A Gentle Introduction To The American Invitational Mathematics Examination

A Gentle Introduction to the American Invitational
Mathematics Examination
Scott A. Annin
California State University, Fullerton
March 16, 2014

ii
Contents
1 Algebraic Equations 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Distance-Rate-Time Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Systems of Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2 Combinatorics 15
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 Permutations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.1 Permutations with Repetition . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.2 Permutations without Repetition . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 Combinations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.1 Combinations without Repetition . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.2 Combinations with Repetition . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.4 More Challenging Combinatorics Problems . . . . . . . . . . . . . . . . . . . . . . . 23
2.5 Some Combinatorial Identities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.6 The Inclusion-Exclusion Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3 Probability 43
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.2 Properties of Probability Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
iii
iv CONTENTS
3.3 Examples: Tournaments, Socks, and Dice . . . . . . . . . . . . . . . . . . . . . . . . 49

3.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4 Number Theory 57
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.2 Fundamental Theorem of Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.3 Greatest Common Divisor and Least Common Multiple . . . . . . . . . . . . . . . . 62
4.4 Modular Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.5 Divisibility Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5 Sequences and Series 75

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.2 Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.2.1 Arithmetic Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.2.2 Geometric Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.3 Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.3.1 Arithmetic Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.3.2 Geometric Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.4 Some Other Useful Sequences and Series . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.5 Additional Examples of Sequences and Series . . . . . . . . . . . . . . . . . . . . . . 90
5.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
6 Logarithmic and Trigonometric Functions 97

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.2 Logarithmic Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.3 Trigonometric Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
6.4 Putting Logarithmic and Trigonometric Functions Together . . . . . . . . . . . . . . 116
6.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
7 Complex Numbers and Polynomials 121

CONTENTS v
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

7.2 The Algebra of Complex Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
7.3 The Geometry of Complex Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
7.4 Basic Definitions and Facts about Polynomials . . . . . . . . . . . . . . . . . . . . . 133
7.5 Polynomials with Complex Roots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
7.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
8 Plane Geometry 149

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
8.2 Triangles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
8.2.1 Congruent and Similar Triangles . . . . . . . . . . . . . . . . . . . . . . . . . 162
8.2.2 Area of a Triangle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
8.3 Circles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
8.4 Geometrical Concepts in the Complex Plane . . . . . . . . . . . . . . . . . . . . . . . 176
8.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
9 Spatial Geometry 183

9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
9.2 Rectangular Boxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
9.3 Cylinders, Cones, and Spheres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
9.4 Tetrahedra and Pyramids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
9.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
10 Hints for the Exercises 193

10.1 Hints for Chapter 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
10.1.1 Hints to Get Started . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
10.1.2 More Extensive Hints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
vi CONTENTS

11 Solutions to Exercise Sets 217

11.1 Chapter 1 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
Preface
A Gentle Approach to the American Invitational Mathematics Examination is a celebration of math-

ematical problem solving at the level of the high school American Invitational Mathematics Exami-
nation (AIME). The book is intended, in part, as a resource for comprehensive study for the AIME
competition to be used by students, coaches, teachers, and mentors preparing for the exam.
However, this manuscript is also written for anyone who enjoys solving problems, exploring recre-
ational mathematics, and sharpening their critical thinking skills. Readers who study this book are
destined to see their problem solving skills enhanced. It is not intended just for the exceptionally
brilliant or for those who are intensely competitive, even though the AIME is a competition. Rather,
anyone who enjoys thinking about interesting problems in mathematics should enjoy this book.
One of the author’s aims in writing this book is to make the problems accessible and interesting to
a wide audience and to lure in those who feel perhaps that the AIME problems are not for them.
Therefore, even if you are not involved in the AIME as a student or mentor or do not have confidence
in your own skills in mathematical problem solving, as long as you have an interest mathematical
problem solving and a desire to explore fun mathematical problems at the high school level, you
should find this book engaging. You may surprise yourself with what you can do!
The AIME is the second in a series of exams known collectively as the American Mathematics
Competitions (AMC). In the next couple of pages, we will describe the series of exams comprising
the AMC, including the focus of this manuscript, the AIME, and offer some words of advice and
encouragement. After that, the book will survey a large compendium of topics and problems from
past AIMEs that will form the basis for learning, discussion, and discovery.
The American Mathematics Competitions

In 1950, the Mathematical Association of America launched the AMC as a means of “strengthening
the mathematical capabilities of our nation’s youth”. In 2008, nearly 60 years later, almost a
half-million middle and high school students across the United States participated in the program,
representing over 5,000 schools.
The AMC consists of a series of examinations covering a broad spectrum of topics. The problems
appearing in these examinations can all be solved by using precalculus methods. The exams become
progressively more difficult, beginning with the AMC 10 and AMC 12 in February each year1 ,
1 There is also an AMC 8 exam given each year in November that is only open to students in eighth grade or
vii
viii CONTENTS
followed by the AIME in March or early April, the United States of America Mathematical Olympiad
(USAMO) later in April, and the International Mathematics Olympiad (IMO) in July. With the
help of a three-week Mathematical Olympiad Summer Program (MOSP) hosted at the University
of Nebraska each June for the six members of the United States’ IMO team, the United States has
performed exceptionally well in the IMO. The United States fielded its first IMO team in 1974, and
placed second. The number of countries participating in the IMO is usually between 80 and 90. The
United States has placed first in total team score at the IMO four times and, on 26 occasions, has
finished in the top three.
The goals of the AMC, however, reach far beyond success in the IMO. There have been disturbing
trends in the United States in recent years in mathematics education at the middle and high school
levels. Many students in the U.S. under-perform in mathematics compared to their counterparts in
other countries. The AMC examinations are designed to involve students of all mathematical ability,
not just the brightest few. The AMC 10 and AMC 12 exams, in particular, contain some problems
that do not require extraordinarily creative thought. The hope is that such problems can serve as
a confidence boost to many students, and that participation in the exam can foster enthusiasm for
mathematics in general. The problems, by and large, are fun and engaging. They usually do not
require heavy machinery or excessive computations. Instead, they call on a student to analyze ideas,
to think logically, and in the case of some of the harder problems, to find a clever approach.
About the AMC 10, AMC 12, and the American Invitational Mathematics Exam
The AMC 10 exam is open to all students in 10th grade or below, while the AMC 12 is open to all
students in 12th grade or below. Both exams consist of 25 multiple-choice problems to be solved
without a calculator in 75 minutes. Students receive 6 points for each correct answer, 1.5 points for
each blank answer, and no points for a wrong answer. The AMC 10 and AMC 12 exams are both
offered twice every year. Therefore, an 11th or 12th grader can attempt the AMC 12 exam twice,
while a 10th grader could attempt the AMC 10 twice, the AMC 12 twice, or take each exam once.
Students who score at least 120 points or finish in the top 2.5% on the AMC 10 exam or who
score at least 100 points or finish in the top 5% on the AMC 12 exam are invited to take the
American Invitational Mathematics Exam. The AIME is a 15 problem exam in which each answer
is an integer in the range 000-999, inclusive. As with the AMC 10 and AMC 12 exams, the AIME
competition is offered twice each year. Throughout this book, problems from the first version of a
given year’s exam with be denoted by “AIME” and problems from the second version of a given
year’s exam will be denoted by “AIME-2”. Since students mark their answers on a scanned answer
sheet, it is imperative that all three digits of the answer are accurately recorded. This includes the
possibility of leading zeros in the answer. For instance, the answer 46 should be marked as 046 on
the answer sheet. Students have three hours to complete the exam, and no calculators are allowed.
The problems appearing on the AIME are noticeably more challenging than those on the AMC
10 and AMC 12 exams overall, but still, they do not require knowledge of calculus. Some of the
most common themes in the AIME problems are algebra, combinatorics (i.e. counting), probability,
number theory, complex numbers, polynomials trigonometry, and geometry. However, any topic at
the high school level is fair game, and problems on the AIME often require students to use ideas
from multiple mathematics disciplines to obtain the solution.
younger. That exam is not the subject of this volume.

CONTENTS ix
Students who achieve high scores on both the AMC and AIME may then be selected to compete in
the United States of America Mathematical Olympiad (USAMO). Selection criteria are not fixed,
but roughly 500 students nationwide are chosen each year. The USAMO is a six problem essay
competition spanning two days. The problems contained in the USAMO are extremely challenging
and require great ingenuity and creative thinking. From the USAMO participants, six are chosen to
represent the United States at the International Mathematics Olympiad (IMO) each summer. Several
texts and resources have been developed on the subject of the USAMO and IMO competitions. The
AIME, however, has been comparatively under-represented in materials and resources for coaches,
teachers, and students. The present volume aims to make some headway in this area by providing
a resource for teachers and students preparing for the AIME.
An Integer Answer from 000 to 999

We indicated that every answer to every question appearing on the AIME is a three digit integer in
the range 000 to 999. At first, this may seem to place severe limitations on the breadth of questions
that are suitable for the exam. However, there are several standard ways that questions whose
answers do not meet this criterion are typically modified so that the answers do comply. Here are
just a few:
• If the actual answer x is a whole number larger than 999, the question can be modified to ask
for the left-most three digits of x.
• Alternatively, if the actual answer x is a whole number larger than 999, the question might
ask for the remainder when x is divided by 1000.
• For some negative answers x, the question can simply ask for the absolute value, |x|.
• If the actual answer x is a real number in the interval [0, 1000], the question might ask for
either the ceiling of x, dxe, or the floor of x, bxc. Recall that
dxe denotes the smallest integer greater than or equal to x
and
bxc denotes the largest integer less than or equal to x.
x
• If the actual answer is a fraction, say , the question will usually ask the solver to first simplify
y
x m
the fraction to “lowest terms”, = , where m and n are relatively prime integers with n > 0.
y n
This is accomplished by removing all common factors that arise in both the numerator x and
denominator y. Finally, the question can ask for some integer combination of m and n (such
as m + n). The subject of relatively prime integers is taken up in Chapter 4 of the book (see
Section 4.3). Reducing the fraction as described above is necessary in order to ensure the
problem has a unique answer.
√
• Some AIME questions will ask for the value of m appearing under a radical, such as m. In
such cases, it is customary for the question to require us to√pull out√from under the radical
any square factors that occur. For instance, we can rewrite 75 as 5 3. The same applies to
x CONTENTS
√ √ √
any root. For instance, we can rewrite 4 162 = 4 81 · 2 = 3 4 2. In general, by extracting all
factors
√ of the form pk (for a prime p and positive integer k ≥ 2) from an expression of the form
k
N , we can obtain a unique form for the answer to an AIME problem.
√ Here is an excerpt
m−n
from an example from Chapter 5: “...can be written in the form , where m, n, and p
p
are positive integers and m is not divisible by the square of any prime. Find 100m + 10n + p.”
The requirement that m not be divisible by the square of any prime ensures its unique value,
as any square factors must be extracted and then simplified with n and p.
Even though the answer to every AIME problem is an integer in the range 000–999, AIME
competitors should still be prepared to do hand calculations involving approximations of com-
monly seen mathematical values. While it is impossible to give a complete list, we briefly
enumerate a few of these values here:
x Approximate Value of x
e 2.71828
π 3.14159
ln 2 0.69315
ln
√3 1.09861
√2 1.41421
3 1.73205
The number of decimal places required to obtain a sufficiently accurate approximation depends
on the problem, but certainly two or three digits to the right of the decimal point would be
an excellent start. This is by no means a complete list, √ but it is a start and can help to
approximate other needed values as well, such as π 2 or 6 or ln 8, and so on.
The Structure of this Book

The first nine chapters in the book are arranged according to mathematical topic, sometimes broken
into sections, each of which begins with some of the important facts and theorems about that topic
that are commonly required to solve AIME problems. These summaries are not intended to serve
as a review of everything important about the topic, but rather, attention has been focused on
those aspects that are likely to arise in solving AIME problems. Moreover, in an effort to keep the
discourse focused on problem solving and avoid extending the length of the manuscript, it is not our
intention to prove rigorously all of the results that are summarized herein. Of course, understanding
why a result is true can often assist the problem solver in applying it effectively, and so far as this can
be achieved, some informal discussion of the results will usually be given. Throughout the chapters,
some problems that relate to the topic at hand are provided, along with solutions. Most of these
problems are taken directly from past AIMEs. They have been carefully chosen to display a wide
range of strategies, subtopics, and difficulty. It is worth noting that many AIME problems require
ideas from multiple branches of mathematics. We have made every effort in this book, however, to
place each problem in the chapter of the topic deemed most crucial in understanding the solution.
Finally, each chapter concludes with a collection of additional problems on the topic from past
AIMEs. Again, a variety of exercises have been chosen, generally of gradually increasing difficulty.
Readers are encouraged to try to work out solutions to these problems.
CONTENTS xi
For readers who are not sure how to begin on a particular exercise, Chapter 10 is a collection of
hints for the exercises that are designed to help the reader get started. It is always more beneficial
for an individual to arrive at an independent solution to a problem, rather than simply reading a
published solution. The hints provide the reader an opportunity to enjoy some assistance getting
started, but without spoiling the full solution.
The hints provided are divided into two groups. First, each problem has a simple hint designed
merely to get the reader started on one possible track towards a solution. Such hints are usually
not too detailed, but rather, provide just enough information to give the reader an idea where to
begin. These hints appear in the subsection “Hints to get Started”. Then, second, each problem also
has a more significant hint (also in Chapter 10) to guide the reader who would benefit from a more
comprehensive strategy and plan of solution. These are contained in the subsection “More Extensive
Hints”. The reader seeking guidance on how to solve a problem in the exercises is encouraged first
to consult only the “Hints to get Started” and then review the “More Extensive Hints” if the first
hint is insufficient. In this way, we hope that the book, with its layers of hints and advice, is viewed
as “Gentle”, as the title suggests.
Full solutions to all exercises are contained in the final chapter, Chapter 11. Of course, many
problems have multiple solutions, and any solution that leads to the correct answer is a good one,
even if it is not the one (or, in some cases, two) given in this book. Readers should resist the
temptation to merely read solutions. For readers seeking to hone problem solving skills, it is better
for the reader to take hints (Chapter 10) or, if they must, simply peek at the solution in small
doses at a time, all the while trying to engender as much of the solution as they can on their own.
Other readers who are taking a more leisurely interest in the material in the book may be content to
enjoy the solutions in this manuscript more freely – solutions have been written to be as accessible
as possible, often with an eye towards increased accessibility and understandability over and above
unlikely or unappreciated cleverness.
Advice for Success on the AIME

The casual reader who is using the AIME as a source of fun and recreation can view the present
manuscript as a tour guide for a mathematical vacation. For more serious AIME contenders and
competitors, however, the road to success requires a lot of practice. That, too, can be found within
these pages. While every problem in the book is solved, the AIME student has an opportunity to
approach each one as a challenge and as a source of training.
The practice afforded by this book can be experienced by an individual in isolation if desired, but
there are also ways to enhance the experience of problem solving by involving others in the learning
process. Indeed, math clubs or workshops at school could devote some time to the problems, perhaps
turning AIME problems into a little team competition event. Alternatively, teachers could challenge
their students with a problem or two out of this book from time to time.
The bottom line is that the AIME problems have the power to change the culture of high school
mathematics. AIME preparation doesn’t have to take a back seat to the mundane world of high
school curriculum standards. There is room for fun and exploration, and some kids are screaming
for it (by screaming about a boring, rigid structure of learning in their mathematics classes), while
others might be if only they knew what they are missing. Let us use the AMC program, including the
AIME, to ignite a fire within today’s young people, old people, and innovative thinkers everywhere
xii CONTENTS
who love mathematics or want to grow to love mathematics.
A Word from the Author

If you are reading this book, it means you already have an interest in mathematics competitions,
problem solving, creative thinking, and the beauty of mathematics. For that, I applaud you. As
a former AIME contestant and current workshop leader and AIME promoter, I can affirm the
value of participating in this competition or spending time to ponder the problems. Regardless of
your involvement or outcome on the exam, or even if you never enroll as an AIME contestant, the
exposure to fun and interesting mathematics, the excitement of discovery, and the time and energy
you invest into problems appearing on the AIME all serve to strengthen you as a problem solver, a
critical thinker, and a lover of mathematics. The AIME requires you to pursue ideas creatively, and
this effort will take you far beyond the normal expectations of the typical high school mathematics
classroom. By exploring the problems contained in the AIME, you are honing your mathematical
skills for a lifetime, and you are challenging yourself to be the best you can be in mathematics.
In writing this book, I have made every effort to present the mathematics as clearly as possible, in
a way that I myself would teach it in front of a room full of students. I have not always opted to
present the shortest, neatest, or most clever solution to a problem. Rather, I have opted always to
present a solution that is as accessible as possible, one that a typical student can understand and
might say “I could have come up with that”. In this way, students will be less likely to confront
discouragement by a solution that they feel is out of reach and hard to imagine themselves finding
on their own.
I have tried to avoid reliance on heavy machinery, deep theorems, or undue abstraction. I have
intentionally kept the mathematical proofs in this book in the background, unless they truly reveal
insight into how to solve problems and think about the subject matter that they address. I have
written each solution you find in this volume on my own, following my own logic, notations, and
ideas. In some cases, I have presented more than one solution to a given problem if there are
interesting or valuable insights gained by doing this. It is quite possible that the reader may from
time to time discover a different solution to problems in this book that they prefer over the ones
I presented. That is, of course, perfectly acceptable and expected, as different people may prefer
different solutions. It is possible to find other published solutions to the AIME problems in this
book either by contacting the American Mathematics Competitions office or via the “Art of Problem
Solving” website, www.artofproblemsolving.com.
Acknowledgements
I am grateful to reviewers who have found typographical or mathematical errors in the book and
pointed them out to me. I have made every effort to write the exposition carefully and cleanly,
but for whatever errors that remain in the manuscript, I take full responsibility. I am grateful to
my home institution, California State University, Fullerton, for the support and encouragement to
produce this book. I am also grateful to the support I received from the office of the American
Mathematics Competitions in Lincoln, Nebraska. I received much advice and encouragement, as
well as countless resources to use in preparing the book, from AMC Director Dr. Steven Dunbar
and his helpful and always cheerful and ready-to-help staff. Many thanks also to Don Albers of the
CONTENTS xiii
Mathematical Association of America, for his encouragement to produce this book and his confidence
in me to do so.
I also wish to express my gratitude to colleague Steven Davis who kindled my interest in the AIME
as my co-speaker and co-author in a variety of venues devoted to problem solving over the past
several years. My journey with the AIME began as a student, and I must acknowledge the support
and encouragement that two of my Lincoln East Junior and Senior High School teachers, Leona
Penner and Jerry Beckmann, gave me for several years of my training. Without their contagious
enthusiasm, I would not be the AIME lover or the mathematician that I am today.
I have been given the opportunity to present many of the solutions in this book at workshops
and conferences, and I am grateful for this chance to expose them to a broad audience. Venues
that have been especially open to these presentations include the National Council of Teachers of
Mathematics, the California Mathematics Council, and the Problem Solving Seminar at California
State University, Fullerton. For the latter participation, I gratefully acknowledge my colleague and
friend, Dr. Bogdan Suceava.
Finally, I wish to express my forever thankfulness to the two people who have loved me and supported
me throughout all of my endeavors in life, my parents Arthur and Juliann Annin. They greeted me
at their doorstep with open arms when I came to stay with them during my sabbatical leave. Much
of the actual work on this project was completed while I was under their roof. For their endless love
and encouragement, I am forever in their debt. Mom and Dad, you are my best friends, and I love
you.
Chapter 1
Algebraic Equations
“Your problem may be modest; but if it challenges your curiosity and brings into play
your inventive faculties, and if you solve it by your own means, you may experience
the tension and enjoy the triumph of discovery. Such experiences at a susceptible age
may create a taste for mental work and leave their impact on mind and character for
a lifetime.”
- Polya, “How to Solve It”
1.1 Introduction
Mathematics is a language. It consists of symbols and notation that must be established and agreed
upon before meaningful communication can occur. If mathematics is to be useful in addressing real-
world problems, it must be able to transcend the language we speak in. That is, the problems we
wish to solve must be interpreted, translated into mathematical language, and solutions converted
back into the language we speak in. Many problems in the AIME competition require precisely these
steps to be carried out.
One of the main components of the language of mathematics is the equation. An equation is an
entity that involves numbers, variables, and relationships among them. Equations commonly arise
in every branch of mathematics, including all of the subjects covered in this book: number theory,
geometry, algebra, trigonometry, polynomials, and so on. Therefore, before we examine AIME
problems that involve these other mathematical topics, it will be helpful to have a chapter devoted
first and foremost to the study and solution of algebraic equations, without dealing with any deeper
mathematical notions underlying the problems. At the same time, this will give us an opportunity to
practice the language of mathematics; that is, we will practice converting problems into equations,
solving those equations, and then converting the solutions back into useable real-world information.
Here is a good example to illustrate the above strategy.
1
2 CHAPTER 1. ALGEBRAIC EQUATIONS
Example 1.1.1. (2009 AIME-2, Problem #1) Before starting to paint, Bill had 130 ounces of
blue paint, 164 ounces of red paint, and 188 ounces of white paint. Bill painted four equally sized
stripes on the wall, making a blue stripe, a red stripe, a white stripe, and a pink stripe. Pink is a
mixture of red and white, not necessarily in equal amounts. When Bill finished, he had equal amounts
of blue, red, and white paint left. Find the total number of ounces of paint Bill had left.
Solution: We must introduce some variables in order to manufacture mathematical equations from
the information given in the problem. These variables should represent either known or unknown
quantities important to the problem. What variables do we need here? The amount of paint (in
ounces) required to paint one of Bill’s stripes is an unknown quantity. Let us call it x. Since the
pink stripe is a mixture of some red paint and some white paint, let r denote the amount of red
paint (in ounces) used in the pink stripe, and let w denote the amount of white paint (in ounces)
used in the pink stripe.
Now we can step up some equations that express relationships among the variables. First,
x = r + w, (1.1)
since the total amount of paint used for the pink stripe is x, which is the sum of the contribution of
red paint, r, and the contribution of white paint, w.
Next, since Bill ends up with equal amounts of blue, red, and white paint, we have
130 − x = 164 − x − r = 188 − x − w.
It may be more transparent for some readers to see this chain of equalities written as three separate
equalities:
130 − x = 164 − x − r, (1.2)
130 − x = 188 − x − w, (1.3)
164 − x − r = 188 − x − w. (1.4)
From Equation (1.2), we find that r = 34, and from Equation (1.3), we find that w = 58. Therefore,
using Equation (1.1),
x = r + w = 34 + 58 = 92.
Thus, Bill ends up with 130 − x = 130 − 92 = 38 ounces of blue paint. Moreover, since he finishes
with the same amount of blue, red, and white paint, he has 38 ounces of each color of paint, for a
total of 3 · 38 = 114 ounces of paint. ANSWER : 114. 2
Remark: We actually did not even use Equation (1.4) here, but this is a regular phenomenon
in problem solving. Students are often confronted with information beyond that which is needed
to solve a problem. It is simply a case of a toolbox that contains more tools than one actually
needs in order to solve the problem at hand, and this is indeed a common situation that occurs
not just in mathematics, but in real life. One challenge of problem solving is to decide which tool
is appropriate to use in which circumstance. In this case, no harm would be done if we had also
contemplated Equation (1.4), but it would have perhaps unnecessarily complicated the solution we
presented above.
1.1. INTRODUCTION 3
The high school mathematics curriculum includes a firm grounding in percentages. Not surprisingly,
therefore, we find problems about percentages well-represented in the AIME competition. These
problems usually lead to algebraic equations that must be solved. Our next example provides an
illustration. Note that, in contrast to Example 1.1.1 above, in this case the necessary variables
required for solving have already been introduced in the statement of the problem.
Example 1.1.2. (2011 AIME, Problem #1) Jar A contains four liters of a solution that is 45%
acid. Jar B contains five liters of a solution that is 48% acid. Jar C contains one liter of a solution
that is k% acid. From jar C, m/n liters of the solution is added to jar A, and the remainder of the
solution in jar C is added to jar B. At the end both jar A and jar B contain solutions that are 50%
acid. Given that m and n are relatively prime1 positive integers, find k + m + n.
Solution: We can keep track of how much acid is in each jar before and after the solution in jar C
has been reallocated. Jar A begins with (0.45)(4) = 1.8 L of acid. Jar B begins with (0.48)(5) = 2.4
L of acid. The reallocation of solution from jar C adds
m k km
=
n 100 100n
liters of acid to jar A, and it adds

m k k km
1− = −
n 100 100 100n
m
liters of acid to jar B. Thus, after reallocation, jar A contains 4 + liters of solution, of which
n
km
1.8 + is acid. Since we are given that 50% of the final solution in jar A is acid, we conclude
100n
that m km
(0.5) 4 + = 1.8 + . (1.5)
n 100n
Applying the same reasoning with jar B, we conclude that
m k km
(0.5) 6 − = 2.4 + − . (1.6)
n 100 100n
km
Both Equations (1.5) and (1.6) contain the expression , so solving each for this expression and
100n
equating the results, we have
m k m
(0.5)(4 + ) − 1.8 = 2.4 + − (0.5) 6 − . (1.7)
n 100 n
In simplifiying Equation (1.7), we see that m/n cancels, leaving us to solve easily for k. We obtain
m 2
k = 80. Replacing this value of k in Equation (1.5), we quickly find that = , which is consistent
n 3
with the requirement that m and n be relatively prime positive integers. Therefore, we have m = 2
and n = 3. We conclude that k + m + n = 80 + 2 + 3 = 85 = 085. ANSWER : 085. 2
1 The notion of relatively prime positive integers is discussed in Chapter 4.
1.2 Distance-Rate-Time Problems
One of the popular classes of word problems in the AIME involves an equation well-known from
physics. Problems involving the relationship between the distance travelled by an object in a certain
period of time moving at a specified rate (or velocity) are quite common in the AIME competition,
not to mention day-to-day life. The basic relationship is given by the formula
Distance = Rate × Time . (1.8)
Caution: One must take care to ensure consistency of units in applying this formula. For instance,
if an object has a rate (or velocity) measured in miles per hour, and if the time over which the object
travels is given in seconds, then one must either convert the time from units of seconds into hours,
or convert the rate from units of miles per hour into miles per second. Precisely this issue will arise
in Examples 1.2.2 and 1.2.3 below.
Let us now explore some problems that apply Equation (1.8).
Example 1.2.1. (2008 AIME, Problem #3) Ed and Sue bike at equal and constant rates.
Similarly, they jog at equal and constant rates, and they swim at equal and constant rates. Ed covers
74 kilometers after biking for 2 hours, jogging for 3 hours, and swimming for 4 hours, while Sue
covers 91 kilometers after jogging for 2 hours, swimming for 3 hours, and biking for 4 hours. Their
biking, jogging, and swimming rates are all whole numbers of kilometers per hour. Find the sum of
the squares of Ed’s biking, jogging, and swimming rates.
Solution: Let us begin by establishing some variables to represent the relevant rates (in kilometers
per hour). Let b denote the biking rate of Ed and Sue, let j denote the jogging rate of Ed and Sue,
and let s denote the swimming rate of Ed and Sue. Using Equation (1.8), we have
2b + 3j + 4s = 74 and 4b + 2j + 3s = 91. (1.9)
Therefore, we have three unknowns, but only two equations. Normally, such a system of linear
equations cannot have a unique solution for b, j, and s, but because we are also given that b, j, and
s are whole numbers, it will turn out that a unique solution is forthcoming.
We must compute b2 + j 2 + s2 . The strategy is to eliminate one of the variables by using the two
equations above. For instance, by doubling the first equation, we can solve for 4b in each equation
and set the results equal to each other:
148 − 6j − 8s = 4b = 91 − 2j − 3s.
Hence,
4j + 5s = 57, (1.10)
or
4j = 57 − 5s. (1.11)
Equation (1.10) will clearly have a limited number of solutions in positive whole integers for j and
s. Furthermore, in Equation (1.11), note that 57 − 5s will be a positive multiple of 4 only if s = 1, 5,
or 9. The corresponding values of j are j = 13, 8, or 3, respectively. For each pair (s, j), we can
solve for b by using either equation in (1.9). The results are summarized in the table below:
1.2. DISTANCE-RATE-TIME PROBLEMS 5
s j b
1 13 15.5
5 8 15
9 3 14.5
Only the second option consists entirely of whole number values, so we conclude that s = 5, j = 8,
and b = 15. The sum of the squares of these rates is
s2 + j 2 + b2 = 25 + 64 + 225 = 314.
ANSWER : 314. 2
Example 1.2.2. (2012 AIME, Problem #4) Butch and Sundance need to get out of Dodge. To
travel as quickly as possible, each alternates walking and riding their only horse, Sparky, as follows.
Butch begins by walking while Sundance rides. When Sundance reaches the first of the hitching posts
that are conveniently located at one-mile intervals along their route, he ties Sparky to the post and
begins walking. When Butch reaches Sparky, he rides until he passes Sundance, then leaves Sparky
at the next hitching post and resumes walking, and they continue in this manner. Sparky, Butch,
and Sundance walk at 6, 4, and 2.5 miles per hour, respectively. The first time Butch and Sundance
meet at a milepost, they are n miles from Dodge, and they have been traveling for t minutes. Find
n + t.
Solution: We can look at the n mile journey from Dodge as consisting of two parts, one part walking
and one part riding, for each of the two characters in the problem. Let us begin by introducing the
variable k to denote the number of miles that Butch rides Sparky. (Of course, the choice of Butch,
rather than Sundance, is arbitrary here.) This means that Butch walks n − k miles. Therefore,
Sundance has walked for k miles and ridden Sparky for n − k miles. We can compute the total
amount of time T (in hours) that has elapsed from the point of view of either Butch’s journey or
Sundance’s journey. Be careful to note that the variable t given in the problem is measured in
minutes, and therefore, t = 60T , since T is measured in hours.
From Butch’s point of view, if we denote the time (in hours) elapsed with Butch walking as TBW
and the time (in hours) elapsed with Butch riding Sparky as TBR , then we have
n−k k
T = TBW + TBR = + . (1.12)
4 6
Similarly, if we denote the time (in hours) elapsed with Sundance walking as TSW and the time (in
hours) elapsed with Sundance riding Sparky as TSR , then we have
k n−k
T = TSW + TSR = + . (1.13)
2.5 6
Equating (1.12) and (1.13), we find that
n−k k k n−k
+ = + .
4 6 2.5 6
Rearranging, this becomes
n−k 7k
= .
12 30
Eliminating fractions, we obtain

