-intercept b:
• If b is positive, then the line described by (2.18) crosses the
y-axis “above” the origin.
• If b is negative, then the line described by (2.18) crosses the
y-axis “below” the origin.
• If b is 0, then the line described by (2.18) crosses the y-axis at
the origin.
 2.5. SOLVING A LINEAR EQUATION IN ONE VARIABLE 43

• As b increases or decreases, the point at which the line de-

scribed by (2.18) crosses the y-axis moves “up” or “down”,
respectively.
Similar comments can be made about the x- and y-intercepts in the
intercept form of the line equation
x y
+ =1
a b

2.5. Solving a Linear Equation in One Variable

After all this long preparation, let’s now “solve” linear equations!
We’ll again start with two examples:
“A pool is being filled with water at a constant rate of
50 `/min (liters per minute), starting at noon. If the
pool already contained 1,000 liters of water at noon,
how much water will be in the pool after 30 minutes?”

“A pool is being filled with water at a constant rate

of 50 `/min, starting at noon. If the pool already
contained 1,000 liters of water at noon, what time
will it be when the pool contains 2,500 liters?”
We can see that the time t (in minutes since noon) and the water
volume V in the pool (in liters) are related in both examples by the
equation
(2.19) V = 50t + 1000
since the “slope” is the rate of 50 `/min, and the “V -intercept” is the
1,000 liters of water already in the pool at noon. Now the first example
above is just an arithmetic problem: We replace t by 30 and obtain the
answer V = 50 · 30 + 1000 = 2500 (in liters). On the other hand, the
second problem gives V = 2500 and asks for a solution of the equation
(2.20) 50t + 1000 = 2500
This requires “undoing” the operation of “computing the volume from
the time” as explained in section 1.6, i.e., it requires rewriting the
equation by isolating t. We undo (2.20) in two steps:
50t + 1000 = 2500 turns into 50t = 1500, and
50t = 1500 turns into t = 30
This then gives the answer: “It will be 12:30 p.m. when the pool con-
tains 2,500 liters.”
44 2. LINEAR EQUATIONS

More precisely, what we are saying is that the following three equa-
tions are equivalent (namely, have the same solutions in t):
(2.21) 50t + 1000 = 2500
(2.22) 50t = 1500
(2.23) t = 30
But why do we know that they are equivalent? Notice that we really
have to show two things:
(1) Any solution to equation (2.21) is also a solution to equa-
tion (2.23), and
(2) any solution to equation (2.23) is also a solution to equa-
tion (2.21).
Let’s address item (1) first: Equation (2.22) is obtained from equa-
tion (2.21) by subtracting 1000 “on both sides”. But, if we replace t
in equation (2.21) by any number which makes both sides of the equa-
tion equal, then this equality remains true if we subtract 1000 from
both sides. (After all, the two sides of equation (2.21) represent the
same number before the subtraction, so they must still represent the
same number after the subtraction!) The same is true in going from
equation (2.22) to equation (2.23): If we replace t in equation (2.22)
by any number which makes both sides of the equation equal, then this
equality remains true after we divide by 50 on both sides.
The argument for item (2) is similar, but with a crucial difference:
We first have to show that if we replace t in equation (2.23) by any
number which makes both sides of the equation equal, then this equal-
ity is true before we divide both sides by 50. There is only one way
to show this: Note that equation (2.22) can be obtained from equa-
tion (2.23) by multiplying both sides by 50! Now we can argue as for
item (1): If we replace t in equation (2.23) by any number which makes
both sides of the equation equal, then this equality remains true if we
multiply both sides by 50, and the resulting equation is equation (2.22).
A similar argument shows that any solution to equation (2.22) is a so-
lution to equation (2.21), since equation (2.21) can be obtained from
equation (2.22) by adding 1000 to both sides of the equation. Both
steps of the argument for item (2) can be viewed as “undoing” the
corresponding steps of the argument for item (1).
This leads us to two crucial observations about “solving equations”
in general:
Proposition 2.6. (1) Any solution to an equation A = B
(where A and B are algebraic expressions) remains a solution
 2.5. SOLVING A LINEAR EQUATION IN ONE VARIABLE 45

if we add or subtract the same algebraic expression on both

sides of the equation; and vice versa, any solution to the equa-
tion after adding or subtracting the same algebraic expression
is also a solution of the original equation A = B.
(2) Any solution to an equation A = B (where A and B are alge-
braic expressions) remains a solution if we multiply or divide
both sides of the equation by the same nonzero algebraic ex-
pression; and vice versa, any solution to the equation after
multiplying or dividing by the same nonzero algebraic expres-
sion is also a solution of the original equation A = B. (Here,
an algebraic expression is nonzero if it cannot equal 0 no mat-
ter what numbers replace the letters in the expression.)

Here are two fine points which you can ignore on a first reading,
but which are important in general:

Remark 2.7. There is small caveat here: If the algebraic expres-

sion C in Proposition 2.6, which we add, subtract, multiply or divide
by on both sides of an equation cannot be computed for all values of its
variables, then our proposition only holds for those values at which C
can be computed.
There is also an extension of Proposition 2.6 (2): If the algebraic
expression C equals 0 for some but not all replacements of letters in C
by numbers, then Proposition 2.6 (2) remains true for those values at
which C does not equal 0.

Let’s first look at the caveat. Here are two very simple examples:
(1) If we add x1 to both sides of the equation x = 0, say, then the new
equation x+ x1 = x1 obviously does not have the same solutions as x = 0;
but the proposition still holds for all values of x at which x1 is defined,
namely, whenever x is nonzero: Both equations x = 0 and x + x1 = x1
are
√ false for all nonzero x, as the proposition states. (2) If we add
x to
√ both sides √ of the equation x = −1, say, then the new equation
x + x = −1 + x obviously does not have the same solutions√as
x = −1; but the proposition still holds for all values of x at which x
is defined,
√ √ whenever x is non-negative: Both x = −1 and
namely,
x + x = −1 + x are false for all non-negative x as the proposition
states.
As for the extension of Proposition 2.6 (2), here is a very simple ex-
ample: The equation x = 1 obviously does not have the same solutions
as the equations x2 = x (obtained by multiplying both sides by x).
However, if we only look at the solutions to both equations where x
46 2. LINEAR EQUATIONS

is nonzero, then they do indeed have the same solutions, namely, only
x = 1.
Now for the proof of Proposition 2.6: Adding to, subtracting from,
multiplying by a nonzero expression, or dividing by a nonzero expres-
sion, both sides of an equation “preserves” all solutions of the original
equation; and since these steps can be reversed by subtracting, adding,
dividing, and multiplying, respectively, we also don’t get any extra
solutions.
Notice that the crucial part in Proposition 2.6 is the “vice versa”
part: Not only do we “preserve” a solution by adding, subtracting,
multiplying or dividing both sides of an equation; we also don’t get
any new solution. Of course, in part (2) of Proposition 2.6, we defi-
nitely need the restriction that we don’t multiply or divide by 0 (or an
expression which could equal 0): Dividing by 0 is not allowed anyway,
and multiplying an equation A = B by 0 leads to the equation 0 = 0,
which almost always has more solutions than the original equation,
namely, any number is a “solution” to 0 = 0. A more sophisticated ex-
ample (and an example of a very common error) is to divide both sides
of the equation x2 = x by x to “arrive” at the equation x = 1. Clearly,
x may be zero, so this would not be allowed under Proposition 2.6,
and it is also easy to see that Proposition 2.6 fails in this example:
The equation x2 = x has two solutions, namely, 0 and 1, whereas the
equation x = 1 only has the solution 1. (However, as noted above, the
extension of our proposition still allows us to state that the nonzero
solutions of x2 = x and x = 1 agree!)
More generally, why am I putting so much emphasis on the “vice
versa” part? After all, it’s obvious that you shouldn’t multiply both
sides of an equation by 0 while solving an equation. The reason is that
this “vice versa” part will become much more subtle later on and leads
to frequent mistakes. For example, if we square an equation A = B,
then any solution to the original equation A = B is still a solution to
the equation A2 = B 2 after squaring, but not vice versa, as you can see
from the simple example x = 1: It has only one solution, namely, 1; but
the equation x2 = 1 (i.e., after squaring) has two solutions, namely, 1
and −1. So as we extend Proposition 2.6 later on (e.g., in Proposi-
tion 5.3) to allow more “operations” on equations, we always have to
make sure that the “vice versa” part also holds, since it can fail in
subtle ways!
Let’s now vary the pool problem a bit to arrive at the general form
of a linear equation:
 2.5. SOLVING A LINEAR EQUATION IN ONE VARIABLE 47

“A pool is being filled at a constant rate of 50 `/min,

starting at noon, at which time it already contains
1,000 liters of water. If that same pool had contained
1,600 liters of water at noon and had been filled at
the rate of only 30 `/min, starting at noon, it would
have been full at the same time. At what time will the
pool be full? How much water can the pool hold?”
If we still let t be the time (in minutes) since noon, the expression
50t + 1000 describes the water volume under the first scenario, while
the expression 30t + 1600 describes it under the second scenario. The
problem tells us that the pool is full (and so contains the same volume
of water) at the same time t, so we arrive at the equation
(2.24) 50t + 1000 = 30t + 1600
Now we can use Proposition 2.6 three times to “isolate” t: Equa-
tion (2.24) is equivalent to
20t + 1000 = 1600
by subtracting 30t from both sides. Next we subtract 1000 from both
sides to arrive at the equivalent equation
20t = 600
Finally, we divide by 20 on both sides to arrive at the solution
t = 30
which yields the answer “The pool will full at 12:30 p.m.” A small
further computation shows that the pool can hold 2,500 liters (since
50 · 30 + 1000 = 30 · 30 + 1600 = 2500).
There is one other way to solve our pool problems, namely, graph-
ically. Let’s return to the first equation we tried to solve:
(2.20) 50t + 1000 = 2500
Consider the equation describing the water volume in terms of the
time:
(2.19) V = 50t + 1000
We can view solving (2.20) as determining the t-coordinate of the
point P = (30, 2500) where the line described by equation (2.19) inter-
sects the horizontal line V = 2500. (See Figure 2.12.)
Let’s turn to the other equation we solved in this section:
(2.24) 50t + 1000 = 30t + 1600
48 2. LINEAR EQUATIONS

Figure 2.12. Solving the equation 50t + 1000 = 2500 graphically

Again, there is a nice graphical solution. We can describe the water

volume in terms of the time for the two different rates as follows:
(2.25) V = 50t + 1000 for the rate of 50 `/min
(2.26) V = 30t + 1600 for the rate of 30 `/min
The problem now asked for the time at which the volume is the same,
i.e., to determine the t-coordinate of the point P = (30, 2500) where
the two lines described by the equations (2.25) and (2.26) intersect.
(See Figure 2.13.)
A general linear equation has the form
(2.27) m0 x + b0 = m1 x + b1
and can be viewed graphically as finding the x-coordinate of the point
at which the lines described by
(2.28) y = m0 x + b0 and y = m1 x + b1
 2.5. SOLVING A LINEAR EQUATION IN ONE VARIABLE 49

Figure 2.13. Solving the equation 50t + 1000 = 30t +

1600 graphically

intersect (as, e.g., in Figure 2.12, where one line is horizontal, or Fig-
ure 2.13, where neither line is horizontal).
Algebraically speaking, we first subtract (up to two times) to “get
all the x on one side” and “all the constants on the other side”. In order
to see what might happen, it’s easier to view this problem graphically
as finding the x-coordinate of the point where the two non-vertical
lines described by the equations in (2.28) intersect. There are now
three cases for the two lines described by the equations in (2.28):
Case 1: The lines intersect in a single point P , say: Then the two
lines must have different slope, i.e., m0 6= m1 . In that case, there is
exactly one solution to the equation (2.27), namely the x-coordinate
50 2. LINEAR EQUATIONS

of P . Algebraically, we can find this value in three steps:

(m0 − m1 ) x + b0 = b1 (by subtracting m1 x)
(m0 − m1 ) x = b1 − b0 (by subtracting b0 )
b 1 − b0
x= (by dividing by m0 − m1 )
m0 − m1
Here, the last step is possible since m0 6= m1 and thus m0 − m1 6= 0.
A real-life example of such a situation would be the following:
“Two trains leave the same city, each heading in the
same direction at a constant speed of 80km/h and
120km/h, respectively. At noon, the first train is at a
distance of 140km from the city, and the second train
at a distance of 100km. At what time will the second
train pass the first?
It is easy to see that solving this problem amounts to solving the
equation 80t + 140 = 120t + 100, and, after some algebra, one obtains
the solution t = 1 (where the time t is measured in hours since noon).
This solution is shown in Figure 2.14 as the point P at the intersection
of the lines `1 and `0 .
Case 2: The lines are parallel and have no point in common: Then
the two lines must have the same slope, i.e., m0 = m1 , but different
y-intercept, i.e., b0 6= b1 . In that case, there is no solution to the
equation (2.27). Algebraically, we can see this in three steps:
(m0 − m1 ) x + b0 = b1 (by subtracting m1 x)
(m0 − m1 ) x = b1 − b0 (by subtracting b0 )
0 = b1 − b0 (since m0 = m1 )
The last equation is always false and thus has no solution x. A similar
real-life example of such a situation would be the following:
“Two trains leave the same city, both heading in the
same direction at a constant speed of 120km/h. At
noon, the first train is at a distance of 160km from
the city, and the second train at a distance of 100km.
At what time will the second train pass the first?
It is easy to see that solving this problem amounts to solving the
equation 120t + 160 = 120t + 100, which has no solution: The sec-
ond train will never pass the first. This situation is also shown in
Figure 2.14: The two lines `2 and `0 do not intersect.
Case 3: The lines are parallel and in fact coincide: Then the two
lines must have the same slope and the same y-intercept, i.e., m0 =
 2.5. SOLVING A LINEAR EQUATION IN ONE VARIABLE 51

Figure 2.14. Solving the three trains problems graphically

m1 and b0 = b1 . In that case, every value of x is a solution to the

equation (2.27). Algebraically, we can see this again in three steps:

(m0 − m1 ) x + b0 = b1 (by subtracting m1 x)

(m0 − m1 ) x = b1 − b0 (by subtracting b0 )
0=0 (since m0 = m1 and b0 = b1 )

The last equation is always true, and so any value of x is a solution.

A similar (somewhat non-sensical) real-life example of such a situation
would be the following:
“Two trains leave the same city, both heading in the
same direction at a constant speed of 120km/h. At
52 2. LINEAR EQUATIONS

noon, both trains are at a distance of 100km from the

city. At what time will one train pass the other?
It is easy to see that solving this problem amounts to solving the
equation 120t + 100 = 120t + 100, and that the trains will always be at
the same distance from the city. This situation is shown in Figure 2.14:
The travel of both trains is described by the same line `0 .

2.6. Solving Simultaneous Linear Equations in Two or More

Variables
Let’s start again with an example:
“I’m thinking of two numbers. The larger of the two
numbers is less than twice the smaller number by 3.
The sum of the two numbers is 18. What are my
numbers?”
Now it is, of course, possible to solve this problem using only what
we already know, namely, linear equations in one variable: If I call the
smaller number x, then the larger number can be expressed as 2x − 3
(since I’m told that “the larger of the two numbers is less than twice
the smaller number by 3”). The other piece of information then tells
us about the sum of the two numbers, namely, that
x + (2x − 3) = 18
which can be rewritten as 3x − 3 = 18 and solved to x = 7 as discussed
in the previous section. But it may, and will, not always be so easy to
express one unknown quantity in terms of another; so an alternative
and more general approach is to call the two numbers x and y (and let’s
say x is the smaller of the two). Now we can translate the information
from the problem into two equations
y = 2x − 3
(2.29)
x + y = 18
We can rewrite these equations into two line equations in general form
as
−2x + y + 3 = 0
(2.30)
x + y − 18 = 0
Solving these equations now means that we have to find (all the) values
for the unknowns x and y which make both equations true.
There are a number of different methods to solve such simultaneous
linear equations as we’ll call them, i.e., several equations mentioning
several unknowns (but where none of the unknowns is raised to any
 2.6. SOLVING LINEAR EQUATIONS IN TWO OR MORE VARIABLES 53

power, or occurs in a denominator or in a product with another un-

known) so that each equation is equivalent to a line equation in general
form.
Geometrically, solving two simultaneous linear equations in two un-
knowns means that we need to find the point(s) of intersection of two
lines in the plane. (If the lines are not parallel, there will be exactly
one point of intersection.) Figure 2.15 shows the two lines from our
example problem and the point of intersection P = (7, 11).

Figure 2.15. Solving the two equations 2x + y + 3 = 0

and x + y − 18 = 0 graphically

Algebraically, there are a number of different ways to solve simulta-

neous linear equations, and experience will tell you and your students
which method is more efficient in a particular situation. E.g., starting
from (2.29), it would be easiest to use the substitution method : We
already know that y = 2x − 3, so we could simply replace y in the
other equation x + y = 18 by 2x − 3 and obtain the single equation
54 2. LINEAR EQUATIONS

x + (2x − 3) = 18 in a single unknown, which can then be solved as in

the previous section. (This is essentially how we solved this problem
at first at the beginning of the current section.)
We could also have started from (2.30), say, and found out from the
equation x + y = 18 that x = 18 − y (by subtracting y on both sides);
now we can replace x by 18 − y in the other equation −2x + y + 3 = 0
to arrive at the equation −2(18 − y) + y + 3 = 0 in a single variable,
which can then be simplified to 3y − 33 = 0 and solved for y as in
the previous section (namely, to y = 11). Clearly, the second way of
using the substitution method is somewhat more cumbersome here, but
the guiding principle is always the same: Rewrite one of the equations
so that one variable occurs alone on one side and not at all on the
other side, and then substitute it in the other equation to arrive at
an equation in a single variable. Once we have found the value of one
unknown, it’s easy to find the value for the other unknown. E.g., in
the substitution of the previous paragraph, we obtain x = 7 and then
compute y = 2·7−3 = 11. In the substitution of the current paragraph,
we obtain y = 11 and then compute x = 18 − 11 = 7.
Starting from (2.30), there is a very different way to solve the equa-
tions called the elimination method. The idea is that the equation
x + y − 18 = 0 is equivalent to 2x + 2y − 36 = 0 (by multiplying
both sides of the equation by 2). Now we can add this new equation
2x + 2y − 36 = 0 and the other equation −2x + y + 3 = 0 and arrive at
(2x + 2y − 36) + (−2x + y + 3) = 0 + 0. After simplifying to 3y − 33 = 0,
we see that we have “eliminated” one unknown, namely, x, and so have
an equation we can again solve by the method of the previous section.