30(n − k) = 84k,
or
30n = 114k.
Cancelling a common factor of 6 from both sides, we obtain 5n = 19k. Since n must be a whole
number, this equation shows that n must be a multiple of 19. Thus, the smallest possible value of n
is n = 19, with corresponding value k = 5. Substituting these values into Equation (1.12) or (1.13)
above, we conclude that
13
T = ,
3
but this is measured in hours. Thus,
t = 60T = 260.
Hence, the final answer is
n + t = 19 + 260 = 279.
ANSWER : 279. 2
Example 1.2.3. (2008 AIME, Problem #12) On a long straight stretch of one-way single-lane
highway, cars all travel at the same speed and all obey the safety rule: the distance from the back
of the car ahead to the front of the car behind is exactly one car length for each 15 kilometers per
hour of speed or fraction thereof. (Thus the front of a car travelling 52 kilometers per hour will be
four car lengths behind the back of the car in front of it.) A photoelectric eye by the side of the
road counts the number of cars that pass in one hour. Assuming that each car is 4 meters long and
that the cars can travel at any speed, let M be the maximum whole number of cars that can pass the
photoelectric eye in one hour. Find the quotient when M is divided by 10.
Solution: To maximize the whole number of cars that can pass the photoelectric eye in a given time
frame, the speed of each car should be a multiple of 15 kilometers per hour. Note that the rates given
in the problem are measured in kilometers per hour, but the car length is given in meters. Therefore,
we shall convert the rates into units of meters per hour. Let us say that the cars travel with a speed
of 15,000k meters per hour. In this case, the distance between the front of two consecutive cars will
be 4(k + 1) meters, since there are k car lengths between the back of one car and the front of the
next car. Using Equation (1.8), we compute the time that elapses between consecutive cars that
pass the photoelectric eye:
Distance 4(k + 1)
Time = = hours per car.
Velocity 15000k
To obtain the number of cars per hour that pass the photoelectric eye, we take the reciprocal:
15000k 3750k
Number of Cars per Hour = = .
4(k + 1) k+1
By having the first car passing the eye immediately at the start of the hour, the whole number of
cars that are able to pass the eye in one hour is

3750k
Whole Number of Cars per Hour = ,
k+1
1.2. DISTANCE-RATE-TIME PROBLEMS 7
k
where dxe denotes the “ceiling” of x, the smallest integer greater than or equal to x. Since <1
k+1
but becomes arbitrarily close to 1 by taking k very large, for sufficiently large k we can obtain

3750k
= 3750.
k+1
(In fact, k ≥ 3750 achieves this.) Thus, M = 3750, so that M/10 = 375. ANSWER : 375. 2
Equation (1.8) uses the rate of change of an object’s distance over a period of time to find the total
change in distance during the time interval. If one considers other types of rates of change, it is
possible to obtain other relevant formulas involving the rate of change of some quantity, a given time
interval, and the total change in the quantity in question over the time interval. For instance, we can
consider a group of workers manufacturing some product (e.g. hangers). Their “productivity rate”
might be measured in terms of the amount of product produced per unit time interval (e.g. 100
hangers produced per hour). One can then compute the total amount of the product manufactured
by using the equation
Amount of Product Manufactured = Productivity Rate × Time . (1.14)
If the workers producing hangers labor for five hours, then at a productivity rate of 100 hangers per
hour, they will produce 100 · 5 = 500 hangers during the five hours.
Now let us consider an example from the AIME competition.
Example 1.2.4. (2007 AIME-2, Problem #4) The workers in a factory produce widgets and
whoosits. For each product, production time is constant and identical for all workers, but not nec-
essarily equal for the two products. In one hour, 100 workers can produce 300 widgets and 200
whoosits. In two hours, 60 workers can produce 240 widgets and 300 whoosits. In three hours, 50
workers can produce 150 widgets and m whoosits. Find m.
Solution: Staying consistent to our theme in this chapter, we begin by introducing some variables.
There are a lot of numbers weighing on this problem. It would be useful to simply determine how
long it takes for one worker to make one widget, and the time it takes for one worker to make one
whoosit. Therefore, we proceed as follows.
Let x denote the time (in hours) required for one worker to produce one widget, and let y denote
the time (in hours) required for one worker to produce one whoosit. We wish to write an equation,
in terms of x and y, for the amount of time required (given as one hour) for 100 workers to produce
300 widgets and 200 whoosits. Note that 100 workers can produce one widget in x/100 hours.
Thus, to produce 300 widgets requires 300 · (x/100) = 3x hours. Similarly, 200 whoosits requires
200 · (y/100) = 2y hours. The total time required to produce both is 3x + 2y hours. Thus, we are
given that
3x + 2y = 1.
Similarly, for two hours, the given information tells us that
x y
240 · + 300 · = 4x + 5y = 2.
60 60
The system of equations

3x + 2y = 1 and 4x + 5y = 2 (1.15)
can be easily solved. For instance, we can use the method of elimination by multiplying the first
equation by 4 and the second equation by 3 to get
12x + 8y = 4 and 12x + 15y = 6.
Subtracting the former equation from the latter eliminates x, and we obtain 7y = 2, so that y = 2/7.
Then plugging this result into any of the equations obtained above, we quickly deduce that x = 1/7.
Now, in three hours, we have
x y
150 · +m· = 3.
50 50
That is,
1 m 2
3· + · = 3.
7 50 7
Hence,
150 + 2m
= 3.
350
Therefore,
150 + 2m = 1050,
so that 2m = 900. Hence, we conclude that m = 450. ANSWER : 450. 2
1.3 Systems of Equations
In several examples in the previous two sections, we encountered systems of algebraic equations (for
instance, see Equations (1.2)–(1.4), (1.9), or (1.15)), and we were required to find a simultaneous
solution to this system. The AIME commonly contains problems involving systems of equations, but
the systems can often be much more complicated, involving numerous equations that are sometimes
non-linear. It takes practice and experience with these problems to successfully solve them; few
specific guidelines are available and a bit of cleverness is often required. Let us explore a couple of
examples.
Example 1.3.1. (2000 AIME, Problem #7) Suppose that x, y, and z are three positive numbers
1 1 1 m
that satisfy the equations xyz = 1, x + = 5, and y + = 29. Then z + = , where m and n
z x y n
are relatively prime positive integers. Find m + n.
Solution: The danger in this problem is to find oneself going round in circles trying to plug one of
the given equations into another. Moreover, one should avoid the temptation to try to determine
explicitly the values of x, y, and z. The bit of cleverness that makes this problem much easier is to
multiply the equations
1 1 1 m
x + = 5, y + = 29, z+ =
z x y n
together to obtain
1 1 1 m 145m
x+ y+ z+ = 5 · 29 · = .
z x y n n
1.3. SYSTEMS OF EQUATIONS 9
Expanding the left side, we obtain eight terms:

1 1 1 1 145m
xyz + x + y + z + + + + = .
x y z xyz n
On the left side, we can substitute
1 m 1 1
xyz = 1, = − z, = 5 − x, = 29 − y
y n z x
to obtain m 145m
1 + x + y + z + (29 − y) + − z + (5 − x) + 1 = .
n n
That is,
m 145m
36 + = ,
n n
or
144m
= 36.
n
We conclude that
m 36 1
= = ,
n 144 4
so that m = 1 and n = 4. Thus, m + n = 1 + 4 = 5 = 005. ANSWER : 005. 2
Next we study an example that consists of a larger system, but consisting of linear equations.
Example 1.3.2. (1989 AIME, Problem #8) Assume that x1 , x2 , . . . , x7 are real numbers such
that
x1 + 4x2 + 9x3 + 16x4 + 25x5 + 36x6 + 49x7 = 1 (1.16)
4x1 + 9x2 + 16x3 + 25x4 + 36x5 + 49x6 + 64x7 = 12 (1.17)
9x1 + 16x2 + 25x3 + 36x4 + 49x5 + 64x6 + 81x7 = 123. (1.18)
Find the value of
16x1 + 25x2 + 36x3 + 49x4 + 64x5 + 81x6 + 100x7 . (1.19)
Solution: There are far more variables than equations in this linear system, so it will be impossible
to solve explicitly for each variable x1 , x2 , . . . , x7 . Of course, by taking linear combinations of
these equations2 , we can obtain additional equations. In fact, a promising strategy is to try and
form Equation (1.19) by taking a linear combination of the three given equations (1.16), (1.17), and
(1.18). To see how to do this effectively, note that all coefficients in all equations are perfect squares.
Moreover, if we concentrate on a given variable xi (for some i = 1, 2, 3, 4, 5, 6, 7), observe that the
coefficients of the three given equations take the form n2 , (n + 1)2 , and (n + 2)2 , and the coefficient
of xi in the requested equation is (n + 3)2 . It may be useful, for instance, to line up the terms of
Equations (1.16) – (1.18) one above another into columns. In this context, we are seeking to perform
2 A linear combination of a system of equations is an equation obtained by taking multiples of some (or all) of the
equations in the system and adding the resulting equations together.

addition “vertically” column by column. Therefore, we seek to find constants a, b, and c such that
for all n = 1, 2, 3, . . . , 7, we have
an2 + b(n + 1)2 + c(n + 2)2 = (n + 3)2 .
Expanding this equation, we obtain
(a + b + c)n2 + (2b + 4c)n + (b + 4c) = n2 + 6n + 9.
Equating the coefficients on both sides of the equation, we find that
a + b + c = 1, 2b + 4c = 6, b + 4c = 9.
Using the latter two equations, we find quickly that b = −3 and c = 3. Therefore, from the first
equation, we deduce that a = 1. We conclude that by adding Equation (1.16) to −3 times Equation
(1.17) and 3 times Equation (1.18), we will arrive the desired expression on the left side. Hence,
16x1 + 25x2 + 36x3 + 49x4 + 64x5 + 81x6 + 100x7 = (1)(1) + (−3)(12) + (3)(123) = 1 − 36 + 369 = 334.
ANSWER : 334. 2
Now that we have studied a variety of examples from past AIME competitions that utilize algebraic
equations, we invite the reader to try some additional AIME problems. This chapter, and every
chapter, will conclude with a collection of such problems. We have supplied at least one solution
to each of the problems in Chapter 11. However, the interested reader who runs into difficulty may
prefer to instead consult Chapter 10, which includes two layers of hints. One layer provides the
reader with a brief idea of how to begin thinking about a solution, and the deeper layer provides
the reader with a more guided solution, but leaves the details and some of the more straightforward
aspects to the reader. This may be preferable for some readers over simply reading our solution.
1.4 Exercises
Hints begin on Page 194. Solutions begin on Page 217.
1. (2008 AIME, Problem #1) Of the students attending a school party, 60% of the students
are girls, and 40% of the students like to dance. After these students are joined by 20 more
boy students, all of whom like to dance, the party is now 58% girls. How many students now
at the party like to dance?
2. (2008 AIME-2, Problem #2) Rudolph bikes at a constant rate and stops for a five-minute
break at the end of every mile. Jennifer bikes at a constant rate which is three-quarters the
rate that Rudolph bikes, but Jennifer takes a five-minute break at the end of every two miles.
Jennifer and Rudolph begin biking at the same time and arrive at the 50-mile mark at exactly
the same time. How many minutes has it taken them?
3. (2007 AIME, Problem #2) A 100 foot long moving walkway moves at a constant rate of 6
feet per second. Al steps onto the start of the walkway and stands. Bob steps onto the start
1.4. EXERCISES 11
of the walkway two seconds later and strolls forward along the walkway at a constant rate of 4
feet per second. Two seconds after that, Cy reaches the start of the walkway and walks briskly
forward beside the walkway at a constant rate of 8 feet per second. At a certain time, one of
these three persons is exactly halfway between the other two. At that time, find the distance
in feet between the start of the walkway and the middle person.
4. (1992 AIME, Problem #3) A tennis player computes her win ratio by dividing the number
of matches she has won by the total number of matches she has played. At the start of a
weekend, her win ratio is exactly 0.500. During the weekend, she plays four matches, winning
three and losing one. At the end of the weekend, her win ratio is greater that 0.503. What is
the largest number of matches she could have won before the weekend began?
5. (1993 AIME, Problem #3) The table below displays some of the results of last summer’s
Frostbite Falls Fishing Festival, showing how many contestants caught n fish for various values
of n.
n 0 1 2 3 ··· 13 14 15
number of contestants who caught n fish 9 5 7 23 ··· 5 2 1
In the newspaper story covering the event, it was reported that
• the winner caught 15 fish;

• those who caught 3 or more fish averaged 6 fish each;
• those who caught 12 or fewer fish averaged 5 fish each.
What was the total number of fish caught during the festival?
6. (1986 AIME, Problem #4) Determine 3x4 + 2x5 if x1 , x2 , x3 , x4 , and x5 satisfy the system
of equations given below:
2x1 + x2 + x3 + x4 + x5 = 6
x1 + 2x2 + x3 + x4 + x5 = 12
x1 + x2 + 2x3 + x4 + x5 = 24
x1 + x2 + x3 + 2x4 + x5 = 48
x1 + x2 + x3 + x4 + 2x5 = 96.
7. (2004 AIME, Problem #5) Alpha and Beta both took part in a two-day problem-solving
competition. At the end of the second day, each had attempted questions worth a total of 500
points. Alpha scored 160 points out of 300 points attempted on the first day, and scored 140
out of 200 points attempted on the second day. Beta, who did not attempt 300 points on the
first day, had a positive integer score on each of the two days, and Beta’s daily success ratio
(points scored divided by points attempted) on each day was less than Alpha’s on that day.
Alpha’s two-day success ratio was 300/500 = 3/5. The largest possible two-day success ratio
that Beta could have achieved is m/n, where m and n are relatively prime positive integers.
What is m + n?
8. (2001 AIME, Problem #8) Call a positive integer N a 7-10 double if the digits of the
base-7 representation of N form a base-10 number that is twice N . For example, 51 is a 7 − 10
double because its base-7 representation is 102. What is the largest 7 − 10 double?
9. (2004 AIME-2, Problem #5) In order to complete a large job, 1000 workers were hired,
just enough to complete the job on schedule. All the workers stayed on the job while the first
quarter of the work was done, so the first quarter of the work was completed on schedule.
Then 100 workers were laid off, so the second quarter of the work was completed behind
schedule. Then an additional 100 workers were laid off, so the third quarter of the work was
completed still further behind schedule. Given that all workers work at the same rate, what is
the minimum number of additional workers, beyond the 800 workers still on the job at the end
of the third quarter, that must be hired after three-quarters of the work has been completed
so that the entire project can be completed on schedule or before?
10. (1990 AIME, Problem #6) A biologist wants to calculate the number of fish in a lake. On
May 1 she catches a random sample of 60 fish, tags them, and releases them. On September 1
she catches a random sample of 70 fish and finds that 3 of them are tagged. To calculate the
number of fish in the lake on May 1, she assumes that 25% of these fish are no longer in the
lake on September 1 (because of death and emigrations), that 40% of the fish were not in the
lake May 1 (because of birth and immigrations), and that the number of untagged fish and
tagged fish in the September 1 sample are representative of the total population. What does
the biologist calculate for the number of fish in the lake on May 1?
11. (2004 AIME-2, Problem #6) Three clever monkeys divide a pile of bananas. The first
monkey takes some bananas from the pile, keeps three-fourth of them, and divides the rest
equally between the other two. The second monkey takes some bananas from the pile, keeps
one-fourth of them, and divides the rest equally between the other two. The third monkey
takes the remaining bananas from the pile, keeps one-twelfth of them and divides the rest
equally between the other two. Given that each monkey receives a whole number of bananas
whenever the bananas are divided, and the numbers of bananas the first, second, and third
monkeys have at the end of the process are in the ratio 3 : 2 : 1, what is the least possible
total for the number of bananas?
12. (2010 AIME, Problem #9) Let (a, b, c) be a real solution of the system of equations
x3 − xyz = 2, y 3 − xyz = 6, z 3 − xyz = 20.
m
The greatest possible value of a3 + b3 + c3 can be written in the form n, where m and n are
relatively prime positive integers. Find m + n.
13. (1984 AIME, Problem #10) Mary told John her score on the American High School Math-
ematics Examination (AHSME), which was over 80. From this, John was able to determine the
number of problems Mary solved correctly. If Mary’s score had been any lower, but still over
80, John could not have determined this. What was Mary’s score? (Recall that the AHSME
consists of 30 multiple-choice problems and that one’s score, s, is computed by the formula
s = 30 + 4c − w, where c is the number of correct and w is the number of wrong answers;
students are not penalized for problems left unanswered.)
14. (1987 AIME, Problem #10) Al walks down to the bottom of an escalator that is moving
up and he counts 150 steps. His friend, Bob, walks up to the top of the escalator and counts 75
1.4. EXERCISES 13
steps. If Al’s speed of walking (in steps per unit time) is three times Bob’s speed, how many
steps are visible on the escalator at any given time? (Assume that this number is constant.)
15. (1990 AIME, Problem #15) Find ax5 + by 5 if the real numbers a, b, x, and y satisfy the
equations
ax + by = 3,
ax2 + by 2 = 7,
ax3 + by 3 = 16,
ax4 + by 4 = 42.
Chapter 2
Combinatorics
“Mathematics may be defined as the economy of counting. There is no problem in the

whole of mathematics which cannot be solved by direct counting.”
- Ernst Mach
2.1 Introduction
Combinatorics is a broad term that refers to the mathematics of counting. Many problems in
the AIME competitions ask students to determine in how many ways something can be done. In
addition, many problems in the AIME that initially give no hint of combinatorics turn out to require
counting techniques to obtain the answer. Indeed, it is common for the essence of an AIME problem
to remain hidden until one becomes absorbed in it, and the mathematics at the core of the problem
can often come from a different subject matter than it may seem at first. It often turns out that
combinatorics is at the heart of an initially non-combinatorics looking problem, and therefore, this
chapter is one of the most important in the text.
Although many problems on counting at the AIME level can be posed in an elementary way, com-
binatorics can be a slippery subject. In the theory of counting there are very few formulas and
set procedures that are widely applicable, and with each problem, a careful analysis must be made
anew. Most attempts to place counting strategies into a rigid pedagogical apparatus begin by asking
two fundamental questions about the arrangements of objects that are being counted:
Question 2.1.1. Does the order of the objects matter?
Question 2.1.2. Is repetition of objects allowed?
These questions give rise to four possible scenarios: (1) order of objects matters and repetition of
objects is allowed, (2) order of objects matters and repetition of objects is not allowed, (3) order
of objects does not matter and repetition of objects is allowed, and (4) order of objects does not
matter and repetition of objects is not allowed. Arrangements of objects in which the order of the
15
16 CHAPTER 2. COMBINATORICS
objects matters are called permutations, while arrangements in which the order of the objects does
not matter are called combinations. In the next sections, therefore, we study permutations and
combinations, both with and without repetition and give examples that model each scenario.
2.2 Permutations
2.2.1 Permutations with Repetition
In this section, we consider permutations of objects in which repetition of objects is allowed. A typical
example of a problem here is that of counting license plates, say with eight capital letters. In this case,
the “objects” in question are the letters. The license plates “ABCDEFGH” and “DGACBHFE” are
clearly not the same, even though the same set of letters is used for each. Thus, the order of the letters
in the arrangements matters (i.e. we are talking about permutation). Moreover, “CFHCHDAH” is
a valid license plate, despite the multiple occurrences of the letters “C” and “H”. This means that
repetition of objects is allowed.
The general formula that applies to counting problems in this situation is given by the following:
The number of ways to make an ordered arrangement of k objects

(with repetition allowed) from a set of n different objects is nk .
The visualization of this principle is seen below.
|n choices | |n choices | |n choices| |n choices|

...
slot 1 slot 2 slot k − 1 slot k
In this case, the object chosen for one slot is independent from the object chosen in any other
slot. Therefore, the choices that we have in each slot are unaffected by choices made in other slots.
Therefore, the Multiplication Principle applies:
Theorem 2.2.1. (The Multiplication Principle) If a task has independent subtasks T1 , T2 , . . . ,

Tk that can be completed in n1 , n2 , . . . , nk ways, respectively, then the total number of ways to
complete the task is the product n1 n2 . . . nk .
Remark: In the special case n1 = n2 = · · · = nk = n, we recover the expression nk for the number
of ways to complete the main task.
Example 2.2.2. In how many ways can 24 high school students be assigned to five faculty advisors,
if each student receives exactly one advisor?
Solution: Line up the 24 students in a row, with one slot associated with each student. There
are five ways to fill each slot with a faculty advisor. So there are five choices associated with
each slot. Since there are a total of 24 slots, we have 524 ways to carry out the process. It is
important to recognize here that the students are choosing the faculty advisers, not the other way
2.2. PERMUTATIONS 17
around. If one tries to solve the problem by having faculty members choose the students, it is
much more complicated because the number of students that each faculty member can advise is not
predetermined. 2
The next example, Example 2.2.4, is taken from the 2007 AMC 10 Competition and is another
illustration of permutations with repetition. This time, however, we will be required to subtract out
unwanted solutions. We do this via the Subtraction Rule:
Theorem 2.2.3. (The Subtraction Rule) If among a total collection of n objects that meet a
given condition, k of the objects do not meet this condition, then the number of objects meeting the
condition is n − k.
Example 2.2.4. (AMC 10 A 2007, Problem #12) Two tour guides are leading six tourists.
The guides decide to split up. Each tourist must choose one of the guides, but with the stipulation
that each guide must take at least one tourist. How many different groupings of guides and tourists
are possible?
Solution: If we begin by ignoring the stipulation, then each tourist has two choices for a tour
guide. The same “tour guide” can be repeatedly chosen by the tourists, so repetition is allowed.
Furthermore, the order of arrangements matters in the sense that it matters which tourists go with
each tour guide. So we have 26 = 64 possible outcomes to this selection. However, two configurations
violate the stipulation that each guide must take at least one tourist; namely, if all tourists happen
to choose the same guide. Thus, by subtraction, we obtain the final answer: 64 − 2 = 62. 2
We will find occasion to apply this Subtraction Rule again later in this chapter (see Examples 2.4.6
and 2.4.7). It is, indeed, a frequently used tool in the art of counting.
2.2.2 Permutations without Repetition
Here is a question of a somewhat different nature: How many eight-letter license plates consist of
eight distinct capital letters? In this case, we are not allowed to “repeat” the use of any letter, so
repetition of objects is no longer allowed. Another typical example of this situation occurs when
making arrangements of people in a line (such as a photograph arrangement), since a person cannot
be “repeated” in more than one spot in the line.
To determine the number of eight-letter license plates in which all letters are distinct, we draw eight
slots. In the first, we can use any of the 26 letters. In the second, we can use any letter other than
the one already used in the first slot (25 choices). The third slot will have 24 choices, and so on.
26!
Thus, the answer is 26 · 25 · 24 · 23 · 22 · 21 · 20 · 19 = . This example illustrates the following
18!
general principle:
The number of ordered arrangements of k objects (with no repetition) from a set
n!
of n different objects is n · (n − 1) · (n − 2) · · · · · (n − k + 1) = , provided k ≤ n.
(n − k)!
Obviously, if k > n, it is impossible to form such an arrangement.
For our example above with license plates, we have n = 26 and k = 8.
To visualize the principle described above, we can use the following picture:
|n choices | |n − 1 choices | |n − k + 2 choices| |n − k + 1 choices|

...
slot 1 slot 2 slot k − 1 slot k
Example 2.2.5. How many ways are there to choose a president, vice-president, secretary, and
treasurer from a group of nine committee members?
Solution: We can represent the four jobs with four slots. There are nine people who can fill the
“president slot”. After this, there are only eight ways left to fill the “vice-president slot”. Then
there are seven ways to fill the “secretary slot” and six ways to fill the “treasurer slot”. Hence, the
answer is 9 · 8 · 7 · 6 = 3024. We arrive at this answer in the notation above with n = 9 and k = 4. 2
Of course, sometimes additional constraints are placed on the arrangements under construction, so
that the general principle in the box above must be modified. However, Theorem 2.2.1 still governs
the solution process. A nice illustration of this can be found in the following problem from the AMC
10 competition from 2008.
Example 2.2.6. (AMC 10 B 2008, Problem #21) Ten chairs are evenly spaced around a round
table and numbered clockwise from 1 through 10. Five married couples are to sit in the chairs with
men and women alternating, and no one is to sit either next to or across from his/her spouse. How
many seating arrangements are possible?
Solution: Label the five men M1 , M2 , M3 , M4 , and M5 , with spouses (respectively) W1 , W2 , W3 ,

W4 , and W5 . Since all ten individuals are distinct from one another, no repetition is allowed. We can
use the Multiplication Principle (Theorem 2.2.1) if we break down the task of seating the 10 people
around the table into subtasks. One task is to decide whether the men occupy the even-numbered
chairs or the odd-numbered chairs. There are two choices for this subtask. Next, the men must be
positioned in the five available seats. Since no restrictions are in place at this point, we have 5! ways
to do this subtask (first man to be seated has 5 chairs to choose from, the second man has 4 chairs
to choose from, and so on). Without loss of generality, imagine we have seated the men as shown in
Figure 2.1(a) below.
Figure 2.1: (a): Without loss of generality, seat the five men in the odd-numbered seats as shown.
(b): With the men positioned in the odd-numbered seats, the even-numbered seats must be occupied
by women as shown. Once a woman is chosen in one of these positions, the remaining occupants of
the remaining even-numbered chairs are determined with no further choices.
2.3. COMBINATIONS 19
The final subtask is to seat the women in the five open seats, subject to the constraint that each
woman must not sit next to or across from her spouse. We can actually picture the valid arrangements
(see Figure 2.1(b)).
As Figure 2.1(b) shows, there are only two ways to seat the five women. So this final subtask can be
completed in only two ways. Applying the Multiplication Principle (Theorem 2.2.1), we conclude
that there are 2 · 5! · 2 = 480 arrangements. 2
2.3 Combinations
2.3.1 Combinations without Repetition
Here is another problem similar to Example 2.2.5: From a group of nine committee members, in
how many ways can a sub-committee of four members be chosen? Here, we have not designated
titles for the four jobs of the four members who are singled out. Although similar in some respects
to Example 2.2.5, this problem is an example of a different type of counting problem, a situation
in which the order of objects does not matter. Therefore, we are now talking about combinations
rather than permutations.
A typical problem illustrating this situation is the following: How many eight-card poker hands can
be made from a standard deck of 52 cards? Notice that with a single deck, no card appears more
than once, so that repetition of objects (i.e. cards) is not allowed here. Moreover, a player is free to
discard from their hand in any order, so the order in which the cards appear in the hand does not
matter.
Example 2.3.1. How many eight-card poker hands can be made from a standard deck of 52 cards?
Solution: Drawing slots for such a problem would appear to be misleading, since a row of slots
suggests an ordering. However, we can begin by temporarily assuming that the order does matter.
If we keep track of the order in which our eight poker cards are dealt to us, we can draw eight slots
in which to put the cards. There are 52 playing cards, so the number of ordered arrangements of
52!
eight of them is , according to the previous section.
44!
Now, how do we compensate for the fact that the order of the cards actually does not matter?
To understand this, imagine an eight-card poker hand that consists specifically of the 2-9 of clubs.
From Section 2.2, we conclude that there are 8! different ordered arrangements of these eight specific
52!
cards. Our total above counts each of these arrangements separately, when in fact we should
44!
have only counted the hand with the 2–9 of clubs one time.
For each unordered selection of eight cards, there are 8! ordered arrangements of those cards. This
52!
means that our answer above, which counted ordered arrangements, is inflated by a factor of
44!
8!. To obtain the correct answer for the number of unordered arrangements of eight cards, we must
52!
therefore divide our previous answer by 8!: there are possible eight-card poker hands. 2
8!44!
n!
Expressions of the form like we just obtained arise so often in counting problems that we
k!(n − k)!
have a special notation and phrase to describe them.

1 n n!
Notation. If n and k are nonnegative integers with k ≤ n, we write to mean ,
k k!(n − k)!
pronounced “n choose k”. This quantity is often referredto asa binomial coefficient. Note that
0 n
since 0! = 1 conventionally, we have = 1. In fact, = 1 for all nonnegative integers n.
0 0

52
In this language, the number of different eight-card poker hands is , or “52 choose 8”. The
8
word “choose” suggests that we are only selecting the objects, not that we are putting them in order.
Let us summarize this discussion:
Given n distinct objects,
the
number of unordered selections (without repetition)
n
of k of the objects is , provided that k ≤ n.
k
We now return to discuss the modification to Example 2.2.5 that we mentioned above.
Example 2.3.2. From a group of nine committee members, in how many ways can a sub-committee
of four members be chosen?
Solution: We are making an unordered selection of four people from a group of size nine. The

9 9!
number of ways to do this is simply = = 126. 2
4 4!5!

n
A short-cut for computing : Because these expressions arise so frequently in the AIME,
k
it is useful
for this
timed, calculator-free test to be aware of the simplest possible procedure for
n
computing by hand, since factorial expressions often involve large numbers. Assume that
k
k ≤ n − k. Then a quick cancellation process shows that

n n! n · (n − 1) · (n − 2) · · · (n − k + 1)
= = .
k k!(n − k)! 1 · 2 · 3···k
If k > n − k, then simply reverse the roles of k and n − k in this simplification. For example we have

9 9! 9·8·7·6
= = = 126.
4 4!5! 1·2·3·4
Informally, we “factorial down” k steps on the top from n, and we “factorial up” k steps on the
bottom from 1.
Example 2.3.3. (AIME 2005-2, Problem #1) A game uses a deck of n different cards, where
n is an integer and n ≥ 6. The number of possible sets of 6 cards that can be drawn from the deck
is 6 times the number of possible sets of 3 cards that can be drawn. Find n.
1 Other notations, common in calculators and computers, include Ckn , C(n, k), and n Ck .
2.3. COMBINATIONS 21
Solution: While it is not explicitly stated, the sets of cards drawn from the deck are unordered.
That is, no record is made of the order in which the cards are drawn in forming the sets. This is in
keeping with the usual interpretation of a set of cards as an unordered collection in which the holder
is free to play cards in any desired sequence.

n
With this interpretation, the number of possible sets of six cards from a deck of n cards is ,
6

n
while the number of possible sets of three cards from a deck of n cards is . We are given that
3

n n
=6 . Using Equation (2.6), we have
6 3
n! n!
=6· .
6!(n − 6)! 3!(n − 3)!
Simplifying, we obtain (n − 3)! = 720(n − 6)!, and cancelling (n − 6)! from both sides, this reduces
to (n − 3)(n − 4)(n − 5) = 720. With a little trial-and-error, it can be seen that the only value of n
that satisfies this equation is n = 13 = 013. ANSWER : 013. 2
2.3.2 Combinations with Repetition
Imagine that you are given instructions at your household to bring home 18 pieces of fruit from a
grocery store that sells only apples, bananas, and oranges. However, you were not told how many
of each type of fruit should be bought! The question is: how many different options do you have?
You could buy six of each fruit, or you could buy 18 apples, or you could buy seven apples, eight
bananas, and four oranges, or a host of other possibilities. Let us assume that there is an ample
supply of each type of fruit when you get to the store (regular shoppers will recognize that this
is a questionable assumption). We will also assume that all apples are identical, all bananas are
identical, and all oranges are identical.
Because an ample supply of identical fruits is available, repetition of fruits is allowed. On the other
hand, the order in which you select the fruits from the store is immaterial and forgotten by the time
you get home. For instance, if you pick out seven apples, then four oranges, and then eight bananas,
the outcome is the same as if you first pick out eight bananas, then seven apples, and finally four
oranges. Hence, the order in which the fruits are selected does not matter.
To see how many selections are possible, let us reconsider the examples mentioned above. If we were
to purchase seven apples, eight bananas, and four oranges, we could represent this by the following
sequence of “stars and bars”:
∗ ∗ ∗ ∗ ∗ ∗ ∗| ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗| ∗ ∗ ∗ ∗
A selection of six of each fruit, on the other hand, would appear as
∗ ∗ ∗ ∗ ∗ ∗ | ∗ ∗ ∗ ∗ ∗ ∗| ∗ ∗ ∗ ∗ ∗ ∗,
while a selection of 18 apples would appear as
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ||
The two bars are used to create “bins” for the three different types of fruit. The number of stars
in a particular fruit’s bin tells us how many of that type of fruit is selected. In this way, we have a
one-a-one correspondence between permissible selections of fruit and sequences of stars and bars.
Therefore, to determine the number of permissible selections of fruit, we can just as well count
the number of sequences of stars and bars. The number of bars is one less than the number of
different types of objects, while the number of stars is the total number of objects to be selected.
In our example, we have 3 − 1 = 2 bars and 18 stars. Hence, we have a string of stars and bars
of
length
18 + 2 = 20, in which there are two spaces that are filled by bars. There are hence
20
= 190 ways to choose two spaces in which to place the bars. The other spaces are then
2
automatically selected as stars. Each distinct placement of bars (and hence stars) corresponds to a
distinct selection of fruits. The observations made in this example can be stated generally as follows:
Given an unlimited supply of k types of identical objects, wecan make an

n+k−1
unordered selection of n objects from among these types in ways.
k−1
k − 1 is the number
Here, of bars and n is the number of stars. Thus, we can think of the formula
“stars plus bars”
as .
“bars”
Example 2.3.4. How many ways are there to give nine identical stuffed animals to seven children?
Solution: In this problem, each child is assigned a “bin” in which to place his or her stuffed
animals. Since we need seven bins, we will need six bars. Each stuffed animal is represented as a
star,
sowe have nine stars and six bars. The number of ways to arrange nine stars and six bars is
15
= 5005.
6
Remark: If all of the stuffed animals in Example 2.3.4 were distinguishable from one another, there
would be 79 ways to distribute them, according to Section 2.2 (see Example 2.2.2). 2
We conclude this section with a couple more examples of the stars and bars method.
Example 2.3.5. How many quadruples (x, y, z, w) of nonnegative integers satisfy
x + y + z + w = 2013? (2.1)
Solution: When a person sees a problem like this for the first time, they may be a bit surprised by
the revelation that the stars and bars method is applicable. But it is true for any equation of the
type shown in Equation (2.1) in which nonnegative integer solutions are desired. Each variable x,
y, z, and w receives a bin, and the number of stars appearing in each variable’s bin can be used to
represent the value of that variable (see Figure 2.2).
2.4. MORE CHALLENGING COMBINATORICS PROBLEMS 23
Figure 2.2: The values of x, y, z, and w correspond to the number of stars placed in the four bins
separated by three bars, respectively.

2016
Here, we have three bars and 2013 stars, giving a total of = 1,363,558,560 solutions to
3
the equation. 2
Our final example in this section builds on the ideas of Example 2.3.5 and appeared in the 1998
AIME competitiion.
Example 2.3.6. (AIME 1998, Problem #7) Let n be the number of ordered quadruples
4
X n
(x1 , x2 , x3 , x4 ) of positive odd integers that satisfy xi = 98. Find .
i=1
100
Solution: The important first step is to represent each positive odd integer xi in the form xi =
2ki + 1, where each ki is a nonnegative integer. Plugging these expressions for xi into the summation
given in the problem, we obtain
98 = x1 + x2 + x3 + x4
= (2k1 + 1) + (2k2 + 1) + (2k3 + 1) + (2k4 + 1)
= 2(k1 + k2 + k3 + k4 ) + 4.
Therefore, k1 + k2 + k3 + k4 = 47. To find the number of solutions to this equation such that k1 , k2 ,
k3 , and k4 are nonnegative, we can apply the method of stars and bars in like fashion to Example
2.3.5. Using three bars to separate the four variables and letting the number of stars in the ith bin
be the value
of ki , we have 47 stars and 3 bars. The number of different ways to arrange them is
50 n
n= = 19, 600, so the answer to the problem is = 196. ANSWER : 196. 2
3 100
2.4 More Challenging Combinatorics Problems
The difficulty with most combinatorics problems is that most often they do not fall cleanly into one
and only one of the four types of problems discussed in Sections 2.2 and 2.3. That is certainly the
case with the vast majority of AIME problems that require combinatorial analysis. In many AIME
problems, a complex problem must be broken into simpler cases for which the techniques of the
previous four sections can be applied. We will consider several AIME examples to illustrate this
momentarily. We begin, however, with a classic example.
Example 2.4.1. How many ways are there to arrange the letters of the word “MISSISSIPPI”?
Remark: In this problem, it is hard to determine the answers to Questions 2.1.1 and 2.1.2 raised
at the outset of this chapter regarding repetition of objects and order of objects in arrangements.
On the one hand, an M before an S is different than an S before an M. However, there are multiple
copies of S, and the relative order of multiple copies of S with respect to one another is immaterial.
Moreover, S is repeated four times, while M is not repeated at all!
Solution #1: One way to proceed is to model a solution along the lines of Example 2.3.1. First
pretend that all 11 letters of the word are distinct. If that were the case, there would be 11!
arrangements of the letters. Next, we have to divide by the number of various arrangements of
repeated letters that we just assumed were distinct. There are 2! ways to arrange the two P’s with
respect to each other, 4! ways to arrange the four I’s with respect to each other, and 4! ways to
11!
arrange the four S’s with respect to each other. Hence, there are = 34, 650 arrangements
2! · 4! · 4!
of the letters in the word “MISSISSIPPI”.
Solution #2: There is an alternative solution that many students find equally appealing that
proceeds as follows. Imagine 11 slots in a line waiting to receive the letters of “MISSISSIPPI”. We
proceed to choose slots in which to place each of the four distinct letters of the word, one by one.
It does not matter in what order we choose to place the four letters M, I, S, and P. Let us decide
(arbitrarily) to place M first, then I, then S, and then P. We could view the process as consisting of
four subtasks and organize a table showing the number of ways to complete each subtask:
Subtask # Description # of Ways

to Complete
11
1 Place M in one of 11 available spaces
1
10
2 Place four I’s in four of 10 remaining spaces
4
6
3 Place four S’s in four of 6 remaining spaces
4
2
4 Place two P’s in two remaining spaces
2
Using the Multiplication Principle (Theorem 2.2.1) with the values in the last column of the table
above, we once again arrive at the answer:

11 10 6 2
= 11 · 210 · 15 · 1 = 34, 650.
1 4 4 2
Note that if we chose to arrange the letters in place in a different order, the binomial coefficients
that would arise in the last column above would be different, but the product, and hence the answer,
would of course be the same. 2
Example 2.4.2. (2000 AIME-2, Problem #5) Given eight distinguishable rings, let n be the
number of possible five-ring arrangements on the four fingers (not the thumb) of one hand. The
order of rings on each finger is significant, but it is not required that each finger have a ring. Find
the leftmost three nonzero digits of n.
Solution: Let us follow the model used in Solution #2 to Example 2.4.1 and develop a table of
subtasks required to make a five-ring arrangement as prescribed in the problem. As in Example
2.4.1, there is some variability in how these subtasks are viewed or ordered, but let us outline the
following reasonable approach:
Subtask # Description # of Ways

toComplete
8
1 Choose five rings to use in the arrangement
5
8
2 Decide how many rings to place on each finger
5
3 Choose where to position the five chosen rings 5!
Let us explain the values in the last column of the table. For the first subtask, since
we have eight
8
rings, we must choose five of them to occur in the arrangement. There are = 56 ways to
5
complete this task. For our next task, let us decide upon the number of rings that will go on each
finger. We are not concerned with which ring goes where yet, just how many rings go on each finger.
We can think of each finger as a “bin”, ready to receive rings, with three bars to separate the four
fingers. Five rings will be placed
in thebins,which we can represent as stars. By the method of
5+3 8
stars and bars, we there have = = 56 ways to complete this task. Finally, we
3 3
must place the five chosen rings from the first subtask onto the fingers in place of the stars chosen
in the second subtask. There are five places to put the first ring, four places to put the second ring,
three places to put the third ring, and so on, for a total of 5! = 120 ways to complete this third (and
final) subttask. Applying the Multiplication Principle (Theorem 2.2.1) gives

8 8
n= · · 5! = 56 · 56 · 120 = 376, 320.
5 5
Therefore, the leftmost three nonzero digits of n are 376.
Remark: Of course, there are different ways to formulate the subtasks involved in making an
acceptable arrangement of rings. For instance, it is not necessary to choose the five rings involved in
the arrangement first (subtask 1 in the table above). Instead, the number of rings per finger can be
determined first (subtask 2 in the table above), and then one can simply choose one of eight rings for
the first spot, one of seven remaining rings for the second spot, and so on. In essence, this combines
subtasks 1 and 3 from the table above. Notice here that we are choosing rings for pre-determined
spots, whereas in the solution above, we are choosing spots for pre-determined rings. This is an
acceptable variation to the solution. ANSWER : 376. 2
With a more complex counting problem, it is often plausible to decompose the problem into a few
simpler problems that are either independent from one another, or dependent on one another in
ways that can be measured. In the latter situation, we may be able to apply the Inclusion-Exclusion
Principle (see Section 2.6 below), and in the former situation, we can use the Sum Rule.
Theorem 2.4.3. (The Sum Rule) If one wishes to compute the number of objects meeting one
of r mutually disjoint2 possible conditions, and there are nj objects meeting the jth condition, then
the total number of objects is n1 + n2 + · · · + nr .
Theorem 2.4.3 applies nicely when using an analysis of cases to solve a counting problem. The
idea is that sometimes it is not easy to count the total number of objects meeting some particular
condition, but if one breaks the objects into manageable subsets that are easier to count, the Sum
Rule can supply the final answer. The trick is to find the most effective decomposition of the objects
into subsets, as there are often many possibilities. Let us consider an example.
Example 2.4.4. (2007 AIME, Problem #10) In the 6 × 4 grid shown, 12 of the 24 square
are to be shaded so that there are two shaded squares in each row and three shaded squares in each
column. Let N be the number of shadings with this property. Find the remainder when N is divided
by 1000.
Solution: We will apply the Sum Rule after decomposing this problem into four mutually exclusive3
cases that are determined by the number of shaded rows that are in common in the first two columns
(which could be 0, 1, 2, or 3).

6
Note that for each of the four columns, there are = 20 ways to shade three squares in that
3
column. However, the requirements given in the problem for the shadings make the choices for
shaded squares in one column dependent on the choices made in other columns. Let us assume that
the first three squares of the first column are shaded:
X
X
X
Now consider the second column. Of course, with no information other than the shaded squares in
the first column, any three of the six squares of the second column may be shaded, but the choice
of shaded squares in the second column will greatly impact the number of ways to shade the third
and fourth columns. A manageable way to proceed is to explore four cases, distinguished according
to how many common rows of shaded squares the first two columns share.
2 Mutually disjoint conditions are conditions such that no more than one of them can hold for a given object
simultaneously.
3 In mutually exclusive cases, each object being counted belongs to one and only one of the cases.
Case 1: We assume that the shaded squares of the second column all occur in the same rows as the
shaded squares of the first column:
X X
X X
X X
In this case, there is only one way to complete the shading of the last two columns. (What is it?)
Case 2: We assume

that the first two columns share shaded squares in two of the first three rows.
3
There are = 3 ways to choose which two of the first three rows are shaded in the second
2
3
column, and there are = 3 ways to choose which one of the last three rows are shaded in the
1
second column. We have established one such choice here:
X X
X X
X
X
For this configuration, there are only two ways to complete the shading of the last two columns:
X X X X
X X X X
X X X X
and
X X X X
X X X X
X X X X
Applying the Multiplication Principle (Theorem 2.2.1) to the choices made in Case 2, there are
3 · 3 · 2 = 18 shadings in this case.
Case 3: We assume

that the first two columns share a shaded square in one of the first three rows.
3
There are = 3 ways to choose which one of the first three rows are shaded in the second
1
3
column, and there are = 3 ways to choose which two of the last three rows are shaded in the
2
X X
X
X
second column. We have established one such choice here:
X
X
Now consider the third column. The first square cannot be shaded, while the last
square
must be
4
shaded. This leaves four squares, of which two must be shaded. Thus, we have = 6 ways to
2
shade the third column. Once this is one, the last column has only one acceptable shading. Hence,
in Case 3, we have found by the Multiplication Principle (Theorem 2.2.1) a total of 3 · 3 · 6 = 54
legal shadings.
Case 4: We assume that the first two columns share no common shaded rows. In this case, the last
three squares of the second column must be shaded. This is shown in the 6 × 4 grid here:
X
X
X
X
X
X

6
Now, the shading of squares in the third column can be done arbitrarily (there are = 20
3
ways to do this), and this forces the last column to be shaded in the three rows that are unshaded
in the third column. Therefore, we have 20 shadings in this case.
Applying the Sum Rule (Theorem 2.4.3), we add the number of shadings obtained in the four cases
above to obtain 1 + 18 + 54 + 20 = 93 shadings. But these four cases were predicated on the initial
assumption that the first three squares of the first column are shaded. In fact, there are 20 ways to
shade this column, and each of them will yield 93 acceptable ways to complete the shading of the
last three columns. Multiplying once more, we find that N = 93 · 20 = 1860. Therefore, when N is
divided by 1000, the remainder is 860. ANSWER : 860 2
Example 2.4.5. (2008 AIME-2, Problem #12) There are two distinguishable flagpoles, and
there are 19 flags, of which 10 are identical blue flags, and 9 are identical green flags. Let N be the
number of distinguishable arrangements using all of the flags in which each flagpole has at least one
flag and no two green flags on either pole are adjacent. Find the remainder when N is divided by
1000.
Solution: Since the flagpoles are distinguishable, we shall refer to them as Flagpole A and Flagpole
B. One approach to this problem involves a large case-by-case analysis according to the number of
green flags that are placed on Flagpole A (which could be any number from zero to nine)4 . The
reader who wishes to experiment with this labor-intensive approach can certainly do so, but since
4 Considering green flags on Flagpole A is arbitrary. One could equally well consider blue flags and/or Flagpole B
if preferred.
many of the cases are similar, in terms of efficiency it will take less space to consider an alternate
solution that takes advantage of such similarities. That is the approach that we will take here.
For any configuration of flags on the two flagpoles, we let g denote the number of green flags on
Flagpole A, and we let b denote the number of blue flags on Flagpole A. Then we must have 9 − g
green flags and 10 − b blue flags on Flagpole B. Because no two green flags on either pole are
adjacent, there must be at least one blue flag placed between any two green flags. Thus, we must
have b ≥ g − 1 (to satisfy this requirement on Flagpole A) and we must have 10 − b ≥ 8 − g (to
satisfy this requirement on Flagpole B). Putting these two inequalities together, we find that
−1 ≤ b − g ≤ 2.
Equivalently,
g − 1 ≤ b ≤ g + 2. (2.2)
Now, the g green flags on Flagpole A partition the flagpole into g+1 sections (the spaces surrounding
the g green flags), and we can let xi denote the number of blue flags in the ith section. Thus, we
have
x1 + x2 + · · · + xg+1 = b,
with the condition that
x2 , x3 , · · · , xg ≥ 1.
We set x0i = xi − 1 (for i = 2, 3, · · · , g) and x01 = x1 and x0g+1 = xg+1 to obtain the equation
x01 + x02 + · · · + x0g+1 = b − g + 1,
subject to nonnegative integer solutions for x01 , x02 , .... The number of solutions is

b+1
. (2.3)
g
At the same time, we need to consider Flagpole B. The 9 − g green flags partition Flagpole B into
10 − g sections (the spaces surrounding the 9 − g green flags), and we can let yi denote the number
of blue flags in the ith section. Thus, we have
y1 + y2 + · · · + y10−g = 10 − b,
with the condition that

y2 , y3 , . . . , y9−g ≥ 1.
Proceeding as we did for Flagpole A, we can show that the number of valid arrangements of flags
on Flagpole B is
11 − b
. (2.4)
9−g
Applying the Multiplication Principle (Theorem 2.2.1), Equations (2.3) and (2.4) dictate that the
number of ways to arrange g green flags on Flagpole A and b blue flags on Flagpole A (and the rest
of the flags on Flagpole B) is
b+1 11 − b
.
g 9−g
Next, we must sum this product over all valid choices of g and b. Note that 0 ≤ g ≤ 9, and
from Equation (2.2), we have g − 1 ≤ b ≤ g + 2. However, some of these choices are invalid,
namely (g, b) ∈ {(0, −1), (0, 0), (9, 10), (9, 11)}. These are impossible since each flagpole must have
at least one flag. Thus, we can subtract four terms corresponding to these four pairs (g, b) from the
summation
9 g+2
X X b+1 11 − b
= 2860. (2.5)
g 9−g
g=0 b=g−1
Hence, our final answer is

0 12 1 11 11 1 12 0
N = 2860 − − − − = 2310.
0 9 0 9 9 0 9 0
Thus, the remainder when N is divided by 1000 is 310.

Remark: The observant reader may notice that the terms of the summation in Equation (2.5)
corresponding to g = 0, 1, 2, 3, and 4 are identical to the terms obtained for g = 9, 8, 7, 6, and 5,
respectively. This should not be surprising, since the same rules govern the arrangement of flags
on each of the two flagpoles. For instance, g = 0 means that no green flags are placed on Flagpole
A. The number of such arrangements is identical to the number of arrangements in which no green
flags appear on Flagpole B, which is the case g = 9. Thus, it somewhat simplifies our calculation in
Equation (2.5) to note that
9 g+2 4 g+2
X X b+1 11 − b X X b+1 11 − b
=2 = 2860.
g 9−g g 9−g
g=0 b=g−1 g=0 b=g−1
Looking for such symmetries in counting problems is often a useful, even sometimes necessary, skill
in solving certain problems. ANSWER : 310 2
Example 2.4.6. (2004 AIME, Problem #6) An integer is called snakelike if its decimal rep-
resentation a1 a2 . . . ak satisfies ai < ai+1 if i is odd and ai > ai+1 if i is even. How many snakelike
integers between 1000 and 9999 have four distinct digits?
Solution: There are multiple ways to proceed. In this solution, we will determine:
1. the number of snakelike integers having four distinct digits (including the possibility that the
left-most digit is zero), and
2. the number of snakelike integers beginning with zero and followed by three distinct nonzero
digits.
We will subtract the number of snakeline integers meeting condition (2) from those meeting condition
(1). This is an application of the Subtraction Rule (Theorem 2.2.3).
For collection (1) above,
let us select four distinct digits, without concern for the order that they
10
are chosen in. There are = 210 ways to do this. Suppose the digits are a, b, c, and d, with
4
a < b < c < d. Then we can construct all possible snakelike numbers that use these four digits: acbd,
adbc, bcad, bdac, and cdab. Therefore, we have five such snakelike numbers. By the Multiplication
Principle (Theorem 2.2.1), there are 210 · 5 = 1050 four-digit snakelike numbers with distinct digits,
including the possibility that 0 is a chosen digit and is placed in the leftmost position of the number.

9
How many of these four-digit snakelike numbers have 0 as left-most digit? There are = 84
3
ways to choose distinct digits b, c, and d with 0 < b < c < d. Now we can easily construct all
four-digit snakelike numbers with distinct digits, beginning with 0, and using the digits b, c, and
d: 0cbd and 0dbc. Therefore, we have two such snakelike numbers. Appealing to the Multiplication
Principle (Theorem 2.2.1) once more, we obtain 84 · 2 = 168 four-digit snakelike numbers with
distinct digits and 0 as left-most digit. Therefore, we have 168 numbers in collection (2). Finally,
using Theorem 2.2.3, we subtract the 168 snakelike numbers with left-most digit 0 from the 1050
four-digit snakelike numbers with distinct digits obtained above (since we are only interested in
counting snake-like numbers between 1000 and 9999): 1050 − 168 = 882.
Remark: It is possible to arrive at this answer without subtraction by counting the snakelike
numbers in two cases: (1) snakelike numbers where 0 is present, and (2) snakelike numbers where
0 is not present. The reader is invited to verify that there are 252 snakeline numbers where 0 is
present and 630 snakelike numbers where 0 is absent. ANSWER : 882. 2
Let us consider another example in which the Subtraction Rule figures prominently.
Example 2.4.7. (AIME 2010-2, Problem #11) Define a T -grid to be a 3 × 3 matrix which
satisfies the following two properties:
1. Exactly five of the entries are 1’s, and the remaining four entries are 0’s.
2. Among the eight rows, columns, and long diagonals (the long diagonals are {a13 , a22 , a31 } and
{a11 , a22 , a33 }), no more than one of the eight has all three entries equal.
Find the number of distinct T -grids.
Solution: We will count the number of distinct 3 × 3 matrices that meet requirement (1) and then
subtract the 3 × 3 matrices in this collection that to meet requirement (2). The number of 3 × 3
fail
9
matrices that meet requirement (1) is clearly = 126, the number of ways to choose 5 out of
5
the 9 positions in which to place the 1’s (leftover 4 spots receive the 0’s). From this total, we will
subtract those 3 × 3 matrices that fail to meet requirement (2). Below we illustrate examples of
types of 3 × 3 matrix that meet (1) but fail (2):
1 1 1 1 1 1 1 1 1 1 0 1 1 0 1
1 0 0 , 0 1 0 , 0 0 0 , 1 0 1 , 0 1 0
1 0 0 0 0 1 1 1 0 1 0 0 1 0 1
All of the matrices that meet (1) but fail (2) can be described as similar to one of the five examples
above. We summarize this observation in the table below:
Violation # Description Number of Matrices

1 One row and one column of 1’s 9
2 One diagonal and one row or column or 1’s 12
3 One row of 0’s and one row of 1’s 18
4 One column of 0’s and one column of 1’s 18
5 Two diagonals of 1’s 1
We leave to the reader to confirm the numbers in the last column of this table. By the Sum Rule
(Theorem 2.4.3), we must add the results in the last column of the table. Hence, we have
9 + 12 + 18 + 18 + 1 = 58
matrices meeting requirement (1) and failing requirement (2). Hence, the total number of T -grids
is 126 − 58 = 68 = 068. ANSWER : 068. 2
Remark: Here is an alternate case analysis to obtain the number of matrices that satisfy (1) but
not (2). Such matrices can be classified into one of two mutually exclusive types: (a) one threesome
of 0’s and one threesome of 1’s (Cases 3 and 4 in the table), and (b) two threesomes of 1’s (Cases
1,2, and 5 in the table). For type (a), the threesome of 1’s must occupy one of the six rows or
columns (a long diagonal of 1’s would prevent a threesome of 0’s from existing), and once chosen,
the threesome of 0’s must occupy one of the two remaining rows or columns (same type, row or
column, as the threesome of 1’s). Finally, the last three entries of the matrix to be filled must
contain two 1’s and one 0, and there are three ways to place these values in those three entries. The
Multiplication Principle concludes that we have 6 · 2 · 3 = 36 violators in case (a). We leave to the
reader to establish that there are exactly 22 violators of type (b). With 36 + 22 = 58 violators, we
obtain 126 − 58 = 68 T -grids.
2.5 Some Combinatorial Identities
In this section, we wish to record a few of the most important identities that relate the binomial
coefficients. In some cases, the identities can be verified computationally from the expression

n n!
= . (2.6)
k k!(n − k)!
In other cases, a so-called combinatorial argument, which consists of computing the number of ways
to perform a given task in two different ways that reflect the two sides of the identity, is most
effective. In still other cases, mathematical induction can be used to verify the statement. While
we will not take the space here to derive the identities presented in detail, the reader may wish to
spend some time thinking about them. Any textbook on discrete mathematics or combinatorics will
provide a deeper discussion of the proofs of these identities. We begin with the famous Binomial
Theorem.
Theorem 2.5.1. (The Binomial Theorem) Let x and y be real numbers. For every nonnegative
integer n, we have
n
X n
(x + y)n = xn−k y k .
k
k=0
2.5. SOME COMBINATORIAL IDENTITIES 33
The Binomial Theorem has some useful special cases. Most notably, if x = y = 1, we obtain the
relation
n
n
X n n n n n
2 = = + + ··· + + . (2.7)
k 0 1 n−1 n
k=0
One nice way to justify Equation (2.7) combinatorially is to count the number of subsets T of an
n-element set S. On the one hand, each of the n elements of S independently has two choices
(belong to T or do not belong to T ), giving 2n by the Multiplication
Principle.
On the other hand,
n
the subsets T can be grouped according to size k (and there are ways to form a subset of
k
size k from n available elements).
The next result is easily verified by using Equation (2.6).

n n
Proposition 2.5.2. Let k and n be nonnegative integers with n ≥ k. Then = .
k n−k

n
The final identity in this section, Pascal’s Formula, relates the value of to the values
k
n−1 n−1
and . Again, this result can be verified directly from Equation (2.6) or with
k k−1
a combinatorial argument. To save space, we shall not reproduce either verification here.
Proposition 2.5.3. (Pascal’s Formula) Let k and n be positive integers with n ≥ k. Then

n n−1 n−1
= + .
k k k−1
Let us conclude this section with a couple of applications of the results in this section to problems
from the AIME competition.
Example 2.5.4. (2000 AIME-2, Problem #7) Given that
1 1 1 1 1 1 1 1 N
+ + + + + + + =
2!17! 3!16! 4!15! 5!14! 6!13! 7!12! 8!11! 9!10! 1!18!
N
find the greatest integer that is less than .
100
Solution: Examining the fractions appearing on both sides of the relation,we see
that if the
n! n
numerators were 19!, we would have expressions of the form = . Therefore,
k!(n − k)! k
multiplying the given relation through by 19!, we have that

19 19 19
19N = + + ··· + ,
2 3 9
which, according to Proposition (2.5.2) is the same as

19 19 19
19N = + + ··· + .
17 16 10
Adding the two formulas above yields

19 19 19 19
38N = + + ··· + + .
2 3 16 17
Applying Equation (2.7), we can rewrite this as

19 19 19 19
38N = 219 − − − − = 219 − 1 − 19 − 19 − 1.
0 1 18 19
Solving for N yields N = 13, 796, so the answer is 137. ANSWER : 137. 2
Remark: One must be careful in computing 219 . One possibility is to view it as the calculation of
219 = 29 · 210 = 512 · 1024. This can be done by hand, but care should be exercised in doing so.
The reader might consider if there are any tricks that could be used to shrink the magnitude of the
numbers occuring in the calculation.
Example 2.5.5. (1989 AIME, Problem #2) Ten points are marked on a circle. How many
distinct convex polygons of three or more sides can be drawn using some (or all) of the ten points as
vertices? (Polygons are distinct unless they have exactly the same vertices.)
Solution: A convex polygon of three or more sides is constructed by selecting three or more of the
ten points marked on the circle. The convex polygon is uniquely determined by the choice of points.
The order in which the points of the polygon are selected is immaterial, so we will use combinations,
rather than permutations. We can select any number of points from 3 to 10. Therefore, the number
of distinct polygons is

10 10 10 10 10 10 10 10
+ + + + + + + ,
3 4 5 6 7 8 9 10
and using Equation (2.7), we can write this expression as

10 10 10
210 − − − = 1024 − 1 − 10 − 45 = 968.
0 1 2
Remark: If we denote the ten points of the circle as 1, 2, 3, . . . , 10, one may recognize that each
convex polygon of three or more sides corresponds uniquely to a subset of S = {1, 2, 3, . . . , 10} of
size three or more. Thus, we are really just counting the number of subsets of S of size 3 or more.
ANSWER : 968. 2
2.6 The Inclusion-Exclusion Principle
Recall that the Sum Rule (Theorem 2.4.3) provides a simple means of computing the total number
of objects lying in one of various mutually disjoint sets: simply sum the number of objects in each
of the sets. In this section, our primary concern is: What if the sets are not mutually disjoint?
2.6. THE INCLUSION-EXCLUSION PRINCIPLE 35
Before we proceed, let us introduce a little bit of set theoretic notation. For a finite set X, we use
|X| to denote the number of elements in the set X. Given two sets X and Y , we write X ∪ Y for the
union of X and Y , which is the set consisting of all elements that belong to X or to Y or to both.
Finally, we write X ∩ Y for the intersection of X and Y , which is the set consisting of all elements
belonging to X and to Y . The notions of union and intersection can be naturally extended to any
finite number of sets.
Now suppose we want to count the number of states in the United States that start with the letter
“A” or start with the letter “C”. To do this, we can simply count the number of states beginning
with A,
A = {Alabama, Alaska, Arizona, Arkansas},
count the number of states beginning with C,
C = {California, Colorado, Connecticut},
and add them. Therefore, the answer to this question is
|A ∪ C| = |A| + |C| = 4 + 3 = 7.
Notice that the sets A and C contain no common elements, so no elements of the sets A or C have
been counted multiple times in arriving at the value of |A ∪ C|.
If two sets A and B contain common elements, we must be a bit more careful in computing |A ∪ B|.
For instance, we could ask: How many states in the United States begin or end with the letter “A”?
The set A above contains the states that begin with the letter “A”, and we can define B to be the
set containing those states that end with the letter “A”:
B = {Alabama, Alaska, Arizona, California, Florida, Georgia, Indiana, Iowa, Louisiana, Minnesota,
Montana, Nebraska, Nevada, North Carolina, North Dakota, Oklahoma, Pennsylvania,
South Carolina, South Dakota, Virginia, West Virginia}.
Note that |A| = 4 and |B| = 21. However, |A ∪ B| = 22 6= |A| + |B|. That is, 22 states’ names either
begin or end with the letter “A”. Notice that |A| + |B| = 4 + 21 = 25, which is three greater than
the correct answer. Why? The reason is that there are three states, Alabama, Alaska, and Arizona,
which belong to both A and B. Therefore, these three states contributed one element to both A
and B, and therefore are counted twice in |A| + |B|. Not coincidentally, |A ∩ B| = 3 in this case,
and if we subtract three from |A| + |B|, we obtain the correct answer. Thus,
Theorem 2.6.1. (The Inclusion-Exclusion Principle for Two Sets) Given two finite sets A
and B, we have
|A ∪ B| = |A| + |B| − |A ∩ B|.
Figure 2.3 illustrates the Inclusion-Exclusion Principle for two sets.