Alternatively, we could have subtracted the first equation −2x + y +

3 = 0 in (2.30) from the second equation x + y − 18 = 0 to arrive at
(x+y −18)−(−2x+y +3) = 0−0, and after simplifying to 3x−21 = 0,
we have again “eliminated” one unknown, namely, y this time, and so
have an equation we can again solve by the method of the previous
section.
In either way of eliminating one unknown, we can solve the result-
ing equation in one unknown and then replace the solution for this
unknown in one of the original equations to now solve for the other
unknown. E.g., in the last paragraph, once we have solved the new
equation 3x − 21 to x = 7, we can solve for y by solving 7 + y − 18 = 0
and arrive at y = 11.
Note that the elimination method introduces a new way to rewrite
equations. In Proposition 2.6, we had discussed rewriting a single equa-
tion by adding or subtracting the same expression to or from both sides,
 2.6. SOLVING LINEAR EQUATIONS IN TWO OR MORE VARIABLES 55

or by multiplying or dividing both sides by the same nonzero expres-

sion on both sides. The elimination method requires a new operation
on two simultaneous equations:
Proposition 2.8. (1) Any solutions to two simultaneous equa-
tions A = B and C = D (where A, B, C, and D are alge-
braic expressions) remain solutions if we add or subtract one
equation to (or from) the other, i.e., the simultaneous solu-
tions of A = B and C = D are still simultaneous solutions
of the simultaneous equations A + C = B + D and C = D,
and also simultaneous solutions of the simultaneous equations
A − C = B − D and C = D. (Similarly, for later reference,
if C and D are nonzero expressions, then any solutions to two
simultaneous equations A = B and C = D (where A, B, C,
and D are algebraic expressions) remain solutions if we multi-
ply or divide one equation by the other, i.e., the simultaneous
solutions of A = B and C = D are still simultaneous solutions
of the simultaneous equations AC = BD and C = D, and also
simultaneous solutions of the simultaneous equations CA = D B

and C = D.)
(2) Conversely, any simultaneous solutions of the equations A +
C = B +D and C = D (or A−C = B −D and C = D, respec-
tively) are also simultaneous solutions of the original equations
A = B and C = D. (Similarly, again for later reference, if C
and D are nonzero expressions, then any simultaneous solu-
tions of the equations AC = BD and C = D (or CA = D B
and
C = D, respectively) are also simultaneous solutions of the
original equations A = B and C = D.)
The proof of the part (1) of Proposition 2.8 is simple: If we replace
the variables in A, B, C, and D by any fixed numbers, then the truth
of the equations A = B and C = D with the variables replaced by
these fixed numbers implies the truth of the equations A + C = B + D,
A−C = B −D, AC = BD and CA = D B
. For part (2) of Proposition 2.8,
we have to work just a little harder. E.g., for the case A+C = B+D and
C = D, we use the fact that we can “reverse” the operation of “adding
equations” by “subtracting”: From A + C = B + D and C = D, we
obtain (A + C) − C = (B + D) − D, which simplifies to the desired
equation A = B. Similarly, for the case AC = BD and C = D, we use
the fact that we can “reverse” the operation of “multiplying equations”
by “dividing”: From AC = BD and C = D, we obtain AC C
= BDD
,
which simplifies to the desired equation A = B, given that C and D
are nonzero expressions.
56 2. LINEAR EQUATIONS

Let’s now return to our geometric intuition for solving two simul-
taneous linear equations in two unknowns again to see what else can
happen. We had said that, geometrically, solving two simultaneous
linear equations in two unknowns amounts to finding the point(s) of
intersection of the two lines described by the two equations. This is
easy by either of the two algebraic methods (substitution method or
elimination method) described above as long as the two lines are not
parallel and therefore intersect in exactly one point. Let’s vary our
original problem a bit as follows:
“I’m thinking of two numbers. The sum of the two
numbers is 12. The sum of twice each of the two
numbers exceeds 24 by 3. What are my numbers?”
We can quickly translate this problem into the following two simul-
taneous equations
x + y = 12
(2.31)
2x + 2y − 3 = 24

If we tried to solve (2.31) by the substitution method, we could

rewrite the first equation as x = 12 − y and thus the second equa-
tion as 2(12 − y) + 2y − 3 = 24. But this last equation simplifies to
24 − 3 = 24, which is always false. Since this last equation has no si-
multaneous solutions x and y, neither do the original equations (2.31),
by Proposition 2.8 (2).
Or we could have tried to solve (2.31) by the elimination method,
multiplying the first equation by 2 to obtain 2x + 2y = 24, and then
subtracting that from the second equation to arrive at (2x+2y)−(2x+
2y − 3) = 24 − 24, which simplifies to 3 = 0, again an equation which
is always false. And this implies that the original equations (2.31) do
not have any simultaneous solutions, again by Proposition 2.8 (2).
So what happened? If we rewrite (2.31) as line equations in slope-
intercept form
y = −x + 12
(2.32)
y = −x + 13.5

we quickly see that these lines are parallel and do not intersect; so
it is also geometrically clear that they cannot have any simultaneous
solutions.
Finally, it is possible that the two lines from two simultaneous linear
equations coincide, in which case there are infinitely many simultaneous
solutions. Here is a somewhat non-sensical example:
 2.6. SOLVING LINEAR EQUATIONS IN TWO OR MORE VARIABLES 57

“I’m thinking of two numbers. The sum of the two

numbers is 12. The sum of twice each of the two
numbers equals 24. What are my numbers?”
This problem can be expressed by the two simultaneous equations
x + y = 12
(2.33)
2x + 2y = 24
But these equations represent the same line, with the equation y =
−x + 12 in slope-intercept form, so any solution x is possible as long
as y is then chosen so as to satisfy y = −x + 12.
Finally, the above carries over to simultaneous linear equations in
more than two unknowns. Here is an example:
“Richard spends $35 on two books, a CD, and a DVD.
Susan can only afford to pay $25 for a CD and a
DVD. And Tom pays just $20 for a book and a DVD.
Assuming that each book costs the same, each CD
costs the same, and each DVD costs the same, what
is the price of each item?”
Let’s denote the price of a book by b, the price of a CD by c, and
the price of a DVD by d. Then we arrive at the equations
2b + c + d = 35
(2.34) c + d = 25
b + d = 20
representing in turn the total cost of Richard’s, Susan’s, and Tom’s
purchases, respectively.
It is “possible” to visualize a solution to this problem in three-
dimensional coordinate space: Each linear equation then represents a
plane in three-dimensional space, and a solution is given by any point
at which all three planes intersect. Obviously, this is rather hard to
draw (or even visualize), so school mathematics generally approaches
this kind of problem only algebraically by one of our two methods,
the substitution method or the elimination method, or a combination
of the two. Here are two (of several) possible ways to solve (2.34)
algebraically:
(1) The second and third equation of (2.34) allow us to express
both c and b in terms of d by subtracting d on both sides of
each equation, arriving at
c = 25 − d
b = 20 − d
58 2. LINEAR EQUATIONS

Now we can replace c and b in the first equation of (2.34) and

obtain the following equation in only one unknown:
2(20 − d) + (25 − d) + d = 35
This equation simplifies to
65 − 2d = 35
and gives us the solution d = 15 as in the previous section.
Now we can also compute
c = 25 − d = 25 − 15 = 10
b = 20 − d = 20 − 15 = 5
(2) We can start by subtracting the second equation of (2.34) from
the first and arrive at the equation
(2b + c + d) − (c + d) = 35 − 25
which simplifies to the following equation in only one unknown:
2b = 10
This gives us the solution b = 5. Now we can use the third
equation of (2.34) to compute
d = 20 − b = 20 − 5 = 15
and then use the second equation of (2.34) to compute
c = 25 − d = 25 − 15 = 10
In either case, we see that a book costs $5, a CD $10, and a
DVD $15.
 CHAPTER 3

Order and Linear Inequalities

3.1. Ordering the Numbers

All the number systems we will consider here (except for the com-
plex numbers, which we will touch upon only briefly in subsection 5.3.2)
carry a natural “order”, which formalizes the very intuitive idea of
“more” and “less”. This concept is introduced very early for the whole
numbers: If Mary has seven apples and John has five, then Mary has
more apples than John. Using like denominators, one can carry over
this concept to fractions; and using place value, to positive decimals.
On the other hand, the concept of “more” and “less” is somewhat
counterintuitive for negative numbers and doesn’t always conform with
everyday language: If Susan owes $5 and Bob owes $8, then Bob “owes
more” than Susan; but if one considers their “savings” and allows this
to be negative in the case of debt, then Susan has “more” savings
than Bob. A better model for negative numbers and their order is
temperature: −20◦ is less than −10◦ and also less than +10◦ (even
though 20 is “bigger” than 10). Another intuitive way to think about
the order of all numbers (positive, negative, or 0) is on the number
line: Being less than a number means being to the left of the point on
the number line representing that number.
Such concepts are expressed in mathematics by inequalities, which
is somewhat of a misnomer: We express that two numbers, say, 2 and 3,
are unequal by writing 2 6= 3, which is literally an “inequality”; but one
usually understands an inequality to be a statement involving one of
the symbols <, >, ≤, or ≥. We will start by “formally” using only <,
and we will consider a > b to be merely an abbreviation for b < a.
Here are some basic properties of the ordering < for all numbers a,
b, and c:
(irreflexivity) a < a is false
(transitivity) a < b and b < c implies a < c
(comparability) a < b or a = b or a > b
So the first property (irreflexivity) states that no number is less than
“itself” (thus the name of the property: “reflexive” means “referring
59
60 3. ORDER AND LINEAR INEQUALITIES

to itself”). The second property (transitivity) states that from two

facts one can conclude a third: If I know that b is greater than a
but less than c, then certainly a is less than c. The third property
(comparability) states that whenever two numbers a and b are not
equal, then one must be less than the other.
What is a bit confusing is that there is an alternative way to state
properties of the ordering of the numbers, using the symbol ≤. (Again,
a ≥ b is simply an abbreviation for b ≤ a, so “formally”, we don’t use ≥
at all!) Now there are two ways to deal with the symbol ≤:
On the one hand, we could simply consider a ≤ b as an abbreviation
by letting
(3.1) a≤b abbreviate a < b or a = b
and so the symbol ≤ gives no additional meaning which we couldn’t
express before.
On the other hand, we could replace the symbol < by the symbol ≤
and only deal with the latter. Then the basic properties of the ordering
under ≤ become the following, for all numbers a, b, and c:
(reflexivity) a≤a
(transitivity) a ≤ b and b ≤ c implies a ≤ c
(antisymmetry) a ≤ b and b ≤ a implies a = b
(comparability) a ≤ b or a ≥ b
So now the first property (reflexivity) states that every number is less
than or equal to “itself” (thus again the name of the property). The
second property (transitivity) states that from two facts one can con-
clude a third: If I know that b is greater than or equal to a but less
than or equal to c, then certainly a is less than or equal to c. The third
property (antisymmetry) states that if a number is both less than or
equal to, and greater than or equal to, another number, then the two
numbers are indeed equal. The last property (comparability) states
that for any two numbers, one is less than or equal the other, or vice
versa.
Now, assuming that we have only introduced the symbol ≤ so far
(but not <), we can think of < as an abbreviation by letting
(3.2) a<b abbreviate a ≤ b and a 6= b
So, in summary, it doesn’t really matter which of the two symbols, <
or ≤, we introduce first; we can simply define the other from it. And
in fact the properties of ≤ can be proved from the properties of < and
vice versa:
 3.2. ORDER AND THE ARITHMETICAL OPERATIONS 61

Exercise 3.1. (1) Show that assuming the three properties

of < and the definition of ≤ by (3.1), we can prove the four
properties of ≤.
(2) Conversely, show that assuming the four properties of ≤ and
the definition of < by (3.2), we can prove the three properties
of <.

Two more remarks are in order: In light of the comparability prop-

erties, we don’t need any symbols like 6< or 6≤, since we can rewrite,
e.g., a 6< b as a ≥ b (i.e., as b ≤ a), and a 6≤ b as a > b (i.e., as b < a).
Finally, there are many different ways to express order in English,
using words like “exceed”, “not”, etc.:

Exercise 3.2. For each of the statements a < b, a ≤ b, a > b, and

a ≥ b, find at least four different ways to express these in the English
language, for quantities a and b.

3.2. Order and the Arithmetical Operations

So far, we have considered the ordering of the numbers and the
arithmetical operations separately. But they are closely connected in-
deed, as we will now explore in detail. These connections will be es-
sential in “solving inequalities” later on.
First of all, recall the following trivial properties linking equality
of numbers and the arithmetical operations. (These were crucial in
establishing Propositions 2.6 and 2.8!)

(3.3) a = b implies a + c = b + c
(3.4) a = b implies a − c = b − c
(3.5) c 6= 0 and a = b implies ac = bc
a b
(3.6) c 6= 0 and a = b implies =
c c

In fact, not only do these properties hold, but it’s also not hard to see
that their converses hold; i.e., the right-hand side of each statement im-
plies the left-hand side (always assuming that c 6= 0 in (3.5) and (3.6)).
This is since we can “undo” the addition in (3.3) using (3.4); the sub-
traction in (3.4) using (3.3); the multiplication in (3.5) using (3.6); and
the division in (3.6) using (3.5).
62 3. ORDER AND LINEAR INEQUALITIES

There are now the following similar properties linking the order <
of the numbers and the arithmetical operations:
(3.7) a < b implies a + c < b + c
(3.8) a < b implies a − c < b − c
(3.9) c > 0 and a < b implies ac < bc
a b
(3.10) c > 0 and a < b implies <
c c
Why do properties (3.7)–(3.10) hold? This is easy and intuitive to
see for properties (3.7)–(3.8): Adding or subtracting c to both sides of
an inequality does not change its truth. Similarly, for properties (3.9)–
(3.10), multiplying or dividing both sides of an inequality by a positive
number c does not change its truth.
There are also more visual proofs: The implications (3.7) and (3.8)
hold since we can think of them as amounting to “shifting” a and b on
the number line to the right or left by an equal distance, given by c.
The implications (3.9) and (3.10) hold since we can think of them as
amounting to “scaling” a and b on the number line by the positive
factor or divisor c.
Again, we have the converses of these properties (the “right-to-left”
direction of these implications), “undoing” the operations as for (3.3)–
(3.6), still under the assumption that c > 0 in (3.9) and (3.10).
We can also state similar properties and their converses using ≤,
except that we still assume c > 0 in (3.13) and (3.14):
(3.11) a ≤ b implies a + c ≤ b + c
(3.12) a ≤ b implies a − c ≤ b − c
(3.13) c > 0 and a ≤ b implies ac ≤ bc
a b
(3.14) c > 0 and a ≤ b implies ≤
c c
And again we have the converses of these properties (the “right-to-left”
direction of these implications), “undoing” the operations as for (3.3)–
(3.6), still under the assumption that c > 0 in (3.13) and (3.14).
But this leads to an obvious question, and one which is at the root of
many errors when solving inequalities: Why do we need the restriction
that c > 0 in properties (3.9)–(3.10) and (3.13)–(3.14)? We start with
the following easy but crucial observations:
(3.15) a < b implies − a > −b
(3.16) a ≤ b implies − a ≥ −b
 3.2. ORDER AND THE ARITHMETICAL OPERATIONS 63

Note that (3.15) and (3.16) hold not only for positive numbers a and b,
but for 0 and negative numbers as well!
In order to see why (3.15) holds, we distinguish three cases:
Case 0 ≤ a < b: Then b lies to the right of a on the number line,
and both are ≥ 0. Since taking the negative of a number means to
reflect the points on the number line about 0, −b will now lie to the
left of a on the number line, and both are ≤ 0. (See Figure 3.1.)

-b -a 0 a b

Figure 3.1. The proof of (3.15) when 0 ≤ a < b

Case a < b ≤ 0: Then b lies to the right of a on the number line,

and both are ≤ 0. Since taking the negative of a number means to
reflect the points on the number line about 0, −b will now lie to the
left of −a on the number line, and both are ≥ 0. (See Figure 3.2.)

a b 0 -b -a

Figure 3.2. The proof of (3.15) when a < b ≤ 0

Case a < 0 < b: Then b lies to the right of 0, and a to the left of 0,
on the number line. Since taking the negative of a number means to
reflect the points on the number line about 0, −b will now lie to the
left of 0, and −a to the right of 0, on the number line. (See Figure 3.3.)

-b a 0 -a b

Figure 3.3. The proof of (3.15) when a < 0 < b

In each case, we have established −b < −a as required for (3.15).

The proof for (3.16) is similar.
Using (3.15), we can now prove versions of (3.9) and (3.10) as well
as (3.13) and (3.14) when c < 0:
64 3. ORDER AND LINEAR INEQUALITIES

(3.17) c < 0 and a < b implies ac > bc

a b
(3.18) c < 0 and a < b implies >
c c
(3.19) c < 0 and a ≤ b implies ac ≥ bc
a b
(3.20) c < 0 and a ≤ b implies ≥
c c
Each of (3.17)–(3.20) follows from the corresponding version when c >
0 by applying (3.15) or (3.16). E.g., (3.17) is shown as follows, assuming
c < 0:
a<b implies a(−c) < b(−c)
(3.21) implies a(−1)(−c) > b(−1)(−c)
implies ac > bc
The first implication follows by (3.9) (using that −c > 0), and the
second by (3.15) (using that −(a(−c)) = a(−1)(−c)), while the third
just evaluates (−1)(−c) = c.
A more visual proof of (3.17), closely following (3.21), goes as fol-
lows: Multiplying an inequality a < b by a negative number c amounts
to first scaling a and b on the number line by the positive number −c
(as in the first line of (3.21)), followed by a reflection of the number
line about the number 0 (as in the second line of (3.21)).
Again, we have the converses of (3.17)–(3.20) (i.e., the “right-to-
left” implications also hold, always assuming that c < 0) since we can
“undo” the multiplication or division by c by dividing or multiplying
by 1c . This uses that c < 0 implies 1c < 0, which follows by dividing
both sides of c < 0 by 1c twice.

3.3. Linear Inequalities in One Unknown

Equipped with the properties from the last section, it is now easy to
solve linear inequalities in one unknown, i.e., inequalities where both
sides are linear expressions. Let’s start with an example:
Little Red Riding Hood falls asleep in the forest, 5km
away from her grandmother’s house. The next morn-
ing, she walks toward her grandmother’s house at a
constant speed of 3km/h for a while. If she is now at
most 2km away from her grandmother’s house, how
long will she have walked?
 3.3. LINEAR INEQUALITIES IN ONE UNKNOWN 65

Let’s denote the time (in hours, from when she starts walking) by t
and the distance from her grandmother’s house (in kilometers) by s.
Then we have the following equation for time and distance:
(3.22) s = −3t + 5
The problem tells us that this quantity is at most 2 at the end of the
walk, leading to the inequality
−3t + 5 ≤ 2
Now we can first subtract 5 from both sides (using (3.12)) to obtain
−3t ≤ −3
and then divide by −3 (using (3.20)) to arrive at the solution
(3.23) t≥1
Thus she will have walked at least one hour.
Note that solving linear inequalities is thus very similar to solving
linear equations, the one difference being that whenever we multiply or
divide by a negative number, we have to invert the inequality sign. This
is reflected in the following proposition, which closely follows Proposi-
tion 2.6:
Proposition 3.3. (1) Any solution to an inequality A < B
(where A and B are algebraic expressions) remains a solu-
tion if we add or subtract the same algebraic expression on
both sides of the inequality; and vice versa, any solution to
the inequality after adding or subtracting the same algebraic
expression is also a solution of the original inequality A < B.
(2) Any solution to an inequality A < B (where A and B are alge-
braic expressions) remains a solution if we multiply or divide
both sides of the inequality by the same positive algebraic ex-
pression; and vice versa, any solution to the inequality after
multiplying or dividing by the same positive algebraic expres-
sion is also a solution of the original inequality A < B. (Here,
an algebraic expression is positive if it is positive no matter
what numbers replace the letters in the expression.)
(3) Any solution to an inequality A < B (where A and B are al-
gebraic expressions) is a solution of the inequality AC > BC
(or CA > BC
) obtained by multiplying (or dividing, respectively)
both sides of the inequality by the same negative algebraic ex-
pression and switching the inequality sign; and vice versa, any
solution to the inequality AC > BC (or CA > B C
) after multi-
plying (or dividing, respectively) by the same negative algebraic
66 3. ORDER AND LINEAR INEQUALITIES

expression and switching the inequality sign is also a solution

of the original inequality A < B. (Here, an algebraic expres-
sion is negative if it is negative no matter what numbers re-
place the letters in the expression.)
These results also hold with < and > switched; with < and > re-
placed by ≤ and ≥, respectively; and with < and > replaced by ≥ and ≤,
respectively.
The proof of Proposition 3.3 is basically the same as for Proposi-
tion 2.6: Adding or subtracting an algebraic expression on both sides
of an inequality, or multiplying and dividing both sides of an inequal-
ity by a nonzero algebraic expression, does not affect the solutions by
the results of Section 3.2 and the fact that all these operations on in-
equalities can be “undone”, except for the caveat that multiplication
or division by a negative number flips the inequality symbol.
Here is now a slightly more complicated example:
Little Red Riding Hood falls asleep in the forest, 5km
away from her grandmother’s house. The next morn-
ing, she starts walking toward her grandmother’s house
at a constant speed of 3km/h. When she starts walk-
ing, the big bad wolf starts along the same trail at a
constant speed of 6km/h from a distance of 8km from
the grandmother’s house. How long will Little Red
Riding Hood be safe from the wolf?
Let’s denote the time (in hours, from when she starts walking) by t
and the distance from her grandmother’s house (in kilometers) by sR
and sW for Little Red Riding Hood and the big bad wolf, respectively.
Then we have the following equation relating time and distance for
Little Red Riding Hood:
sR = −3t + 5
And for the big bad wolf, the equation relating time and distance reads:
sW = −6t + 8
The problem tells us that we want to find the times t when
sR < sW
since Little Red Riding Hood is only safe from the wolf as long as she
is closer to her grandmother’s house than the wolf. This leads to the
inequality
(3.24) − 3t + 5 < −6t + 8
 3.3. LINEAR INEQUALITIES IN ONE UNKNOWN 67

Now we can first subtract 5 from both sides (using (3.8)) to obtain
−3t < −6t + 3
then add 6t to both sides (using (3.7)) to obtain
3t < 3
and finally divide by 3 (using (3.10)) to arrive at the solution
(3.25) t<1
Thus she will be safe at any time less than one hour.
There are two other nice ways (in addition to (3.23) and (3.25)) to
represent the solutions to inequalities like (3.22) and (3.24):
• We can represent the solutions as intervals using the notation
from subsection 1.5.1: The solutions to (3.22) are all t ≥ 1,
and so the solution set can be written as [1, ∞). Similarly, the
solutions to (3.24) are all t < 1, and so the solution set can be
written as (−∞, 1).
• We can represent the solutions on the number line, by let-
ting the points corresponding to solutions be denoted by the
thicker part of the number line, and by using solid dots for
interval endpoints which are a solution, and circles for interval
endpoints which are not a solution. The solutions to (3.22)
and (3.24) on the number line are now pictured in Figures 3.4
and 3.5, respectively.