In the case where A and B have no common elements, we have |A ∩ B| = 0 and there is no need to
subtract (i.e. “exclude”) anything from |A| + |B|. There is also a three-set version of the Inclusion-
Exclusion Principle that is important to be aware of.
Figure 2.3: The shaded region indicates A ∩ B. Its |A ∩ B| elements are counted twice in |A| + |B|
and must be subtracted out once in the Inclusion-Exclusion Principle.
Theorem 2.6.2. (The Inclusion-Exclusion Principle for Three Sets) Given three finite sets
A, B, and C, we have
|A ∪ B ∪ C| = |A| + |B| + |C| − |A ∩ B| − |A ∩ C| − |B ∩ C| + |A ∩ B ∩ C|.
Figure 2.4 illustrates the Inclusion-Exclusion Principle for three sets.
Figure 2.4: The shaded regions 1,2, and 3 contain elements counted twice in |A| + |B| + |C|, while
the elements in shaded region 4 are counted three times in |A| + |B| + |C|. Theorem 2.6.2 ensures
that each element of A ∪ B ∪ C is counted exactly once.
Of course, Theorems 2.6.1 and 2.6.2 can be extended to any finite number of sets. Therefore, it is
worth recording the general Inclusion-Exclusion formula for n sets:
Theorem 2.6.3. (The Inclusion-Exclusion Principle for n Sets) Given finite sets A1 , A2 ,
· · · , An , we have  
n n
[ X X
(−1)k+1 

Ai = Ai ∩ · · · ∩ Ai  . (2.8)
1 k
i=1 k=1 1≤i1 <···<ik ≤n
The complexity of the notation appearing in Equation (2.8) may take a moment to understand. The
left-hand side is simply the size of the union of the n given sets:
n
[
Ai = A1 ∪ A2 ∪ · · · ∪ An .

i=1
On the right side, the notation

X

Ai 1 ∩ · · · ∩ Ai
k
1≤i1 <···<ik ≤n
means that we should evaluate the sum of the sizes of all possible intersections of k of the sets. Once
we have done that for each fixed integer k with 1 ≤ k ≤ n, then according to the outside sum over
k from k = 1 to k = n, we are to sum the results for all k with 1 ≤ k ≤ n and take negative signs
whenever k is even. More discussion of series can be found in Chapter 5.
Now let us consider a few examples.
Example 2.6.4. Among 123 high school students attending a particular school, 32 of them enjoy
basketball and golf, 21 of them enjoy basketball and swimming, 15 of them enjoy golf and swimming,
and 11 of them enjoy all three sports. If 27 of the students do not enjoy any of these sports and the
same number of students enjoy each of the three sports, how many students at the high school enjoy
golf ?
Solution: Let B denote the set of students who enjoy basketball, let G denote the set of students
who enjoy golf, and let S denote the set of students who enjoy swimming. Since 27 students do not
enjoy any of the sports, we know that 123 − 27 = 96 students do enjoy at least one sport. That is,
|B ∪ G ∪ S| = 96. Hence, by Theorem 2.6.2, we have that
96 = |B| + |G| + |S| − |B ∩ G| − |B ∩ S| − |G ∩ S| + |B ∩ G ∩ S|.
We are given
|B ∩ G| = 32, |B ∩ S| = 21, |G ∩ S| = 15, and |B ∩ G ∩ S| = 11.
Incorporating the fact that |B| = |G| = |S|, we obtain
96 = 3|G| − 32 − 21 − 15 + 11,
from which we deduce that

96 + 32 + 21 + 15 − 11 153
|G| = = = 51.
3 3
Thus, 51 students at the school enjoy golf. 2
Example 2.6.5. How many five-digit numbers greater than 10,000 begin with an odd digit and have
no digit equal to 0, end with an even digit and have no digit equal to zero, or have all digits identical?
Solution: Let A denote the set of five-digit numbers that begin with an odd digit and have no digit
equal to 0, let B denote the set of five-digit numbers that end with an even digit and have no digit
equal to zero, and let C denote the set of five-digit numbers that have all digits identical. We are
seeking to determine
|A ∪ B ∪ C| = |A| + |B| + |C| − |A ∩ B| − |A ∩ C| − |B ∩ C| + |A ∩ B ∩ C|.
We must determine each quantity on the right-hand side of this equation.

By the Multiplication Principle (Theorem 2.2.1), we have
|A| = 5 · 94 ,
since the first digit must be 1,3,5,7, or 9, and each other digit can be any digit from 1-9. Similarly,
|B| = 4 · 94 .
(Here, the last digit must be 2, 4, 6, or 8 (four choices).) Next,
|C| = 9.
Now
|A ∩ B| = 5 · 93 · 4,
since the first and last digit have five and four choices, respectively, and the middle three digits each
have nine choices. Finally, the reader should check that
|A ∩ C| = 5, |B ∩ C| = 4, |A ∩ B ∩ C| = 0.
Therefore, by Theorem 2.6.2, we have
|A ∪ B ∪ C| = 5 · 94 + 4 · 94 + 9 − 5 · 93 · 4 − 5 − 4 + 0 = 44, 469.
2
Our final example in this section will carry the lion’s share of work required to solve an AIME
problem (Example 3.2.6) involving probability that will be discussed in the next chapter.
Example 2.6.6. Each unit square of a 3-by-3 unit-square grid is to be colored either blue or red.
How many 3-by-3 unit-square grids contain a 2-by-2 red square?
Solution: We will use the following notation. Let E denote the set of all 3 × 3 unit square grids
that contain a 2 × 2 red square. Further, let
A = the set of grids that contain a 2-by-2 square in the upper left that is all red,
B = the set of grids that contain a 2-by-2 square in the upper right that is all red,
C = the set of grids that contain a 2-by-2 square in the lower left that is all red,
D = the set of grids that contain a 2-by-2 square in the lower right that is all red.
Then we have
E = A ∪ B ∪ C ∪ D.
Figure 2.5: The four 3-by-3 unit-square grids shown satisfy the requirements for sets A, B, C, and
D, respectively.
Applying Theorem 2.6.3 with n = 4, we have the explicit expression

|E| = |A ∪ B ∪ C ∪ D|
= |A| + |B| + |C| + |D| − |A ∩ B| − |A ∩ C| − |A ∩ D| − |B ∩ C| − |B ∩ D| − |C ∩ D|
+ |A ∩ B ∩ C| + |A ∩ B ∩ D| + |A ∩ C ∩ D| + |B ∩ C ∩ D| − |A ∩ B ∩ C ∩ D|.
Now
|A| = |B| = |C| = |D| = 25 ,
since four red squares are required to form the 2-by-2 red square in question, but the remaining five
squares in the grid can be colored either blue or red independently, thus giving two choices in each
of five squares.
Next, we have
|A ∩ B| = |B ∩ D| = |C ∩ D| = |A ∩ C| = 23 ,
since each of these intersections requires six of the nine squares in the 3-by-3 grid to be colored red,
leaving the remaining three squares open to either blue or red color. But
|A ∩ D| = |B ∩ C| = 22 ,
since each of these intersections requires seven of the nine squares to be colored red, leaving only
two remaining squares open to either blue or red color.
We have
|A ∩ B ∩ C| = |A ∩ B ∩ D| = |A ∩ C ∩ D| = |B ∩ C ∩ D| = 2,
since each of these intersections requires eight of the nine squares in question to be colored red,
leaving only one square open to blue or red color choice. Finally, only the 3 × 3 grid that is
exclusively colored with red squares belongs to all four sets:
|A ∩ B ∩ C ∩ D| = 1.
Therefore, we find that

|E| = 4 · 25 − 4 · 23 − 2 · 22 + 4 · 2 − 1 = 95.
2
The Inclusion-Exclusion Principle can be useful in problems from all different corners of mathematics.
On the AIME competition, it is usually used in the service of a problem that involves some other
mathematical subject, such as probability or number theory. For instance, the preceding example
will be extended to an AIME problem on probability in Example 3.2.6, while Example 4.3.3 in
Chapter 4 will provide an application in the area of number theory.
2.7 Exercises
1. (2007 AIME-2, Problem #1) A mathematical organization is producing a set of commem-

orative license plates. Each plate contains a sequence of five characters chosen from the four
letters in AIME and the four digits 2007. No character may appear in a sequence more times
than it appears among the four letters in AIME or the four digits in 2007. A set of plates in
N
which each possible sequence appears exactly once contains N license plates. Find .
10
2. (2012 AIME-2, Problem #3) At a certain university, the division of mathematical sciences
consists of the departments of mathematics, statistics, and computer science. There are two
male and two female professors in each department. A committee of six professors is to
contain three men and three women and must also contain two professors from each of the
three departments. Find the number of possible committees that can be formed subject to
these requirements.
3. (2004 AIME, Problem #3) A convex polyhedron P has 26 vertices, 60 edges, and 36 faces,
24 of which are triangular, and 12 of which are quadrilaterals. A space diagonal is a line
segment connecting two non-adjacent vertices that do not belong to the same face. How many
space diagonals does P have?
4. (2012 AIME, Problem #3) Nine people sit down for dinner where there are three choices
of meals. Three people order the beef meal, three order the chicken meal, and three order
the fish meal. The waiter serves the nine meals in random order. Find the number of ways
in which the waiter could serve the meal types to the nine people so that exactly one person
receives the type of meal ordered by that person.
5. (1983 AIME, Problem #10) The numbers 1447, 1005, and 1231 have something in com-
mon: each is a 4-digit number beginning with 1 that has exactly two identical digits. How
many such numbers are there?
6. (2009 AIME-2, Problem #6) Let m be the number of five-element subsets that can be
chosen from the set of the first 14 natural numbers so that at least two of the five numbers are
consecutive. Find the remainder when m is divided by 1000.
7. (2006 AIME-2, Problem #4) Let (a1 , a2 , a3 , . . . , a12 ) be a permutation of (1, 2, 3, . . . , 12)
for which
a1 > a2 > a3 > a4 > a5 > a6 and a6 < a7 < a8 < a9 < a10 < a11 < a12 .
An example of such a permutation is (6, 5, 4, 3, 2, 1, 7, 8, 9, 10, 11, 12). Find the number of such
permutations.
2.7. EXERCISES 41
8. (2007 AIME-2, Problem #6) An integer is called parity-monotonic if its decimal repre-
sentation a1 a2 a3 . . . ak satisfies ai < ai+1 if ai is odd, and ai > ai+1 if ai is even. How many
four-digit parity-monotonic integers are there?
9. (2005 AIME, Problem #5) Robert has 4 indistinguishable gold coins and 4 indistinguish-
able silver coins. Each coin has an engraving of a face on one side, but not on the other. He
wants to stack the eight coins on a table into a single stack so that no two adjacent coins are
face to face. Find the number of possible distinguishable arrangements of the 8 coins.
10. (1997 AIME, Problem #10) Every card in a deck has a picture of one shape – circle,
square, or triangle, which is painted in one of three colors – red, blue, or green. Furthermore,
each color is applied in one of three shades – light, medium, or dark. The deck has 27 cards,
with every shape-color-shade combination represented. A set of three cards from the deck is
called complementary if all of the following statements are true:
(i) Either each of the three cards has a different shape or all three of the cards have the same
shape.
(ii) Either each of the three cards has a different color or all three of the cards have the same
color.
(iii) Either each of the three cards has a different shade or all three of the cards have the same
shade.
How many different complementary three-card sets are there?

11. (2009 AIME, Problem #10) The Annual Interplanetary Mathematics Examination (AIME)
is written by a committee of five Martians, five Venusians, and five Earthlings. At meetings,
committee members sit at a round table with chairs numbered from 1 to 15 in clockwise order.
Committee rules state that a Martian must occupy chair 1 and an Earthling must occupy chair
15. Furthermore, no Earthling can sit immediately to the left of a Martian, no Martian can
sit immediately to the left of a Venusian, and no Venusian can sit immediately to the left of
an Earthling. The number of possible seating arrangements for the committee is N (5!)3 . Find
N.
12. (1986 AIME, Problem #13) In a sequence of coin tosses one can keep a record of the
number of instances when a tail is immediately followed by a head, a head is immediately
followed by a head, etc. We denote these by T H, HH, etc. For example, in the sequence
HHT T HHHHT HHT T T T of 15 coin tosses we observe that there are five HH, three HT ,
two T H, and four T T subsequences. How many different sequences of 15 coin tosses will
contain exactly two HH, three HT , four T H, and five T T subsequences?
13. (2002 AIME-2, Problem #9) Let S be the set {1, 2, 3, . . . , 10}. Let n be the number of
sets of two non-empty disjoint subsets of S. (Disjoint sets are defined as sets that have no
common elements.) Find the remainder obtained when n is divided by 1000.
14. (2007 AIME-2, Problem #13) A triangular array of squares has one square in the first
row, two in the second, and, in general, k squares in the kth row for 1 ≤ k ≤ 11. With the
exception of the bottom row, each square rests on two squares in the row immediately below,
as illustrated in the figure. In each square of the eleventh row, a 0 or a 1 is placed. Numbers
are then placed into the other squares, with the entry for each square being the sum of the
entries in the two squares below it. For how many initial distributions of 0’s and 1’s in the
bottom row is the number in the top square a multiple of 3?
FIGURE from 2007 AIME-2 BOOKLET (Problem 13) GOES HERE !!!
15. (2001 AIME, Problem #14) A mail carrier delivers mail to the nineteen houses on the
east side of Elm Street. The carrier notices that no two adjacent houses ever get mail on the
same day, but that there are never more than two houses in a row that get no mail on the
same day. How many different patterns of mail delivery are possible?
Chapter 3
Probability
“It’s not that I’m so smart; it’s just that I stay with problems longer.”
- Albert Einstein
3.1 Introduction
Probability is everywhere in our daily lives. Whether you are watching a weather forecast for the
chance of rain, checking the likelihood that your airplane flight will arrive late based on the time
and date of travel, or pondering the odds of your favorite basketball team winning their next game,
you are exploring probability. It comes as no surprise that probability problems are prolific in the
AIME competition. In this chapter, we will explore some of the many examples.
The present chapter may be viewed merely as a continuation of the previous chapter on combina-
torics. That is because probability often involves delicate counting arguments. Out of an array of
different possible outcomes (whose number must be counted), we must count the number of outcomes
that comply with some requirement. More precisely, probability is concerned with the likelihood
that a particular outcome or set of outcomes, called an event, occurs when an experiment is con-
ducted. An experiment has a set of all possible outcomes, called the sample space. Typically we
denote the set of outcomes comprising the event by E and the sample space by S. Therefore, E is
a subset of S. It is not necessary for the sample space S to be finite, but for most AIME problems,
it is. Recall from Section 2.6 that the number of elements in a set S is conventionally denoted by
|S|. Mathematically, if we assume that each outcome of the experiment is equally likely, then the
probability that the event occurs when the experiment is conducted is
|E|
p(E) = . (3.1)
|S|
For instance, the probability that a “3” is obtained when a fair die is rolled once is 1/6, since there
are six equally-likely outcomes, one of which is “3”.
43
44 CHAPTER 3. PROBABILITY
It is important to realize that Equation (3.1) only holds if the various outcomes of the experiment
are equally likely. Sometimes it is not even clear what the outcomes of the experiment are! To
illustrate this, consider an example.
Example 3.1.1. Suppose two fair dice are rolled. What is the probability that the sum of the dice
is 10?
Solution: The possible sums that can be rolled on two dice are 2,3,4,. . . ,12, which seemingly gives
us 11 possible outcomes. However, these 11 outcomes are not equally likely and Equation (3.1)
cannot be applied. It is far more likely, for instance, that the dice sum to 7 than that the dice sum
to 2. This is because a sum of 2 is achieved only when each die has 1 showing, whereas there are a
host of ways to roll a sum of 7.
How can we find the probability that the two dice sum to 10 in this case? The answer is to change
the sample space from the set of possible sums on the two dice to the set of pairs of values rolled on
each die:
S ={(1, 1), (1, 2), . . . , (1, 6), (2, 1), (2, 2), . . . , (2, 6), . . . , (5, 1), (5, 2), . . . , (5, 6), (6, 1), (6, 2), . . . , (6, 6)}.
Now we have |S| = 36, and all 36 outcomes in S are equally likely. We are not interested in just one
outcome now, since there are multiple pairs (a, b) with a + b = 10. In fact, several different outcomes
of the experiment comprise the desired event E:
E = {(4, 6), (5, 5), (6, 4)}.
Applying Equation (3.1), the probability that the sum obtained when two fair dice are rolled is
|E| 3 1
p(E) = = = .
|S| 36 12
2
3.2 Properties of Probability Functions
Equation (3.1) shows quite clearly the combinatorial nature of probability problems. To compute
p(E) in the case of an experiment in which there are finitely many equally likely outcomes, we
must count the number of outcomes of the experiment that lie in E and in S. Our next goal is to
enumerate a few properties of the probability function p(E). Before we collect this list, however, we
need to clarify a couple of points. First, recall from Section 2.6 that the symbols ∪ and ∩ appearing
between the sets in some parts of Theorem 3.2.1 below denote set-theoretic union and intersection,
respectively. Second, the notion of independent events which appears in the last item in Theorem
3.2.1 is defined as follows. Two events E1 and E2 in a sample space S are said to be independent if
the occurrence of one of the events when the experiment is performed is unaffected by the occurrence
(or lack thereof) of the other. For example, two flips of a fair coin are independent from one another,
since the result of one flip does not affect the other.
Theorem 3.2.1. Let S denote a finite sample space for an experiment, and let E ⊆ S. Then
3.2. PROPERTIES OF PROBABILITY FUNCTIONS 45
1. 0 ≤ p(E) ≤ 1 and p(S) = 1.

2. If E denotes the set S − E, called the complement of E, then
p(E) = 1 − p(E).
3. If E1 and E2 are subsets of S, then
p(E1 ∪ E2 ) = p(E1 ) + p(E2 ) − p(E1 ∩ E2 ).
4. If E1 and E2 are subsets of S such that E1 ∩ E2 = ∅, then we have
p(E1 ∪ E2 ) = p(E1 ) + p(E2 ).
5. If E1 and E2 are independent events, then
p(E1 ∩ E2 ) = p(E1 )p(E2 ).
Most of the parts of Theorem 3.2.1 are easy to prove using Equation 3.1 as extensions of correspond-
ing statements known to hold for sizes of finite sets.
Note that part (4) of Theorem 3.2.1 is the “probability version” of the Sum Rule (and can be
extended to any finite number of mutually disjoint events), and part (5) of Theorem 3.2.1 is the
“probability version” of the Multiplication Principle (Theorem 2.2.1) and can be extended to any
finite number of mutually independent events.
In Chapter 2, we discussed how it can sometimes be profitable in a counting problem to count
whatever objects are not desired and subtract from a total; this is the Subtraction Rule (Theorem
2.2.3). Using part (2) of Theorem 3.2.1, we see that the same technique can be useful in the area of
probability as well. Our next example is a quick illustration of this.
Example 3.2.2. Suppose two fair dice are rolled. What is the probability that the sum of the dice
is not 10?
Solution: Referring to Example 3.1.1 above, if we continue to let E be the event that the two dice
sum to 10, then we are seeking to find p(E). By Theorem 3.2.1, we conclude that
1 11
p(E) = 1 − p(E) = 1 − = .
12 12
2
Now let us consider a variation of Example 3.1.1.
Example 3.2.3. (2006 AIME-2 Problem #5) When rolling a certain unfair six-sided die with
faces numbered 1, 2, 3, 4, 5, and 6, the probability of obtaining face F is greater than 1/6, the prob-
ability of obtaining the face opposite face F is less than 1/6, the probability of obtaining each of
the other faces is 1/6, and the sum of the numbers on each pair of opposite faces is 7. When two
such dice are rolled, the probability of obtaining a sum of 7 is 47/288. Given that the probability of
obtaining face F is m/n, where m and n are relatively prime positive integers, find m + n.
Solution: Let F 0 denote the face opposite face F . The probability of obtaining a face other than
face F and face F 0 is 4 · 61 = 23 . Hence, p(obtaining face F ) + p(obtaining face F 0 ) = 13 . Write
1 1
p(obtaining face F ) = +x and p(obtaining face F 0 ) = − x,
6 6
where x > 0. Now, there are six ways to obtain a sum of 7 on the two dice:
(1, 6) (2, 5) (3, 4) (4, 3) (5, 2) (6, 1).

0
Two of these
six pairs
involve faces F and F , and the probability of obtaining either of these pairs
1 1
is +x − x , where we have appealed to the independence of the value rolled on the two
6 6
individual dice from one another and applied part (5) of Theorem 3.2.1. The probability of obtaining
any of the other four pairs is 1/36. Summing the six probabilities of the six pairs, we find that

1 1 1 47
4· +2 +x −x = .
36 6 6 288
A brief algebraic manipulation yields
1 5
− x2 = .
36 192
Hence, r r
1 5 1 1
x= − = = .
36 192 576 24
1 1 1 5
Therefore, the probability of obtaining face F is + x = + = . Therefore, m = 5 and
6 6 24 24
n = 24, and so m + n = 5 + 24 = 29 = 029. ANSWER: 029. 2
As we study the next examples, observe how the probability computations involved essentially require
solving a problem in combinatorics.
Example 3.2.4. (1995 AIME, Problem #3) Starting at (0, 0), an object moves in the coordinate
plane via a sequence of steps, each of length one. Each step is left, right, up, or down, all four equally
likely. Let p be the probability that the object reaches (2, 2) in six or fewer steps. Given that p can
be written in the form m/n, where m and n are relatively prime positive integers, find m + n.
Solution: It clearly requires at least four steps for the object to reach (2, 2), starting at (0, 0). Since
it is impossible for the object to be at (2, 2) after five steps1 , we are computing the probability that
the object rests at (2, 2) after either 4 or 6 steps. Let
E1 = the event that the object is at (2, 2) after four steps,
and let
E2 = the event that the object is at (2, 2) after six steps.
We must compute
p(E1 ∪ E2 ) = p(E1 ) + p(E2 ) − p(E1 ∩ E2 ).
1 After an odd number of moves, the sum of the two coordinates of the object’s location must be odd.
3.2. PROPERTIES OF PROBABILITY FUNCTIONS 47
In the ensuing discussion, let R denote a right move, let L denote a left move, let U denote an
upward move, and let D denote a downward move. In order to reach (2, 2) after four
steps, the
4
object’s path must include exactly two R moves and two U moves. There are = 6 such
2
paths, out of a total of 44 = 256 possible paths of length four (since there are four equally likely
6 3
choices for each of the four moves). Hence, p(E1 ) = = .
256 128
Next consider E2 , the event that the object is at (2, 2) after six moves. The requires either three R
moves, oneL move,
and two
Umoves, or it requires two R moves, three U moves, and one D move.
6 3 2
There are = 60 of each of these types of moves. Since there are a total of 46
3 1 2
2 · 60 15
possible paths of length 6, we have p(E2 ) = = .
46 512
Finally, we must subtract p(E1 ∩ E2 ). A six-step move that belongs to both E1 and E2 would be
required to have two R moves and two U moves in the first four moves, followed by either U D, DU ,
RL, or LR moves. The first four moves areindependent of the last two moves, so the number of
4 24 3
moves belonging to both E1 and E2 is · 4 = 24. Thus, p(E1 ∩ E2 ) = 6 = .
2 4 512
We conclude from part (3) of Theorem 3.2.1 that
3 15 3 24 3
p(E1 ∪ E2 ) = p(E1 ) + p(E2 ) − p(E1 ∩ E2 ) = + − = = .
128 512 512 512 64
Hence, m = 3 and n = 64, so that m + n = 3 + 64 = 67 = 067. ANSWER: 067. 2
Example 3.2.5. (2008 AIME, Problem #9) Ten identical crates each have dimensions 3 ft ×
4 ft × 6 ft. The first crate is placed flat on the floor. Each of the remaining nine crates is placed,
in turn, flat on top of the previous crate, and the orientation of each crate is chosen at random. Let
m
n be the probability that the stack of crates is exactly 41 ft tall, where m and n are relatively prime
positive integers. Find m.
Solution: At first glance, this problem may not appear to be a combinatorics problem, since we
are not directly asked to count anything. In fact, it may seem at first to be a problem in spatial
geometry, but a little further thought reveals that this is not the case. Since the orientation of the
crates is randomly selected, there is an equal likelihood that each crate stands 3 feet tall, 4 feet
tall, and 6 feet tall. It will help simplify things to imagine that the crates are all distinguishable.
(Maybe, for instance, each crate is a unique color.) Let us number the crates2 from 1 to 10 and
denote the height of the ith crate by hi for i = 1, 2, . . . , 9, 10. There are 310 different values for
the 10-tuple (h1 , h2 , . . . , h9 , h10 ), since each hi is equal to either 3, 4, or 6, and the values of hi are
independent over all i. In order to determine the probability that the stack of crates is 41 feet tall,
we must determine for how many of the 10-tuples we have h1 + h2 + · · · + h10 = 41.
Since hi ≥ 3 for each i, we can simplify the problem by letting Hi = hi − 3 for each i = 1, 2, . . . , 9, 10.
Then we are seeking to determine the number of solutions to the equation
H1 + H2 + · · · + H10 = 11,
2 We could not number the crates if the crates were not distinguishable.
where Hi = 0, 1, or 3 for each i. From this equation, observe that Hi = 3 must hold for at least one
value of i but cannot hold for more than three values of i. Let us consider each of these cases in
turn.
Case 1: Hi = 3 for one value of i. In this case, there are 10 ways to determine for which i we have
Hi = 3. The sum of the remaining nine values of Himust be 8, so exactly eight of the remaining
9
nine values of i must give Hi = 1. There are = 9 ways to choose which values of i have
8
Hi = 1. By the Multiplication Principle (Theorem 2.2.1), the number of 10-tuples generated in this
case is 10 · 9 = 90.

10
Case 2: Hi = 3 for two values of i. In this case, there are = 45 ways to determine for
2
which i we have Hi = 3. The sum of the remaining eight values of H i must be 5, so exactly five
8
of the remaining eight values of i must give Hi = 1. There are = 56 ways to choose which
5
values of i have Hi = 1. By the Multiplication Principle (Theorem 2.2.1), the number of 10-tuples
generated in this case is 45 · 56 = 2520.

10
Case 3: Hi = 3 for three values of i. In this case, there are = 120 ways to determine for
3
which i we have Hi = 3. The sum of the remaining seven values of H
i must be 2, so exactly two
7
of the remaining seven values of i must give Hi = 1. There are = 21 ways to choose which
2
values of i have Hi = 1. By the Multiplication Principle (Theorem 2.2.1), the number of 10-tuples
generated in this case is 120 · 21 = 2520.
Adding the results of all three cases, according to the Sum Rule, we find that there are 2520 +
2520 + 90 = 5130 different 10-tuples (h1 , h2 , . . . , h9 , h10 ) whose components sum to 41. Therefore,
the probability that the stack of crates is exactly 41 feet tall is
5130 33 · 190 190
10
= 10
= 7 . (3.2)
3 3 3
Therefore, m = 190. ANSWER: 190. 2
Remark: Note in (3.2) that we had to decompose the numerator, m, and denominator, n, into
small numerical factors and then cancel in order to ensure that the numerator and denominator are
relatively prime, as required in the problem.
Our final example in this section is an easy extension of Example 2.6.6.

Example 3.2.6. (2001 AIME-2, Problem #9) Each unit square of a 3-by-3 unit-square grid
is to be colored either blue or red. For each square, either color is equally likely to be used. The
probability of obtaining a grid that does not have a 2-by-2 red square is m/n, where m and n are
Solution: In Example 2.6.6, we determined the number of 3-by-3 unit-square grids that contain
a 2-by-2 red square. The reader may recall that if E is the set of such unit-square grids, we have
3.3. EXAMPLES: TOURNAMENTS, SOCKS, AND DICE 49
|E| = 95. If we let S denote the sample space of all possible colored 3-by-3 grids, then |S| = 29 = 512,
since each of the nine unit squares contained in the grid is independently colored one of two colors.
Thus, the probability that a given 3-by-3 unit square grid contains a 2-by-2 red square is
|E| 95
p(E) = = .
|S| 512
We are given that p(E) = m/n. Since

95 417
p(E) = 1 − p(E) = 1 − = ,
512 512
and 417 and 512 are relatively prime, we conclude that m = 417 and n = 512. Therefore, m + n =
417 + 512 = 929. ANSWER : 929. 2
3.3 Examples: Tournaments, Socks, and Dice
Several AIME problems in past years have involved tournaments in which each team plays each
other team exactly once. Let us consider one such example here. More examples appear in the
exercise set.
Example 3.3.1. (2001 AIME-2, Problem #11) Club Truncator is in a soccer league with six
other teams, each of which it plays once. In any of its 6 matches, the probabilities that Club Truncator
will win, lose, or tie are each 1/3. The probability that Club Truncator will finish the season with
more wins than losses is m/n, where m and n are relatively prime positive integers. Find m + n.
Solution #1: Let S denote the set of all possible outcomes (consisting of results of six games) of
Club Truncator’s season, and let E denote the subset of outcomes in which Club Truncator ends
the season with more wins than losses.
The table below summarizes the possible numbers of wins, losses, and ties that Club Truncator can
achieve such that it has more wins than losses. The last column, which counts the number of ways
in which the given win-loss-tie scenario can occur, is explained below.
Wins Losses Ties # of ways this can occur

1 0 5 6
2 0 4 15
3 0 3 20
4 0 2 15
5 0 1 6
6 0 0 1
2 1 3 60
3 1 2 60
4 1 1 30
5 1 0 6
3 2 1 60
4 2 0 15
If w represents the number of wins, ` represents the number of losses, and t represents the number
of ties, then we have w + ` + t = 6, and the number of ways to garner w wins, ` losses, and t ties
6!
in six games is by the same reasoning used to solve the “MISSISSIPPI” problem in Example
w!`!t!
2.4.1. For instance, the number of ways that Club Truncator can win four games, lose zero games,
and tie two games is the same as the number of ways to arrange the letters WWWWTT in string
6!
of length six, which is 4!2! = 15. Summing the numbers in the last column of the table above gives
us the number of ways that Club Truncator can win more games than it loses:
|E| = 6 + 15 + 20 + 15 + 6 + 1 + 60 + 60 + 30 + 6 + 60 + 15 = 294.
The total number of possible outcomes to Club Truncator’s season is |S| = 36 = 729 (since each
of the six games has three equally-likely possible outcomes, and outcomes of different games are
independent from one another). Thus, the probability that Club Truncator finishes the season with
more wins than losses is
|E| 294 98
p(E) = = = .
|S| 729 243
Hence, m = 98 and n = 243, we have m + n = 341.
Solution #2: An alternative approach to the solution above is to more explicitly define three
events of interest as follows:
E1 = {event that Club Truncator finishes the season with more wins than losses},
E2 = {event that Club Truncator finishes the season with less wins than losses},
and
E3 = {event that Club Truncator finishes the season with the same number of wins and losses}.
We are seeking to find p(E1 ). Note by symmetry that p(E1 ) = p(E2 ). Moreover,
1 = p(E1 ) + p(E2 ) + p(E3 ) = 2p(E1 ) + p(E3 ).
Hence,
1 − p(E3 )
p(E1 ) = .
2
We leave the calculation of p(E3 ) as an exercise–it involves computing
p(E3 ) = p(0 wins and 0 losses) + p(1 win and 1 loss)

+ p(2 wins and 2 losses) + p(3 wins and 3 losses).
ANSWER: 341. 2
Example 3.3.2. (1991 AIME, Problem #13) A drawer contains a mixture of red socks and
blue socks, at most 1991 in all. It so happens that, when two socks are selected randomly without
replacement, there is a probability of exactly 1/2 that both are red or both are blue. What is the
largest possible number of red socks in the drawer that is consistent with this data?
Solution: Let x denote the number of red socks in the drawer, and let y denote the number of blue
socks in the drawer. We are given that x + y ≤ 1991. The probability of drawing two red socks,
without replacement, out of the drawer is
x x−1
p= · .
x+y x+y−1
The probability of drawing two blue socks, without replacement, out of the drawer is
y y−1
q= · .
x+y x+y−1
We are given that
1 x(x − 1) y(y − 1)
=p+q = + .
2 (x + y)(x + y − 1) (x + y)(x + y − 1)
Multiplying through by 2(x + y)(x + y − 1), this becomes
(x + y)(x + y − 1) = 2[x(x − 1) + y(y − 1)],
and with a little algebraic manipulation, we find that
x + y = (x − y)2 .
Thus, x + y must be a perfect square not exceeding 1991. The largest such perfect square is 442 .
Thus, if x − y = 44, we have x + y = 1936, from which we can quickly deduce that x = 990. It is
easy to see that if x − y < 44, then x < 990. Hence, the largest possible number of red socks in the
drawer is 990. ANSWER : 990. 2
One tool commonly used in probability problems where the experiment involves multiple tasks to
be carried out in succession leading to an array of outcomes is the tree diagram. (See Figure
3.1.) Often the outcomes from one task within the experiment impact the probability values for
outcomes associated with later tasks in the experiment. Starting from the top of the tree, branches
extend downward with labels corresponding to the first task in the experiment. One can write the
probability of each possible outcome of the first task next to the branches. Then, from the ends of
each of these branches, new branches can be drawn extending further down the tree showing the
outcomes corresponding to the second task. Again, the probabilities associated with each outcome
can be labeled on these branches.
Keep in mind that these probability values may depend on what outcome of the first task is selected.
The tree can be extended to any number of tasks, although generally tree diagrams are only utlized
if the number of tasks and outcomes is reasonably small. Otherwise, another more sophosticated
way of viewing the experiment may be appropriate.
We will use a tree diagram to solve the next example, which is found in the AMC 10th grade exam.
Other problems suited to tree diagram analysis can be found in the exercises at the end of the
chapter.
Example 3.3.3. (AMC 10 B 2008, Problem #16) Two fair coins are to be tossed once. For
each head that results, one fair die is to be rolled. What is the probability that the sum of the die
rolls is odd? (Note that if no die is rolled, the sum is 0.)
Figure 3.1: A tree diagram illustrates all outcomes of an experiment involving multiple tasks, with
the first task on the first (top) layer of the tree, the second task on the next layer, and so on. The
probabilities of each outcome of each task may be written along the branches.
Solution: We will construct a tree diagram that contains two levels. The upper portion of the tree
indicates the three possible outcomes from tossing the two fair coins (0 heads, 1 head, or 2 heads).
The lower portion of the tree shows the possible sums resulting from rolling the number of dice
specified by the coin tosses. Next to each outcome from each portion of the experiment, we label
branches with their probability values (which the reader can verify).
Figure 3.2: Tree diagram for an experiment consisting of two tasks: flipping two fair coins, following
by rolling 0, 1, or 2 fair dice (according to the number of heads obtained) and summing the values
on the dice.
We are only concerned with those branches that lead to odd sums at the bottom. For example,
if 0 heads are flipped, the sum is 0 (which is even), and we ignore this part of the tree (far left).
The middle branch (from the top) has a probability value of 1/2, and each branch beneath it on
the lower portion has probability 1/6. Thus, we have a probability of 21 · 16 = 12 1
of obtaining
any of these outcomes, three of which we are interested in. Finally, the right branch (from the
top) has a probability value of 1/4, and the odd sums that are possible (3, 5, 7, 9, and 11) occur
1 2 3 2 1
with probabilities 18 , 18 , 18 , 18 , and 18 , respectively. Hence, each of these outcomes occurs with
1 2 3 2 1
probabilities 72 , 72 , 72 , 72 , and 72 , respectively. Summing over all outcomes we are considering, we
obtain the final answer:

1 1 2 3 2 1 1 1 3
3 + + + + + = + = .
12 72 72 72 72 72 4 8 8
2
Remark: The reader may find other ways to solve this problem that do not involve the tree diagram,
but many solutions to this problem are essentially equivalent to what we have done here by organizing
the outcomes systematically and computing the probabilities of each outcome of interest. The tree
diagram is a handy book-keeping device that can help reduce confusion that might otherwise occur
on multi-task probability problems such as this.
Our final example in this section is meant to prepare the way for an AIME example (see Example
5.3.4) that relies on geometric series in conjunction with probability theory. Readers already versed
in geometric series may prefer to skip ahead and study Example 5.3.4 now.
Example 3.3.4. Dave rolls a fair six-sided die until a six appears for the first time. Independently,
Linda rolls a fair six-sided die until a six appears for the first time. What is the probability that
Linda rolls her die exactly k times and that the number of times Dave rolls his die is k − 1, k, or
k + 1? (Your expression will depend on k and will be different for k ≥ 2 than for k = 1.)
Solution: We can let E1 denote the event that Linda rolls her die exactly k times, and we can let
E2 be the event that Dave rolls his die k − 1, k, or k + 1 times. We wish to compute p(E1 ∩ E2 ),
and since E1 and E2 are independent events, part (5) of Theorem 3.2.1 guarantees that
p(E1 ∩ E2 ) = p(E1 )p(E2 ).
Let us consider rolling a six as a “success” (occurs with probability 1/6) and not rolling a six as a
“failure” (occurs with probability 5/6). In order for Linda to roll her die exactly k times, she must
experience k − 1 failures followed by a success. Since each roll of the die is independent from all
other rolls, we can therefore conclude that
k−1
5 1
p(E1 ) = · .
6 6
Next, we must consider E2 . Observe that E2 consists of three types of outcomes: outcomes with
k − 2 failures followed by a success (denote this set by E21 ), outcomes with k − 1 failures followed
by a success (denote this set by E22 ), and outcomes with k failures followed by a success (denote
this set by E23 ). Since
E2 = E21 ∪ E22 ∪ E23
and
E21 ∩ E22 = E21 ∩ E23 = E22 ∩ E23 = ∅,
a three term generalization of part (4) of Theorem 3.2.1 shows that3 for k ≥ 2,
k−2 k−1 k
5 1 5 1 5 1
p(E2 ) = p(E21 ) + p(E22 ) + p(E23 ) = · + · + · .
6 6 6 6 6 6
3 Observe that in the case k = 1, we have p(E21 ) = 0.
Therefore, for k ≥ 2, we have
p(E1 ∩ E2 ) = p(E1 )p(E2 )

k−1 k−2 k−1 k
5 1 5 1 5 1 5 1
= · · + · + ·
6 6 6 6 6 6 6 6
2 2k−3 2k−2 2k−1
1 5 5 5
= + + .
6 6 6 6
On the other hand, if k = 1, we can directly compute that

1 1 5 11
p(E1 ) = and p(E2 ) = p(E22 ) + p(E23 ) = + = ,
6 6 36 36
11
so that p(E1 ∩ E2 ) = in this case. 2
216
3.4 Exercises
1. (2004 AIME-2, Problem #2) A jar has 10 red candies and 10 blue candies. Terry picks
two candies at random, then Mary picks two of the remaining candies at random. Given that
the probability that they get the same color combination, irrespective of order, is m/n, where
m and n are relatively prime positive integers, find m + n.
2. (2000 AIME-2, Problem #3) A deck of forty cards consists of four 1’s, four 2’s, . . . , and
four 10’s. A matching pair (two cards with the same number) is removed from the deck. Given
that these cards are not returned to the deck, let m/n be the probability that two randomly
selected cards also form a pair, where m and n are relatively prime positive integers. Find
m + n.
3. (1998 AIME, Problem #4) Nine tiles are numbered 1, 2, 3, . . . , 9, respectively. Each of
three players randomly selects and keeps three of the tiles, and sums those three values. The
probability that all three players obtain an odd sum is m/n, where m and n are relatively
prime positive integers. Find m + n.
4. (2010 AIME-2, Problem #4) Dave arrives at an airport which has twelve gates arranged
in a straight line with exactly 100 feet between adjacent gates. His departure gate is assigned
at random. After waiting at that gate, Dave is told the departure gate has been changed to
a different gate, again at random. Let the probability that Dave walks 400 feet or less to the
m
new gate be a fraction , where m and n are relatively prime positive integers. Find m + n.
n
5. (1989 AIME, Problem #5) When a certain biased coin is flipped 5 times, the probability
of getting heads exactly once is not equal to 0 and is the same as that of getting heads exactly
twice. Let i/j, in lowest terms, be the probability that the coin comes up heads exactly 3
times out of 5. Find i + j.
3.4. EXERCISES 55
6. (2001 AIME, Problem #6) A fair die is rolled four times. The probability that each of the
final three rolls is at least as large as the roll preceding it may be expressed in the form m/n,
where m and n are relatively prime positive integers. Find m + n.
7. (1996 AIME, Problem #6) In a five-team tournament, each team plays one game with
every other team. Each team has a 50% chance of winning any game it plays. (There are no
m
ties.) Let be the probability that the tournament will produce neither an undefeated team
n
nor a winless team, where m and n are relatively prime integers. Find m + n.
8. (2010 AIME, Problem #4) Jackie and Phil have two fair coins and a third coin that comes
4
up heads with probability . Jackie flips the three coins, and then Phil flips the three coins.
7
m
Let be the probability that Jackie gets the same number of heads as Phil, where m and n
n
9. (1993 AIME, Problem #7) Three numbers, a1 , a2 , a3 , are drawn randomly and without
replacement from the set {1, 2, 3, . . . , 1000}. Three other numbers, b1 , b2 , b3 , are then drawn
randomly and without replacement from the remaining set of 997 numbers. Let p be the
probability that, after a suitable rotation, a brick of dimensions a1 × a2 × a3 can be enclosed in
a box of dimensions b1 × b2 × b3 , with the sides of the brick parallel to the sides of the box. If
p is written as a fraction in lowest terms, what is the sum of the numerator and denominator?
10. (1990 AIME, Problem #9) A fair coin is to be tossed 10 times. Let i/j, in lowest terms,
be the probability that heads never occur on consecutive tosses. Find i + j.
11. (2006 AIME-2, Problem #10) Seven teams play a soccer tournament in which each team
plays every other team exactly once. No ties occur, each team has a 50% chance of winning
each game it plays, and the outcomes of the games are independent. In each game, the winner
is awarded 1 point and the loser gets 0 points. The total points are accumulated to decide
the ranks of the teams. In the first game of the tournament, team A beats team B. The
probability that team A finishes with more points than team B is m/n, where m and n are
12. (2003 AIME-2, Problem #13) A bug starts at a vertex of an equilateral triangle. On each
move, it randomly selects one of the two vertices where it is not currently located, and crawls
along a side of the triangle to that vertex. Given that the probability that the bug moves to its
starting vertex on its tenth move is m/n, where m and n are relatively prime positive integers,
find m + n.
13. (2002 AIME-2, Problem #12) A basketball player has a constant probability of .4 of
making any given shot, independent of previous shots. Let an be the ratio of shots made to
shots attempted after n shots. The probability that a10 = .4 and an ≤ .4 for all n such that
1 ≤ n ≤ 9 is given to be pa q b r/(sc ), where p, q, r, and s are primes, and a, b, and c are positive
integers. Find (p + q + r + s)(a + b + c).
14. (1999 AIME, Problem #13) Forty teams play a tournament in which every team plays
every other team exactly once. No ties occur, and each team has a 50% chance of winning
any game it plays. The probability that no two teams win the same number of games is m/n,
where m and n are relatively prime positive integers. Find log2 n.
15. (1995 AIME, Problem #15) Let p be the probability that, in the process of repeatedly
flipping a fair coin, one will encounter a run of 5 heads before one encounters a run of 2 tails.
Given that p can be written in the form m/n, where m and n are relatively prime positive
integers, find m + n.
Chapter 4
Number Theory
“I feel my training in mathematics provided me with the invaluable ability to apply

logic, reason, and careful quantitative, as well as qualitative, analysis to my work.
These thought processes are desired and applicable to almost any field.”
- Denise Cammarata, Senior Engineer
4.1 Introduction
Anyone who has spent time perusing the problems in the AIME or other mathematical competitions
will be struck by how frequently there is a problem in the given year’s exam that refers to that year.
It is a common practice of question writers for these competitions to include the current year in a
problem or two, and so it is no surprise that coaches and mentors encourage their AIME (and other)
students to know the prime factorization for the given year before entering the competition.
Let us briefly recall the notion of a prime number and a divisor of an integer. We say that a nonzero
integer a is a divisor (or factor) of the integer b (or simply, a divides b), sometimes written a|b, if
there exists an integer r such that b = ar. A prime number is a positive integer p ≥ 2 whose only
positive divisors are 1 and p. Alternatively, p is prime if it is a positive integer with exactly two
distinct positive divisors.
Primes can be considered the “building blocks” for all integers. This idea is codified in the Fun-
damental Theorem of Arithmetic. Although many students inherently know what it says, it is so
central and important in number theory that it bears repeating.
57
58 CHAPTER 4. NUMBER THEORY
4.2 Fundamental Theorem of Arithmetic
The Fundamental Theorem of Arithmetic makes precise the notion that all integers can be built as
a product of primes, and that there is essentially only one way in which this can be done for each
integer.
Theorem 4.2.1. (Fundamental Theorem of Arithmetic) Every integer n ≥ 2 can be factored
as a product of prime integers:
n = pa1 1 pa2 2 pa3 3 . . . pakk ,
where p1 , p2 , . . . , pk are distinct primes, and a1 , a2 , . . . , ak are positive integers. Moreover, such a
factorization is unique up to the order in which the primes are listed.
The proof of this theorem is a classic example of the technique of strong induction and can be found
in texts on number theory, discrete mathematics, or strategies of proof. Let us observe that, with n
as in Theorem 4.2.1, any positive divisor d of n must take the form
d = pb11 pb22 · · · pbkk , (4.1)
where 0 ≤ bi ≤ ai for each i. Hence, there are ai + 1 possible values of bi . These possible values of
bi are mutually independent (over all values of i) from one another. Hence, by the Multiplication
Principle (Theorem 2.2.1), we have the following well-known result in number theory.
Theorem 4.2.2. With the notation as above, the number of positive divisors of n = pa1 1 pa2 2 · · · pakk
is
τ (n) = (a1 + 1)(a2 + 1) · · · (ak + 1). (4.2)
It is important not to overlook the significance of the uniqueness of the prime factorization of the
positive integer. Indeed, the uniqueness portion of the theorem is often the key to solving problems
in number theory on the AIME. Before we consider problems from the AIME, let us examine a
couple of problems that very aptly reflect the use of the Fundamental Theorem of Arithmetic.
Example 4.2.3. Let p be a prime number. Find all integers x such that x2 − 1 = 17p.
Solution: Notice that 17p is already expressed as a product of primes. Since x2 − 1 = 17p, the
uniqueness part of the Fundamental Theorem of Arithmetic requires that the factorization of x2 − 1
into a product of primes be the same. Writing x2 − 1 = (x − 1)(x + 1), note that x + 1 6= 1 and
x − 1 6= 1 since x2 − 1 > 17. Thus, either x + 1 or x − 1 must equal ±17 and the other must equal
±p.
Case 1: x − 1 = p and x + 1 = 17. This is impossible, since x + 1 = 17 implies that p = x − 1 = 15,

and 15 is not prime.
Case 2: x − 1 = 17 and x + 1 = p. In this case x + 1 = 19. Thus, x = 18 and p = 19.
Case 3: x − 1 = −p and x + 1 = −17. In this case, x = −18 and p = 19.
Case 4: x − 1 = −17 and x + 1 = −p. This is impossible, since x − 1 = −17 implies that
x + 1 = −15, so that p = 15, a contradiction.
4.2. FUNDAMENTAL THEOREM OF ARITHMETIC 59
From the analysis above, we see that the integers x that solve the equation are x = ±18. 2
The next example and variations of it are common in problem-solving forums focused on number
theory.
Example 4.2.4. Find the number of zeros at the end of the number 100!.
Solution: Another way to ask this question is: “How many factors of 10 are there in 100! ?” Since
10 = 2 · 5, the number of factors of 10 in 100! is determined by the number of factors of 2 and the
number of factors of 5. In fact, if we use the Fundamental Theorem of Arithmetic to express 100!
as a product of prime powers, say
100! = 2a1 3a2 5a3 7a4 · · · ,
where a1 , a2 , a3 , . . . are nonnegative integers, then the number of zeros in 100! will be determined
by the values of a1 and a3 , which dictate the number of factors of 2 and 5, respectively. Since every
even number contributes at least one factor of 2, and only multiples of 5 contribute factors of 5,
observe that a1 > a3 . In other words, if we imagine creating factors of 10 from factors of 2 and 5,
we encounter a shortage of factors of 5. Thus, the number of factors of 10 is precisely equal to a3 ,
the number of factors of 5 in 100!. Each of the following numbers contains at least one factor of 5:
5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100.
This list contains 20 numbers. However, the numbers 25, 50, 75, and 100 each contain an additional
factor of 5 (since they are each divisible by 25). Therefore, we actually have 24 factors of 5 in the
number 100!. Hence, the number of zeros at the end of 100! is 24. 2
As the solution presented to Example 4.2.4 illustrates, one can easily determine the number of factors
of 5 in 100! by enumerating all occurrences of 5 in its prime factorization. For larger numbers, this
technique may be less effective. The next example illustrates a more general strategy:
Example 4.2.5. For a given prime p, how many factors of p occur in the prime factorization of
N !?
Solution: To answer this, observe that every pth positive integer is divisible by p, every p2 th positive
integer is divisible by p2 , and so on. Hence, we can compute the number of factors of p in N ! by the
formula:
∞
X N
Number of factors of a prime p in N ! = , (4.3)
pk
k=1
where for any real number x, the notation bxc denotes the greatest integer less than or equal to x, or
more briefly, the “floor” of x. There are only finitely many nonzero terms in this sum, since pk > N
for all sufficiently large k. 2
Applying Equation (4.3) to solve Example 4.2.4 above, we can obtain the number of factors of 5 in
100! via
∞
X 100 100 100
= + = 20 + 4 = 24.
5k 5 25
k=1
Now let us employ Equation (4.3) to solve another problem. It is a slightly more complicated
variation on Example 4.2.4.
Example 4.2.6. Find the number of zeros at the end of the base-6 expansion of the number 100!.
Solution: We can write 100! uniquely in the form
100! = a0 + 6a1 + 62 a2 + · · · + 6k ak , (4.4)
where each ai belongs to {0, 1, 2, 3, 4, 5}. The base-6 representation of 100! is ak ak−1 . . . a2 a1 a0 . The
key observation is that the number of zeros at the end of this expression is precisely equal to the
number of factors of 6 that divide 100!. (For instance, if a0 = a1 = 0, then from (4.4), we see that
62 divides 100!.) Therefore, we rephrase the problem to ask: How many factors of 6 occur in 100!?
The analysis now proceeds along the same lines as that used in Example 4.2.4. Each factor of 6
requires one factor of 2 and one factor of 3. We will have less factors of 3 than 2, so we simply count
the factors of 3. This is achieved via Equation (4.3):
∞
X 100
= 33 + 11 + 3 + 1 = 48.
3k
k=1
Hence, there are 48 zeros at the end of the base-6 expansion of the number 100!. 2
Now let us consider some examples from the AIME.
Example 4.2.7. (2000 AIME, Problem #1) Find the least positive integer n such that no
matter how 10n is expressed as a product of two positive integers, at least one of these two integers
contains the digit 0.
Solution: We have 10n = 2n 5n . The two positive integers whose product is 10n must therefore be
built with powers of 2’s and 5’s. If one of the integers contains both a 2 and a 5, it will surely end
in a 0, so the only way that our integers can avoid ending in 0 is to take the integers to be 2n and
5n . We must simply find the smallest n such that one of these these integers contains a zero digit.
Since we do not expect n to be very large, we can simply enumerate a few cases:
n 2n 5n Conclusion
1 2 5 Neither contains a 0
8 256 390625 58 contains a 0
Thus, the answer is n = 8 = 008. ANSWER : 008. 2
Several number theory problems occurring on the AIME competition involve summing positive
integers. For this reason, some of the basic formulas for finite sums of positive integers will often
prove to be useful. This is a finite series, a topic that belongs most properly to the next chapter.
4.2. FUNDAMENTAL THEOREM OF ARITHMETIC 61
However, the formulas in Theorem 5.4.2, especially the first one,

m
X m(m + 1)
n = 1 + 2 + ··· + m = , (4.5)
n=1
2
will be useful in several problems appearing in this chapter and the exercises. Here is an example.
Example 4.2.8. (2009 AIME-2, Problem #4) A group of children held a grape-eating contest.
When the contest was over, the winner had eaten n grapes, and the child in kth place had eaten
n + 2 − 2k grapes. The total number of grapes eaten in the contest was 2009. Find the smallest
possible value of n.
Solution: From the given information, the winner eats n grapes, the second place child eats n − 2
grapes, the third place child eats n − 4 grapes, and so on. Thus, if ` denotes the number of children
participating in the contest, then the total number of grapes eaten, which is given as 2009, can be
expressed as
`
X
n + (n − 2) + (n − 4) + · · · + (n + 2 − 2`) = (n + 2 − 2k) = 2009.
k=1
Rearranging the left-hand side, we have
`n + [(−2) + (−4) + · · · + (2 − 2`)] = 2009,
or
`n + (−2)[1 + 2 + · · · + (` − 1)] = 2009.
Applying Equation (4.5) with m = ` − 1, we have
(−2)(` − 1)`
`n + = 2009,
2
or
`n − `(` − 1) = 2009.
Therefore, `(n−`+1) = 2009. As a result, we conclude that ` divides 2009. The (positive) divisors of
2009, which correspond to the possible values of `, can be determined using Equation (4.1), together
with the prime factorization 2009 = 72 · 41. We summarize the results in the table below. Note
that we are able to solve for the values of n in the last column by using the values in the first two
columns.
` n−`+1 n
1 72 · 41 2009
7 7 · 41 293
72 41 89
41 72 89
7 · 41 7 293
72 · 41 1 2009
From this table, we see that the smallest possible value of n is 89 = 089. ANSWER : 089. 2
4.3 Greatest Common Divisor and Least Common Multiple
Given two integers a and b, not both zero, the greatest common divisor of a and b is the largest
positive integer that divides both a and b. The greatest common divisor of a and b is often written
gcd(a, b). Observe that every nonzero integer d divides 0, so that for all integers b 6= 0, we have
gcd(0, b) = |b|.
A very important situation in number theory occurs when gcd(a, b) = 1. In this case, we say that a
and b are relatively prime. This is another way of saying that, when the prime factorizations of a
and b supplied by the Fundamental Theorem of Arithmetic are written down, they share no common
primes in their factorizations. There are many useful applications of the concept of relatively prime
integers. One nice connection with divisibility is the following:
Proposition 4.3.1. If a, b, and c are integers such that a and b are relatively prime, and a divides
bc, then a divides c.
Proposition 4.3.1 can be proven by using the uniqueness portion of the Fundamental Theorem of
Arithmetic (Theorem 4.2.1) and is left to the reader.
A useful tool related to relatively prime integers is the Euler-phi function, ϕ, defined on each
positive integer n to be the number of integers a with 1 ≤ a ≤ n such that gcd(a, n) = 1. That is,
for each positive integer n, we have

ϕ(n) := {a : a is an integer with 1 ≤ a ≤ n such that gcd(a, n) = 1}.
(4.6)
The reader can readily verify that ϕ(6) = 2, ϕ(20) = 8, and ϕ(77) = 60, for instance. In addition,
using the Subtraction Rule, we can easily check that for each prime p we have
ϕ(p) = p − 1, (4.7)
and more generally, for each positive integer k,
ϕ(pk ) = pk − pk−1 . (4.8)
To understand Equation (4.8), note that we are simply counting all integers a with 1 ≤ a ≤ pk and
removing the pk−1 integers in this range that contain a factor of p; namely, p, 2p, 3p, . . . , pk−1 · p.
Using a similar argument, which the reader is encouraged to think through, the Inclusion-Exclusion
Principle (Theorem 2.6.1) and the Subtraction Rule (Theorem 2.2.3) can be invoked to show that
if p and q are distinct primes, then ϕ(pq) = pq − p − q + 1.
Notice that we can write this as
ϕ(pq) = pq − p − q + 1 = (p − 1)(q − 1) = ϕ(p)ϕ(q).
More generally, if k and ` are nonnegative integers, the formula
ϕ(pk q ` ) = ϕ(pk )ϕ(q ` ) (4.9)

4.3. GREATEST COMMON DIVISOR AND LEAST COMMON MULTIPLE 63
follows easily from the Inclusion-Exclusion Principle and the Subtraction Rule, together with Equa-
tion (4.8). (It is actually true that for any two relatively prime positive integers n1 and n2 , we have
ϕ(n1 n2 ) = ϕ(n1 )ϕ(n2 ). However, we will have no need for this more general result in this text, and
it takes too much space to prove here.)
We will use the Euler-phi function and Equations (4.6), (4.7), (4.8), and (4.9) in several of our next
examples.
Example 4.3.2. How many ordered pairs of integers (a, b) with 1 ≤ a ≤ b ≤ 100 have
gcd(a, b) = 12?
Solution: Since both a and b must be multiples of 12 and 1 ≤ a ≤ b ≤ 100, we may write a = 12k
and b = 12`, where k and ` are integers with 1 ≤ k ≤ ` ≤ 8 and gcd(k, `) = 1. For each fixed value of
`, the number of possible choices for k is φ(`). Summing over all integer values of ` with 1 ≤ ` ≤ 8,
we obtain the answer:
X8
ϕ(`) = 1 + 1 + 2 + 2 + 4 + 2 + 6 + 4 = 22.
`=1
2
Example 4.3.3. (2005 AIME-2, Problem #4) Find the number of positive integers that are
divisors of at least one of 1010 , 157 , 1811 .
Solution: Let us determine the number of divisors of each of the three numbers given in this
problem. Write 1010 = 210 510 . According to Theorem 4.2.2, 1010 has (10 + 1)(10 + 1) = 121 divisors.
Next, write 157 = 37 57 . In this case, Theorem 4.2.2 tells us that there are (7 + 1)(7 + 1) = 64
divisors. Finally, we can write 1811 = 211 322 , so that 1811 has (11 + 1)(22 + 1) = 276 divisors.
We must recognize, however, that some of the divisors counted for these three numbers are the same,
so that 121 + 64 + 276 = 461 is an over-count of the actual number of divisors. We must apply
the Inclusion-Exclusion Principle (Theorem 2.6.2) to account for numbers that have been counted
multiple times.
Let
A = {positive divisors of 1010 }, B = {positive divisors of 157 }, C = {positive divisors of 1811 }.
We have already determined that
|A| = 121, |B| = 64, and |C| = 276.
To compute |A ∩ B|, note that the divisors that are common to both 1010 and 157 are precisely
divisors of gcd(1010 , 157 ) = 57 . Since there are eight divisors of 57 according to Theorem 4.2.2, we
have
|A ∩ B| = 8.
Similarly, since gcd(1010 , 1811 ) = 210 , there are 11 common divisors of both 1010 and 1811 , so
|A ∩ C| = 11.
Moreover, since gcd(157 , 1811 ) = 37 , there are eight common divisors of both 157 and 1811 , so that
|B ∩ C| = 8.
Finally, observe that there is only one common divisor of all three numbers, namely 1 itself, so that
|A ∩ B ∩ C| = 1.
Hence, by the Inclusion-Exclusion Principle, we have
|A ∪ B ∪ C| = 121 + 64 + 276 − 8 − 11 − 8 + 1 = 435.
ANSWER : 435. 2
Example 4.3.4. (2004 AIME, Problem #8) Define a regular n-pointed star to be the union of
n line segments P1 P2 , P2 P3 , . . . , Pn P1 such that
• the points P1 , P2 , . . . , Pn are coplanar and no three of them are collinear.