-3 -2 -1 0 1 2 3

Figure 3.4. The solution set of (3.22) on the number line

Here is one final example, illustrating how more than one inequality
for a quantity works:
Little Red Riding Hood falls asleep in the forest, 5km
away from her grandmother’s house. The next morn-
ing, she walks toward her grandmother’s house at a
constant speed of 3km/h for some time. How long
68 3. ORDER AND LINEAR INEQUALITIES

-3 -2 -1 0 1 2 3

Figure 3.5. The solution set of (3.24) on the number line

will she have walked when she is between 3km and

2km from her grandmother’s house?
The equation relating time and distance for Little Red Riding Hood
still reads
s = −3t + 5
where s is the distance from grandmother’s house (in kilometers) and t
the time from when she starts walking (in hours). The information
from the problem now tells us that the distance is between 2 and 3,
leading to the inequalities
(3.26) 2 ≤ −3t + 5 and 3 ≥ −3t + 5
which is usually written more compactly as
(3.27) 2 ≤ −3t + 5 ≤ 3
Note that this compact notation only works since the inequalities in
(3.26) are linked by the word “and”! In general, you can abbreviate
the statement “A ≤ B and B ≤ C” by “A ≤ B ≤ C”; and similarly
for A < B < C, A ≥ B ≥ C, and A > B > C. More about this in a
minute!
We can perform the same operations on the inequality (3.27) to
solve it: First subtract 5 on all sides to obtain
−3 ≤ −3t ≤ −2
and then divide by −3 on all sides to arrive at the solution
2
(3.28) 1≥t≥
3
So Little Red Riding Hood walks between 40 minutes and one hour.
Again, we can represent the solution set by an interval, namely, [ 32 , 1],
or on the number line as in Figure 3.6.
 3.4. ESTIMATION AND APPROXIMATION 69

-3 -2 -1 0 2/3 1 2 3

Figure 3.6. The solution set of (3.27) on the number line

Note that in this last example, the set of t which are not a solution
can be described by the statement
2 > −3t + 5 or 3 < −3t + 5
which is frequently (but incorrectly!) abbreviated as 3 < −3t + 5 <
2. This last statement makes no sense at all since it would imply
that 3 < 2. The “real” problem is that one cannot write the set of
“non”-solutions of (3.27) as a single interval; rather, it consists of two
intervals, namely, (−∞, 32 ) and (1, ∞). Using our set notation from
subsection 1.5.1, we can also write the set of “non”-solutions of (3.27)
as (−∞, 32 ) ∪ (1, ∞).
We will return to more complicated inequalities later on and then
discuss possible solution sets of inequalities in more detail. For now,
we turn to one common application of inequalities, estimation and
approximation.
3.4. Estimation and Approximation
In a “purely mathematical” setting, the number 2 means exactly
that: It’s the number 2, no more and no less. And writing it differently,
e.g., as 2.0 or 2.00, does not change that fact.
But in the “real world”, we can’t measure quantities exactly; we can
only measure them approximately. And there often is a “convention”
that 2 means 2 rounded to the nearest integer (i.e., 2 denotes a quan-
tity x which satisfies the inequality 1.5 ≤ x < 2.5), whereas 2.0 means 2
rounded to the nearest tenth (i.e., 2.0 denotes a quantity x which sat-
isfies the inequality 1.95 ≤ x < 2.05), and 2.00 means 2 rounded to the
nearest hundredth (i.e., 2.00 denotes a quantity x which satisfies the
inequality 1.995 ≤ x < 2.005).
A better and more precise way to write these would be to write
2 ± .5; 2 ± .05; and 2 ± .005, respectively. One advantage is that it
allows us to write things like 2 ± .1 (i.e., 2 “rounded to the closest
70 3. ORDER AND LINEAR INEQUALITIES

fifth” or, more precisely, a quantity x which satisfies the inequality

1.9 ≤ x < 2.1). Another advantage is that when reading 120, we can’t
really be sure whether this means 120 “rounded to the closest integer”
(i.e., a quantity x which satisfies the inequality 119.5 ≤ x < 120.5), or
120 “rounded to the closest ten” (i.e., a quantity x which satisfies the
inequality 115 ≤ x < 125); whereas, reading it as 120 ± .5 or 120 ± 5,
respectively, clarifies which of these two we mean.
There are now some quirks about estimation and approximation
which are frequently swept under the rug in school mathematics but
which are important to keep in mind for you as a teacher, and for your
students to at least be aware of, especially in the age of ubiquitous
pocket calculators which can spew out a number with eight significant
digits regardless of how imprecise the input was!
We’ll restrict ourselves here to just one example with two parts:
Given a rectangle of length 8.4cm and width 5.2cm,
what are (1) its perimeter, and (2) its area? Now
suppose that length and width are only given to the
nearest tenth, what are the possible values for (1) its
perimeter and (2) its width?
Your first inclination may be to reach for a pocket calculator and
to compute the perimeter as
p = 2 · 8.4 + 2 · 5.2 = 27.2
and the area as
A = 8.4 · 5.2 = 43.68
so, purely mathematically, the perimeter is exactly 27.2cm and the area
exactly 43.68cm2 . But in “real life”, the length l measuring 8.4cm and
the width w measuring 5.2cm means that
8.35 ≤ l < 8.45 and 5.15 ≤ w < 5.25
For the perimeter, this implies that
27 = 2 · 8.35 + 2 · 5.15 ≤ 2 · l + 2 · w < 2 · 8.45 + 2 · 5.25 = 27.4
and so p = 27.2 ± .2 in our earlier notation. Similarly, for the area,
43.0025 = 8.35 · 5.15 ≤ l · w < 8.45 · 5.25 = 44.3625
and so the area can vary almost all the way between 43cm2 and 44.4cm2 !
(Using the ± notation here is very messy. And in general, it’s quite
hard to give the exact bounds.)
We therefore see that especially computing the area as (exactly)
43.68cm2 is highly misleading! The general rule of thumb is that one
 3.4. ESTIMATION AND APPROXIMATION 71

should never give more significant digits for the output than for the in-
puts: Under this rule of thumb, we would have computed the perimeter
as 27cm and the area as 44cm2 (rounding each to two significant dig-
its), which is not quite accurate by our computations above (since the
latter suggests that 26.5 ≤ p < 27.5 and 43.5 ≤ A < 44.5), but at least
it is “in the right ballpark” for the accuracy of the answer.
 CHAPTER 4

The Concept of a Function, and Functions Closely

Related to Linear Functions

4.1. The Concept of a Function

In Chapter 2 on linear equations, we already saw examples of how
one quantity (let’s denote it by y) varies depending on another quantity
(which we will call x for now). In particular, the equation for a (non-
vertical) line can always be written in the form
(4.1) y = mx + b
for some fixed parameters m and b. In such a situation, we call x the
independent variable, and y the dependent variable, since we consider
how “y varies as x varies”.
First note that in the particularly simple case of (4.1) (and as long
as m 6= 0), the choice of which variable to call dependent and which
independent is somewhat arbitrary from a mathematical point of view
since (4.1) can be rewritten as
y b
x= −
m m
However, the “real life” context might make one choice more “natural”
than the other: Say, x might be the time elapsed since a train left
the station at a constant speed, and y might be the distance of the
train from the station: It would then be more natural to consider the
distance as time varies, rather than the time as distance varies, since
time “can’t be made flexible” whereas the distance of the train can
vary in a different pattern if the speed equals a different constant, or
is no longer constant at all.
Putting this minor point aside, let’s concentrate on the concept of
how a quantity y varies depending on how another quantity x varies:
Mathematics usually expresses such a situation by the concept of a
function
(4.2) y = f (x)
First of all, note that this notation may already cause some confu-
sion: f (x) does not mean that we multiply quantity f by quantity x,
72
 4.1. THE CONCEPT OF A FUNCTION 73

but rather that, as a computer scientist might put it, we apply some
“rule” f to the input x in order to determine the output y.
The set of possible values for the quantity x is called the domain
of the function; it is determined either mathematically by when the
equation y = f (x) “makes sense” or “is defined”, or by the “real life”
context as to when it makes sense for the “real life” quantity x. The
critical feature of a function is, however, the fact that for each argu-
ment x in the domain of the function, there is exactly one value for y
such that y = f (x) holds, as already suggested by the way we write
this. But note that we could have written the relation between the
quantities x and y differently, and it wouldn’t have been so easy to
check whether the relation defines a function: E.g., the relation
(4.3) x − y2 = 0
can be expressed as
x = y2
and so we can think of it as describing a function for the quantity x
as y varies. (The fact that I “switched” the roles of x and y here is
deliberate: The choice of what to call the variables is also completely
arbitrary, and one should not always choose the same letter for the
independent and the dependent variable, respectively!) On the other
hand, (4.3) does in general not describe y as a function of x: E.g.,
setting x = 1, both y = 1 and y = −1 satisfy (4.3). However, we can
resolve this “problem” by restricting the allowable values of y: If we
specify in addition to (4.3) that y ≥ 0, then (4.3) can be rewritten as
√
y= x
and y is a function of x. All this should convince you that when a
relationship between quantities is specified, determining whether or
not this relationship can be described as one quantity being a function
of another may depend on additional constraints being imposed, and
sometimes this is needed in order to obtain a function. And, of course,
the “real life” context may impose such or other constraints.
So let’s look at the concept of a function more abstractly.1 We’ll
also introduce some more handy notation going along with functions.
1In these lecture notes, we’ll restrict ourselves to functions from numbers to
numbers. More generally, one can consider functions from any set of objects to
any other set of objects. E.g., one could define a function with domain the set of
all Americans, which on the input “person” returns the value “eligible to vote” or
“not eligible to vote”. As long as we assume that for each American, exactly one
of “eligible to vote” or “not eligible to vote” is true, this would also be a function,
albeit admittedly much more abstract.
74 4. FUNCTIONS

First of all, let’s fix two sets of numbers X and Y ; we’ll call them the
domain and the codomain 2 of our function f . We then use

(4.4) f :X→Y

to denote that f is a function with domain X and codomain Y . Next,

for each argument x (i.e., each number x in the domain X), we have
to specify exactly one number y (called the value of f at x) in the
codomain Y ; we’ll then denote this number y as f (x). To expand on
our previous notation, we could write
f :X→Y
(4.5)
x 7→ f (x)

So in this notation, the function f (x) = 2x + 1 could be written as

f :R→R
x 7→ 2x + 1

but we could also restrict the domain to the set of positive real numbers,
say, and write

f : (0, ∞) → R
x 7→ 2x + 1

or to the set of whole numbers and write

f :Z→Z
x 7→ 2x + 1

Note in this notation that the two arrows are slightly different:
In the second line, we use the special arrow 7→ to indicate that x is
“mapped to” y; so a function is sometimes also called a map or a
mapping.
Often, in school mathematics, functions are given by simple formu-
las, such as f (x) = 2x + 1 or f (x) = x2 . But they do not have to be so
simple! Here are three “perfectly reasonable” functions for which there

2While the notion of a domain is completely standard, the notion of a codomain

is not, and often either no name is used for it at all, or it is sometimes called the
“range”, which, however, conflicts with our later notation.
 4.1. THE CONCEPT OF A FUNCTION 75

are no “easy formulas”:

(
1 if x ∈ Z
(4.6) f (x) =
0 otherwise
(
x if x ∈ Q
(4.7) g(x) =
0 otherwise
(4.8) h(x) = the largest integer n ≤ x
Note that in each of (4.6)–(4.8), we clearly have a function in the sense
in which we defined it at the beginning of the chapter, with domain
the set of all real numbers: For each argument x, there is exactly one
value y satisfying each of the three statements, respectively.
Another way to represent a function, apart from a “formula” or a
description of a function as above, is in the form of a table. Table 4.1
shows some values for the functions f , g and h. The problem with
a table is obviously that it can only show the values of a function at
a very small number of values; and in the case of the function g, the
table would not distinguish between g and the function√ ĝ(x) = x, for
example, if we hadn’t added the irrational number 2 to our table in
spite of the fact that it looks a bit out of place!
√
x -1 -0.5 0 0.5 1 2
f (x) 1 0 1 0 1 0
g(x) -1 -0.5 0 0.5 1 0
h(x) -1 -0.5 0 0 1 1

Table 4.1. A table for the functions described in (4.6)–(4.8)

A third, and much more useful, way to represent a function is in

the form of a graph. First of all, the graph of a function is the set of
all points (x, f (x)) in the two-dimensional plane, where x varies over
all arguments x in the domain of the function. Figures 4.1 and 4.2
show the graphs of f and h. The graph of g cannot be represented so
easily: Since every irrational number x is arbitrarily close to a rational
number x0 (e.g., the decimal representation of x truncated at any fixed
number of digits), we have f (x) = 0 and f (x0 ) = x0 for arguments x
and x0 which can be arbitrarily close together, and that is not really
possible to draw.
At the beginning of the section, we defined a function as map-
ping each argument x in the domain X to exactly one value y in the
76 4. FUNCTIONS

Figure 4.1. The graph of the function f from (4.6)

Figure 4.2. The graph of the function greatest integer

function h from (4.8)

codomain Y . There is a visual way to check this property whether

any given picture (“curve”) in the plane is the graph of a function:
To check whether for any argument x in the domain X, there is only
one value y, we perform the “Vertical Line Test” by drawing a vertical
 4.2. RANGE, ONTO AND 1–1 FUNCTIONS, AND INVERSE 77

line at x and checking how often the curve intersects the vertical line.
If the line and the curve intersect only once (for each argument x),
then the curve “passes the Vertical Line Test” and is the graph of a
function. This is because this unique point of intersection (x, y), say,
tells us that the value of the function f at x is y. See Figure 4.3 for
an example of a curve (described by the equation x = y 2 ) which does
not pass the Vertical Line Test: The line given by x = 1 intersects the
curve at the points (1, 1) and (1, −1). Algebraically, it’s
√ also easy to
see that the curve has two parts, one described by
√y = x (the “upper
half of the curve”), the other described by y = − x (the “lower half of
the curve”); and both halves individually pass the Vertical Line Test,
as they should,
√ since they
√ are the graphs of the two functions given
by y = x and y = − x, with domain the set of nonnegative real
numbers.

Figure 4.3. The Vertical Line Test for the curve x = y 2

with the line x = 1

4.2. Range of a Function, Onto and One-to-One Functions,

and Inverse of a Function
We now discuss some properties of functions, leading up to the
concept of “undoing” or “inverting” a function (i.e., given the value y,
finding the argument x). Of course, this may not always be possible,
as we will see.
78 4. FUNCTIONS

4.2.1. Range of a Function, and Onto Functions. First of

all, let the range of a function f : X → Y be the set of all y in the
codomain Y of f which are of the form f (x) for some argument x in
the domain X of f ; we will denote the range of f by ran(f ). (Similarly,
we will denote the domain of f by dom(f ).)
Sometimes, such as in the example of the function f : R → R, x 7→
2x + 1, the range of f is the same as the codomain: This is because
any y in the codomain R can be written as 2x + 1 by simply setting
x = y−1 2
. On the other hand, the range may consist of only a single
number, such as in the case of the function f : R → R, x 7→ 1: Here
only the number 1 is of the form f (x), and in this case, we can choose x
arbitrarily and will always obtain the value 1. In general, the range is a
subset of the codomain, i.e., by definition of the codomain of a function
f : X → Y , any number of the form f (x) for some x in the domain X
must be in the codomain Y . There are many other possibilities for the
range of a function: E.g., in the examples (4.6)–(4.8), we have that the
range of f is {0, 1}, the range of g is Q, and the range of h is Z. This is
easy to see from Figures 4.1 and 4.2 for the functions f and h. For the
function g, we distinguish two cases for g(x): If x is a rational number,
then so is g(x) (which just equals x), and any rational number y satisfies
y = g(y), so all rational numbers must be in the domain of g. On the
other hand, if x is irrational, then g(x) equals 0, which is rational; so
the range of g is exactly the set of all rational numbers. In general,
determining the range of a function is a very complicated problem!
We call a function f : X → Y “onto” (or onto Y , to be more
precise) if the codomain and the range of the function f are the same.
E.g., as remarked above, the function f : R → R, x 7→ 2x + 1 is onto,
while the function f : R → R, x 7→ 1 is not. None of the functions in
the examples (4.6)–(4.8) is onto as we saw above.

4.2.2. 1–1 Functions. Next, we call a function f : X → Y one-

to-one (or 1–1 for short) if, for each value y in the codomain Y , there
is at most one x in the domain X with y = f (x). We can rephrase this
in several equivalent ways:
(1) The function f is 1–1 if, for each value y in the range of f ,
there is exactly one x in the domain X with y = f (x).
(2) The function f is 1–1 if, for all y in the codomain Y , whenever
we have y = f (x) = f (x0 ) for arguments x and x0 in the
domain X, then actually x = x0 .
(3) A visual way of checking whether a function is 1–1 is the Hor-
izontal Line Test: Draw a horizontal line at the value y and
 4.2. RANGE, ONTO AND 1–1 FUNCTIONS, AND INVERSE 79

check how often the graph of the function intersects the ver-
tical line. If the line and the graph intersect at most once
(for each value y), then the graph “passes the Horizontal Line
Test” and is the graph of a 1–1 function. This is because that
point of intersection (x, y), say, tells us that the value of the
function f equals y at at most one value x. (See Figure 4.4
for the example of the function f : R → R, x 7→ x2 which fails
the Horizontal Line Test, e.g., at the line y = 1, and thus is
not 1–1. If we were to restrict the function f to the domain
[0, ∞) (i.e., the “right half” of the graph), or to the domain
(−∞, 0] (i.e., the “left half” of the graph), then the function
would pass the Horizontal Line Test and thus would be 1–1.)

Figure 4.4. The Horizontal Line Test for the curve y =

x2 with the line y = 1

4.2.3. Inverse of a Function. Next we will investigate when we

can “undo” or “invert” a function f : X → Y in the sense that we
can find a function g : Y → X such that the statements y = f (x) and
x = g(y) are equivalent. If such a function g exists, we will call it the
inverse function of f and denote it by f −1 .3
3Note that the “exponent” −1 here does not denote that we take the “(−1)th
power”, i.e., f −1 (x) does not mean f (x)
1
! Later on, in trigonometry, when we
80 4. FUNCTIONS

What are the conditions now under which the inverse function of a
function f : X → Y exists? Note that we must ensure two things:
(1) For each y in the codomain Y of f , there is an argument x in
the domain X of f such that y = f (x); and
(2) for each y in the codomain Y of f , there is not more than one
argument x in the domain X of f such that y = f (x).
Now observe that condition (1) is exactly the condition that f is onto;
and condition (2) is the condition that f is 1–1! These two conditions
can be combined into one condition:
(3) For each y in the codomain Y of f , there is exactly one argu-
ment x in the domain X of f such that y = f (x).
It should be clear that condition (3) is exactly the condition on the ex-
istence of an inverse function which we are looking for: Condition (1)
is necessary since otherwise, for some y, we will not be able to define
f −1 (y). Condition (2) is necessary as well since otherwise, for some y,
we will not know which value x to choose for f −1 (y). On the other
hand, condition (3) is also sufficient to ensure the existence of an in-
verse function f −1 : Y → X: Given y in Y , condition (3) ensures the
existence of exactly one x in X such that f (x) = y, and so it is both
natural and inevitable to define f −1 (y) = x.
There is also a very nice visual way to find (the graph of) the inverse
function of a function f : Recall that the graph of a function f : X → Y
is the set of all pairs (x, y) such that y = f (x), whereas the graph of
the inverse function f −1 : Y → X is the set of all pairs (y, x) such
that x = f −1 (y), i.e., such that y = f (x). So the graph of f −1 is
obtained from the graph of f by simply interchanging the coordinates
of each point (x, y)! Graphically, this corresponds to reflecting the
graph of f along the “main diagonal” described by y = x; but, of
course, this works only after we rename the variables from x = f −1 (y)
to y = f −1 (x). Figures 4.5 and 4.6 show two examples.
We conclude with another example of our concepts for functions
from this section: Let X = {0, 1, 2} and Y = {1, 2, 3} be two finite
sets of numbers. Let f and g be two functions from X to Y , defined

commonly write expressions like sin2 x in place of (sin x)2 , this notation can lead to
serious ambiguities, since we won’t know how to interpret sin−1 x unless the context
clarifies this, and I always recommend to use the special notation arcsin x for the
inverse of the sine function in place of sin−1 x; but we’ll leave these intricacies aside
for now.
 4.2. RANGE, ONTO AND 1–1 FUNCTIONS, AND INVERSE 81

Figure 4.5. The graphs of the function y = 3x − 3 and

its inverse function y = 31 x + 1

Figure 4.6. The graphs

√ of the function y = x3 and its
inverse function y = 3 x
82 4. FUNCTIONS

by
f (0) = 1 g(0) = 1
f (1) = 3 g(1) = 3
f (2) = 3 g(2) = 2
We can represent f and g visually as shown in Figure 4.7.

f g

0 1 0 1

f g
1 2 1 2

2 3 2 g 3
f

Figure 4.7. The finite functions f and g from X =

{0, 1, 2} to Y = {1, 2, 3}

It is now easy to see that f is neither onto nor 1–1: f is not onto
since 2 is in the codomain but not in the range of f ; and it is not 1–1
since f (1) = f (2). It is therefore impossible to find an inverse function
for f : Given the value 2 in the codomain Y of f , we can’t find any x
in the domain X of f with f (x) = 2; furthermore, for the value 3 in
the codomain Y of f , there are two different x in the domain X of f
with f (x) = 3. So it is not possible to find exactly one value for f −1 (y)
for either y = 2 or y = 3. On the other hand, for each y in Y , there is
exactly one x in X such that g(x) = y, so we can define g −1 : Y → X
with
g −1 (1) = 0
g −1 (2) = 2
g −1 (3) = 1
and g thus has an inverse function.
4.3. Some Functions Closely Related to Linear Functions
So far, we have mainly dealt with one kind of function, linear func-
tions, which can be expressed as f (x) = mx + b for fixed parameters m
 4.3. SOME FUNCTIONS CLOSELY RELATED TO LINEAR FUNCTIONS 83

and b. In this section, we will encounter some functions closely related

to linear functions.
4.3.1. Step Functions. These are functions whose graph is hor-
izontal except at certain points where it “jumps” from one value to
another. We already saw an example of a step function in Figure 4.2;
it is usually denoted by y = bxc, defined by
(4.8) bxc = the largest integer n ≤ x
and called the greatest integer function (sometimes also the “floor func-
tion”). There is a similar function called the least integer function (or
sometimes the “ceiling function”); it is defined by
(4.9) dxe = the smallest integer n ≥ x
Its graph is shown in Figure 4.8.