• each of the n line segments intersects at least one of the other line segments at a point other
than an endpoint,
• all of the angles at P1 , P2 , . . . , Pn are congruent,
• all of the n line segments P1 P2 , P2 P3 , . . . , Pn P1 are congruent, and
• the path P1 P2 . . . Pn P1 turns counterclockwise at an angle of less than 180◦ at each vertex.
There are no regular 3-pointed, 4-pointed, or 6-pointed stars. All regular 5-pointed stars are similar,
but there are two non-similar regular 7-pointed stars. How many non-similar regular 1000-pointed
stars are there?
Solution: This problem appears to involve geometric aspects, but as we shall see momentarily, at
its core this is a problem in number theory. The regular n-pointed star has n vertices that can be
viewed as equally-spaced points around a given circle. To draw such a star, we begin at any vertex
and proceed clockwise (without loss of generality) to connect every jth vertex by a line segment for
some positive integer j with j < n. Note that j is fixed throughout the star construction because of
the conditions given in the problem. Since each line segment must intersect another line segment,
j 6= 1 and j 6= n − 1. If j and n share a common factor d > 1, then the sequence of line segments
drawn will return to the initial vertex prior to completing the star. Moreover, connecting every jth
vertex results in exactly the same regular n-pointed star as connecting every (n − j)th vertex. Using
these facts, then, we can determine how many values of j result in non-similar regular n-pointed
stars. Figure 4.1 shows the 5, 6, 7, and 8-pointed stars.
We have already observed that only values of j with j < n such that gcd(j, n) = 1 can be used. There
are ϕ(n) such values of j. However, two of these values of j are 1 and n − 1, which we eliminated.
Therefore, we have ϕ(n) − 2 values of j that can produce regular n-pointed stars. However, the
number of stars actually created is only half of this, since j and n − j (which are distinct, since
if j = n − j, then gcd(j, n) = j 6= 1) both generate the same star. Hence, the number of regular
ϕ(n) − 2
n-pointed stars is .
2
4.3. GREATEST COMMON DIVISOR AND LEAST COMMON MULTIPLE 65
Figure 4.1: There is one 5-pointed star, no 6-pointed stars, two 7-pointed stars, and one 8-pointed
star.
In this case, we have n = 1000 = 23 · 53 . Thus, applying Equations (4.8) and (4.9), we obtain
ϕ(1000) = ϕ(23 )ϕ(53 ) = (23 − 22 )(53 − 52 ) = (8 − 4)(125 − 25) = 400.
Thus, the number of regular 1000-pointed stars is
ϕ(1000) − 2 398
= = 199.
2 2
ANSWER : 199. 2
In some instances, it is necessary to consider the greatest common divisor of a set of more than two
integers. For instance, given three integers a, b, and c (not all zero), we can ask how to compute
gcd(a, b, c), the greatest common divisor of a, b, and c. Of course, one could choose to define this as
a two-step process where the greatest common divisor of only two integers at a time are considered.
For instance, we could define
gcd(a, b, c) = gcd(gcd(a, b), c),
but then questions will arise about the uniqueness of the answer (e.g. does gcd(a, b, c) = gcd(c, a, b)
?) and about the practicality of the iterative process in the case of a large set of integers. Fortunately,
there is another approach tied more directly to the prime factorizations of the integers in question.
In terms of prime factorizations, if a and b are positive integers with a = pa1 1 pa2 2 · · · pakk and b =
pb11 pb22 · · · pbkk , where the pi are distinct primes, ai ≥ 0, and bi ≥ 0 for all i, then
min(a1 ,b1 ) min(a2 ,b2 ) min(ak ,bk )
gcd(a, b) = p1 p2 · · · pk . (4.10)
With this viewpoint, we can extend to a formula for the greatest common divisior of three or more
integers in a natural way. For instance, for positive integers a, b (as above), and c = pc11 pc22 · · · pckk
(with each ci ≥ 0), we have
min(a1 ,b1 ,c1 ) min(a2 ,b2 ,c2 ) min(ak ,bk ,ck )
gcd(a, b, c) = p1 p2 · · · pk . (4.11)
Finally in this section, we discuss the concept of the least common multiple. Given two nonzero
integers a and b, the least common multiple of a and b is the smallest positive integer that is
divisible by both a and b. It is often written as lcm(a, b). If we write the prime factorizations of a
and b as a = pa1 1 pa2 2 . . . pakk and b = pb11 pb22 . . . pbkk , where each pi is a prime, ai ≥ 0, and bi ≥ 0 for all
i, then
max(a1 ,b1 ) max(a2 ,b2 ) max(ak ,bk )
lcm(a, b) = p1 p2 · · · pk . (4.12)
Combining this with the expression for gcd(a, b) given in Equation (4.10), we can see that
ab = gcd(a, b) · lcm(a, b).
This follows from the fact that
min(ai , bi ) + max(ai , bi ) = ai + bi
for each i.
As with the greatest common divisor, we can extend the least common multiple to more than two
integers by generalizing Equation (4.12) as follows. Using the same notation introduced for Equation
(4.11), we have
max(a1 ,b1 ,c1 ) max(a2 ,b2 ,c2 ) max(ak ,bk ,ck )
lcm(a, b, c) = p1 p2 · · · pk . (4.13)
Let us consider a couple of examples from the AIME.

Example 4.3.5. (1998 AIME, Problem #1) For how many values of k is 1212 the least common
multiple of the positive integers 66 , 88 , and k?
Solution: Let us begin by writing the numbers in this problem as products of primes. We have
1212 = 224 312 , 66 = 26 36 , and 88 = 224 .
We wish to determine how many values of k satisfy
lcm(26 36 , 224 , k) = 224 312 .
Using Equation (4.12) to combine the first two numbers on the left-hand side, this becomes
lcm(224 36 , k) = 224 312 .
We can apply Equation (4.12) once more to see that k must contain exactly 12 factors of 3 and no
more than 24 factors of 2 in its prime factorization. That is, provided k = 2j 312 with j an integer
satisfying 0 ≤ j ≤ 24, then the maximum powers of 2 and 3 occurring among the three numbers
26 36 , 224 , and k will be 24 and 12, respectively, as required. There are 25 different values of j
meeting the requirement, so the answer here is 25 = 025. ANSWER : 025. 2
Example 4.3.6. (1987 AIME, Problem #7) Let [r, s] denote the least common multiple of
positive integers r and s. Find the number of ordered triples (a, b, c) of positive integers for which
[a, b] = 1000, [b, c] = 2000, and [c, a] = 2000.
Solution: Since the numbers [a, b], [b, c], and [c, a] contain only factors of 2 and 5, we conclude that
the only prime factors of a, b, and c are 2 and 5. Thus, we can write
a = 2a1 5a2 , b = 2b1 5b2 , c = 2c1 5c2 ,
where a1 , a2 , b1 , b2 , c1 , c2 are nonnegative integers. From the fact that [a, b] = 1000 = 23 53 , [b, c] =
2000 = 24 53 , and [c, a] = 2000 = 24 53 , we see from Equation (4.12) that
max(a1 , b1 ) = 3, max(b1 , c1 ) = 4, max(c1 , a1 ) = 4, (4.14)

4.4. MODULAR ARITHMETIC 67
max(a2 , b2 ) = 3, max(b2 , c2 ) = 3, max(c2 , a2 ) = 3. (4.15)

We now must determine the number of different ordered 6-tuples (a1 , a2 , b1 , b2 , c1 , c2 ) satisfying (4.14)
and (4.15), since there is a one-to-one correspondence between such 6-tuples and ordered triples
(a, b, c). Let us begin by using (4.14) to determine the number of distinct choices for (a1 , b1 , c1 ).
Since max(a1 , b1 ) = 3 and max(b1 , c1 ) = 4, we must have c1 = 4. We can consider two cases.
Case 1: a1 = 3 and b1 ≤ 3. Hence, there are four triples (a1 , b1 , c1 ) in this case; namely:
(3, 0, 4), (3, 1, 4), (3, 2, 4), and (3, 3, 4).
Case 2: a1 ≤ 2 and b1 = 3. Thus, we have the three triples (0, 3, 4), (1, 3, 4), and (2, 3, 4) in this
case.
Putting the two cases together, we have a total of seven different triples (a1 , b1 , c1 ).
Now let us use (4.15) to determine the number of triples (a2 , b2 , c2 ).
Case 1: a2 = 3. In this case, (4.15) dictates that either b2 = 3 or c2 = 3, and the other variable can
be 0,1,2, or 3. There are seven such triples (a2 , b2 , c2 ); namely, (3, 3, 0), (3, 3, 1), (3, 3, 2), (3, 3, 3),
(3, 2, 3), (3, 1, 3), and (3, 0, 3).
Case 2: a2 < 3. In this case, (4.15) dictates that b2 = c2 = 3. Hence, we obtain the three triples
(2, 3, 3), (1, 3, 3), and (0, 3, 3) in this case.
Putting these two cases together, we have a total of ten different triples (a2 , b2 , c2 ).
Since the triples (a1 , b1 , c1 ) and (a2 , b2 , c2 ) can be chosen independently, we use the Multiplication
Principle (Theorem 2.2.1) to obtain 7 · 10 = 70 different 6-tuples (a1 , a2 , b1 , b2 , c1 , c2 ). Thus, the
answer is 70 = 070. ANSWER : 070. 2
4.4 Modular Arithmetic
Anyone who has used a clock to tell time has experience with modular arithmetic. If it is 11 o’clock
now, then four hours from now, it will be 3 o’clock. The idea is that although 11 + 4 = 15, modular
arithmetic (in this case “modulo 12”) allows us to “reduce” 15 to 3. We write 15 ≡ 3 (mod 12).
In general, we fix a positive integer n ≥ 2 and proceed as follows. For integers a and b, we write
a ≡ b (mod n) if n divides a − b. We say that a and b are congruent modulo n. For a fixed
positive integer n ≥ 2, every integer a is congruent (mod n) to exactly one number in the set
{0, 1, 2, . . . , n − 1}. Modular arithmetic respects basic arithmetic operations as follows.
Proposition 4.4.1. Fix a positive integer n ≥ 2, and let a, b, c, and d be integers. Then
1. If a ≡ b ( mod n) and c ≡ d ( mod n), then a ± c ≡ b ± d ( mod n).
2. If a ≡ b ( mod n) and c ≡ d ( mod n), then ac ≡ bd ( mod n).

3. If a ≡ b ( mod n) and c is any positive integer, then ac ≡ bc ( mod nc).

a b n
4. If a ≡ b ( mod n) and d is a positive common divisor of a, b, and n, then ≡ ( mod ).
d d d
5. If a ≡ b ( mod n) and m ≥ 2 and m divides n, then a ≡ b ( mod m).
These properties are easy to verify directly from the definition of congruent modulo n given above,
but that is not our primary concern here. Instead, let us consider some examples.
Example 4.4.2. (2010 AIME, Problem #2) Find the remainder when
9 × 99 × 999 × · · · × 99 · · · 9}
| {z
999 90 s
is divided by 1000.
Solution: Formulated in terms of modular arithmetic, we are seeking to determine the value of
9 × 99 × 999 × · · · × 99 · · · 9}
| {z (mod 1000) .
999 90 s
According to property (2) in Proposition 4.4.1, we can compute each of the factors in the product
modulo 1000 separately first, and then multiply the results. Note that, modulo 1000, each of the
numbers 999, 9999, 99999, 999999, and so on reduces to 999. Moreover, 999 ≡ −1 (mod 1000).
Thus, we have
9 × 99 × 999 × · · · × 99 · · · 9} ≡ 9 · 99 · 999 · 9999 · · ·

| {z (mod 1000)
999 90 s
≡ 9 · 99 · (−1) · (−1) · · · (mod 1000)
997
≡ 9 · 99 · (−1) (mod 1000)
≡ 9 · 99 · (−1) (mod 1000)
≡ −891 (mod 1000)
≡ 109. (mod 1000)
Hence, the answer is 109. ANSWER : 109. 2

Example 4.4.3. (1996 AIME, Problem #9) A bored student walks down a hall that contains
a row of closed lockers, numbered 1 to 1024. He opens the locker numbered 1, and then alternates
between skipping and opening each closed locker thereafter. When he reaches the end of the hall, the
student turns around and starts back. He opens the first closed locker he encounters, and then alter-
nates between skipping and opening each closed locker thereafter. The student continues wandering
back and forth in this manner until every locker is open. What is the number of the last locker he
opens?
Solution: Notice that the student will open exactly half of the remaining closed lockers on each pass
by the lockers. To understand what is going on, it is helpful to examine the first several passes. On
the first pass, he opens every odd-numbered locker. On the second pass, he begins with the locker
4.4. MODULAR ARITHMETIC 69
numbered 1024 and opens lockers numbered 1024, 1020, 1016, . . . , 12, 8, and 4. That is, he opens
lockers whose number is a multiple of 4. Which lockers are still closed? They are even-numbered
lockers that are not multiples of 4. That is, lockers with number n where n ≡ 2 (mod 4) are still
closed. Mathematically, note that values of n such that n ≡ 2 (mod 4) are precisely the same as
values of n such that n ≡ 2 or 6 (mod 8). On the third pass, those lockers whose values of n satisfy
n ≡ 2 (mod 8) will be opened, while lockers numbered with n such that n ≡ 6 (mod 8) remain
closed. Now n ≡ 6 (mod 8) is equivalent to n ≡ 6 or 14 (mod 16). Since the fourth pass begins with
the larger-numbered lockers, the lockers with value n such that n ≡ 14 (mod 16) are opened on the
fourth pass, while the lockers with value n such that n ≡ 6 (mod 16) remain closed. Continuing
as before, note that n ≡ 6 (mod 16) is equivalent to n ≡ 6 or 22 (mod 32). On the fifth pass,
lockers with values of n such that n ≡ 6 (mod 32) are opened, while lockers with values of n such
that n ≡ 22 (mod 32) remain closed. It is possible to continue with this analysis in like manner
through the remaining passes to obtain the answer, but we wish to illustrate another way to obtain
the answer from this point in the process. There are only 32 lockers between 1 and 1024 that are
labelled with n such that n ≡ 22 (mod 32):
22, 54, 86, 118, 150, 182, 214, 246, 278, 310, 342, 374, 406, 438, 470, 502, 534,
566, 598, 630, 662, 694, 726, 758, 790, 822, 854, 886, 918, 950, 982, 1014.
At this point, the student is at the far end of the hall (where the larger numbered lockers are). One
can either directly work out the sequence in which the lockers are opened from this point forward,
or one can observe that, after five more passes, the last locker to be opened will be the 22nd one
in the sequence (by essentially repeating the process that led to the conclusions after the first five
passes), counted from right-to-left. The 22nd number in the list above, counted from right-to-left, is
342. ANSWER : 342. 2
Example 4.4.4. (2001 AIME-2, Problem #10) How many positive integer multiples of 1001
can be expressed in the form 10j − 10i , where i and j are integers and 0 ≤ i < j ≤ 99?
Solution: We are seeking the number of positive integers m such that 1001m = 10j −10i . Factoring
on each side, we have
7 · 11 · 13 · m = 10i (10j−i − 1).
By the uniqueness of prime factorizations guaranteed by the Fundamental Theorem of Arithmetic,
10i (10j−i − 1) must be divisible by 7, 11, and 13. None of the primes 7, 11, or 13 divide 10i for any
i since the only prime factors of 10i are 2 and 5. Therefore, 7,11, and 13 must each divide 10j−i − 1.
In other words, the integers i and j must satisfy 0 ≤ i < j ≤ 99 such that
10j−i ≡ 1 (mod 7), 10j−i ≡ 1 (mod 11), 10j−i ≡ 1 (mod 13).
By direct calculation, we see that for j − i = 1, 2, 3, 4, 5 we have 10j−i 6≡ 1 (mod 7). However,
106 ≡ 36 ≡ 23 ≡ 1 ( mod 7),
106 ≡ (−1)6 ≡ 1 ( mod 11),

and
106 ≡ (−3)6 ≡ (−27)2 ≡ (−1)2 ≡ 1 ( mod 13).
Now, using part (2) of Proposition 4.4.1, we can conclude that for a positive integer k, we have
10k ≡ 1 (mod 7) , 10k ≡ 1 (mod 11), 10k ≡ 1 (mod 13)
if and only if k is a multiple of 6. Hence, we are looking for how many ways j − i can equal a positive
multiple of 6 if 0 ≤ i < j ≤ 99. The results appear in the table below:
j−i Number of pairs (i, j) with 0 ≤ i, j ≤ 99

6 94
12 88
18 82
24 76
30 70
36 64
42 58
48 52
54 46
60 40
66 34
72 28
78 22
84 16
90 10
96 4
Summing the numbers in the second column, we obtain the answer

94 + 88 + 82 + 76 + 70 + 64 + 58 + 52 + 46 + 40 + 34 + 28 + 22 + 16 + 10 + 4 = 784.
ANSWER : 784. 2
4.5 Divisibility Tests
We conclude this chapter with a brief section highlighting some basic divisibility tests that are
often useful in solving AIME-level problems in number theory, but are also surprisingly helpful in
everyday math situations, and on a larger scale, they are very important in one of the most significant
modern-day applications of number theory, the field of cryptography. Let N be a positive integer
with (unique) decimal representation
N = a0 + 10a1 + 100a2 + 1000a3 + · · · + 10k ak , (4.16)
where each ai belongs to the set {0, 1, 2, . . . , 9}. The first result shows that there is no need to have
a direct divisibility test for each positive integer m:
Theorem 4.5.1. For relatively prime positive integers m and n, we have
N is divisible by mn if and only if N is divisible by both m and n.
4.5. DIVISIBILITY TESTS 71
For instance, if we are interested in knowing if N is divisible by 6, Theorem 4.5.1 tells us that we
only need to check if N is divisible by 2 and 3. We will not take the space to prove Theorem 4.5.1
here, but the implications of it are that it is sufficient to have divisibility tests for divisors of the
form m = pk for a prime p and positive integer k. Generally, as p and k get larger, such tests are
less practical.
We will only present a few of the easiest tests here. The basic idea is to view Equaton (4.16) modulo
pk and reduce the powers of 10 to manageable values (modulo pk ). Therefore, we are relying on the
notions and notations related to modular arithmetic discussed in Section 4.4.
Divisibility by 2: Note that

N ≡ a0 (mod 2) ,
so we have:
Proposition 4.5.2. Let N be a positive integer with decimal representation given in Equation
(4.16). Then N is divisible by 2 if and only if a0 is divisible by 2.
That is, N is even if and only if the right-most digit of N is even.
Divisibility by 3: We have
N ≡ a0 + a1 + a2 + · · · + ak (mod 3) ,
since 10j ≡ 1 (mod 3) for each nonnegative integer j. Thus:
(4.16). Then N is divisible by 3 if and only if the sum of the digits of N is divisible by 3.
N ≡ a0 + 10a1 (mod 4) ,
which implies that:
(4.16). Then N is divisible by 4 if and only if the two-digit number formed by its two right-most
digits is divisible by 4.
Divisibility by 5: This is similar to divisibility by 2, since
N ≡ a0 (mod 5) .
Therefore:
(4.16). Then N is divisible by 5 if and only if a0 is divisible by 5.
Divisibility by 7: This one is not so easy. Several tests are available, but in general require multiple
steps. Here is one of the easiest tests to explain: Given N as above, divisibility by 7 can be reduced
to the same question for the number
N1 := −2a0 + a1 + 10a2 + 100a3 + · · · + 10k−1 ak .
The process can then be repeated for N1 and as many times as necessary until we can determine
the divisibility by 7 of the reduced number by “inspection”.
N ≡ a0 + 10a1 + 100a2 (mod 8) ,
which implies that:

(4.16). Then N is divisible by 8 if and only if the three-digit number formed by its three right-most
digits is divisible by 8.
N ≡ a0 + a1 + a2 + · · · + ak (mod 9) ,
since 10j ≡ 1 (mod 9) for each nonnegative integer j. Thus:

(4.16). Then N is divisible by 9 if and only if the sum of the digits of N is divisible by 9.
N ≡ a0 − a1 + a2 − · · · + (−1)k ak (mod 11) ,
since 10j ≡ (−1)j (mod 11) for each nonnegative integer j. Thus:
(4.16). Then N is divisible by 11 if and only if the alternating sum a0 − a1 + a2 − · · · of the digits
of N is divisible by 11.
4.6 Exercises
1. (2007 AIME-2, Problem #2) Find the number of ordered triples (a, b, c) where a, b, and c
are positive integers, a is a factor of b, a is a factor of c, and a + b + c = 100.
2. (2006 AIME-2, Problem #3) Let P be the product of the first 100 positive odd integers.
Find the largest integer k such that P is divisible by 3k .
4.6. EXERCISES 73
3. (2008 AIME, Problem #4) There exist unique positive integers x and y that satisfy the
equation x2 + 84x + 2008 = y 2 . Find x + y.
4. (1989 AIME, Problem #3) Suppose n is a positive integer and d is a single digit in base
10. Find n if
n
= 0.d25d25d25 . . . .
810
5. (2006 AIME, Problem #4) Let N be the number of consecutive 0’s at the right end of
the decimal representation of the product 1!2!3!4! · · · 99!100!. Find the remainder when N is
divided by 1000.
6. 1994 AIME, Problem #5) Given a positive integer n, let p(n) be the product of the nonzero
digits of n. (If n has only one digit, then p(n) is equal to that digit.) Let
S = p(1) + p(2) + p(3) + · · · + p(999).
What is the largest prime factor of S?
7. (2006 AIME-2, Problem #7) Find the number of ordered pairs of positive integers (a, b)
such that a + b = 1000 and neither a nor b has a zero digit.
8. (1999 AIME, Problem #7) There is a set of 1000 switches, each of which has four positions,
called A, B, C, and D. When the position of any switch changes, it is only from A to B, from
B to C, from C to D, or from D to A. Initially each switch is in position A. The switches are
labeled with the 1000 different integers 2x 3y 5z , where x, y, and z take on the values 0, 1, . . . , 9.
At step i of a 1000-step process, the ith switch is advanced one step, and so are all the other
switches whose labels divide the label on the ith switch. After step 1000, has been completed,
how many switches will be in position A?
9. (2008 AIME, Problem #7) Let Si be the set of all integers n such that 100i ≤ n < 100(i+1).
For example, S4 is the set {400, 401, 402, . . . , 499}. How many of the sets S0 , S1 , S2 , . . . , S999
do not contain a perfect square?
10. (2004 AIME-2, Problem #8) How many positive integer divisors of 20042004 are divisible
by exactly 2004 positive divisors?
11. (1993 AIME, Problem #9) Two thousand points are given on a circle. Label one of the
points 1. From this point, count 2 points in the clockwise direction and label this point 2.
From the point labeled 2, count 3 points in the clockwise direction and label this point 3. (See
figure.) Continue this process until the labels 1, 2, 3 . . . , 1993 are all used. Some of the points
on the circle will have more than one label and some points will not have a label. What is the
smallest integer that labels the same point as 1993?
FIGURE from 1993 AIME, Problem 9 GOES HERE !!!

12. (2000 AIME, Problem #11) Let S be the sum of all numbers of the form a/b, where a
and b are relatively prime positive divisors of 1000. What is the greatest integer that does not
exceed S/10?
13. (1987 AIME, Problem #11) Find the largest possible value of k for which 311 is expressible
as the sum of k consecutive positive integers.
14. (2005 AIME, Problem #12) For positive integers n, let τ (n) denote the number of positive
integer divisors of n, including 1 and n. For example τ (1) = 1 and τ (6) = 4. Define S(n) by
S(n) = τ (1) + τ (2) + · · · + τ (n).
Let a denote the number of positive integers n ≤ 2005 with S(n) odd, and let b denote the
number of positive integers n ≤ 2005 with S(n) even. Find |a − b|.
15. (1992 AIME, Problem #15) Define a positive integer n to be a factorial tail if there is
some positive integer m such that the decimal representation of m! ends with exactly n zeros.
How many positive integers less than 1992 are not factorial tails?
Chapter 5
Sequences and Series
“Mathematics can be characterized as the science of patterns. Finding a pattern is a

powerful problem-solving strategy.”
- National Council of Teachers of Mathematics
5.1 Introduction
Formally, a sequence of elements belonging to a set S is a function f : N → S, where N =

{1, 2, 3, . . . } is the set of positive integers. Informally, however, we usually just represent the sequence
as a list of terms, a1 , a2 , a3 , . . . . Here, an stands for the value of f (n) in the formal definition given
above. Sometimes the notation for a sequence is abbreviated even further to simply (an )n∈N or (an ).
The definition of a sequence given formally above can be modified in order for the beginning term
to be a0 , or in fact, ai for any integer i. The first term of a sequence is most commonly a1 or a0 ,
however. Any finite subset of (an )n∈N is sometimes called a finite sequence.
In the case of an infinite sequence where S = R, the set of real numbers, we can discuss the issue
of convergence of the sequence of real numbers. Loosely speaking, we say that (an ) converges to a
real number a if the terms an can be made arbitrarily close1 to a for sufficiently large n. While it
is possible to make this notion much more precise, that is not our objective here.
In the case of a sequence of real numbers, it is often of interest to add the terms of the sequence.
Of course, since the sequence contains infinitely many values, it is certainly possible for this sum to
be undefined. Whether meaningful or not, the sum of the terms of a sequence is formally referred
to as a series:
X∞
S= an .
n=1
1 The distance between two real numbers a and b is measured by computing the absolute value of their difference,
|a − b|.
75
76 CHAPTER 5. SEQUENCES AND SERIES
Of course, it may be of interest only to sum certain terms of the sequence, in which case the range of
values of n given in this summation can be modified accordingly. If S can be computed and results
in a finite value, we say that the series converges. Otherwise, the series diverges. Formally, the
X∞
series S = an is said to converge if the sequence (Sk ) of its partial sums
n=1
k
X
Sk := an
n=1
converges. A divergent series is not necessarily the result of obtaining an infinitely large (unbounded)
X∞
sum, as the sequence (−1)n demonstrates: the partial sums Sk oscillate between −1 (for k odd)
n=1
and 0 (for k even); thus, the sequence (Sk )k∈N of partial sums does not converge.
A necessary condition for a series to converge is that the terms an approach zero as n becomes
arbitrarily large. However, this condition alone does not guarantee convergence; the harmonic series
∞
X 1
is well-known to diverge, despite having terms approaching zero as n → ∞.
n=1
n
Many problems in the AIME competition involve sequences and series. In some cases, terms of a
sequence may be given, while in others, a recursive definition2 for the terms of the sequence may
be given instead. Since series are only understood in the context of sequences, let us begin our
discussion in this chapter with sequences.
5.2 Sequences
Problems involving sequences often require the solver to discover a pattern. Such a pattern, if
present, may not be apparent at the outset of the problem. In order to find the pattern, we offer
the following:
General Strategy Tip: When working with sequences, it is often worthwhile to enumerate the
first several terms of the sequence to see whether or not a pattern emerges after a few terms.
The two most commonly occurring types of sequences in the AIME are arithmetic sequences and
geometric sequences, so we will concentrate mainly on these in the next few pages. Both of these
types of sequences exhibit familiar patterns, as we discuss below, that make them easily recognizable
after enumerating the first few terms in accordance with the strategy tip above.
2 In a recursive definition, one obtains the terms a of the sequence via a formula relating a to previous terms
n n
an−1 , an−2 , . . . of the sequence.
5.2. SEQUENCES 77
5.2.1 Arithmetic Sequences
A sequence of real numbers (an )n∈N is called an arithmetic sequence if there is a constant k such
that an − an−1 = k for all n ≥ 2. The value k is sometimes referred to as the common difference
for the arithmetic sequence.
Another way of describing an arithmetic sequence is to list its terms as
a1 , a1 + k, a1 + 2k, a1 + 3k, a1 + 4k, .... (5.1)
Generally speaking, the nth term of an arithmetic sequence with common difference d and first term
a1 is
an = a1 + (n − 1)k.
Example 5.2.1. (AIME 2005, Problem #2) For each positive integer k, let Sk denote the
increasing arithmetic sequence of integers whose first term is 1 and whose common difference is k.
For example, S3 is the sequence 1, 4, 7, . . . . For how many values of k does Sk contain the term
2005?
Solution: The terms of the sequence Sk are
1, 1 + k, 1 + 2k, 1 + 3k, ....
We must determine for how many values of k we can write 2005 = 1 + nk for some positive integer
n. Subtracting 1 from both sides, this equation is equivalent to nk = 2004. Therefore, a necessary
and sufficient condition for solvability of this equation for the positive integer k is that k is a divisor
of 2004 = 22 · 3 · 167. Recall from Theorem 4.2.2 that the number of divisors of pa1 1 pa2 2 . . . pakk is
(a1 + 1)(a2 + 1) · · · (ak + 1). Thus, the number of positive divisors of 2004 is (2 + 1)(1 + 1)(1 + 1) =
3 · 2 · 2 = 12. Therefore, we have 12 values of k for which we can solve the equation 2005 = 1 + nk
for positive integer n. Hence, the answer is 12 = 012. ANSWER : 012. 2
Example 5.2.2. (AIME 2003-2, Problem #8) Find the eighth term of the sequence
1440, 1716, 1848, . . .
whose terms are formed by multiplying the corresponding terms of two arithmetic sequences.
Solution #1: Let us denote the terms of one of the arithmetic sequences by
a, a + k, a + 2k, a + 3k, ..., (5.2)
and the terms of the other arithmetic sequence by
b, b + `, b + 2`, b + 3`, .... (5.3)
We are given that

ab = 1440, (5.4)
(a + k)(b + `) = 1716, (5.5)
(a + 2k)(b + 2`) = 1848, (5.6)

and we wish to determine the value of the eighth term,
(a + 7k)(b + 7`) = ab + 7(a` + bk) + 49k`.
Subtracting (5.4) from (5.5), we obtain
a` + bk + k` = 276, (5.7)
and subtracting (5.4) from (5.6), we obtain
2a` + 2bk + 4k` = 408,
or equivalently,
a` + bk + 2k` = 204. (5.8)
Subtracting (5.7) from (5.8), we deduce that k` = −72. Substituting this conclusion into (5.7), we
find that a` + bk = 348. Therefore,
(a + 7k)(b + 7`) = ab + 7(a` + bk) + 49k` = 1440 + 7(348) + 49(−72) = 348.
Remark: In both this solution and the next one, it is not necessary to compute the terms in the
two arithmetic sequences explicitly. ANSWER : 348. 2
Solution #2: As in Solution #1, let us denote the terms of one of the arithmetic sequences by (5.2)
and (5.2), respectively. Thus, the nth term in the sequence formed by multiplying the corresponding
terms of the two arithmetic sequences is
f (n) := (a + (n − 1)k)(b + (n − 1)`) = (k`)n2 ,
which is a quadratic function of n, say
f (n) = c0 + c1 n + c2 n2 ,
for some real constants c0 , c1 , c2 . Since we are given f (1) = 1440, f (2) = 1716, and f (3) = 1848,
we can solve for the constants c0 , c1 , c2 above and then evaluate f (8). We leave the details to the
reader to fill in. 2
5.2.2 Geometric Sequences
The condition defining an arithmetic sequence, namely, that the difference in value of consecutive
terms is constant, can be modified by requiring that the ratio of consecutive terms maintain a
constant value. The resulting sequence is known as a geometric sequence.
More precisely, a sequence of nonzero real numbers (an )n∈N is called a geometric sequence if there
an
is a nonzero constant r such that = r for all n ≥ 2. The constant r is sometimes referred
an−1
5.2. SEQUENCES 79
to as the common ratio for the geometric sequence. Of course, if (an ) is a geometric sequence of
integers, then the common ratio r must be a rational number.
Listing the first few terms of a geometric sequence explicitly, we have
a1 , a1 r, a1 r2 , a1 r3 , a1 r4 , ....
In fact, we have the general relation

an = a1 rn−1
for determining the nth term of a geometric sequence. Already with this limited background on
geometric sequences, we are ready to solve the following AIME problem.
Example 5.2.3. (AIME 2012-2, Problem #2) Two geometric sequences a1 , a2 , a3 , . . . and
b1 , b2 , b3 , . . . have the same common ratio, with a1 = 27, b1 = 99, and a15 = b11 . Find a9 .
Solution: Let the common ratio of both sequences be denoted by r. We have
a15 = a1 r14 = 27r14 and b11 = b1 r10 = 99r10 .
Hence, we are given that

27r14 = 99r10 ,
which can be simplified to
3r4 = 11.
We actually do not need to solve this for r, since we are asked to find
2
11 27 · 121
a9 = a1 r8 = 27(r4 )2 = 27 = = 3 · 121 = 363.
3 9
ANSWER : 363. 2
As Example 5.2.3 shows, some AIME problems involving sequences involve just a basic knowledge
to solve, in this case the definition of a geometric sequence. Now let us consider a trickier exam-
ple involving a sequence that requires the simultaneous consideration of arithmetic and geometric
properties.
Example 5.2.4. (AIME 2003, Problem #8) In an increasing sequence of four positive integers,
the first three terms form an arithmetic progression, the last three terms form a geometric progression,
and the first and fourth terms differ by 30. Find the sum of the four terms.
Solution: Since the first three terms form an arithmetic progression, we can write the first three
terms as a, a + k, and a + 2k. Alternatively, since the last three terms form a geometric progression,
we may write the last three terms as b, br, and br2 . Either approach is possible. We will take the
latter approach and invite the interested reader to pursue the former (which actually turns out to
be simpler!).
Since the last three terms form an increasing sequence of positive integers, we have b ≥ 1 and r > 1,
and owing to the fact that the first three terms form an arithmetic progression, the first term, x,
must be such that b − x = br − b. Thus, the first term is x = 2b − br = b(2 − r) > 0, from which we
deduce that r < 2. Therefore, we have the increasing sequence of positive integers
2b − br, b, br, br2 .
In order for this to be an increasing sequence consisting of positive integers, it is necessary that r is
a rational number with 1 < r < 2. We are also given that
30 = br2 − (2b − br),
or
2 30
r +r− 2+ = 0.
b
Using the quadratic formula and the fact that r must be positive, we deduce that
√
q
−1 + 9 + 120 b −1 + 1b 9b2 + 120b
r= = .
2 2
Since
√ the terms of the sequence are all integers, we know that r must be rational. This requires that
9b2 + 120b be an integer. In other words, 9b2 + 120b must be a perfect square. In particular, each
prime factor occuring in 9b2 + 120b must occur an even number of times. From this observation, it
follows that b must be divisible by 3, say b = 3c (for some integer c ≥ 1). Hence,
9b2 + 120b = 81c2 + 360c = 9(9c2 + 40c)
must be a perfect square. Hence,

9c2 + 40c
must be a perfect square (larger than (3c)2 ). Writing
9c2 + 40c = (3c + t)2 ,
for some integer t ≥ 1, we can attempt to solve for t. Simplifying, we have
40c = 6ct + t2 .
Note that
40
t< ≈ 6.67 < 7,
6
so we can simply try the list of values of t dictated by these restrictions:
t c
1 1/34
2 1/7
3 9/22
4 1
5 5/2
6 9
5.3. SERIES 81
There are only two integer values for c given in this table. If c = 1, then b = 3, and the reader can
quickly verify that r = 3, contrary to 1 < r < 2. Thus, we must have c = 9 and b = 27, and the
reader can quickly verify that r = 4/3.
Hence, the sum of the four terms is
16
2b − br + b + br + br2 = 3b + br2 = 3 · 27 + 27 · = 81 + 48 = 129.
9
ANSWER : 129. 2
5.3 Series
Consider a sequence (an )n∈N of real numbers. A series consists of a sum of terms from the sequence
(an )n∈N . If we wish to sum the terms of the sequence from the rth term to the sth term (with
integers r ≤ s), inclusive, we denote the series by
s
X
an .
n=r
Such a series, consisting of a sum of a finite number of terms, is referred to as a finite series. A
common example of a finite series is
m
X
n = 1 + 2 + 3 · · · + m.
n=1
Of course, every finite series converges. For instance, it is well-known (and we will review in Section
5.4) that we have
m
X m(m + 1)
n = 1 + 2 + 3 + ··· + m = . (5.9)
n=1
2
On the other hand, if we wish to sum all terms of a series beginning with the rth term, we write
∞
X
an .
n=r
This is an infinite series. To explain with mathematical rigor how one can sum an infinite collection
of numbers in a finite amount of time requires some work. This is typically done in calculus courses
by using the concept of a sequence of partial sums. In the case of an infinite series, it is possible
for the infinite sum to diverge; that is, the sum of the terms may not result in a finite value. For
instance, if an = n for all positive integers n, then
∞
X ∞
X
an = n = ∞.
n=1 n=1
We say such a series diverges to ∞. It stands to reason that any infinite series that converges to
a finite value must have terms an that become arbitrarily close to zero as n gets arbitrarily large.
While this condition is necessary for convergence, it alone is not sufficient. For instance, even the
∞
X 1
harmonic series fails to converge. However, it is an intriguing result in analysis that
n=1
n
∞
X 1 π2
= ,
n=1
n2 6
and hence converges. More generally, it is a well-known result from calculus that, for any real
number p, the sequence
∞
X 1
n=1
np
converges if and only if p > 1.

Let us next consider arithmetic and geometric series. These series arise as sums of terms of arithmetic
and geometric sequences.
5.3.1 Arithmetic Series
Suppose (an )n∈N is an arithmetic sequence with common difference k. It is easy to see that an infinite
arithmetic series cannot converge except in the trivial case where a1 = 0 and k = 0. However, we can
consider the case of a finite arithmetic series. In the calculation below, we will appeal to Equation
(5.9) with m − 1 in place of m on the fourth line.
m
X
an = a1 + a2 + · · · + am
n=1
= a1 + (a1 + k) + (a1 + 2k) + · · · + (a1 + (m − 1)k)
= ma1 + k(0 + 1 + 2 + · · · + (m − 1))
(m − 1)m
= ma1 + k
2
m
= (2a1 + k(m − 1))
2
m
= (a1 + (a1 + k(m − 1)))
2
m
= (a1 + am ).
2
Notice that
m
a1 + am 1 X
= an , (5.10)
2 m n=1
so that the average of the first and last term of an arithmetic series is the same as the average of all
terms of the series. With a little thought, this observation should not be surprising. Let us consider
an example from the AIME.
5.3. SERIES 83
Example 5.3.1. (2012 AIME, Problem #2) The terms of an arithmetic sequence add to 715.
The first term of the sequence is increased by 1, the second term is increased by 3, the third term is
increased by 5, and in general, the kth term is increased by the kth odd positive integer. The terms
of the new sequence add to 836. Find the sum of the first, last, and middle terms of the original
sequence.
Solution: Using the notation above this problem, there are m terms in the sequence. Thus,
715 + [1 + 3 + 5 + · · · + (2m − 1)] = 836.
Now the sum of the first m odd positive integers is m2 . (You can either observe this for small values
of m until you see the pattern, or you can prove it rigorously using the same technique discussed
below part (1) of Theorem 5.4.2 below.) Thus,
715 + m2 = 836.
That is,
m2 = 121.
We conclude that m = 11. So we are being asked to compute a1 + a6 + a11 . Observe from Equation
(5.10) that
11
a1 + a11 1 X 715
= an = .
2 11 n=1 11
Hence,
2
a1 + a11 = · 715.
11
Now observe that because the sequence is arithmetic, the middle term a6 is the average of the first
and last terms:
1
a6 = (a1 + a11 ).
2
Thus, we have

3 3 2 715
a1 + a6 + a11 = (a1 + a11 ) = · · 715 = 3 · = 195.
2 2 11 11
ANSWER : 195 2
5.3.2 Geometric Series
Next we consider the case of a geometric series. Suppose (an )n∈N is a geometric sequence with
common ratio r 6= 0. Then we can see easily that an = a1 rn−1 for all positive integers n. The sum
of the first m terms of a geometric sequence is
m
X
an = a1 + a2 + · · · + am
n=1
= a1 + a1 r + a1 r2 + · · · + a1 rm−1
= a1 (1 + r + r2 + · · · + rm−1 ).
Observing that
rm − 1 = (1 + r + r2 + · · · + rm−2 + rm−1 )(r − 1),
provided that r 6= 1 we can write
m
X rm − 1
an = a1 . (5.11)
n=1
r−1
(Of course, if r = 1, then the sum of the first m terms of the geometric series is simply ma1 .)
Now, if |r| < 1, then as m becomes arbitrarily large, we see that rm tends to zero. Therefore, we
conclude that if |r| < 1, then the infinite geometric series converges to
∞
X −1 a1
an = a1 = . (5.12)
n=1
r−1 1−r
In contrast, if |r| > 1, then since the terms of the sequence (an )n∈N become unbounded, the geometric
series diverges.
Now let us consider some examples.
Example 5.3.2. (2011 AIME-2, Problem #5) The sum of the first 2011 terms of a geometric
sequence is 200. The sum of the first 4022 terms is 380. Find the sum of the first 6033 terms.
Solution: Using the notation in Equation (5.11), we have that

r2011 − 1 r4022 − 1
a1 = 200 and a1 = 380.
r−1 r−1
Thus, we have
200 a1 380
= = 4022 .
r2011 − 1 r−1 r −1
Thus,
200(r4022 − 1) = 380(r2011 − 1). (5.13)
4022 2011 2 4022
Observe that r = (r ) , so we can express r − 1 as a difference of two perfect square
numbers and thus factor it as follows:
r4022 − 1 = (r2011 − 1)(r2011 + 1).
Substituting this into Equation (5.13) and simplifying the result, we have
200(r2011 + 1) = 380.
Hence, we have
r2011 = 0.9.
Hence, we have
a1 200 200
= 2011 = = −2000.
r−1 r −1 −0.1
Thus, according to Equation (5.11), the sum of the first 6033 terms of the geometric series is
r6033 − 1

a1 6033 200
0.93 − 1 = (−2000)(−0.271) = 542.

a1 = (r − 1) =
r−1 r−1 −0.1
ANSWER : 542. 2
5.3. SERIES 85
Example 5.3.3. (2002 AIME-2, Problem #11) Two distinct, real, infinite geometric series
each have a sum of 1 and have the same second term. The third
√ term of one of the series is 1/8, and
m−n
the second term of both series can be written in the form , where m, n, and p are positive
p
integers and m is not divisible by the square of any prime. Find 100m + 10n + p.
Remark: √ Before we begin, recall from the Preface that the form of AIME answers involving radicals,
such as m, must be clarified in order that a unique answer is obtained. Here, the requirement
that m not be divisible by the square of any prime ensures its unique value, as any square factors
occuring under the radical must first be extracted and then simplified with the rest of the expression
to obtain unique values m, n, and p.
Solution #1: We are being asked to find the second term of the series, so let us use x to denote
this quantity. One of the series has third term 1/8, so if we denote by r the common ratio of this
series, we have
1
r= .
8x
Hence, the first term of this series is a = 8x2 . From the fact that this series sums to 1, Equation
(5.12) implies that
8x2
1 = 1.
1 − 8x
A quick algebraic rearrangement of this expression gives
64x3 = 8x − 1.
Therefore,
64x3 − 8x + 1 = 0. (5.14)
To find the roots of Equation (5.14), we might first look for rational roots. To do this, there is a
useful Rational Roots Test – see Theorem 7.4.6. In this case, the Rational Roots Test implies that
the only candidates for rational roots of Equation (5.14) are
1 1 1 1 1 1
x = ±1, ± , ± , ± , ± , ± , ± .
2 4 8 16 32 64
We can test each of these 14 candidates, or we can simply observe by inspection (if we ignored the
Rational Roots Test) that x = 1/4 is a solution of this equation. Using long division of polynomials3
to divide 64x3 − 8x + 1 by x − 14 , we find that
1 1
64x3 − 8x + 1 = (x − )(64x2 + 16x − 4) = 4(x − )(16x2 + 4x − 1).
4 4
Applying the quadratic equation to the latter factor, we find that the three roots of Equation (5.14)
are √ √
1 5−1 − 5−1
x= , x= , .
4 8 8
3 Additional background on polynomials, division, roots, and more can be found in Chapter 7.
Of these options, only √

5−1
x=
8
has the correct form given in the problem. Therefore, m = 5, n = 1, and p = 8. We conclude that
100m + 10n + p = 518. ANSWER : 518. 2
Solution #2: This is only a slight variation on Solution #1. One could express the series whose
third term is 1/8 by assuming the first term is a, and the common ratio is r. Then we have that the
first series sums to
a
a + ar + ar2 + ar3 + · · · = a(1 + r + r2 + r3 + · · · ) = = 1. (5.15)
1−r
Here, we have ar2 = 1/8, or 8ar2 = 1. From Equation (5.15), we see that a = 1 − r, so that
8(1 − r)r2 = 1.
Therefore,
8r3 − 8r2 + 1 = 0.
Once more, we have a cubic equation similar to Equation (5.14) to solve. Once we determine
the three solutions for r, we can then quickly compute the second term of the series, which is
ar = (1 − r)r = r − r2 and find the one that meets the proper form given in the problem. We leave
these details to the reader.
Remark: This problem is a little tricky in that we were told there are two infinite, geometric series
in question. However, as both solutions above show, there is really only a need to consider the series
whose third term is 1/8. From that information alone, we are able to answer this question. In this
problem especially, and all AIME problems, we need to be careful not to be misled by extraneous
information.
Here is an example that combines a working knowledge of geometric series with probability theory.
The probability portion of the analysis was already carried out in Example 3.3.4.
Example 5.3.4. (2009 AIME-2, Problem #8) Dave rolls a fair six-sided die until a six appears
for the first time. Independently, Linda rolls a fair six-sided die until a six appears for the first time.
m
Let m and n be relatively prime positive integers such that is the probability that the number of
n
times Dave rolls his die is equal to or within one of the number of times Linda rolls her die. Find
m + n.
Solution: Let k denote the number of times Linda rolls her die. In Example 3.3.4, we determined
that, for k ≥ 2, the probability that Linda rolls her die k times and that Dave rolls his die within
one of k is 2 2k−3 2k−2 2k−1
1 5 5 5
+ + .
6 6 6 6

1 1 5 1
Moreover, for the case k = 1, we determined that the probability is + · . To compute
6 6 6 6
m
, we must sum these probabilities over all positive integers k. We will split the infinite series that
n
5.4. SOME OTHER USEFUL SEQUENCES AND SERIES 87
5
arises into two infinite series according to whether the number of factors of in a given term is even
6
or odd: X ∞ 2 2k−3 2k−2 2k−1
m 1 1 5 1 1 5 5 5
= + · + + +
n 6 6 6 6 6 6 6 6
k=2
2 2 ∞ 2 2k−3 2k−2 2k−1
1 1 5 X 1 5 5 5
= + + + +
6 6 6 6 6 6 6
k=2
2 ∞ 2k−3 2k−2 2k−1
1 5 X 5 5 5
= 1+ + + +
6 6 6 6 6
k=2
5
Careful inspection of the terms added inside the braces reveals that the odd powers of are summed
6
twice each, whereas the even powers are summed only once. Thus, we can reorganize the terms within
5
the braces by grouping the odd powers of together and doing likewise for the even powers. Thus,
6
we have
2 " 2 4 # " 2 4 #
m 1 5 5 5 5 5
= 1+ + + ··· + 2 1+ + + ··· .
n 6 6 6 6 6 6
2
5 25
Applying Equation (5.12) to r = = , we find that
6 36
2 4
5 5 1 36
1+ + + ··· = 25 = ,
6 6 1 − 36 11
so that 2
m 1 36 5 36 1 96 8
= +2 = · = .
n 6 11 6 11 36 11 33
Therefore, m = 8 and n = 33, so m + n = 8 + 33 = 41 = 041. ANSWER : 041. 2
5.4 Some Other Useful Sequences and Series
The Fibonacci sequence
Perhaps no sequence is more famous than the much celebrated Fibonacci sequence. The sequence
begins with the two terms f0 = 0 and f1 = 1, and all subsequent terms of the sequence are derived
from the recurrence relation
fn = fn−1 + fn−2 , n ≥ 2.
Therefore, we can quickly enumerate the first several terms of the Fibonacci sequence:
0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377, 610, 987, 1597 . . . . (5.16)
A more advanced treatment of the theory of recurrence relations actually reveals a “stand-alone”
formula for fn :
" √ !n √ !n #
1 1+ 5 1− 5
fn = √ − . (5.17)
5 2 2
√
1+ 5
The expression appearing in this formula is often known as the “golden ratio”, and it is
2 √
truly remarkable to find the integers fn defined in terms of complicated expressions involving 5.
The Fibonacci sequence arises in a variety of mathematical applications, and is present and accounted
for in several AIME problems back through history. A good case in point is the following example.
Example 5.4.1. (1998 AIME, Problem #8) Except for the first two terms, each term of the
sequence 1000, x, 1000 − x, . . . is obtained by subtracting the preceding term from the one before that.
The last term of the sequence is the first negative term encountered. What positive integer x produces
a sequence of maximum length?
Solution: When it is unclear how to proceed on a problem involving sequences or series, it is

advisable to start by enumerating the first several terms to see if a pattern emerges. In this case,
we have
a1 = 1000, a2 = x, a3 = 1000 − x, a4 = 2x − 1000, a5 = 2000 − 3x, a6 = 5x − 3000,
a7 = 5000−8x, a8 = 13x−8000, a9 = 13000−21x, a10 = 34x−21000, a11 = 34000−55x, . . .

The coefficients arising in the terms of this sequence clearly involve the terms of the Fibonacci
sequence (5.16). Although not necessary for completing the solution, we pause to note that

fn−1 x − 1000fn−2 , n = 2, 4, 6, . . .
an = = (−1)n (fn−1 x − 1000fn−2 ).
1000fn−2 − fn−1 x, n = 3, 5, 7, . . .
Mathematical induction can be invoked to prove this formula, but we will not take space to provide
this proof here. Let us create a table showing the restictions on x required for the terms an to be
nonnegative:
n an Requirement on x for an ≥ 0
3 1000 − x x ≤ 1000
4 2x − 1000 x ≥ 500
5 2000 − 3x x ≤ 666
6 5x − 3000 x ≥ 600
7 5000 − 8x x ≤ 625
8 13x − 8000 x ≥ 616
9 13000 − 21x x ≤ 619
10 34x − 21000 x ≥ 618
11 34000 − 55x x ≤ 618
From the results in this table, we see that the first 11 terms of the sequence are all nonnegative if
and only if x = 618. Thus, the answer is 618. ANSWER : 618. 2
5.4. SOME OTHER USEFUL SEQUENCES AND SERIES 89
Sums of powers of integers

∞
X
The geometric series arn−1 is a classic example of an infinite series that converges to a finite
n=1
value, provided the common ratio r lies in the interval (−1, 1). This was recorded in Equation (5.12).
However, there are also some finite series whose values are useful to know for the AIME competition
because they arise frequently. We memorialize them in our next result.
Theorem 5.4.2. For each positive integer m, we have the following:
m
X m(m + 1)
1. n = 1 + 2 + 3 + ··· + m = ,
n=1
2
m
X m(m + 1)(2m + 1)
2. n2 = 12 + 22 + 32 + · · · + m2 = ,
n=1
6
m
X m2 (m + 1)2
3. n3 = 13 + 23 + 33 + · · · + m3 = .
n=1
4
All three parts of this theorem can be proved by induction on m, but for the first part, there is a
m
X
clever proof credited to Gauss that goes as follows: Let S = n. Then
n=1
S= 1 + 2 + 3 + · · · + (m − 2) + (m − 1) + m
S = m + (m − 1) + (m − 2) + · · · + 3 + 2 + 1
Summing both sides of both equations, vertically term by term, we arrive at
2S = (m + 1) + (m + 1) + · · · + (m + 1) = m(m + 1),
and dividing both sides by 2 completes the proof. The first two parts of Theorem 5.4.2 arise rather
frequently in solutions to AIME problems, as can be seen in several of the examples and exercises
in Chapters 4 and 5.
Telescoping Series
In some instances, series with rather complicated expressions turn out to be surprisingly elegant
after some simplification is done. One of the best examples of this is a telescoping series, in which
∞
X
the terms of the series an have the form an = bn − cn for some real numbers bn and cn in such
n=1
a way that the bn and cn values exhibit some cancellation. For instance, consider the series
∞
X 1 1 1 1 1 1 1 1 1
− = 1− + − + − + − + ....
n=1
n n+1 2 2 3 3 4 4 5
Notice that except for the “1” in the first parenthesized term on the right side of this equation,
the value that is subtracted from each parenthesized term is subsequently added back in within the
following parenthesized term. Thus, we conclude that
∞
X 1 1
− = 1.
n=1
n n+1
It is important to note, however, that the parenthesized terms an = bn − cn must tend to zero (as
n → ∞) for convergence to occur. Thus, for example, the series
∞
X
((n + 1) − n) = (2 − 1) + (3 − 2) + (4 − 3) + · · ·
n=1
might at first appear to converge to −1 by using cancellation, but since an = (n + 1) − n = 1, we of

course conclude that
X∞ X∞
((n + 1) − n) = 1=∞
n=1 n=1
diverges. AIME problems that exploit telescoping behavior are given below in Examples 5.5.2 and
5.5.3.
5.5 Additional Examples of Sequences and Series
Despite the mathematical importance of arithmetic and geometric sequences and series, the majority
of the problems about sequences and series that appear in the AIME competition are less standard.
In this section, we investigate some examples to illustrate the wide variety of sequences and series
that have appeared in past AIME competitions.
Example 5.5.1. (2008 AIME-2, Problem #6) The sequence (an ) is defined by
a2n−1
a0 = 1, a1 = 1, and an = an−1 + for n ≥ 2.
an−2
The sequence (bn ) is defined by
b2n−1
b0 = 1, b1 = 3, and bn = bn−1 + for n ≥ 2.
an−2
b32
Find .
a32
Solution #1: To see what is going on, we enumerate the first few terms of the sequence (an ) in
5.5. ADDITIONAL EXAMPLES OF SEQUENCES AND SERIES 91
hopes of recognizing a pattern. We see that
a21 12
a2 = a1 + =1+ = 2,
a0 1
a2 22
a3 = a2 + 2 =2+ = 6,
a1 1
a2 62
a4 = a3 + 3 =6+ = 24,
a2 2
a2 242
a5 = a4 + 4 = 24 + = 120,
a3 6
and so on. From the pattern, we might suspect that an = n! for all nonnegative integers n. A careful
mathematician would insist on a rigorous proof at this stage, and in order to provide one, we can
proceed by induction as follows:
Proof that an = n! for all nonnegative integers n by induction: Clearly a0 = 1 = 0! and

a1 = 1 = 1!, so the base cases work. Now if we assume that an−1 = (n − 1)!, then
a2n−1
an = an−1 +
an−2
[(n − 1)!]2
= (n − 1)! +
(n − 2)!
= (n − 1)! + (n − 1)!(n − 1)
= (n − 1)!n
= n!,
as required.
Similarly, we can compute the first several terms of the sequence (bn ):
b21 32
b2 = b1 + =3+ = 12,
b0 1
b2 122
b3 = b2 + 2 = 12 + = 60,
b1 3
b2 602
b4 = b3 + 3 = 60 + = 360,
b2 12
b2 3602
b5 = b4 + 4 = 360 + = 2520.
b3 60
1
Again looking for a pattern, we might suspect that bn = (n + 2)! for all nonnegative integers n.
2
We leave the inductive verification that this is indeed the correct formula for bn to the reader as an
exercise similar to our proof of the formula for an given above. Thus, we conclude that
1
b32 2 · 34! 1
= = · 34 · 33 = 17 · 33 = 561.
a32 32! 2
Solution #2: Another way to reach the answer via pattern recognition is to compute that
b1 b2 b3 b4 b5
= 3, = 6, = 10, = 15, = 21, ....
a1 a2 a3 a4 a5
We recognize the values 3, 6, 10, 15, 21, . . . as the so-called triangular numbers obtained by summing
the first finitely many positive integers. In particular, here we have
m+1
bm X (m + 1)(m + 2)
= n= ,
am n=1
2
b32
from which we can readily obtain the numerical value = 561 as well.
a32
Solution #3: Note that

an an−1
=1+ ,
an−1 an−2
and the fraction on the right is of the same form as the one on the left, so we can apply the formula
iteratively to itself:
an an−1 an−2 an−3 a1
=1+ =2+ =3+ = · · · = (n − 1) + = n.
an−1 an−2 an−3 an−4 a0
Therefore, for all n ≥ 1,

an = nan−1 ,
which is well-known to be the recurrence relation for
an = n!.
Applying the same reasoning to the sequence {bn }, we find that