Figure 4.8. The graph of the least integer function from (4.9)

These two functions are particularly useful to computer scientists,

who use them quite frequently. But they come up in “real life” situa-
tions quite early in school mathematics; in fact, we have already seen
an example which really requires such a function if done carefully:
84 4. FUNCTIONS

“A pay phone charges 50¢ for the first minute and 10¢
for each additional minute of a long-distance call.”
Before, we had restricted the domain of the function describing the
cost of the phone call to the set of positive integers, so we defined this
function as
f : {1, 2, 3, . . . } → R
(4.10)
n 7→ 10n + 40
where n is the length of the phone call (in minutes) and f (n) is its
cost (in cents), and its graph was given in Figure 2.3. But this is really
cheating, since phone calls can be of length any positive real (at least
in principle), and then the phone company rounds up the length of the
phone call to the next integer. So the cost function is really given by
g : (0, ∞) → R
(4.11)
x 7→ 10 · dxe + 40
The graph of the “real” cost function g is given in Figure 4.9.

Figure 4.9. The cost of a phone call from (4.9)

What are the pros and cons of describing the cost function accu-
rately as in (4.11) rather than just as f (x) = 10x + 40 (as in (2.6))?
Certainly, the former is more accurate; it is the actual cost function as
computed by the phone company. On the other hand, f (x) = 10x + 40
is a close approximation, which is much easier to work with: Unlike g,
f is 1–1 and onto and thus has an inverse function; so we can easily
compute the length of a 90-cent phone call, say, using f , or rather f −1 .
If we tried this for an 85-cent phone call, say, we’d still get an “answer”
 4.3. SOME FUNCTIONS CLOSELY RELATED TO LINEAR FUNCTIONS 85

using f −1 , namely 4.5 minutes, and even though the answer would be
wrong (there is no phone call costing exactly 85 cents!), it would tell
us that the phone call would be around 4 or 5 minutes.
So the upshot is that a linear function is sometimes just a useful
approximation to the real function, such as a step function, which is
much harder to deal with. School mathematics is often happy to just
deal with the approximation, but you as teachers should be aware of
this distinction, and the reasons why one uses approximations, since
it may confuse some of your students, and you should then be able to
give at least a “roughly correct” answer.

4.3.2. Piecewise Linear Functions. Another very common kind

of function one encounters in “real life” problems is a “piecewise lin-
ear” function: The graph of such a function is not a line, but consists
of several line segments and is “bent” at the end points of these line
segments, without any “jumps”, however.
Here is an example:
“John swims for half an hour at a constant speed of
2km/h, then jogs for an hour at a constant speed of
8km/h, and finally bikes for half an hour at a constant
speed of 20km/h. Describe the distance traveled by
John in terms of the time elapsed.”
Let’s let t (in hours) be the time elapsed since John starts, and s (in
km) his distance traveled since then. Then the graph of s as a function
of t is as shown in Figure 4.10.
Let’s first describe this graph verbally: John first swims from time
zero hours to time one-half hour at a speed of 2km/h, covering a dis-
tance of one kilometer. Then he jogs for one hour (from time one-half
hour to time one-and-a-half hours) at a speed of 8km/h, covering an
additional eight kilometers, for a total distance of nine kilometers for
now. Finally, he bikes for half an hour (from time one-and-a-half hours
to time two hours) at a speed of 20km/h, covering an additional ten
kilometers, for final total distance of nineteen kilometers.
Let’s translate this into an algebraic description of our function:

2t
 if 0 ≤ t ≤ 12
(4.12) s(t) = 1 + 8(t − 21 ) if 12 ≤ t ≤ 1 12
9 + 20(t − 1 1 )

if 1 21 ≤ t ≤ 2
2

So in the intervals [0, 12 ], [ 12 , 1 21 ], and [1 12 , 2], s behaves like the linear

functions s = 2t, s = 1 + 8(t − 12 ), and s = 9 + 20(t − 1 12 ), respectively,
as can be seen, e.g., from the point-slope formula of a linear function,
86 4. FUNCTIONS

Figure 4.10. The distance s traveled by John terms of

the time t elapsed

and as can be seen also from the graph shown in Figure 4.10. Note that
at the points t = 12 , we can compute s both as 2 · 12 and as 1 + 8( 21 − 12 );
so the function s “bends” but does not “jump” at t = 12 , and we are
allowed to use 0 ≤ t ≤ 21 and 12 ≤ t ≤ 1 21 in the first and second case
of (4.12) even though the cases “overlap” at t = 12 . Similarly, at t = 1 12 ,
we can compute s both as 1 + 8(1 12 − 12 ) and as 9 + 20(1 12 − 1 12 ). The
speeds at the various times corresponds to the slopes of the various line
segments in the graph.

4.3.3. The Absolute Value Function. The absolute value func-

tion is a very important special case of a piecewise linear function. In-
formally, it gives the “distance” of a number x from the number 0 on
the number line. Very informally, the absolute value of x is “x without
the minus sign”. As you probably already know, the absolute value
of x is denoted by |x|. The graph of the absolute value function is
shown in Figure 4.11.
The formal definition of the absolute value of x is given by
(
x if x ≥ 0
(4.13) |x| =
−x if x < 0
 4.3. SOME FUNCTIONS CLOSELY RELATED TO LINEAR FUNCTIONS 87

Figure 4.11. The graph of the absolute value function

y = |x|

This simple and elegant definition is the source of endless confu-

sion: If x is negative, then −x is positive; so the second case of (4.13)
always produces a positive value! Therefore, the absolute value of a
number is always non-negative, one of the most important facts about
the absolute value:
(4.14) |x| ≥ 0 for all x
Before we discuss an example of how the absolute value function is
used, let’s state and discuss three important rules about this function,
which hold for all real numbers a and all non-negative real numbers r:
(4.15) |a| = r exactly when a = r or a = −r
(4.16) |a| < r exactly when − r < a < r
(4.17) |a| > r exactly when a < −r or a > r
Why do properties (4.15)–(4.17) hold true? This is easy to “see
intuitively” from the intuitive definition of |x| as the “distance” of x
from 0: The two numbers exactly at distance r from 0 are r and −r.
(If r = 0, there is only one number here, of course.) The expression
|a| < r states that a is at a distance less than r from 0, which clearly
holds exactly for the numbers in the interval (−r, r). (This interval
is empty if r = 0, as there are no numbers at a distance less than 0
from 0.) Finally, the expression |a| > r covers all the cases not included
in the first two cases, i.e., all numbers in the intervals (−∞, −r) and
(r, ∞).
88 4. FUNCTIONS

A second, more visual way to see why properties (4.15)–(4.17) hold

true is to look at the graph in Figure 4.11: E.g., the expression |a| < r
means that we need to look at all values < r on the y-axis, and the
arguments x for which |x| < r correspond to the points on the line
segments from (−r, r) to the origin, and from the origin to (r, r), not
including the points (−r, r) and (r, r) themselves.
A third, more algebraic way to see why properties (4.15)–(4.17)
hold true is to analyze the statements on the left-hand sides in light of
the formal definition of |x| in (4.13): E.g., |a| = r means that either
a ≥ 0 and so |a| = a, forcing a = r; or that a < 0 and so |a| = −a,
forcing −a = r; this gives exactly the two cases on the right-hand side
of (4.15). Similar arguments can be made for (4.16)–(4.17).
The main use of the absolute value lies in the fact that it does
not only allow us to measure the distance of a number from 0 on the
number line, but indeed the distance on the number line between any
two numbers. Namely, the distance between numbers a and b is given
by
|a − b|
The properties (4.15)–(4.17) now immediately translate to properties
of the distance between numbers a and b (for any non-negative real
number r)
(4.18) |a − b| = r exactly when a = b + r or a = b − r
(4.19) |a − b| < r exactly when b − r < a < b + r
(4.20) |a − b| > r exactly when a < b − r or a > b + r
since the right-hand sides of (4.18)–(4.20) are equivalent to the right-
hand sides of (4.15)–(4.17) when we replace a by a − b.
A very common application of the absolute value is in estimating
the accuracy of a measurement: We already saw that in “real life”
applications, a length ` of 10 meters, say, really means a length of
10m±.5m, i.e., that
(4.21) 9.5 ≤ ` ≤ 10.5
(We actually wrote 9.5 ≤ ` < 10.5 before, but let’s ignore that fine
point for the moment; in applications, it makes little difference.) An-
other way to express (4.21) would be
|` − 10| ≤ .5
as can be seen from (4.18)–(4.19). Note that the negation of this state-
ment, i.e., that the true length ` differs from 10m by more than half a
 4.3. SOME FUNCTIONS CLOSELY RELATED TO LINEAR FUNCTIONS 89

meter, is expressed by
|` − 10| > .5
or equivalently by
` < 9.5 or ` > 10.5

4.3.4. Direct and Inverse Variation. We already saw in Chap-

ter 2 that often, the ratio (i.e., the quotient) between two quantities x
and y, say, is a constant, which we’ll call k:
y
(4.22) k=
x
This can be rewritten as
(4.23) y = kx
In this case, we say that y is proportional to x or that y varies directly
with x. An example of this from section 2.2 was constant speed v as
the ratio of distance s traveled over time t:
s
(4.24) v=
t
At other times, we have that the product between two quantities x
and y, say, is a constant, which we’ll call k:
(4.25) k = yx
This can again be rewritten as
k
(4.26) y=
x
In this case, we say that y is inversely proportional to x or that y varies
inversely with x. The speed equation (4.24) also gives us an example
of this if, instead of constant speed, we assume that the distance is
constant and we want to study how the speed varies as we vary the
time. (E.g., you may consider the travel over a fixed distance and
thus consider the speed as depending on the time it takes you to travel
this distance.) Figures 4.12 and 4.13 show the graphs of the functions
in (4.23) and (4.26) for k = 1.
Sometimes, we want to study the relationship between more than
two quantities, and then the concepts of direct and inverse variation are
particularly useful. The so-called “Combined Gas Law” from physics,
describing properties of gas in a sealed container, provides such an
example:
PV
(4.27) =C
T
90 4. FUNCTIONS

Figure 4.12. The graph of the function y = x

Here, P is the pressure (in pascal, say), V the volume (in m3 , say),
T the absolute temperature (in degrees Kelvin), and C is a constant
(which is fixed as long as we only consider a fixed amount of a particular
kind of gas).
The Combined Gas Law can now be rewritten, and stated as saying
the following:
CT
(1) P = : The pressure varies directly with the temperature
V
(at constant volume), and inversely with the volume (at con-
stant temperature).
CT
(2) V = : The volume varies directly with the temperature
P
(at constant pressure), and inversely with the pressure (at con-
stant temperature).
PV
(3) T = : The temperature varies directly with the pressure
C
(at constant volume), and also directly with the volume (at
constant pressure).
4.3. SOME FUNCTIONS CLOSELY RELATED TO LINEAR FUNCTIONS 91

1
Figure 4.13. The graph of the function y = x
 CHAPTER 5

Quadratic Functions, Equations and Inequalities

5.1. Introduction to Quadratic Functions

So far, the only functions we have considered are functions which
are linear functions, or at least fairly close to linear functions. We will
now turn to quadratic functions.
Let’s start with an example. Consider the sequence of shapes in
Figure 5.1.

n=1 n=2 n=3 n=4

Figure 5.1. A sequence of shapes

There are several ways in which we could think of the (n + 1)st

shape being generated from the nth shape; here are two definitions
leading to the same sequence:
(1) The (n + 1)st shape is generated from the nth shape by adding
a new row of squares below the last row of the nth shape,
extending that last row one extra square to the left and one
extra square to the right.
(2) The (n + 1)st shape is generated from the nth shape by adding
two new squares to the right of each row of the nth shape and
then adding one new square above the middle of the new top
row.
Whichever way you define the sequence of shapes, it is important to
define it precisely: Simply letting the student “guess the pattern” is
highly misleading, since there are many different ways to continue the
“pattern”. E.g., we could have a “pattern” which “stabilizes” at the
4th shape, i.e., the nth shape for all n > 4 is the same as the 4th shape.
92
 5.1. INTRODUCTION TO QUADRATIC FUNCTIONS 93

Or, we could have a “pattern” which “oscillates” between the 1st and
the 4th shape, i.e., the 5th shape is the same as the 3rd, the 6th shape
is the same as the 2nd, the 7th shape is the same as the 1st, the 8th
shape is the same as the 2nd again, etc. There is no reason why one
pattern should be considered more “reasonable” than the others! This
is why a careful definition of the intended pattern is essential.
Now let’s look at the perimeter and the area of the nth shape in
Figure 5.1 as defined above (assuming that the side length of each
square is one unit length, and each square has area one unit square).
Table 5.1 below shows these for the first five shapes.

number of shape 1 2 3 4 5
perimeter of shape 4 10 16 22 28
area of shape 1 4 9 16 25

Table 5.1. Perimeter and area of the shapes from Figure 5.1

We will now see that perimeter and area show a very different pat-
tern:
Table 5.1 suggests that the perimeter increases by a fixed amount
(namely, 6) each time. We can also see this, more precisely and for
general n, from our definitions of the shapes: E.g., from the above
definition (1) for the shapes, we see that each time we add to the
pattern, we move the bottom edge down one unit; and then we add
a square on each side of the new bottom row, resulting in an increase
in perimeter by six edges of these two squares. We can also see this
from the above definition (2) for the shapes: Each time we add to the
pattern, we move all the edges on the right over by two units; and then
we add four edges at the top (including for the new top square) and
two edges at the bottom, resulting in an increase in perimeter by six
edges.
Therefore, the perimeter (let’s call it P (n)) of the nth shape is a
linear function of n, with slope 6 and the graph containing the point
(1, 4) (since the 1st shape has perimeter 4). By the point-slope form
for a line equation(see (2.11)), we arrive at the formula
(5.1) P (n) = 6(n − 1) + 4 = 6n − 2
Here is another way of looking at finding the expression for this
function: Let’s define a new function (called the first difference func-
tion) by
(5.2) P 1 (n) = P (n + 1) − P (n)
94 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

So this function P 1 tells us not the function P itself but by how much
the function P increases from n to n + 1. As we showed above, this
increase is constant, namely, we have
(5.3) P 1 (n) = 6
for all whole numbers n > 0. Note that (5.3) (together with (5.2)) does
determine the function P ; e.g., the function P 0 (n) = 6n also satisfies
both (5.3) and (5.2). But if we also specify that
(5.4) P (1) = 4
then we can compute the function P at all whole numbers n > 0
by starting with n = 1 in (5.4) and then repeatedly using that (5.3)
and (5.2) give us
P (n + 1) = P (n) + P 1 (n) = P (n) + 6
to compute P (2), P (3), etc.1
We note that any function P (defined on the positive whole num-
bers) with constant first difference function P 1 must be linear: Suppose
that P 1 (n) = c for all n > 0. Then we can compute
P (2) = P (1) + c
P (3) = P (2) + c = P (1) + 2c
P (4) = P (3) + c = · · · = P (1) + 3c
...
P (n) = P (n − 1) + c = · · · = P (1) + (n − 1)c
...

On the other hand, the increase in area from one shape to the next
is not constant: The increase from the 1st to the 2nd shape is 3, from
the 2nd to the 3rd is 5, from the 3rd to the 4th is 7, etc. So, instead of
the increase being constant, it appears from Table 5.1 that the increase
increases by a constant, namely, 2. We can also see this more precisely
from our definitions of the shapes: E.g., from the above definition (1)
for the shapes, we see that each time we add to the pattern, we add
two more squares to the shape than we did the last time (since the new
bottom row is two units wider than the previous bottom row). And we
can see this from Figure 5.2: The squares marked “new” (in regular
font) are the same number as was added at the previous step; the two
squares marked “new” (in slanted font) are added in addition to the
1Such a definition of a function is sometimes called a definition by recursion.
 5.1. INTRODUCTION TO QUADRATIC FUNCTIONS 95

other new squares; so we do indeed add two more squares at each step
than we did at the previous step.

new

new new new

new new new new new

new new new new new new

n=1 n=2 n=3 n=4

Figure 5.2. Adding new squares in our sequence of shapes

This results in the following equation for the first difference func-
tion A1 of the area A:

(5.5) A1 (n) = A(n + 1) − A(n) = 2n + 1

So, by the argument on the previous page, the first difference function
A1 (n) = A(n + 1) − A(n) is a linear function, since it increases by 2
each time; namely, the so-called second difference function

A2 (n) = A1 (n + 1) − A1 (n)

is constant, namely, 2.
Whenever the second difference function is constant, the first dif-
ference function must be linear, as we saw before; and whenever the
first difference function is linear, then the function is quadratic, as we
will now show.2
First of all, a quadratic function is a function of the form

(5.6) A(n) = an2 + bn + c

2If you happen to know some calculus, note the analogy: We consider here so-
called discrete functions, i.e., functions defined on the positive integers, say, in an
area called discrete mathematics. For such functions, the first difference function
is the analogue of the first derivative of a continuous function, and the second
difference function is the analogue of a continuous function. What we just said was
that a discrete function is linear exactly when its first difference function is constant;
and that a discrete function is quadratic exactly when its second difference function
is constant. Of course, the derivatives do not exist for all continuous functions; but
the difference functions of discrete functions always exist.
96 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

for parameters a, b and c. Now note that (5.6) satisfies

A1 (n) = A(n + 1) − A(n)
= (a(n + 1)2 + b(n + 1) + c) − (an2 + bn + c)
(5.7) = (a(n + 1)2 − an2 ) + (b(n + 1) − bn) + (c − c)
= (an2 + 2an + a − an2 ) + (bn + b − bn)
= (2a)n + (a + b)
and so A1 (n) is a linear function of n. But we have
A(n + 1) = A(n) + (A(n + 1) − A(n)) = A(n) + A1 (n)
and so having A(1) and a formula for A1 (n) completely determines the
function A.
Comparing (5.5) and (5.7), we now have
2n + 1 = (2a)n + (a + b)
and so
2 = 2a
1=a+b
which solves to
a=1
b=0
Now we still have to find c for (5.6). Setting n = 1, we also have
A(1) = 1 = 12 + c
and therefore c = 0. Thus the area of the nth shape in Figure 5.1 is
(5.8) A(n) = n2
Of course, you probably “guessed” the formula for the area all along
from Table 5.1. However, I hope you found the “formal” way of deriving
the formula above instructive, since it works for any sequence of values
with constant second difference; we’ll work out a second example below
to show how this works in a more complicated case. But let’s first prove
the formula (5.8) for the area of the nth shape geometrically: Figure 5.3
shows how to move some of the “old” squares to “new” places in such
a way that the nth shape becomes a square of side length n and thus
of area n2 .
Now, as promised, let’s slightly vary the problem: Consider the
shapes in Figure 5.4. It is now no longer so easy to guess a formula for
 5.1. INTRODUCTION TO QUADRATIC FUNCTIONS 97

new new new

new new old new new

new old new old old new

old old old old old old

n=1 n=2 n=3 n=4

Figure 5.3. Computing the area of the nth shape in

our sequence of shapes from Figure 5.1

n=1 n=2 n=3 n=4

Figure 5.4. A new sequence of shapes

the area of the nth shape. But it’s not hard to see that the areas of
these shapes satisfy
(5.9) A(n + 1) − A(n) = n + 1
and so the second difference is constant 1. Comparing (5.9) with (5.7)
above, we arrive at
n + 1 = (2a)n + (a + b)
and so
1 = 2a
1=a+b
which solves to
1
a=
2
1
b=
2
Now we still need to find c. Setting n = 1 gives
A(1) = 1 = 1
2
· 12 + 21 + c
98 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

and therefore again c = 0. Thus the area of the nth shape in Figure 5.1
is
(5.10) A(n) = 12 n2 + 21 n
Exercise 5.1. Find a formula for the perimeter of the shapes in
Figure 5.4, and prove both it and the area formula (5.10) geometrically
in at least two different ways.
5.2. Solving Quadratic Equations
We call an equation a quadratic equation (in an unknown x) if it is
of the form
(5.11) ax2 + bx + c = 0
for some parameters a, b, c in R. We will also assume from now on that
a 6= 0 since otherwise (5.11) reduces to a linear equation. Note that a
more general equation
ax2 + bx + c = a0 x2 + b0 x + c0
can easily be transformed into the format (5.11) by “subtracting the
right-hand side to the left”; so we can restrict ourselves in general to
equations of the form (5.11).
We will now consider increasingly sophisticated (and more compli-
cated) techniques to solve a quadratic equation, starting with a review
of the binomial laws, which will prove very useful.
5.2.1. Reviewing the Binomial Laws. The binomial laws (for
algebraic expressions a, b and c) state that
(5.12) (a + b)2 = a2 + 2ab + b2
(5.13) (a − b)2 = a2 − 2ab + b2
(5.14) (a + b)(a − b) = a2 − b2
and are simple but very useful consequences of the laws of arithmetic
from Chapter 1.
We will prove them in two different ways, algebraically from the
laws of arithmetic, and geometrically using models for addition and
multiplication.
Let’s start with an algebraic proof of (5.12):
(a + b)2 = (a + b)(a + b) (by definition of squaring)
= a(a + b) + b(a + b) (by distributive law)
= (a2 + ab) + (ba + b2 ) (by distributive law again)
= a2 + 2ab + b2 (by comm./assoc./distrib. laws)
 5.2. SOLVING QUADRATIC EQUATIONS 99

For a geometric proof of (5.12) (for a, b > 0), refer to Figure 5.5.
The area of the big square, which has side length a+b, is (a+b)2 , and is
the sum of the areas of the two smaller squares and the two rectangles,
which have areas a2 , b2 , ba and ab, respectively. The binomial law (5.12)
now follows immediately.

b ba b2

a a2 ab

a b

Figure 5.5. A geometric proof (for a, b > 0) showing

that (a + b)2 = a2 + 2ab + b2

Exercise 5.2. Give similar algebraic and geometric proofs of the

binomials laws (5.13)–(5.14). (For the geometric proofs, assume that
0 < b < a to make geometric sense. Explain why this assumption is
desirable for the geometric proofs.)
5.2.2. Solving Quadratic Equations by Factoring. Let’s start
with an example:
(5.15) x2 + 2x − 3 = 0
Since x2 + 2x − 3 = (x − 1)(x + 3) (by using the distributive property
repeatedly!), this can be rewritten as
(5.16) (x − 1)(x + 3) = 0
Of course, it’s not always so easy to see this immediately, and in fact
it is often impossible to “see” such a “factoring” of the left-hand side
of a quadratic equation at all. There is one useful fact from advanced
algebra3 which often comes in handy: Suppose we have an equation of
3Here I am using the word “algebra” not in the sense of school algebra, but in
the sense in which research mathematicians use it. The proof of this fact is typically
covered in an advanced undergraduate course on “Modern Algebra”.
100 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

the form x2 + bx + c = 0 (i.e., the parameter a equals 1, or, put differ-

ently, x2 is not preceded by a number) and the parameters b and c are
integers, then all the solutions are (positive or negative) integers which
divide the parameter c, or else necessarily involve square roots. So in
the above example (5.15), we know right away, without even factoring,
that the solutions to (5.15) either involve square roots or are among
the numbers 1, −1, 3, −3 (which are the only integers dividing 3).
Now we use a simple fact about the real numbers: Multiplying two
nonzero numbers always produces a nonzero number:
(5.17) A 6= 0 and B 6= 0 implies A · B 6= 0
Equivalently, if a product of two numbers is zero, then one of the factors
has to be zero:
(5.18) A · B = 0 implies A = 0 or B = 0
Both (5.17) and (5.18) hold for all numbers A and B. And since the
converse of these is trivial (a product where one factor is zero must be
zero), we actually have
(5.19) A · B = 0 holds exactly when A = 0 or B = 0
Now let’s apply (5.19) to (5.16):
(x + 1)(x − 3) = 0 holds exactly when x + 1 = 0 or x − 3 = 0
which shows that the equation (5.15) has exactly the two solutions −1
and 3.
More generally, and again using (5.19), the solution(s) to a qua-
dratic equation of the form
(x + A)(x + B) = 0
are −A and −B. (Of course, if A = B then there is only one solution
here.)
Let’s talk about another very simple way to solve quadratic equa-
tions of another special form next.
5.2.3. Solving Quadratic Equations by Taking the Square
Root. Let’s again start with an example: Consider the quadratic equa-
tion
x2 + 2x = 3
Since x2 + 2x + 1 = (x + 1)2 , this can be rewritten as
(x + 1)2 = 4
after adding 1 to both sides. (Of course, it’s not always so easy to
see this immediately, but we will learn a technique in subsection 5.2.4
which shows that this can always be done.) The point of this rewriting
 5.2. SOLVING QUADRATIC EQUATIONS 101

is that now we can “take the square root” on both sides and we arrive
at
x+1 = ±2 (which is an abbreviation for “x + 1 = 2 or x + 1 = −2”)
(Don’t forget the ± here! More about that in a minute.) Subtracting 1
on both sides then yields
x = −1 ± 2 (which is an abbreviation for “x = 1 or x = −3”)
Clearly, “taking the square root” here did the trick, but why was
that justified? This step, and the ± part of it, is a frequent source of
student errors, so let’s look at it more closely: What we really did was
to take the following steps, each resulting in an equivalent statement:
(5.20) (x + 1)2 = 4
(5.21) |x + 1|2 = |2|2
(5.22) |x + 1| = |2|
(5.23) x + 1 = ±2 (i.e., x + 1 = 2 or x + 1 = −2)
Let’s analyze each step carefully: Going from (5.20) to (5.21) uses
that 4 = 22 , that 2 = |2|, and the identity
A2 = |A|2
which holds for all algebraic expressions A and follows immediately
from the definition (4.13) of the absolute value. Going from (5.21)
to (5.22) is a new operation on equations for us: If both sides of an
equation A = B (for algebraic expressions A and B) are non-negative,
then the equations
A=B and A2 = B 2
have the same solutions. Finally, the step from (5.22) to (5.23) follows
immediately from (4.15). Since this sequence of three steps is used
quite frequently, we state it as a proposition:
Proposition 5.3. The solutions to the equation
A2 = B 2
(for algebraic expressions A and B) are the same as the solutions to
the statement
A = B or A = −B
The proof of Proposition 5.3 is the same the proof for the special
case above: The following four statements are equivalent, by the same
102 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

reasoning as for (5.20)–(5.23):

A2 = B 2
|A|2 = |B|2
|A| = |B|
A = B or A = −B
Proposition 5.3 now gives us a general procedure to solve quadratic
equations of the form (x + A)2 = B (for any numbers A and B).
But there is a small wrinkle we haven’t considered yet: What if B
is negative? Since the square (x + A)2 is always non-negative, there
cannot be any solution (at least in the real numbers4) in this case!
So we’ll assume in addition that B ≥ 0. Then the solution(s) to
the quadratic equation (x + A)2 = B are
√ √
x = B − A and x = − B − A
Of course, there will be two different solutions if B is positive, and only
one solution (namely, −A) if B = 0. This shows a particularly simple
way to solve quadratic equations, and we’ll see in subsection 5.2.4 that
we can always reduce the solution of a quadratic equation to this case.
5.2.4. Solving Quadratic Equations by Completing the
Square. In subsection 5.2.2, we saw that sometimes a quadratic equa-
tion x2 + bx + c = 0 can be solved very efficiently, namely, whenever
we “see” how to factor the “polynomial” x2 + bx + c as a product
(x + A)(x + B) for some numbers A and B. But if the solutions to
a quadratic equation are too complicated (e.g., if they involve square
roots), then we need another way to solve the equation, which will al-
ways lead to a solution. Subsection 5.2.3 already gave a hint as to how
we might proceed.
First of all, we can always reduce solving an arbitrary quadratic
equation to an equation of the form5
(5.24) x2 + px + q = 0 (e.g., x2 + 2x − 3 = 0)
simply by dividing the original quadratic equation by the number “be-
fore” x2 (which is usually called the coefficient of x2 ).
Next, we can rewrite (5.24) as
(5.25) x2 + px = −q (e.g., x2 + 2x = −(−3) = 3)
4In
subsection 5.3.2, we’ll briefly cover the complex numbers, in which we can
solve the equation (x + A)2 = B even when B is negative.
5We will always accompany our general considerations in this subsection by an
example to make things clearer.
 5.2. SOLVING QUADRATIC EQUATIONS 103

by subtracting q on both sides.

Now we can add 14 p2 to both sides of (5.25) to arrive at
(5.26) x2 + px + 14 p2 = 41 p2 − q (e.g., x2 + 2x + 1 = 1 + 3 = 4)
and we are in the case of subsection 5.2.3, since we can now rewrite
(5.26) as

2
(5.27) x + 12 p = 14 p2 − q (e.g., (x + 1)2 = 4)
(See Figure 5.6 for a geometric picture of how completing the square
works for the left-hand side of the equation: The area corresponding
to px is split in half; each half is put along a side of a square of area x2 .
Now, in order to get a square again, we need to add a square of area 14 p2
in the corner, to arrive at a square of area (x + 21 p)2 . Of course, this
picture assumes p > 0; but a similar picture can be drawn for the case
p < 0.)

p px
½p ½px
½px ¼p2

x x2 x ½px
½px

x x ½p

Figure 5.6. Completing the square geometrically

As in subsection 5.2.3, we can now solve (5.24) as

q
(5.28) 1
x + 2 p = ± 14 p2 − q (e.g., x + 1 = ±2)
and so
q
(5.29) x= − 21 p ± 1 2
4
p −q (e.g., x = −1 ± 2)
The main “trick” in this way of solving the quadratic equation is
in going from (5.25) to (5.26); this step is often calling completing the
104 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

square since we are adding just the right number in order to make the
left-hand side into an expression which can be written as a square, as
shown geometrically in Figure 5.6.

5.2.5. Solving Quadratic Equations by the Quadratic For-

mula. The format of (5.29) above already suggests that one could
solve a general quadratic equation (with parameters in place of num-
bers) by simply finding a general formula which gives the solution(s)
to the equation. Such a formula does indeed exist, and we will now
deduce it in its most general form. ((5.29) is the formula for the special
form of the quadratic equation in (5.24), i.e., when the coefficient of x2
is 1.)
We start with the general quadratic equation6
(5.30) ax2 + bx + c = 0 (e.g., 2x2 + 7x + 5 = 0)
(and still assume that a 6= 0).
First, we can rewrite (5.30) as
(5.31) ax2 + bx = −c (e.g., 2x2 + 7x = −5)
by subtracting c on both sides, and then as
b c
(5.32) x2 + x = − (e.g., x2 + 72 x = − 25 )
a a
by dividing by a on both sides.
  2
b
Now we can add to both sides of (5.32) to arrive at
2a
 2  2
2 b b b c b2 − 4ac
(5.33) x + 2 x + = − =
2a 2a 2a a 4a2
2 2
(e.g., x2 + 72 x + 74 = 74 − 52 = 49−40 9



16
= 16
)
and we are in the case of subsection 5.2.3, since we can now rewrite
(5.33) as
2
b2 − 4ac

b 7 2

9
(5.34) x+ = 2
(e.g., x + 4
= 16 )
2a 4a
As in subsection 5.2.3, we can now solve (5.30) as
√
b b2 − 4ac
(5.35) x+ =± (e.g., x + 47 = ± 34 )
2a 2a
6Again,
we will accompany our general considerations in this subsection by an
example to make things clearer.
 5.2. SOLVING QUADRATIC EQUATIONS 105

and so arrive at the quadratic formula

√
b b2 − 4ac
(5.36) x = − ±
2a 2a
(e.g., x = − 74 ± 43 , i.e., x = −1 or x = − 52 )
for the solution to the general quadratic equation
(5.30) ax2 + bx + c = 0
We see from the quadratic formula (5.36) that the quantity b2 − 4ac
is crucial; it is called the discriminant of the quadratic equation and
determines the number of solutions (in the real numbers): If the dis-
criminant is positive, then (5.36) gives two solutions. If the discrimi-
b
nant is 0, then (5.36) gives only one solution (namely, − 2a ). Finally, if
the discriminant is negative, then (5.36) gives no solution, which makes
sense if we look back at (5.34): If the discriminant is negative, then
the right-hand side of (5.34) is negative (since the denominator of the
right-hand side is a square and thus positive), whereas the left-hand
side of (5.34) is a square and thus non-negative.
To summarize, there are different ways to solve a quadratic equa-
tion, and there is not one “best” way. While the technique of com-
pleting the square and the quadratic formula will always work, the
techniques we discussed earlier are sometimes more efficient. Further-
more, the quadratic formula is rather complicated, and it’s easy to
memorize it wrong; so at the very least, one should remember along
with the quadratic formula also how to derive it so that one can quickly
reconstruct it.

5.2.6. The number of real solutions of a quadratic equa-

tion. We already saw in the last two subsections that a quadratic
equation
(5.30) ax2 + bx + c = 0
has at most two solutions in the real numbers; and we will see this
again graphically in section 5.4. Algebraically, this fact depended on
Proposition 5.3, which states that for all algebraic expressions A and B,
the equation A2 = B 2 holds exactly when A = B or A = −B holds.
There is another proof of the fact that any quadratic equation can
have at most two real solutions, which uses factoring along the lines of
subsection 5.2.2. Let’s start with an example: Consider the equation
(5.37) x2 + 3x + 1 = 0
106 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

From the √ quadratic formula, we can see that one solution of (5.37)
3 1
is − 2 + 2 5. (Of course, the “other” solution from the formula is
√
− 32 − 12 5, but let’s ignore that for now.) We will now “divide” the
√
expression x2 +3x+1 by the expression x+ 23 − 12 5 (i.e., the expression
“x − solution”, for a reason we’ll see later), similarly to the way long
division works for whole numbers, except that now the “digits” of our
expressions are not numbers between 0 and 9 (as in usual long division)
but expressions of the form ax2 , bx or c where a, b and c are numbers.
This procedure is called polynomial division and works in our example
as shown in Table 5.2.
√
x + ( 23 + 1
5)
√ 2
x + ( 23 − 1
5) x2 + 3x + 1
2
3 1
√
− (x2 + ( 2 + 2 5)x)
√
( 32 − 21 5)x + 1
√
− (( 32 − 21 5)x + 1)
0

3 1
√
Table 5.2. Dividing (x2 + 3x + 1) ÷ (x + 2
− 2
5) by
polynomial division

Let’s review this process of polynomial

√ division carefully: We want
to divide x2 + 3x + 1 by x + ( 32 − 12 5). Just as in long division, we
do it “digit” by “digit”, except that instead of “digits”, we now use
the coefficients: In x2 + 3x + 1 , the coefficient 1 of x2 corresponds to
the “hundreds digit”, the coefficient 3 of x corresponds to the “tens
digit”, and the√ constant coefficient 1 corresponds to the “ones digit”;
3 1
in x + ( 2 + 2 5), the coefficient 1 of x corresponds to the “tens digit”,
√
and the constant coefficient 32 + 12 5 corresponds to the “ones digit”.
We first check “how often x goes√into” x2 and find that it “goes x
times”. So we multiply x + ( 23 + 21 5) by x and subtract the result
√
from x2 + 3x + 1, leaving us with ( 32 − 12 5)x. We now “take down” the
√
coefficient 1 and check again “how often x goes into” ( 23 − 12 5)x; we
√ √ √
find it “goes ( 32 − 12 5) times”, multiply x + ( 23 + 12 5) by 23 − 12 5, and
√
subtract the result from ( 23 − 12 5)x + 1, leaving us with remainder 0.
√
So x2 + 3x + 1 divides evenly into x + ( 32 + 12 5) (i.e., there is no
√
remainder), and the quotient is x + ( 32 − 21 5). We can write this also
 5.2. SOLVING QUADRATIC EQUATIONS 107

as
x2 + 3x + 1 3 1√
 
√
=x+ − 5 or
x + 32 + 12 5 2 2
3 1√ 3 1√
     
2
x + 3x + 1 = x + − 5 x+ + 5
2 2 2 2
So why all this work? We can now use (5.18): The solutions x to
the equation
(5.37) x2 + 3x + 1 = 0
are exactly the solutions to the statement
3 1√ 3 1√
   
x+ + 5 = 0 or x + − 5 =0
2 2 2 2
√
Thus equation (5.37) has exactly two solutions, namely, − 32 − 12 5 and
√
− 23 + 12 5.
So what have we accomplished? We started with a quadratic equa-
tion and one solution for it. We then used polynomial division and
computed another solution, and we also saw that these two solutions
are the only ones!
The point of all this is that this can be done in general, leading us
not only to a way of finding one solution of a quadratic equation from
another, but, more importantly, a proof of the following
Proposition 5.4. Any quadratic equation ax2 + bx + c = 0 (with
a 6= 0) has at most two real solutions. (In fact, it has at most two
complex solutions.)
The parenthetical statement above can easily be seen once we have
covered the complex numbers; so let’s concentrate on the case for the
real numbers: First of all, we may assume (by dividing both sides by a)
that a = 1, so let’s only consider a quadratic equation of the form
(5.38) x2 + px + q = 0
Furthermore, if this equation has no real solution at all, then we’re
done. So let’s fix one real solution and call it x0 . We will now attempt
polynomial division of x2 + px + q by x − x0 and see what happens, as
shown in Table 5.3.
Let’s go over this very abstract polynomial division in detail: We
want to divide x2 + px + q by x − x0 . Again, just as in long division,
we do so “digit” by “digit”, except that instead of “digits”, we now
use the coefficients: In x2 + px + q , the coefficient 1 of x2 corresponds
to the “hundreds digit”, the coefficient p of x corresponds to the “tens
108 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

x + (p + x0 )
2
x − x0 x + px + q
− (x2 − x0 x)
(p + x0 )x + q
− ((p + x0 )x − (p + x0 )x0 )
(p + x0 )x0 + q

Table 5.3. Dividing (x2 + px + q) ÷ (x − x0 ) by poly-

nomial division

digit”, and the constant coefficient q corresponds to the “ones digit”;

in x − x0 , the coefficient 1 of x corresponds to the “tens digit”, and the
constant coefficient −x0 corresponds to the “ones digit”.
We first check “how often x goes into” x2 and find that it “goes x
times”. So we multiply x − x0 by x and subtract the result from
x2 + px + q, leaving us with (p + x0 )x. We now “take down” the
coefficient q and check again “how often x goes into” (p + x0 )x; we find
it “goes (p + x0 ) times”, multiply x − x0 by p + x0 , and subtract the
result from (p + x0 )x + q, leaving us with remainder (p + x0 )x0 + q. So
x2 + px + q divides into x − x0 with remainder (p + x0 )x0 + q, and the
quotient is x + (p + x0 ). We can write this also as
x2 + px + q (p + x0 )x0 + q
(5.39) = (x + (p + x0 )) + or
x − x0 x − x0
(5.40) x2 + px + q = (x + (p + x0 ))(x − x0 ) + ((p + x0 )x0 + q)
Note that (5.40) is obtained from (5.39) simply by multiplying by
x − x0 on both sides. Now let’s use the fact that x0 is a solution of
x2 + px + q = 0: Set x = x0 . Then the left-hand side of (5.40) is equal
to 0. But the left half of the right-hand side of (5.40) is also equal to 0
for x = x0 ; so this forces the remainder (p + x0 )x0 + q to be 0 as well,
i.e., x2 + px + q divides into x − x0 evenly without remainder! This
simplifies (5.40) to
x2 + px + q = (x + (p + x0 ))(x − x0 )
and so, by (5.18), the solutions of x2 + px + q = 0 are the solutions to
the statement
x + (p + x0 ) = 0 or x − x0 = 0
i.e., the solutions are x0 and −(p + x0 ). So there are at most two
solutions as claimed. As a by-product, we have also found a formula
 5.3. SQUARE ROOTS AND THE COMPLEX NUMBERS 109

which allows us to compute one solution of a quadratic equation from

the other. (Note that x0 may equal −(p + x0 ), so there may be only
one single solution.)

5.3. A Digression into Square Roots

and the Complex Numbers
5.3.1. Square Roots. As you can see from the quadratic formula,
solving quadratic equations can necessarily involve square roots, so let’s
briefly review them first.
By definition, a square root of a real number a is a solution to the
equation
(5.41) x2 = a
If a = 0, then x = 0 is a unique solution to this equation. If a is
positive, then√
there are two solutions, namely,
√ a positive number which
we denote √by a, and its additive inverse, − a. (The existence of such
a number a is actually a very deep fact about the real numbers which
we cannot prove here!) In order to make the square root unique for
2
all a ≥ 0, we define the non-negative solution to√ the equation x = a
to be “the” square root of a and denote it by a. (Note that this is
somewhat arbitrary, it is simply a definition of what we mean by “the”
square root of a non-negative number. We’ll get to what a square root
of a negative real number could be later in subsection 5.3.2 when we
discuss the complex numbers.)
Let’s now state some rules about square roots. The simplest ones
are about multiplication and division with square roots: For all non-
negative real numbers a and b, we have
p p p
(5.42) a· b= a·b
p r
a a
(5.43) p =
b b
(where we assume b 6= 0 for (5.43)). These rules follow immediately
from the definition of square roots: Let
p
a = the (unique non-negative) solution x to the equation x2 = a
p
b = the (unique non-negative) solution y to the equation y 2 = b
p
ab = the (unique non-negative) solution z to the equation z 2 = ab
r
a a
= the (unique non-negative) solution w to the equation w2 =
b b
110 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

Then (5.42) simply states that

(5.44) xy = z
and (5.42) states that
x
(5.45) =w
y
Since all the numbers involved are non-negative, we can apply Propo-
sition 5.3: Squaring equations (5.44) and (5.45), we find that they are
equivalent to
 2
2 2 2 x2 x
x y = (xy) and 2 =
y y
respectively, both of which are trivially true.
There is one other useful rule about square roots, which connects
them to absolute value: For any real number a (i.e., also for a negative
real number a), we have
√
(5.46) a2 = |a|
This rather surprising (and at first unexpected) fact can be shown
directly from the definition of square roots and absolute value, distin-
√ the two cases in the definition of absolute value: If a ≥ 0,
guishing
then a2 is the unique non-negative solution x to the equation x2 = a2
(in the unknown x, with parameter a); but x = a is a non-negative so-
lution to this equation, and since the non-negative solution is unique,
√
we must have x = a = |a|. On the other hand, if a < 0, then again a2
is the unique non-negative solution x to the equation x2 = a2 (in the
unknown x, with parameter a!); but x = −a is a non-negative solution
to this equation (since a2 = (−a)2 ), and since the non-negative solution
is unique, we must have x = −a = |a|.
We’ll conclude this subsection with two “non-rules”, which fre-
quently creep into school mathematics. They are the following two
statements:
p p p
The statement a + b = a + b is generally false!
p p p
The statement a − b = a − b is generally false!
Once you realize the above formulas are false, it’s very easy to see why:
Just plug in almost any numbers and check your answer. Remember:
Showing that a “general rule” fails is much easier than showing that
it holds: For a general rule to hold, you must check it in “all” cases,
usually by some “mathematical proof”; in order to show that it fails,
you need just one counterexample!
 5.3. SQUARE ROOTS AND THE COMPLEX NUMBERS 111

5.3.2. The Number i and the Complex Numbers. We now

return to the question of how one might define the square root of a
negative real number a. Our first attempt would be to define it as a
solution x to the equation x2 = a. But there is an obvious problem:
for any real number x, x2 is non-negative and thus cannot equal a. So
there cannot be a real square root of a. But we can extend our number
system again.
Think back to the time when you first learned about the negative
numbers. Back then, it was impossible to solve an equation like x+3 =
2 since that would be the solution to a (non-sensical) problem like
“Mary receives three apples and now has two apples.
How many apples did she have before?”
In this context, introducing negative numbers makes no sense: Mary
cannot have had −1 apples before! But if we expand our model of what
numbers mean to notions like temperature, the equation x + 3 = 2
suddenly makes sense:
The temperature rose three degrees and is now at two
degrees. What was the temperature before?”
We are in a similar situation with solving an equation like x2 = −1.
If we insist on staying within the real numbers, there is no solution.
The difficulty is that there is no simple and visual model for the com-
plex numbers. The usefulness of the complex numbers can be seen re-
ally only in higher mathematics (after most of calculus, in upper-level
undergraduate mathematics and beyond) and in fairly sophisticated
science and engineering applications. Thus it is far from obvious why
one would want to introduce the complex numbers, except for the fact
that it enables one to solve some equations.
In order to introduce the complex numbers, let’s start with the sim-
plest quadratic equation which cannot be solved in the real numbers,
namely, x2 = −1. We introduce a new number i, called the imagi-
nary unit, and define it to be a solution of the equation x2 = −1, i.e.,
we define that i2 = −1. Assuming that all the rules of arithmetic we
have for the real numbers also hold for the complex numbers, we see
that (−i)2 = i2 = −1, and so the equation x2 = −1 has two solu-
tions, i and −i. Furthermore, we can factor the expression x2 + 1 as
(x + i)(x − i).
Let’s now move on to solving a slightly √ more general equation,
2
namely, x = −a for positive a. Set b = a. Since we assumed all
the rules of arithmetic for the complex numbers, we have
(bi)2 = b2 i2 = a(−1) = −a
112 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

and
(−bi)2 = (−b)2 i2 = a(−1) = −a
Thus the equation x2 = −a (for positive a) has the two solutions bi
and −bi. While one does not in general define one of bi or −bi to be
“the” square root of −a, both bi and −bi serve as a good substitute
√ for
the square root of −a. Whenever we write expressions √ like ± −a, it
doesn’t matter anyhow which of bi or −bi we use for −a.
Finally, we’ll define a complex number to be any number of the
form a + bi, where a and b are arbitrary real numbers. Addition,
subtraction and multiplication are defined on the complex numbers
just as expected:
(5.47) (a + bi) + (c + di) = (a + c) + (b + d)i
(5.48) (a + bi) − (c + di) = (a − c) + (b − d)i
(5.49) (a + bi)(c + di) = (ac − bd) + (ad + bc)i
Here, the formula (5.49) for multiplication just uses repeated applica-
tion of the distributive law for the left-hand side, plus the fact that
i2 = −1. Division is somewhat more complicated since one has make
the denominator a real number; this involves a slight trick and some
messy computations:
a + bi (a + bi)(c − di)
=
c + di (c + di)(c − di)
(5.50) (ac + bd) + (bc − ad)i
=
c2 + d 2
ac + bd bc − ad
= 2 + 2 i
c + d2 c + d2
Note that the last line of (5.50) is well-defined as long as c+di 6= 0, i.e.,
as long as at least one of c and d is nonzero, making c2 + d2 nonzero.
So, for c + di 6= 0, the formula (5.50) gives us again an expression of
the form A + Bi for real numbers A and B, i.e., a complex number in
our sense. (It is also possible to solve quadratic equations with complex
parameters a, b and c, but that would require the definition of square
roots of complex numbers, which is beyond the scope of these notes.)
We now return to the quadratic formula
√
b b2 − 4ac
(5.36) x=− ±
2a 2a
for the solution to the general quadratic equation
(5.30) ax2 + bx + c = 0
 5.4. GRAPHING QUADRATIC FUNCTIONS 113

(where a, b and c are now arbitrary real numbers with a 6= 0). The
quadratic formula then gives us one or two complex solutions in all
cases; namely, one solution if b2 − 4ac = 0, and two solutions otherwise.
We conclude with an example: Consider the quadratic equation
(5.51) x2 + 2x + 2 = 0
Here, a = 1 and b = c = 2. So the solutions to (5.51) are
√
2 22 − 8
x=− ±
2 √ 2
−4
= −1 ±
2
2i
= −1 ±
2
= −1 ± i
We can check that these two solutions really work:
(−1 + i)2 + 2(−1 + i) + 2 = (1 − 2i − 1) + (−2 + 2i) + 2
= (1 − 1 − 2 + 2) + (−2 + 2)i = 0
and
(−1 − i)2 + 2(−1 − i) + 2 = (1 + 2i − 1) + (−2 − 2i) + 2
= (1 − 1 − 2 + 2) + (2 − 2)i = 0

5.4. Graphing Quadratic Functions and

Solving Quadratic Equations Graphically
Solving a quadratic equation ax2 + bx + c = 0 corresponds, of
course, to “undoing” the function f (x) = ax2 + bx + c, i.e., to finding
all arguments x at which the function f takes the value 0. (We had
seen this before in section 2.5 with solving a linear equations ax + b =
0, which corresponded to finding all x at which the function g(x) =
ax + b takes the value 0.) As in the case of linear equations, one can
always find the solutions to quadratic equations by graphing (at least
approximately); but the graphs of quadratic functions are naturally a
bit more complicated.
Let’s start with an example from physics:
“An arrow is shot straight up from the ground with
an initial velocity of v0 = 50m/sec. If the height of
the arrow (in meters) at time t (after being fired, in
seconds) is described by
1
(5.52) s = v0 t − gt2
2
114 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

(neglecting friction, where g ≈ 9.8m/sec2 is the earth’s

gravity constant), how high will the arrow go, and
when will it hit the ground again?”

Figure 5.7. The motion of an arrow

The motion of the arrow is described approximately by the graph

of the function

(5.53) s(t) = 50t − 4.9t2

shown in Figure 5.7. (We only show the graph for values of t in the
interval [0, 10.2] since the graph appears to be below the t-axis outside
this interval, which would correspond to the arrow being below the
ground, contradicting the “real life” context.) It is now not hard to
check that the arrow is at maximum height at just over five seconds,
when it reaches a height of approximately 130 meters. Furthermore,
the arrow appears to hit the ground after just over ten seconds.
Could we have computed this without a picture of the graph? Of
course, but that requires a bit of thinking and computation. Let’s
address the second question from the problem first: When does the
arrow hit the ground again? This corresponds to a solution of the
quadratic equation
50t − 4.9t2 = 0
 5.4. GRAPHING QUADRATIC FUNCTIONS 115

since the arrow hits the ground when s equals 0 again. We can solve
this equation by factoring:
50
−4.9t2 + 50t = −4.9t(t −)=0
4.9
and so s equals 0 at time t = 0 (when the arrow is fired) and at time
50
t ≈ 4.9 ≈ 10.2. The latter time (10.2 seconds after being fired) must
be the approximate time when the arrow hits the ground again.
The first question (How high will the arrow go, and when will it
be at maximum height?) is harder to answer.7 The solution uses our
technique of completing the square from subsection 5.2.4:
s = 50t − 4.9t2
50


= −4.9 t2 − 4.9 t


50 25 2 25 2
2


(5.54) = −4.9 t − 4.9
t + 4.9
− 4.9
 
50 25 2 2
+ 25


= −4.9 t2 − 4.9 t + 4.9 4.9
25 2 2
+ 25


= −4.9 t − 4.9 4.9

25 2


Since −4.9 t − 4.9 ≤ 0 for all t, we can now see that the height s
252
cannot exceed 4.9 ≈ 127.5 meters, but that that height is indeed the
25
maximum height and is achieved at 4.9 ≈ 5.1 seconds.
Now let’s do the above in the abstract. Consider a quadratic func-
tion
(5.55) f (x) = ax2 + bx + c
where a 6= 0. We can rewrite this, completing the square, as
 
2 b
f (x) = a x + x + c
a
 2  2 !
b b b
= a x2 + x + − +c
a 2a 2a
(5.56)  2 !
b b b2
= a x2 + x + +c−
a 2a 4a
2 
b2
 
b
=a x+ + c− and so
2a 4a

7Calculus was invented to solve problems like this, but in our easy case, we
don’t need it!
116 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

and so
2
b2
   
b
(5.57) f (x) − c − =a x− −
4a 2a
Setting
b2
   
b
x0 = − and y0 = c −
2a 4a
equation (5.57) simplifies to
(5.58) f (x) − y0 = a(x − x0 )2 or f (x) = a(x − x0 )2 + y0
The graph of the quadratic function is called a parabola, and the
point V with coordinates (x0 , y0 ) is called the vertex of the parabola.
(In the arrow example (5.53) above, the vertex is approximately the
point (5.1, 127.5).)
We are still assuming that a 6= 0 and now need to distinguish two
cases before we can discuss the graph of a quadratic function in more
detail:
Case 1: a > 0: Then a(x − x0 )2 + y0 ≥ y0 for all x, and so the minimal
value of f is y0 , which is achieved only at x = x0 . The equation
describes a parabola which “opens upward”. In order to solve
the quadratic equation
(5.59) a(x − x0 )2 + y0 = 0
we need to distinguish three subcases:
Case 1a: y0 < 0: Then the equation (5.59) has two real solutions
r
y0
x = x0 ± −
a
2
since the graph of the function f (x) = a(x p− yx00 ) + y0
intersects the x-axis at the two points (x0 + − a , 0) and
(x0 − − ya0 , 0). The function achieves its minimum value
p
at the vertex (x0 , y0 ). (See Figure 5.8 for an example with
x0 = 2, y0 = −1, and so the vertex is V = (2, −1); the
graph of f intersects the x-axis at (1, 0) and (3, 0).)
Case 1b: y0 = 0: Then the equation (5.59) has only one real solu-
tion
x = x0
since the graph of the function f (x) = a(x − x0 )2 + y0
intersects the x-axis only at the vertex (x0 , 0), at which
the function also achieves its minimum value. (See Fig-
ure 5.9 for an example with x0 = 2, y0 = 0, so the vertex
 5.4. GRAPHING QUADRATIC FUNCTIONS 117

Figure 5.8. The graph of the quadratic function

f (x) = x2 − 4x + 3

is V = (2, 0), which is also the point where the graph of f

intersects the x-axis.)
Case 1c: y0 > 0: Then the equation (5.59) has no real solutions
since

f (x) = a(x − x0 )2 + y0 ≥ y0 > 0

for all t. The function still achieves its minimum value at

the vertex (x0 , y0 ). (See Figure 5.10 for an example with
x0 = 2, y0 = 1, so the vertex is V = (2, 1), and the graph
of f does not intersect the x-axis.)
Case 2: a < 0: Then a(x − x0 )2 + y0 ≤ y0 for all x, and so the maximal
value of f is y0 , which is achieved only at x = x0 . The equation
describes a parabola which “opens downward”. In order to
solve the quadratic equation

(5.59) a(x − x0 )2 + y0 = 0

we again need to distinguish three subcases:

118 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

Figure 5.9. The graph of the quadratic function

f (x) = x2 − 4x + 4

Figure 5.10. The graph of the quadratic function

f (x) = x2 − 4x + 5
 5.4. GRAPHING QUADRATIC FUNCTIONS 119

Case 2a: y0 > 0: Then the equation (5.59) has two real solutions
r
y0
x = x0 ± −
a
2
since the graph of the function f (x) = a(x p− yx00 ) + y0
intersects the x-axis at the two points (x0 + − a , 0) and
(x0 − − ya0 , 0). The function achieves its maximum value
p
at the vertex (x0 , y0 ). (See Figure 5.11 for an example
with x0 = −2, y0 = 1, and so the vertex is V = (−2, 1);
the graph of f intersects the x-axis at (1, 0) and (3, 0).)

Figure 5.11. The graph of the quadratic function

f (x) = −x2 + 4x − 3

Case 2b: y0 = 0: Then the equation (5.59) has only one real solu-
tion
x = x0
since the graph of the function f (x) = a(x − x0 )2 + y0
intersects the x-axis only at the vertex (x0 , 0), at which
it also achieves its maximum value.
Case 2c: y0 < 0: Then the equation (5.59) has no real solutions
since
f (x) = a(x − x0 )2 + y0 ≤ y0 < 0
120 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

for all t. The function still achieves its maximum value

at the vertex (x0 , y0 ).
Finally, let’s study the effect of varying the three parameters a, x0
and y0 in the equations of a quadratic function

(5.58) f (x) − y0 = a(x − x0 )2 or f (x) = a(x − x0 )2 + y0

Since the vertex of the parabola described by these equations is

always the point (x0 , y0 ), the effect of changing x0 or y0 is clear: It
shifts the whole parabola right or left (when changing x0 ) or up or
down (when changing y0 ), respectively. The effect of changing a is a
bit more subtle. We saw already that if a > 0 then the parabola “opens
upward”, and that it if a < 0 then the parabola “opens downward”.
The effect of increasing or decreasing |a| is now to make the parabola
“steeper” or less “steep”. (Compare Figure 5.8 where a = 1 with
Figure 5.12 where a = 4.)

Figure 5.12. The graph of the quadratic function

f (x) = 4x2 − 16x + 15
 5.5. QUADRATIC INEQUALITIES 121

5.5. Quadratic Inequalities

We conclude this chapter by considering quadratic inequalities.
Let’s vary the arrow problem from section 5.4 a bit:
“An arrow is shot straight up from the ground with
an initial velocity of v0 = 50m/sec. If the height of
the arrow (in meters) at time t (after being fired, in
seconds) is described by

1
(5.52) s = v0 t − gt2
2

(neglecting friction, where g ≈ 9.8m/sec2 is the earth’s

gravity constant), when will the arrow be at least 50
meters above the ground?”
We already saw that the motion of the arrow is described approxi-
mately by the graph of the function

(5.53) s(t) = 50t − 4.9t2

shown in Figure 5.7. So answering the above problem amounts to

solving the quadratic inequality

(5.60) 50t − 4.9t2 ≥ 50

First note a common mistake: We cannot simply rewrite this inequal-

ity as −4.9t2 +p50t − 50 ≥ 0 and “use the quadratic formula” to
−50 ± 502 − 4 · (−4.9) · (−50)
get t ≥ ; this “solution” doesn’t even
2 · (−4.9)
make any sense since it specifies t to be ≥ some number ± some other
number! Instead, we can either use the graph of the function or use
the technique of completing the square.
For a graphical solution, we simply look at the graph in Figure 5.7
and note that the graph is at or above the line s = 50 approximately
between times t = 1 14 seconds and t = 9 seconds.
For an exact solution, we use the “completed square version” of
equation (5.53):

25 2 252


(5.54) s = −4.9 t − 4.9
+ 4.9
122 5. QUADRATIC FUNCTIONS, EQUATIONS AND INEQUALITIES

We can now solve our inequality in a few steps as follows:

25 2 2
+ 25


(5.61) −4.9 t − 4.9 4.9
≥ 50
25 2 2
≥ 50 − 25


(5.62) −4.9 t − 4.9 4.9
2 2 252
50 − 25 − 50 252 − 4.9 · 50

25 4.9
(5.63) t− ≤ = 4.9 =
4.9 −4.9 4.9 4.92
r
25 252 − 4.9 · 50
(5.64) t− ≤
4.9 4.92
r r
25 252 − 4.9 · 50 25 252 − 4.9 · 50
(5.65) − ≤ t ≤ +
4.9 4.92 4.9 4.92
which gives the approximate solution
1.124 ≤ t ≤ 9.08
Looking over the steps in (5.61)–(5.65), we see that the key steps are
in going from (5.63) to (5.64) and from (5.64) to (5.65). The second of
these two steps involves statements (4.15) and (4.16) about the absolute
value, namely, combining the two, that we have
(5.66) |a| ≤ r exactly when − r ≤ a ≤ r
The first of these two steps, going from (5.63) to (5.64), involves a
modification of Proposition 5.3 to the case of inequalities:
Proposition 5.5. Let A be an algebraic expression and r ≥ 0 a
number. Then:
(1) The solutions to the equation
A2 = r
are the same as the solutions to the statement
√ √
A = r or A = − r
(2) The solutions to the inequality
A2 < r
are the same as the solutions to the statement
√ √
− r<A< r
(3) The solutions to the equation
A2 > r
are the same as the solutions to the statement
√ √
A < − r or A > r
 5.5. QUADRATIC INEQUALITIES 123

Proposition follow immediately from (4.15)–(4.17) and from the fact

that for all non-negative numbers a and b:
(5.67) a < b exactly when a2 < b2
We now see that the key step above, going from (5.63) to (5.64),
follows immediately from Proposition 5.5.
It is also not hard to see that we could now carry out a general analy-
sis and solution for quadratic inequalities of the form ax2 + bx + c ≤ 0,
< 0, ≥ 0 or > 0, respectively, in the style of section 5.2 (for an algebraic
solution) or section 5.4 (for a graphical solution).
 CHAPTER 6

Exponential and Logarithmic Functions, Equations

and Inequalities

In this last chapter, we’ll cover the exponential functions and their
inverse functions, the logarithmic functions.

6.1. Exponentiation: A Review of the Definition

There is one common arithmetical operation from school mathemat-
ics which we have mostly ignored so far: exponentiation, or “taking the
power”. This operation is initially introduced as repeated multiplica-
tion, just as multiplication (in the whole numbers) can be viewed as
repeated addition.
The actual definition of exponentiation is quite tricky, and we will
present it here very carefully, in a number of steps.
Given any number a and any nonzero whole number m, we define
the mth power of a to be
m
m
a = a · a · a...a
z }| {
(6.1)
Here (and from now on), we will call a the base and m the exponent
(or the power )1. We read am as “a to the mth power” (or “a to the
mth” for short). For m = 2, we read a2 as “a squared” (or “the square
of a”); and for m = 3, we read a3 as “a cubed” (or “the cube of a”).

This definition of exponentiation as repeated multiplication gives

us immediately the following rules of exponentiation (for now, only for
nonzero whole numbers m and n):
(6.2) am · an = am+n
am
(6.3) n
= am−n (if m > n)
a
(6.4) (am )n = am·n
(6.5) (a · b)m = am · bm
1In the Singapore math books, the exponent is also called the index (plural:
indices).
124
 6.1. EXPONENTIATION: A REVIEW OF THE DEFINITION 125

To see that, e.g., (6.2) is true, simply observe that

m n
m n
a · a = a · a · a···a·a · a · a···a
z }| { z }| {
m+n
= a · a · a···a
z }| {

= am+n
And to see that (6.4) is true, we check
n
z }| {
m m m m
(am )n = a · a · a · · · a · a · a · a · · · a · a · a · a · · · a . . . a · a · a · · · a
z }| { z }| { z }| { z }| {
m·n
= a · a · a···a
z }| {

= am·n
Exercise 6.1. How you would prove (6.3) and (6.5) this way? Why
does it make sense to impose the restriction m > n in (6.3) for now?)
So (6.2)–(6.5) follow immediately from our previous rules for mul-
tiplication. Wouldn’t it be nice to have these rules hold not just for
nonzero whole numbers as exponents? Of course, you know that these
rules are still true, right?
Well, there is a small problem: We haven’t even defined what am
means unless m is a nonzero whole number! We will do this in a
sequence of small steps, each time justifying our definition for more
and more possible exponents.
From now on, we will assume that the base a is nonzero. (This
will make sense in a moment when we see how division by a becomes
involved in our definition.) Now we can define exponentiation for an
exponent which is 0 or a negative integer as follows:
(6.6) a0 = 1
1
(6.7) am = (for any negative integer m)
a−m
First, note here that a−m in (6.7) has already been defined in (6.1)
since −m is a nonzero whole number if m is a negative integer. More
importantly, note that we are forced to define exponentiation this way
if we want the analogue of (6.3) to still hold:
a1 a
a0 = = =1
a1 a
a0 1
am = −m = −m
a a
126 6. EXPONENTIAL AND LOGARITHMIC FUNCTIONS

What about rational exponents? For this, we need to restrict the

base a even further: We’ll assume from now on that a is not only
nonzero but indeed a positive real.2 Now we’ll define exponentiation
for rational exponents as follows (for any integer n and any positive
integer p):

1 √
p
(6.8) ap = a
n
 1
n
(6.9) a p = ap

Of course,√this immediately raises another problem: We haven’t defined

yet what p a means for a positive real number a and a positive integer p.
But (as for the case p = 2, i.e., the square root), we can define the
pth root of a simply as the unique positive solution3 to the equation
xp = a. Again note also that the definition in (6.9) already requires
the definition in (6.8). And again, more importantly, note that we
are forced to define exponentiation this way if we want the analogue
of (6.4) to still hold:

 1 p p
ap = a p = a1 = a
n 1
·n
 1 n
a = a = ap
p p

But there are two hidden problems with these definitions:

(1) It is possible that the rational number np (for an integer n

and a positive integer p) happens to equal an integer m, say.
Now we need to show that the definition in (6.9) coincides
with the definition from (6.1) and (6.6)–(6.7). In that case,
we have m = np and so n = mp. Now we can argue as follows

2We impose this restriction since we’ll be taking roots in a moment. It is

actually possible to define am in a meaningful way for a negative real and some
rational exponents m, but the rules for when this is possible are rather complicated,
so we’ll ignore this case here.
3The fact that such a solution always exists and is unique is hard to show and
beyond the scope of these notes;√ so we’ll simply assume it. For an odd positive
integer p, one can also define p a for negative a this way; but we’ll not do so here.
 6.1. EXPONENTIATION: A REVIEW OF THE DEFINITION 127

(using (6.4) for integer exponents only!):

n  1 n
a p = ap
 1 p·m
= ap
 1 p m
= ap
= am

(2) Similarly, given a rational number r, it can be written in many

different ways as a fraction. So, in order to show that ar is
0
uniquely defined, we really need to show that whenever np = np0
for integers n and n0 and for positive integers p and p0 , then

n n0
p p0
a =a

under the definition in (6.9). This proof is similar to the one

in (1) but quite a bit more tedious since it involves many cases,
so we’ll skip it here.
Finally, can we define ar for irrational exponent r? The answer is a
qualified “yes”; qualified because in order to make this definition pre-
cise, one really needs calculus, which we clearly don’t want to assume
here. Instead, let’s give the intuition.4
For an irrational number r and arbitrary positive real number a, we
define ar as being more and more closely approximated by aq where q
is a rational number close to r. The fact that this definition works is
a “miracle” we have to accept here, since proving it would take us far
beyond the scope of these notes.
Finally, looking back, we really need to return to checking our rules
of exponentiation (6.2)–(6.5) for arbitrary exponents. We state them
again here for reference as a proposition:

4Strictly speaking, what we are doing here is merely what you have already
done for the other arithmetical operations, like addition and multiplication, for
arbitrary real numbers. E.g., the sum r0 + r1 of two irrational numbers r0 and r1 ,
say, is “approximated” more and more “closely” by the sum q0 + q1 of rational
numbers q0 and q1 “close” to r0 and r1 , respectively.
128 6. EXPONENTIAL AND LOGARITHMIC FUNCTIONS

Proposition 6.2. The following exponential rules hold for all pos-
itive real numbers a and b, and for all real numbers r and s:
(6.10) ar · as = ar+s
ar
(6.11) = ar−s
as
(6.12) (ar )s = ar·s
(6.13) (a · b)r = ar · br
If both r and s are integers, then these rules hold for arbitrary nonzero
real numbers a and b.
The formal proof of Proposition 6.2 is quite tedious, so we’ll skip it
here and simply accept the exponential rules as given from now on.
There is one special case of (6.11) which comes up frequently,
namely, the case when r = 0, so we’ll state this as a separate formula:
 s  −t
−s 1 1 t 1 1
(6.14) a = s = or, equivalently: a = −t =
a a a a

6.2. Motivation for Exponentiation

So far, we have only discussed the definition of exponentiation in
fair detail and have left the motivation rather weak: “Exponentiation is
repeated multiplication, just like multiplication is repeated addition.”
Of course, this motivation is not really satisfying; one could now go
on and talk about “repeated exponentiation”, but there is a really
good reason why almost no one does: There are essentially no real-life
phenomena which can be described by “repeated exponentiation”, but
there are quite a few which can be described by exponentiation.
Let’s look at the following two examples with which I’d like to
contrast linear vs. exponential functions:
“A fish in Ann’s aquarium is giving birth to babies.
Every hour, three new babies are born. If she has
five fish at the beginning, how many fish will she have
after n hours?”
“The number of bacteria in a lab dish doubles every
hour. If there are one hundred bacteria at the begin-
ning, how many will there be after n hours?”
Look at the first problem first; Table 6.1 shows the number of fish
during the first few hours.
We see right away that the number of fish increases each hour by
a fixed amount, namely, by 3. Therefore the number F of fish after n
 6.2. MOTIVATION FOR EXPONENTIATION 129

number of hours 0 1 2 3 4 5
number of fish 5 8 11 14 17 20

Table 6.1. The number of fish after n hours

hours is given by the linear function

F (n) = 3n + 5
or, more awkwardly, for reasons which will become apparent in a few
paragraphs, as
(6.15) F (n) = 5 + 3 · n
(Note that this function really only makes sense for whole number
values of n; it doesn’t make sense to talk about there being 9.5 fish
after 1.5 hours! But let’s ignore this side issue for now.)
Now let’s look at the second problem; Table 6.2 shows the number
of bacteria during the first few hours.

number of hours 0 1 2 3 4 5
number of bacteria 100 200 400 800 1,600 3,200

Table 6.2. The number of bacteria after n hours

We see right away that the number of bacteria does not increase by
a fixed amount each hour. In fact, the increase itself increases rather
dramatically: From hour 0 to hour 1, the increase is by only 100,
but from hour 4 to hour 5, the increase is already by 1,600. In fact,
this leads us to the following observation: If we denote the number
of bacteria after n hours by B(n), and consider the “first difference
function” (as defined in section 5.1), namely,
B 1 (n) = B(n + 1) − B(n)
then it turns out that in fact B 1 (n) = B(n), i.e., the first difference
function B 1 equals the original function B. We could now try, as in
section 5.1, to take the second difference function
B 2 (n) = B 1 (n + 1) − B 1 (n)
but, of course, this will give us the same function again! So not only is
the function B not linear; it is also not quadratic, by the discussion in
section 5.1.
130 6. EXPONENTIAL AND LOGARITHMIC FUNCTIONS

On the other hand, you will surely have noticed definite “patterns”
in Tables 6.1 and 6.2: In the former, the additive change is constant;
namely, each hour, the number of fish increases by a summand of 3.
On the other hand, in the latter, the multiplicative change is constant;
namely, each hour, the number of bacteria increases by a factor of 2.
It is then not hard to see that the function B can be described by

(6.16) B(n) = 100 · 2n

Note that there is some similarity between (6.15) and (6.16): The
“starting point” (at n = 0) is 5 and 100, respectively, and shows up
first in the expression on the right-hand side. Next comes the op-
eration symbol for the way in which the increase occurs, additively
in (6.15), and multiplicatively in (6.16). Next comes the “rate of
change”, namely, 3 in (6.15) for the additive change of 3 each hour,
and 2 in (6.16) for the multiplicative change of 2 each hour, respectively.
And finally comes n for the number of times by which this increase oc-
curs, preceded by symbol for the “repeated operation” (namely, multi-
plication as repeated addition in (6.15), and exponentiation as repeated
multiplication in (6.16), respectively).
In both our fish and our bacteria example, it is now easy to compute
the number of fish or bacteria, respectively, at time 10 hours, say: It
is 5 + 3 · 10 = 35 fish and 100 · 210 = 102, 400 bacteria, respectively.5
But, as we mentioned already in section 2.5, such problems are really
problems of arithmetic: In order to solve them, we merely have to
evaluate the numerical expressions 5 + 3 · 10 and 100 · 210 , respectively.
On the other hand, suppose that we are given the number of fish, or
bacteria, respectively, and want to compute the number of hours after
which there are this many fish or bacteria. Suppose I want to know
when there will be 50 fish. This is now a “real algebra” problem and
clearly amounts to solving the equation

5 + 3 · n = 50

by “undoing” the algebraic expression 5 + 3 · n to arrive, step by step,

at the solution

3n = 50 − 5
50 − 5
n= = 15
3

5In
the latter case, of course, we should really write “approximately 100,000
bacteria” since 102,400 gives a false sense of precision for our answer.
 6.3. THE EXPONENTIAL FUNCTIONS 131

So it will be after 15 hours that there will be 50 fish in the aquarium;

and in order to arrive at the solution, we first subtract 5 from both
sides of the equation, and then divide by 3 on both sides.
Let’s try the same for the bacteria: Suppose I want to know after
how many hours there will be 6,400 bacteria? Again, this is now a
“real algebra” problem and clearly amounts to solving the equation
100 · 2n = 6, 400
by “undoing” the algebraic expression 100 · 2n to arrive, step by step,
at the solution
6, 400
2n = = 64
100
n=6
So it will be after 6 hours that there will be 6,400 bacteria in the lab
dish. However, obtaining this solution is conceptually more difficult:
In the first step, we simply divide both sides by 100. But the second
step, going from 2n = 64 to n = 6, is a bit mysterious at this point.
We will learn how to solve such so-called “exponential equations” in
general in section 6.4, when we introduce the logarithmic functions. In
our particular example with small numbers, we can, of course, just use
“guess-and-check”.6

6.3. The Exponential Functions

6.3.1. Properties of Exponential Functions. Before we pro-
ceed to solving exponential equations (such as 2n = 64 in the example
above), let’s talk about exponential functions in general and note some
of their properties.
First of all, for each possible base b > 0, we can define an exponential
function f (x) = bx , which is defined for all real numbers x.7 In the case
that b = 1, we get the rather boring constant function f (x) = 1x = 1,
so we will also assume b 6= 1 for the remainder of this section.
We first look at two typical examples of exponential functions, the
function f (x) = 2x
(see Figure 6.1 for its graph), and the exponential
x
function g(x) = 12 (see Figure 6.2 for its graph).
From these two graphs, we observe some properties which we state
in the following proposition. We will then argue that all exponential
functions “look like one these two functions”, depending on whether
b > 1 or 0 < b < 1, respectively.
6And I had to start with the rather artificial number of 6,400 bacteria to “make
the problem work out”.
7We explained already in section 6.1 why the assumption b > 0 is needed.
132 6. EXPONENTIAL AND LOGARITHMIC FUNCTIONS

Figure 6.1. The graph of the exponential function

f (x) = 2x

Proposition 6.3. For any base b > 0 (with b 6= 1), the exponential
function h(x) = bx has the following properties:
(1) The exponential function h(x) = bx is defined for all real num-
bers x, i.e., the domain of h is the set R of all real numbers.
(2) The graph crosses the y-axis at the point (1, 0).
(3) The graph lies entirely above the x-axis, i.e., bx > 0 for all x.
 6.3. THE EXPONENTIAL FUNCTIONS 133

Figure 6.2. The graph of the exponential function

1 x


f (x) = 2

(4) If b > 1 then

(a) the exponential function h(x) = bx increases as x in-
creases, i.e., if x0 < x1 then bx0 < bx1 ;
(b) the graph increases “without bound” as x grows larger and
larger; and
134 6. EXPONENTIAL AND LOGARITHMIC FUNCTIONS

(c) the graph approaches the x-axis arbitrarily closely as x

grows smaller and smaller (i.e., as x “grows larger and
larger in the negative direction”).
(5) If 0 < b < 1 then
(a) the exponential function h(x) = bx decreases as x in-
creases, i.e., if x0 < x1 then bx0 > bx1 ;
(b) the graph approaches the x-axis arbitrarily closely as x
grows larger and larger; and
(c) the graph increases “without bound” as x grows smaller
and smaller (i.e., as x “grows larger and larger in the
negative direction”).
(6) The exponential function h(x) = bx is 1–1.
(7) The range of the exponential function h(x) = bx is the set
(0, ∞) of all positive real numbers.
In summary, the exponential function h(x) = bx is a 1–1 function with
domain R and range (0, ∞) (the set of all positive real numbers).
Let’s look at these statements one by one and try to argue that
they hold true not only for the special cases b = 2 and b = 12 shown in
Figures 6.1 and 6.2, but indeed for all positive bases b 6= 1.
First of all, (1)–(3) follow by analyzing the definitions in section 6.1.
Assume that b > 1. Then, for (4a), note that x0 < x1 implies
by (6.11) that
bx 1
= bx1 −x0 > 1
bx 0
since x1 − x0 > 0, and so bx1 > bx0 . Next, (4b) and (4c) are somewhat
harder to check: In the special case b = 2, we see that x < 2x for
all x; more generally, for arbitrary b > 1, we have that (b − 1)x < bx .
But now (b − 1)x grows “without bound” as x does, and so must bx .
Finally, (4c) follows by (6.14), since, as x grows smaller and smaller
(i.e., as x “grows larger and larger in the negative direction”), −x grows
larger and larger (“in the positive direction”), but then, by (6.14),
1
bx =
b−x
and this expression gets closer and closer to 0 since b−x grows without
bound.
Now assume that 0 < b < 1. Then, similarly, for (5a), note that
x0 < x1 implies by (6.11) that
bx 1
= bx1 −x0 < 1
bx 0
 6.3. THE EXPONENTIAL FUNCTIONS 135

since x1 − x0 > 0, and so bx1 < bx0 . Next, (5b) and (5c) can be reduced
to (4c) and (4b), respectively: 0 < b < 1 implies 1 < 1b by (3.9). Now

−x
apply (4c) to see that, as x grows larger and larger, bx = 1b gets
arbitrarily close to 0. Similarly, apply (4b) to see that, as x grows
smaller and smaller (i.e., as x “grows larger and larger in the negative

−x
direction”), bx = 1b grows larger without bound.
Furthermore, (6) follows by (4a) and (5a) since x0 < x1 implies
bx0 < bx1 or bx0 > bx1 and thus bx0 6= bx1 .
Finally, (7) can be seen from (4b) and (4c) for b > 1, and from (5b)
and (5c) for 0 < b < 1, respectively.8
There is one issue we need to address still: How does the value of
the base b impact on the graph of the exponential function? Figure 6.3
shows some examples; note that for clarity, we used various dotted lines
for some of the graphs so you can quickly tell them apart. As you can
see from these graphs, for b > 1, the bigger b is, the more steeply the
graph of the exponential function increases. On the other hand, for
0 < b < 1, the smaller b, the more steeply the graph of the exponential
function decreases.
One final note: If you look at other algebra books, or at calculus
books, you will find a special exponential function being mentioned,
often called “the” exponential function. This is the exponential func-
tion f (x) = ex with special base e ≈ 2.718 (which is sometimes called
Euler’s number ). The fact that this base is special does not become
apparent until one studies calculus; and using e as a base before is re-
ally more of a pain than a convenience since e is actually an irrational
number. Therefore, we have chosen to ignore the base e here.

6.3.2. Another Application of Exponential Functions. We

have already seen one application of exponential functions in “real life”,
namely, to bacterial growth in section 6.2. We will briefly cover one
other application here to show you that exponential functions are in-
deed very useful, and not only in mathematics.
We had seen in section 6.2 that the “multiplicative change” of the
number of bacteria from one hour to the next was constant: After each
hour, the number of bacteria doubled. There is a much more “every-
day” phenomenon in which such growth occurs, namely, in interest
computations.
8Technically, we are cheating a bit here: The problem is that we haven’t shown
that there are no “holes” in the range of the exponential function; i.e., that the
range really is the whole interval (0, ∞). But calculus is required to actually show
this carefully; so we’re skipping over this fine point here.
136 6. EXPONENTIAL AND LOGARITHMIC FUNCTIONS

Figure 6.3. The graphs of various exponential functions

The setup for interest computations is typically as follows: You

have a certain amount of capital (or debt) called the principal. At the
end of each fixed time period (say, a year, or a month, or a quarter), a
fixed percentage of the current amount is added. This added amount
is called the interest, and the percentage is called the interest rate (per
time period, e.g., per year). Typically, the interest is compounded, i.e.,
 6.3. THE EXPONENTIAL FUNCTIONS 137

the interest is added to the amount, and next time interest is paid,
interest is also added for the previous interest.
For a simple example, suppose I have $100 in a savings account,
and the bank pays me 5% interest at the end of each year. At the
end of the first year, I then have $105 (namely, the principal of $100
plus 5% of it in interest); at the end of the second year, I have $110.25
(namely, the previous amount of $105 plus 5% of it in interest); etc.
See Table 6.3.

number of years 0 1 2 3 4
amount in account $100.00 $105.00 $110.25 $115.76 $121.55

Table 6.3. Savings after n years

We can now see that my savings grow by a factor of 105% = 1.05

each year. Thus the capital I have is described by the function
(6.17) C(n) = 100 · 1.05n
where n is the number of years I have saved. (This number n must be a
whole number here for the problem to make sense.) More generally, the
capital (or debt), starting with a principal P and under an interest rate
of r% per time period, after n time periods, is given by the function
 r n
(6.18) C(n) = P · 1 +
100
where again the number n of time periods must be a whole number.
Of course, (6.18) can also be used to solve for the principal P :
C
P =
r n


1 + 100
or to solve for the interest rate r of time periods:
 r n
1+ = P/C
100
r p
1+ = n P/C
100 p
r = n P/C − 1
It is also possible to solve for the number n of time periods, but that
requires the logarithmic functions to be introduced in section 6.4.
Another application of exponential functions concerns nuclear de-
cay, but again this will first require logarithmic functions and will be
covered in subsection 6.4.2.
138 6. EXPONENTIAL AND LOGARITHMIC FUNCTIONS

6.4. The Logarithmic Functions and Solving Exponential

Equations
Our goal in this section is to find functions which allow us to solve
exponential equations, i.e., equations of the form
(6.19) bx = c
where b and c are parameters. Compare this to previous situations
where we tried to solve equations of the form
(6.20) b + x = c or
(6.21) b·x=c
We could easily solve (6.20) and (6.21) by “undoing” the operations of
addition and multiplication, respectively:
x=c−b
c
x=
b
In each case, we “undid” the equation by using an “inverse” function,
namely, subtraction to “invert” addition, and division to “undo” mul-
tiplication.
More formally: In (6.20) and (6.21), we consider b to be fixed,
so the functions we are inverting are the two functions f (x) = b + x
and g(x) = b · x; and the inverse functions are f −1 (y) = y − b and
g −1 (y) = yb , respectively. (Of course, when inverting g, we have to
assume that b 6= 0; otherwise the function g(x) = b · x cannot be
inverted.)
This should lead us to the conclusion that in order to solve expo-
nential equations as in (6.19), we need to find an inverse function for
the exponential function f (x) = bx . We saw already that we need to
assume that b > 0 for the exponential function to be defined. Also,
we need to assume that b 6= 1; otherwise, the exponential function
f (x) = bx is a constant function and cannot be inverted. So, for the
rest of this section, we’ll assume that b > 0 and b 6= 1.
6.4.1. Definition and Properties of Logarithmic Functions.
In Proposition 6.3, we observed that for any positive base b 6= 1, the
exponential function is a 1–1 function with domain R and range (0, ∞).
Typically, one sets the codomain (as defined in section 4.1) to be the set
(0, ∞) of positive reals so that the exponential function then becomes
a 1–1 and onto function:
expb : R → (0, ∞)
(6.22)
x 7→ bx
 6.4. LOGARITHMIC FUNCTIONS 139

Here, we have introduced the special symbol expb for the exponential
function with base b.
Now recall from subsection 4.2.3 that a function has an inverse func-
tion exactly when the function is 1–1 and onto! This inverse function of
the exponential function expb (x) = bx is usually called the logarithmic
function to base b; it is defined as
logb : (0, ∞) → R
(6.23)
y 7→ the unique x such that bx = y

Let’s immediately return to solving exponential equations for a

minute: The equation
(6.19) bx = c
can be solved using the logarithmic function logb to arrive at
(6.24) x = logb (c)
But how do we evaluate logb (c) in practice, with actual numbers?
E.g., for the bacterial growth example from section 6.2, in order to
solve
(6.25) 2n = 64
we have to evaluate
(6.26) n = log2 (64)
Your first reaction may be: How do I evaluate log2 (64) on my calcu-
lator? There is no log2 button on my calculator? The only buttons I
see are called “log” and “ln”. (And that is only if you have a scientific
calculator, and even then it may not have both.)
I’ll give two answers here. First of all, in simple cases, you can
compute logarithmic functions “in your head”. To solve (6.26), you
could simply recall the definition of log2 (64):
(6.27) log2 (64) = the unique n such that “2n = 64” holds
This means that you have to find the right exponent n to make the
statement 2n = 64 true. A moment’s thought tells you that that expo-
nent must be 6.
Of course, this method will not always work, so how do you do
this with a calculator? The real problem is that there is a different
logarithmic function logb for each base b, and no calculator gives them
all. (Your calculator could give you the option of choosing both b and x
when computing logb (x), but I’ve never seen any calculator like that!)
Fortunately, there is a simple little trick. Most scientific calculators
provide you with two special logarithmic functions, namely, log and ln.
140 6. EXPONENTIAL AND LOGARITHMIC FUNCTIONS

The first, called log, is really log10 and sometimes called the common
logarithm or the decadic logarithm. Its importance is mainly historic,
as we will explain later on; but since it’s somewhat easier than the
other logarithm on your calculator, we will mainly use it. The other
logarithmic function on your calculator, called ln, is really loge for
the number e ≈ 2.718 mentioned at the end of section 6.3.1 and is
usually called the natural logarithm (which explains the symbol ln); it
is somewhat cumbersome and not all that useful until calculus (when it
becomes very important, so important that most mathematicians call
it “the” logarithmic function and ignore all others!).9
There is now a handy little formula which allows you to compute
any logarithmic function from log10 :
log10 (x)
(6.28) logb (x) =
log10 (b)
This formula is somewhat tedious to deduce, so we’ll skip this here.10
(Incidentally, the same formula also works with e in place of 10, i.e.,
we also have logb (x) = ln(x)
ln(b)
.)
Note that formula (6.28) is exactly what you’re looking for with
a calculator which can handle log10 : The right-hand side only in-
volves computing log10 (and division). E.g., here is how we can com-
pute log2 (64) the hard way, using a calculator:
log10 (64) 1.806
log2 (64) = ≈ =6
log10 (2) 0.301
We will see applications of the logarithmic functions in subsec-
tion 6.4.2 where the use of a calculator (and thus the use of for-
mula (6.28)) is inevitable.
But before we head into applications of the logarithmic functions,
let’s look at some properties of these functions. We summarize them in
the following proposition, which is the counterpart of Proposition 6.3
for the exponential functions. (Figure 6.4 shows the graphs of some
logarithmic functions and of the corresponding exponential functions;
9Some calculators have only one button for logarithmic functions, typically
called log, and now you don’t necessarily know which of the two logarithmic func-
tions above that means, so be careful! You can test them as follows: If log(10)
computes as 1 on your calculator, then you have log10 , since 101 = 10. If it com-
putes as approximately 2.3, then you have loge , since e2.3 ≈ 2.72.3 ≈ 10.
10But if you insist, here is the deduction: First of all, x = blogb (x) =

logb (x)
10log10 (b) as well as x = 10log10 (x) by definition of logb and log10 , respectively.

logb (x)
By (6.12), this gives 10log10 (b)·logb (x) = 10log10 (b) = 10log10 (x) . Since exp10
is 1–1, we obtain log10 (b) · logb (x) = log10 (x) and thus logb (x) = log 10 (x)
log (b) .
10
 6.4. LOGARITHMIC FUNCTIONS 141

Figure 6.4. The graphs of two exponential functions

and the corresponding logarithmic functions

note that for fixed base b, the graphs of the exponential and the loga-
rithmic functions to base b are the reflections of each other along the
“main diagonal” y = x as mentioned in subsection 4.2.3. In order to
better distinguish the graphs of the various functions in Figure 6.4, we
have drawn some of them with “dotted” lines; but note that in reality,
they should all be drawn with “solid” lines.)
Proposition 6.4. For any base b > 0 (with b 6= 1), the logarithmic
function logb has the following properties:
(1) The logarithmic function logb (x) is defined for all positive real
numbers x, i.e., the domain of logb is the set (0, ∞) of all
positive real numbers.
(2) The graph crosses the x-axis at the point (0, 1).
142 6. EXPONENTIAL AND LOGARITHMIC FUNCTIONS

(3) If b > 1 then

(a) the logarithmic function logb (x) increases as x increases,
i.e., if x0 < x1 then logb (x0 ) < logb (x1 );
(b) the graph increases “without bound” as x grows larger and
larger; and
(c) the graph decreases “without bound” as x approaches 0.
(4) If 0 < b < 1 then
(a) the logarithmic function logb (x) decreases as x increases,
i.e., if x0 < x1 then logb (x0 ) > logb (x1 );
(b) the graph decreases “without bound” as x grows larger and
larger; and
(c) the graph increases “without bound” as x approaches 0.
(5) The logarithmic function logb is 1–1.
(6) The range of the logarithmic function logb is the set R of all
real numbers.
In summary, the logarithmic function logb is a 1–1 function with do-
main (0, ∞) (the set of all positive real numbers) and range R.

The various parts of Proposition 6.4 can all be verified similarly to

the way we verified Proposition 6.3; we’ll skip this here for brevity’s
sake.
We finally note the following useful properties of the logarithmic
functions, which follow from properties of the corresponding exponen-
tial functions stated in Proposition 6.2:

Proposition 6.5. The following logarithmic rules hold for all pos-
itive real numbers a 6= 1, all positive real numbers r and s, and all real
numbers y:

(6.29) loga (r · s) = loga (r) + loga (s)

r
(6.30) loga ( ) = loga r) − loga (s
s
(6.31) loga (ry ) = y · loga (r)

We will not prove these here, but merely list them for reference.11

11They actually are not hard to show: In each case, we set r = ax and s = ay .
Firstly, loga (r · s) = loga (r) + loga (s) holds since ax · ay = ax+y . Next, loga ( rs ) =
x
loga (r) − loga (s) holds since aay = ax−y . And finally, loga (ry ) = y · loga (r) holds
since (ax )y = axy .
 6.4. LOGARITHMIC FUNCTIONS 143

6.4.2. Applications of Logarithmic Functions. Let’s now fin-

ish our coverage of the logarithmic functions by looking at some appli-
cations. The problem with these is not that they are particularly dif-
ficult, but that each is typically phrased quite differently, which tends
to confuse folks quite a bit.
6.4.2.1. The slide rule. The slide rule was invented in the 1600’s as
a mechanical calculator and was in wide use until the 1970’s, when it
was quickly crowded out by the modern pocket calculator.

Figure 6.5. A slide rule

The basic idea of the slide rule is to take advantage of the log-
arithmic rules (6.29) and (6.30) in Proposition 6.5 in order to make
(approximate) manual computations easier by converting multiplica-
tion to “graphical addition”, and division to “graphical subtraction”.12
Figure 6.5 shows a basic slide rule with a magnifying sliding cursor,
allowing you to better align numbers on the scales; we will mainly use
the middle two scales of it (called scales C and D on our slide rule) for
now. Note that these two scales are sliding number lines, each from 1
to 10, but not drawn to scale like a usual number line, but such that
each number x between 1 and 10 is drawn at a distance log10 (x) to the
right of 1. Note, therefore, that the distance from 1 to 4 is twice the
distance from 1 to 2 (since log10 (4) = log10 (22 ) = 2 · log10 (2) by (6.31)),
and similarly the distance from 1 to 8 is three times the distance from 1
12
This short subsection is not intended as a full-blown guide to using a slide
rule, but rather to give you the basic idea as to how and why a slide rule works.
You can practice with a virtual slide rule on your computer screen at the web site
http://www.antiquark.com/sliderule/sim/n909es/virtual-n909-es.html.
144 6. EXPONENTIAL AND LOGARITHMIC FUNCTIONS

to 2 (since log10 (8) = log10 (23 ) = 3 · log10 (2) by (6.31)). (You can check
this against scale L at the very bottom, which is the usual number line
from 0.0 to 1.0; its numbers represent the (base-10) logarithms of the
numbers on scale D, so log10 (1) = 0, log10 (2) ≈ 0.3, log10 (4) ≈ 0.6,
log10 (8) ≈ 0.9, and log10 (10) = 1.)
To illustrate the use of the slide rule, let’s start with a simple mul-
tiplication problem: Suppose you want to multiply 2 by 3. The slide
rule would accomplish this in the following steps (see Figure 6.5):
(1) Convert 2 and 3 to log10 (2) and log10 (3), represented by the
length 2 on scale D and the length 3 on scale C, respectively.
(2) Add these two lengths and read off their sum as the length
log10 (2) + log10 (3) = log10 (2 · 3) = log10 (6) on scale D (us-
ing (6.29)).
(3) Convert log10 (6) back to 10log10 (6) = 6.
Similarly, in order to divide 6 by 3, say, we would proceed as follows
(again refer to Figure 6.5):
(1) Convert 6 and 3 to log10 (6) and log10 (3), represented by the
length 6 on scale D and the length 3 on scale C, respectively.
(2) Subtract these two lengths and read off their difference as the
length log10 (6) − log10 (3) = log10 ( 36 ) = log10 (2) on scale D
(using (6.30)).
(3) Convert log10 (2) back to 10log10 (2) = 2.
There are many other computations which can be performed on
a slide rule, but since pocket calculators are nowadays a much more
efficient way to perform the same computations, we’ll leave it at that.
6.4.2.2. Radioactive Decay. The next application concerns radioac-
tive materials: As radioactive material decays, its mass shrinks by a
fixed amount per time period which is proportional to the amount left.
Mathematically, this is similar to the balance of an interest-bearing
savings account, which grows by a fixed amount per time period, as
we discussed in subsection 6.3.2, except that the amount of radioactive
material decreases, whereas the balance of an interest-bearing savings
account increases.
Now, just to confuse you, the decay of a radioactive material is
typically described by giving its half-life, namely, the time period it
takes for the amount of material to decrease to half of its original
amount. E.g., the half-life of plutonium is approximately 24,000 years;
Table 6.4 shows the amount left after multiples of 24,000 years, starting
with 1kg of plutonium. As you can see, the mass of the plutonium
shrinks in half every 24,000 years, as the name “half-life” indicates.
 6.4. LOGARITHMIC FUNCTIONS 145

time (years) 0 24,000 48,000 72,000 96,000 t

t
  24,000
1
mass (kg) 1 0.5 0.25 0.125 0.0625
2

Table 6.4. Radioactive Decay: Approximate mass of

plutonium left, starting with 1kg

But suppose we want to describe the amount of plutonium left af-

t
ter only 1,000 years? Well, after t many years, 24,000 many multiples
of the half-life period 24,000 years will have passed, so there will be
t
  24,000
1
many kg of plutonium left. So, in particular, after 1,000
2
1,000
  24,000   241
1 1
years, there will be = ≈ .9715kg = 971.5g of pluto-
2 2
nium left.
Now the logarithmic function also allows us to solve the inverse
problem: Suppose I want to know after how many years there will be
100g of plutonium left. This then amounts to solving the equation
t
  24,000
1
(6.32) = 0.1
2
which can be undone as
t
= log 1 (0.1)
24, 000 2

giving
t = 24, 000 · log 1 (0.1)
2

log(0.1)
= 24, 000 · (by (6.28))
log(0.5)
−1
≈ 24, 000 ·
−0.301
≈ 24, 000 · 3.32 ≈ 79, 700
So it will take approximately 79,700 years for 90% of the plutonium to
decay (and thus only 10% of it to be left).
6.4.2.3. Earthquakes and the Richter Scale. Let’s look another ap-
plication of logarithms, which at first looks quite different, the Richter
scale for earthquakes. This is done as follows: One first measures the
146 6. EXPONENTIAL AND LOGARITHMIC FUNCTIONS

maximum amplitude A (i.e., how far the ground “swings” from its rest-
ing position) at a distance of 100km from the epicenter;13 the amplitude
is measured in µm (read: micrometer, which is equal to 0.001mm or
0.000001m).
The problem with amplitude is that one has to deal with very large
numbers: The amplitude of a barely noticeable earthquake is about
100µm; the amplitude of a very destructive earthquake is as high as
100,000,000µm = 100m or more. As you can see, it’s not very practical
to use the amplitude; so one uses the (decadic) logarithm of the am-
plitude, called the Richter magnitude (or, more correctly, called local
magnitude)

(6.33) ML = log10 (A)

Thus a barely noticeable earthquake of amplitude 100µm has Richter

magnitude ML = log10 (100) = log10 (102 ) = 2. On the other hand, a
very destructive earthquake of amplitude 100,000,000µm has Richter
magnitude ML = log10 (100, 000, 000) = log10 (108 ) = 8. So as long
as the amplitude is just a power of 10, the Richter magnitude will
be a whole number (namely, the number of 0’s in A when A is ex-
pressed in µm), and the Richter scale was chosen to make the Richter
magnitude a number roughly between 0 and 10. One more quirk: If
you have an earthquake of Richter magnitude M0 , say, then an earth-
quake of Richter magnitude M0 + 1 has ten times the amplitude:
This is because the first earthquake has amplitude A0 , say, where
M0 = log10 (A0 ) or A0 = 10M0 ; so the earthquake of Richter mag-
nitude M0 + 1 has amplitude A1 , say, where M0 + 1 = log10 (A1 ) or
A1 = 10M0 +1 = 10 · 10M0 = 10 · A0 . This explains why an earthquake
of Richter magnitude M0 + 1 is generally much more destructive than
an earthquake of Richter magnitude M0 .
6.4.2.4. Sound Volume and Decibels. The (decadic) logarithm ap-
pears similarly in the measurement of (sound) volume or sound inten-
sity. One first expresses volume in terms of power P (formally defined
in physics as energy per time unit), where the volume, or power, is
expressed in watts (abbreviated W). Again the problem is that com-
monly occurring sound volumes can range from 1W (barely audible) to
1,000,000,000,000 W (one trillion watts, where ear damage occurs even
during short exposure). So it is convenient to define a more practical

13There are formulas which one can use to convert the amplitude measured at
a different distance from the epicenter to the amplitude one would have measured
at a distance of 100km; we will not worry about this here.
 6.4. LOGARITHMIC FUNCTIONS 147

measure of volume (or sound intensity) as

(6.34) LdB = 10 · log10 (P )
When expressed as LdB , the unit of volume is decibels.14
Using the above examples, we then see that a barely audible sound
of power 1W has volume LdB = 10 · log10 (1) = 10 · log10 (100 ) = 0
decibels, whereas a damaging sound of power 1,000,000,000,000 W has
volume LdB = 10 · log10 (1, 000, 000, 000, 000) = 10 · log10 (1012 ) = 10 ·
12 = 120 decibels.

14There is also a unit called bel, which equals 10 decibels and is now rarely
used. It was defined as log10 (P ). In a sense, bels are more similar to the definition
of the Richter magnitude, but for some reason, it is now common practice to use
the unit decibels.
 Bibliography

[1] Gross, Herbert I., Algebra by Example: An Elementary Course, D. C. Heath

& Co., Lexington, Mass., 1978 (out of print).
[2] Milgram, R. James, The Mathematics Pre-Service Teachers Need to Know,
available on-line at
ftp://math.stanford.edu/pub/papers/milgram/FIE-book.pdf, 2005 (es-
pecially “Chapter 8: Discussion of Issues in the Algebra Course”).
[3] Papick, Ira J., Algebra Connections: Mathematics for Middle School Teachers,
Prentice Hall, Upper Saddle River, NJ, 2007.
[4] Parker, Thomas H. and Baldridge, Scott J.. Elementary Mathematics for
Teachers, Sefton-Ash Publishing, Okemos, Mich., 2004.
[5] Szydlik, Jennifer and Koker, John, Big Ideas in Mathematics for Future Mid-
dle Grades Teachers: Big Ideas in Algebra, University of Wisconsin-Oshkosh
(preprint).
[6] Wu, Hung-Hsi, Introduction to School Algebra [DRAFT], available on-line at
http://math.berkeley.edu/~wu/Algebrasummary.pdf, 2005.

148
 Index

Abel, Niels Henrik, vii constant, 11

absolute value, 86 converse (of a statement), 34
additive change, 130 coordinate, 30
additive identity, law of the, 8 coordinate axis, 30
additive inverse, 5 coordinate system, 30
additive inverse, law of the, 8 cube (of a number), 124
Al-Kitab al-Jabr wa-l-Muqabala, vi cubic equation, vii
algebraic expression, 12
evaluating, 12 decadic logarithm, 140
identical, 13 decibels, 147
negative, 66 decimal
nonzero, 45 non-repeating, 16
positive, 65 repeating, 16
rewriting, 12 del Ferro, Scipione, vii
algorithm, vi difference (of sets), 17
argument (of a function), 74 difference function
associative law (of addition), 8 first, 93
associative law (of multiplication), 8 second, 95
direct variation, 89
bacterial growth, 128 direction (of velocity), 23
base, 124 discriminant, 105
binomial laws, 98 distance (on the number line), 88
distributive law, 8
Cardano, Gerolamo, vii domain (of a function), 73, 74
ceiling function, 83
change e (Euler’s number), 135
additive, 130 earthquake, 146
multiplicative, 130 elimination method, 54
codomain (of a function), 74 empty set, 16
coefficient, 102 equality, properties of, 61
common logarithm, 140 equation, 13
commutative law (of addition), 8 equivalent, 43
commutative law (of multiplication), Euler’s number, 135
8 exponent, 124
completing the square, 103 exponential equation, 131, 138
complex numbers, 16, 112 exponential function, 131
compounding, 136 exponential rules, 128
149
150 INDEX

exponentiation logarithmic rules, 142

definition, 124
main diagonal, 80
Ferrari, Lodovico, vii map, 74
floor function, 83 mapping, 74
fractions, 7 measurement division, 8
function, 72, 74 Muhammad ibn Mūsā al-Khwārizmı̄,
vi
gradient, see also slope multiplicative change, 130
graph, 25, 31, 75 multiplicative identity, law of the, 8
greatest integer function, 83 multiplicative inverse, 7
multiplicative inverse, law of the, 8
half-life, 144
higher algebra, vi natural logarithm, 140
history of algebra, vii numerical expression, 8
identity, 13
one-to-one (1–1) function, 78
imaginary unit i, 111
onto function, 78
index (of exponentiation), 124
origin, 30
inequality, 59
integers, 4, 16 pair
intercept, 34, 35 ordered pair of reals, 30
interest, 136 parabola, 116
interest rate, 136 parameter, 10
intersection, 17 partitive division, 8
interval, 16 pattern, 92
inverse function, 79 polynomial division, 106
inverse variation, 89 principal, 136
inversely proportional, 89 proportional, 89
least integer function, 83 quadratic equation, vii, 98
letters in algebra, 8 general solution, 105
line equation quadratic formula, 105
general form, 37 quadratic function, 95
intercept form, 39 quadratic inequality, 121
point-slope form, 36 quartic equation, vii
slope-intercept form, 34, 35 quintic equation, vii
two-point form, 37
linear equation, 30 range (of a function), 78
general, 48 rate (of change), 24
linear equations constant, 24
simultaneous, 52 ratio, 21
linear expression, 29 rational numbers, 7, 16
linear function, 82 real numbers, 16
linear inequality, 64 redursion
logarithm definition by, 94
common, 140 Richter scale, 146
decadic, 140 root, 126
natural, 140
logarithmic function, 139 school algebra, vi
 INDEX 151

set, 15
slide rule, 143
slope, 33, 35
solution set, 67
speed, 23
square (of a number), 124
square root, 109
“step” function, 83
substitution method, 54

Tartaglia, Niccolò Fontana, vii

union, 17
unknown, 9

value (of a function), 74

variable, 9
dependent, 9, 72
independent, 9, 72
velocity, 23
vertex, 116
Viète, François (Vieta), vii

whole numbers, 4, 16