1
bn = (n + 2)!.
2
Hence,
b32 1 34! 1
= · = · 34 · 33 = 17 · 33 = 561.
a32 2 32! 2
ANSWER : 561. 2
Solution #1 in the previous example shows the power of doing “brute force” computations in order
to recognize a pattern. Once the pattern is found and sufficiently verified, it can often lead quickly
to the answer to an AIME problem. Next we tackle two problems that have ties with number theory
(specifically, the Fundamental Theorem of Arithmetic).
1
Example 5.5.2. (2002 AIME, Problem #4) Consider the sequence defined by ak = for
k2 + k
1
k ≥ 1. Given that am + am+1 + · · · + an−1 = , for positive integers m and n with m < n, find
29
m + n.
5.5. ADDITIONAL EXAMPLES OF SEQUENCES AND SERIES 93
Solution: The key observation is to notice that

1 1 1 1
ak = = = − .
k2 + k k(k + 1) k k+1
Therefore, the sum of terms am , am+1 , . . . , an−1 is a telescoping series. We have
1
= am + am+1 + · · · + an−1
29
1 1 1 1 1 1
= − + − + ··· + −
m m+1 m+1 m+2 n−1 n
1 1
= −
m n
n−m
= .
mn
Therefore, we must solve
29(n − m) = mn (5.18)
for positive integers m < n.
Now since 29 is prime, the Fundamental Theorem of Arithmetic (Theorem 4.2.1) guarantees that 29
must divide either m or n. Before reading further, the reader should ponder whether there is a way
to rearrange Equation (5.18) in order to clarify which of m or n is divisible by 29. For instance, one
possible rearrangement of Equation (5.18) is
29n = m(29 + n),
but this does not help us progress on the question of which of m or n is divisible by 29. Here is a
more effective rearrangement of Equation (5.18), and the reader should evaluate why this will be
more effective before reading further:
29m = n(29 − m). (5.19)
Since 29 − m < 29 and 29 is prime, the Fundamental Theorem of Arithmetic (Theorem 4.2.1)
guarantees that 29 must divide n. Let us say n = 29` for some positive integer `. Thus,
m = `(29 − m),
so that 29 − m must divide m. Since 29 − m obviously divides itself, then we conclude that 29 − m
must divide the sum m + (29 − m) = 29. Therefore, 29 − m = 1 or 29 − m = 29. Since m is positive,
the latter possibility is excluded. Therefore, 29 − m = 1, or m = 28. Hence, from Equation (5.19),
we conclude that n = 29 · 28. Thus,
m + n = 28 + 29 · 28 = 30 · 28 = 840.
ANSWER : 840. 2
Example 5.5.3. (2009 AIME, Problem #7) The sequence (an ) satisfies a1 = 1 and
1
5(an+1 −an ) − 1 = 2
n+ 3
for n ≥ 1. Let k be the least integer greater than 1 for which ak is an integer. Find k.
Solution: We have
5
1 n+ 3n + 5
5an+1 −an = 1 + 2 = 3
2 = (5.20)
n+ 3 n+ 3
3n + 2
for n ≥ 1. By replacing n with n − 1, we also have
3n + 2
5an −an−1 = . (5.21)
3n − 1
Multiplying Equations (5.20) and (5.21), we see that factors of 5an cancel, leaving us with
3n + 5
5an+1 −an−1 = .
3n − 1
Applying the same technique again, we obtain
3n + 5 3n − 1 3n + 5
5an+1 −an−2 = 5an+1 −an−1 5an−1 −an−2 = · = .
3n − 1 3n − 4 3n − 4
Continuing in this way, we find in general that
3n + 5
5an+1 −an−t = .
3n + 2 − 3t
Setting t := n − 1, we have
3n + 5
5an+1 −a1 = .
5
Using a1 = 1, we can simplify this to 5an+1 = 3n + 5. Therefore,
an+1 = log5 (3n + 5). (5.22)
We need to find the smallest integer n ≥ 1 (since the question is seeking k > 1 such that ak is an
integer) such that log5 (3n + 5) is an integer. Rephrasing this, we must find the smallest integer
n ≥ 1 such 3n + 5 = 5` for some integer `. The requirement that n ≥ 1 forces ` > 1. We can rewrite
this as
3n = 5(5`−1 − 1).
By the Fundamental Theorem of Arithmetic, we see that 5`−1 − 1 must be a multiple of 3. The
smallest integer ` > 1 for which this holds is ` = 3. Therefore, 3n = 53 − 5 = 120, or n = 40.
Indeed, Equation (5.22) implies that a41 = log5 (125) = 3. The answer is therefore k = 41 = 041.
ANSWER : 041. 2
5.6 Exercises
1. (1999 AIME, Problem #1) Find the smallest prime that is the fifth term of an increasing
arithmetic sequence, all four preceding terms also being prime.
5.6. EXERCISES 95
2. (2008 AIME-2, Problem #1) Let N = 1002 + 992 − 982 − 972 + 962 + · · · + 42 + 32 − 22 − 12 ,
where the additions and subtractions alternate in pairs. Find the remainder when N is divided
by 1000.
3. (2005 AIME-2, Problem #3) An infinite geometric series has sum 2005. A new series,
obtained by squaring each term of the original series, has sum 10 times the sum of the original
series. The common ratio of the original series is m/n, where m and n are relatively prime
positive integers. Find m + n.
4. (2001 AIME-2, Problem #3) Given that
x1 = 211,
x2 = 375,
x3 = 420,
x4 = 523, and
xn = xn−1 − xn−2 + xn−3 − xn−4 when n ≥ 5,
find the value of x531 + x753 + x975 .
5. (1985 AIME, Problem #5) A sequence of integers a1 , a2 , a3 , . . . is chosen so that an =
an−1 − an−2 for each n ≥ 3. What is the sum of the first 2001 terms of this sequence if the
sum of the first 1492 terms is 1985, and the sum of the first 1985 terms is 1492?
10000
X 1
6. (2002 AIME-2, Problem #6) Find the integer that is closest to 1000 .
n=3
n2 − 4
7. (1989 AIME, Problem #7) If the integer k is added to each of the numbers 36, 300, and
596, one obtains the squares of three consecutive terms of an arithmetic sequence. Find k.
8. (1986 AIME, Problem #7) The increasing sequence 1, 3, 4, 9, 10, 12, 13, . . . consists of those
positive integers which are powers of 3 or sums of distinct powers of 3. Find the 100th term
of this sequence (where 1 is the 1st term, 3 is the 2nd term, and so on).
9. (2005 AIME-2, Problem #11) Let m be a positive integer, and let a0 , a1 , . . . , am be a
sequence of real numbers such that a0 = 37, a1 = 72, am = 0, and
3
ak+1 = ak−1 −
ak
for k = 1, 2, . . . , m − 1. Find m.
10. (2000 AIME, Problem #10) A sequence of numbers x1 , x2 , . . . , x100 has the property that,
for every integer k between 1 and 100, inclusive, the number xk is k less than the sum of
the other 99 numbers. Given that x50 = m/n, where m and n are relatively prime positive
11. (2006 AIME-2, Problem #11) A sequence is defined as follows: a1 = a2 = a3 = 1, and, for
all positive integers n, an+3 = an+2 + an+1 + an . Given that a28 = 6090307, a29 = 11201821,
X28
and a30 = 20603361, find the remainder when ak is divided by 1000.
k=1
12. (2004 AIME-2, Problem #9) A sequence of positive integers with a1 = 1 and a9 +a10 = 646
is formed so that the first three terms are in geometric progression, the second, third, and fourth
terms are in arithmetic progression, and, in general, for all n ≥ 1, the terms a2n−1 , a2n , a2n+1
are in geometric progression, and the terms a2n , a2n+1 , a2n+2 are in arithmetic progression.
Let an be the greatest term in this sequence that is less than 1000. Find n + an .
13. (2007 AIME, Problem #11) For each positive integer p, let b(p) denote the unique positive
2007
√ 1 X
integer k such that |k − p| < . For example, b(6) = 2 and b(23) = 5. If S = b(p), find
2 p=1
the remainder when S is divided by 1000.
1995
√ X 1
14. (1995 AIME, Problem #13) Let f (n) be the integer closest to 4
n. Find .
f (k)
k=1
15. (2006 AIME, Problem #15) Given that a sequence satisfies x0 = 0 and |xk | = |xk−1 + 3|
for all integers k ≥ 1, find the minimum possible value of |x1 + x2 + · · · + x2006 |.
Chapter 6
Logarithmic and Trigonometric

Functions
“The man ignorant of mathematics will be increasingly limited in his grasp of the
main forces of civilization.”
- John Kemeny
6.1 Introduction
In this chapter and the next, we will explore several of the most important types of mathematical
functions. Naturally, these functions also occur prominently in the AIME competition. These
include logarithmic functions (and their inverses, exponential functions), trigonometric functions
(and their inverses), and polynomials. The topic of polynomials is rather lengthy and tied closely
to a knowledge of complex numbers, so we will postpone that discussion until the next chapter. We
will begin here with logarithmic functions.
6.2 Logarithmic Functions
An exponential function in its simplest form can be expressed as
f (x) := ax (6.1)
for some fixed, positive real number a > 0 with a 6= 1. Thus, f is a function from R to R+ . High
school algebra textbooks discuss these functions in great detail, so we will limit our discussion here
so that we can move quickly towards AIME examples.
Since no two inputs into the function f in Equation (6.1) yield the same output and every y in R+
97
98 CHAPTER 6. LOGARITHMIC AND TRIGONOMETRIC FUNCTIONS
occurs as an output of f , one can uniquely define the inverse function g : R+ → R such that
(g ◦ f )(x) = x and (f ◦ g)(y) = y
for all real numbers x and all positive real numbers y. This function g is precisely the function
g(y) = loga (y) (read “log base a of y”). Because f and g are inverses, we observe that
loga (y) = x if and only if ax = y. (6.2)
Moreover, we have that

loga (ax ) = x and aloga (y) = y. (6.3)
Figure 6.1: The general shapes of the graphs of f (x) = ax and g(x) = loga (y) are shown for the case
when a > 1.
In some instances the subscript a is omitted from the notation loga (y), and in such cases, one should
assume conventionally that a = 10.
Merely from the definition supplied by Equation (6.2), we are already in a position to solve a nice
problem from the AIME competition. Since the answers to AIME problems must be whole integers,
a common strategy in the questions involving logarithms is to take floor and ceiling values:
bloga xc and dloga xe.
That is the case with our first example.

Example 6.2.1. (2007 AIME, Problem #7) Let
1000
X
N= k(dlog√2 ke − blog√2 kc).
k=1
Find the remainder when N is divided by 1000. (Here bxc denotes the greatest integer that is less
than or equal to x, and dxe denotes the least integer that is greater than or equal to x.)
Solution: The key observation is that for all real numbers x, we have

0 if x is an integer,
dxe − bxc =
1 if x is not an integer.
In particular, for all positive numbers k,

6.2. LOGARITHMIC FUNCTIONS 99

0 if log√2 k is an integer,
dlog√ 2 ke − blog√ 2 kc =
1 if log√2 k is not an integer.
√ m
Taking this further, we observe from Equation (6.2) that log√2 k is an integer if and only if k = 2
for some integer m. Thus, N is effectively√ the sum of the first 1000 positive integers except those
that can be expressed as some power of 2. Restricting to integer values of k between 1 and 1000,
inclusively, we see that log√2 k is an integer if and only if k = 1, 2, 4, 8, 16, 32, 64, 128, 256, or 512.
Thus, N consists of the sum of the first 1000 positive integers, with these ten exceptions, and we
can use Theorem 5.4.2 and Equation (5.11):
1000
!
X
N= k − 1 + 2 + 4 + 8 + 16 + 32 + 64 + 128 + 256 + 512
k=1
1000
X 9
X
= k− 2`
k=1 `=0
(1000)(1001)
= − (210 − 1)
2
= 500500 − 1023 = 499477.
Thus, the remainder when N is divided by 1000 is 477. ANSWER : 477. 2
Most of the time, we need a deeper command of how logarithms work than was required in Example
6.2.1. For this reason, and because it is an important content area in the high school math curriculum
anyway, let us remind ourselves of some of the important properties of logarithms.
Properties of Logarithms
Students are expected to know and to take advantage of the commonly known properties for loga-
rithms in solving problems. Therefore, let us quickly enumerate some of these properties so that we
will be well equipped to solve some examples from the AIME that follow.
Theorem 6.2.2. Let x, y, and b be positive real numbers with b 6= 1, and let p be any real number.
Then we have
1. logb (xy) = logb (x) + logb (y)

x
2. logb = logb (x) − logb (y)
y
3. logb (xp ) = p · logb (x)
Warning: Many students have gone astray by trying to invent alternative properties for the log-
arithm function. Be wary of this. There is no property enabling one to simplify the expression
loga (x + y), nor can the product (loga x)(loga y) be simplified.
Example 6.2.3. (2009 AIME-2, Problem #2)√Suppose that a, b, and c are positive real numbers
such that alog3 7 = 27, blog7 11 = 49, and clog11 25 = 11. Find
2 2 2
a(log3 7) + b(log7 11) + c(log11 25) .
z z
Remark: An un-parenthesized expression of the form xy should always be interpreted as x(y ) .
The expression (xy )z , on the other hand, can be re-written instead as xyz .
Solution: We are asked to sum three terms, one involving a, one involving b, and one involving c.
We shall hope to work on each of them independently by using (6.3), Theorem 6.2.2, and the given
information. For example, we have
log3 7
a(log3 7) = alog3 7 = 27log3 7 = (33 )log3 7 = 33log3 7 = 3log3 (7 ) = 73 = 343.
2 3
Using the same approach on the other two terms, we have

log7 11
b(log7 11) = blog7 11 = 49log7 11 = 72log7 11 = 7log7 11 = 112 = 121
2 2
and
log11 25 √ log 25 √ √
c(log11 25) = clog11 25 = 11 11 = 11 2 log11 25 = 11log11 25 = 25 = 5.
2 1
Therefore,
a(log3 7) + b(log7 11) + c(log11 25) = 343 + 121 + 5 = 469.
2 2 2
ANSWER : 469. 2
Since we just covered geometric sequences in the previous chapter, this would be a nice time to cover
an AIME problem that touches both on geometric sequences and on logarithms. In fact, there is
even a little bit of number theory (albeit very specific and elementary) that can be exploited in this
problem; you have to love AIME problems that span a broad spectrum of mathematics!
Example 6.2.4. (2002 AIME-2, Problem #3) It is given that log6 a + log6 b + log6 c = 6, where
a, b, and c are positive integers that form an increasing geometric sequence and b − a is the square
of an integer. Find a + b + c.
Solution: Since a, b, and c form an increasing geometric sequence, we can write b = ar and c = ar2
for some constant r > 1. The given equation is a sum of logarithms with respect to the same base,
so we are naturally inclined to apply part (1) of Theorem 6.2.2 to write
6 = log6 a + log6 b + log6 c
= log6 (abc).
Thus,
abc = 66
a(ar)(ar2 ) = 66
(ar)3 = 66
(ar)3 = 363
ar = 36
b = 36
Since b − a must be a square of an integer and a is positive, b − a must be one of the following: 1,
4, 9, 16, or 25. That is,
a = 35, 32, 27, 20, or 11. (6.4)
Now,
362
c = ar2 = br = 36r =
a
2 4 4
is an integer, which means that a divides 36 = 2 · 3 . Among the five possible values of a listed in
362
(6.4), only a = 27 works. Then c = = 48. Therefore,
27
a + b + c = 27 + 36 + 48 = 111.
ANSWER : 111. 2
One additional formula that arises frequently in the study of logarithms is the change-of-base formula.
This formula allows one to convert expressions from one logarithmic base to another. The formula
reads:
logb x
Change-of-Base Formula for Logarithms: If a, b 6= 1, loga x = . (6.5)
logb a
As a special case of (6.5), we can substitute x = b and use the fact that logb b = 1 for all positive
b 6= 1 to obtain
1
loga b = . (6.6)
logb a
Clearly, one should have the change-of-base formula (6.5) and its special case (6.6) in mind whenever
a problem with expressions involving different logarithm bases is encountered. Using these formulas,
one can sometimes rewrite a variety of expressions involving different bases as expressions involving
a single base. Once this is accomplished, the other properties in Theorem 6.2.2 can be applied. A
good example of this strategy is found in Example 6.2.5 below.
While many proofs in this text are omitted to save space, the change-of-base formula is quite
important, and its proof is both instructive and brief, so we shall include it.
Proof of the Change-of-Base Formula for Logarithms: Let y := loga x. From Equation (6.2),
we have ay = x. Let us now apply the logarithm function, base b, to both sides:
logb (ay ) = logb x.
Applying part (3) of Theorem 6.2.2, we obtain
y logb a = logb x,
so that
logb x
y= ,
logb a
which is precisely the formula (6.5).
Equations (6.5) and (6.6) can be especially helpful if using a calculator to do calculations with
logarithms, because if the calculation you need involves a logarithm base that is non-standard, then
these equations can be used to convert it to a logarithm base that is standard for calculator use. Of
course, the AIME competition does not allow calculators. Nevertheless the logarithm change-of-base
formulas (6.5) and (6.6) often prove to be very useful. Here is an example.
Example 6.2.5. (2000 AIME-2, Problem #1) The number
2 3
6
+
log4 2000 log5 20006
m
can be written as where m and n are relatively prime positive integers. Find m + n.
n
Solution: The logarithm bases in the two terms appearing here are different, but we suspect that
we might be able to use the change-of-base formula to make them the same, since the number 2000
is common to both terms. This is precisely what we will do:
2 3 2 3
+ = +
log4 20006 log5 20006 6 · log4 2000 6 · log5 2000
1
= (2 · log2000 4 + 3 · log2000 5)
6
1
= log2000 (42 · 53 )
6
1
= log2000 2000
6
1
= .
6
Hence, m = 1 and n = 6, so that m + n = 1 + 6 = 7 = 007. ANSWER : 007. 2
Our next example is a good illustration of how AIME problems oftentimes draw on several areas
of mathematics at once. In particular, the next problem involves geometric sequences, logarithms,
number theory, and solving linear equations.
Example 6.2.6. (2006 AIME, Problem #9) The sequence a1 , a2 , . . . is geometric with a1 = a
and common ratio r, where a and r are positive integers. Given that log8 a1 + log8 a2 + · · · + log8 a12 =
2006, find the number of possible ordered pairs (a, r).
Solution: Using the geometric sequence information from Chapter 5, we have an = arn−1 for each
positive integer n. Moreover, the left side of the given equation
log8 a1 + log8 a2 + · · · + log8 a12 = 2006
suggests that we apply part (1) of Theorem 6.2.2. Therefore,
2006 = log8 a1 + log8 a2 + · · · + log8 a12
= log8 a + log8 ar + · · · + log8 ar11
= log8 a · ar · · · ar11

= log8 a12 r1+2+···+11

= log8 a12 r66 .

Therefore, we conclude that

82006 = a12 r66 ,
or equivalently,
a12 r66 = 23·2006 = 26018 .
We conclude from this that both a and r must be powers of 2. Let us say a = 2x and r = 2y for
nonnegative integers x and y. Then
212x 266y = 26018 ,
which requires that 12x + 66y = 6018. Dividing through by 6, we have
2x + 11y = 1003. (6.7)
Thus, to determine the number of possible pairs (a, r) of positive integers, it suffices to determine
the number of pairs of nonnegative integers (x, y) that satisfy Equation (6.7), or equivalently,
2x = 1003 − 11y. (6.8)
Note that if y is even, then 1003 − 11y must be odd, so that there are no integer solutions for x in
Equation (6.8). On the other hand, if y is odd, then 1003 − 11y must be even, and there will always
be a (unique) integer solution for x in Equation (6.8). Therefore, y must be an odd integer satisfying
0 ≤ 11y ≤ 1003. That is, 1 ≤ y ≤ 91. Hence, y can be any odd integer in the range 1 ≤ y ≤ 91.
There are 46 such values of y, which is half of the integers from 1 to 92 (inclusive). For each of
these 46 permissible values of y, there will be a unique corresponding value for x, determined from
Equation (6.8). Hence, the answer is 46 = 046. ANSWER : 046. 2
Systems of Equations involving Logarithms
In Chapter 1, we examined several problems involving systems of algebraic equations. At that time,
we intentionally skipped over a few AIME examples involving systems of equations because those
equations also involved logarithms. We are now in a position to consider these examples.
Example 6.2.7. (2010 AIME-2, Problem #5) Positive numbers x, y, and z satisfy
xyz = 1081 and (log10 x)(log10 yz) + (log10 y)(log10 z) = 468.
Find p
(log10 x)2 + (log10 y)2 + (log10 z)2 .
Solution: If we take the base 10 logarithm of both sides of xyz = 1081 , we will turn the left side
into a sum of logarithms which more closely resembles a portion of the expression we are trying to
compute:
log10 (xyz) = log10 (1081 )

log10 x + log10 y + log10 z = 81
Since we are asked to find the sums of squares of the three terms appearing on the left side of the
latter expression, let us square both sides:
(log10 x + log10 y + log10 z)2 = 812 = 6561.
Expanding the expression on the left side, we obtain

2 2 2
(log10 x) + (log10 y) + (log10 z) + 2 log10 x log10 y + log10 x log10 z + log10 y log10 z = 6561. (6.9)
To determine the quantity in large brackets on the left-side of (6.9), we use the given information
(log10 x)(log10 yz) + (log10 y)(log10 z) = 468
to obtain
(log10 x)(log10 y + log10 z) + (log10 y)(log10 z) = 468. (6.10)
The left-hand side of (6.10) is precisely the bracketed expression on the left-hand side of (6.9), so
we obtain
(log10 x)2 + (log10 y)2 + (log10 z)2 + 2 · 468 = 6561.
Thus,
(log10 x)2 + (log10 y)2 + (log10 z)2 = 6561 − 2 · 468 = 6561 − 936 = 5625 = 752 .
√
(See the Remark below to see how we arrive at 5625 = 75 without a calculator.) Hence, the answer
is p
(log10 x)2 + (log10 y)2 + (log10 z)2 = 75 = 075.
ANSWER : 075. 2
√
Remark: To determine 5626 quickly without a calculator, simply note that
√ √ √
70 = 4900 < 5625 < 6400 = 80
√
and the fact that 5625 must end in “5” to conclude that the answer is 75.
Example 6.2.8. (2000 AIME, Problem #9) The system of equations
log10 (2000xy) − (log10 x)(log10 y) = 4

log10 (2yz) − (log10 y)(log10 z) = 1
log10 (zx) − (log10 z)(log10 x) = 0
has two solutions (x1 , y1 , z1 ) and (x2 , y2 , z2 ). Find y1 + y2 .
Solution: Let us use the following notation:
a := log10 x, b := log10 y, c := log10 z.
Using these notational substitutions and the properties in Theorem 6.2.2, we can rewrite the given
system of equations as
log10 2000 + a + b − ab = 4
log10 2 + b + c − bc = 1
c + a − ca = 0.
6.3. TRIGONOMETRIC FUNCTIONS 105
Since
log10 2000 = log10 (2 · 1000) = log10 2 + log10 103 = log10 2 + 3,
the first equation in the system can be reduced to
log10 2 + a + b − ab = 1. (6.11)
Comparing this with the second equation in the system, we conclude that
a + b − ab = b + c − bc,
and hence,
a(1 − b) = c(1 − b).
Two cases result: (1) a = c, and (2) b = 1. However, starting with b = 1, Equation (6.11) yields
log10 2 = 0, clearly a contradiction since 100 = 1 6= 2. Hence, we conclude that a = c. Since
c + a − ca = 0, we deduce that 2a = a2 , so that a = 0 or a = 2.
Case a = 0: Using Equation (6.11), we find that
log10 2 = 1 − b,
from which we find that

10 10
y = 10b = 101−log10 2 = = = 5.
10 log 10 2 2
Note that there is only one solution (x1 , y1 , z1 ) such that y1 = 5.
Case a = 2: Substituting this value into Equation (6.11) and simplifying, we obtain
log10 2 = b − 1.
Hence, 10b−1 = 2, from which we deduce that
y = 10b = 10 · 10b−1 = 10 · 2 = 20.
Once more, there is only one solution (x2 , y2 , z2 ) such that y2 = 20.
The y-values of the two solutions we have found from the two cases above are y1 = 5 and y2 = 20.
Hence, y1 + y2 = 5 + 20 = 25 = 025. ANSWER : 025. 2
Other examples involving systems of equations involving logarithms are reserved for the exercises.
Later in this chapter, we will return to consider some additional AIME problems that involve loga-
rithmic functions alongside the trigonometric functions we consider in the next section.
6.3 Trigonometric Functions
There are six standard trigonometric functions that arise in the high school mathematics curriculum:
y(x) = cos x, y(x) = sin x, y(x) = tan x, y(x) = sec x, y(x) = csc x, y(x) = cot x. (6.12)
Figure 6.2: The trigonometric functions can be expressed as ratios of the side lengths of a right
triangle.
Briefly, these six functions can all be related to the side lengths of a right triangle.
With the side lengths indicated in Figure 6.2, we observe that
a b b c c a
cos x = , sin x = , tan x = , sec x = , csc x = , cot x = . (6.13)
c c a a b b
Serious AIME contenders need to be well familiar with all of these functions, including roots, stan-
dard values, graphs, etc. but the first two are the most important. That is because all six of the
functions listed above can be written in terms of cos x and sin x:
sin x 1 1 cos x
tan x = , sec x = , csc x = , cot x = . (6.14)
cos x cos x sin x sin x
General Strategy Comment: When trigonometric functions are present, it is a common strategy
to “convert all trigonometric functions to sine and cosine expressions.” This is often a very good
approach, both for proofs of trigonometric identities as well as practical applications. There are
other instances, as we will see in Solution #1 to Example 6.3.3, where it is preferable to leave
expressions involving the other trigonometric functions unaltered.
The functions y = cos x and y = sin x are both periodic with period 2π and amplitude 1. Thus,
for every integer k, we have
cos x = cos(x + 2πk) and sin x = sin(x + 2πk).
As with any function, the “parent functions” (see Figure 6.3) can be modified in a variety of ways
such as phase change, amplitude, or period. A general form for the cosine function could be expressed
using constants a, b, c, and d as
y(x) = a cos(bx + c) + d,
which has amplitude |a|, period 2π/|b|, and with respect to the graph of a cos bx, it is shifted to the
left on the Cartesian plane by c/b units1 and d units vertically up (or down, if d < 0). A similar
expression can be written for the general sine function. We provide the basic graphs of the sine and
cosine functions in Figure 11 below, but we encourage the reader who is not familiar with standard
material on trigonometry to consult other sources for more information.
Trigonometric Values
AIME contestants are expected to be comfortable with both radian and degree measure of angles.
This includes conversions from one to the other. Since there are 360◦ for every 2π radians, we have
1 Of course, if c/b < 0, then the actual shift is to the right.
Figure 6.3: The graphs of f (x) = cos x and g(x) = sin x.
the following conversion formulas:
Angle Measure Conversion Formulas:

xπ
(a): An angle measure of x◦ is the same as radians.
180
◦
180π
(b): An angle measure of y radians is the same as .
y
If one knows the values of sin x and cos x on the standard values of x (such as x = 0, π/6, π/4, π/3, π/2
radians), then one can find the values of the other four functions at these angles by using the formulas
(6.14) above. The table below summarizes the values of cos x, sin x, and tan x on the angles given
above (in radians):
x radians x◦ cos x sin x tan x

0 0◦ √1 0 0√
π/6 30◦ √3/2 √1/2 1/ 3
π/4 45◦ 2/2 √2/2 √1
π/3 60◦ 1/2 3/2 3
π/2 90◦ 0 1 undefined
The angles listed in the table above all belong to the first quadrant of the Euclidean plane, so
that 0 ≤ x ≤ π/2 (in radians). It is also important to be able to compute trignometric values for
angles lying in other quadrants of the plane, and for angles that lie outside of the fundamental range
0 ≤ x ≤ 2π. The typical approach here is to replace the angle in question with a corresponding
reference angle in the first quadrant. The value of the six trignometric functions at the reference
angle is either the same or the negative of the value of the same functions at the original angle.
Further information about these considerations can be found in any text on trigonometry, and due
to space considerations, we will not delve further into this here. Instead, let us consider an AIME
problem in which the sole knowledge of trigonometry needed is the value of cos(nπ) for all integers
n.
Example 6.3.1. (1998 AIME, Problem #5) Given that
k(k − 1) k(k − 1)π
Ak = cos ,
2 2
find
|A19 + A20 + · · · + A98 |.
k(k − 1)
Solution: For all integers k, is always an integer (since either k or k − 1 will be even).
2
Thus, in the formula for Ak , we are evaluating the cosine function at some integer multiple of π.
Note that
1, if n is an even integer
cos nπ = (6.15)
−1, if n is an odd integer.
k(k − 1)
Note that will be an even integer if and only if k or k − 1 is divisible by 4. Using the
2
notation of modular arithmetic discussed in Chapter 4, we can re-express this as follows:
k(k − 1)
will be an even integer if and only if k ≡ 0 (mod 4) or k ≡ 1 (mod 4) . (6.16)
2
Putting the observation in (6.16) together with Equation (6.15), we conclude that

k(k − 1)
, if k ≡ 0 or 1 (mod 4)


Ak = 2 (6.17)
 − k(k − 1)
, if k ≡ 2 or 3 (mod 4).

2
In computing A19 + A20 + · · · + A98 , we can organize the terms into pairs of terms of the form
A`−1 + A` , where ` is an even integer with 20 ≤ ` ≤ 98, in which one term is positive and one term
is negative. In fact, applying Equation (6.17) to find A`−1 and A` , we can simplify the sum to

`−1 if ` ≡ 0 (mod 4)
A`−1 + A` =
−(` − 1) if ` ≡ 2 (mod 4)
Hence,
A19 + A20 = 19,
A21 + A22 = −21,
A23 + A24 = 23,
.. ..
. = .
Thus,
|A19 + A20 + · · · + A98 | = |19 − 21 + 23 − 25 + · · · + 95 − 97|
= |(−2) + (−2) + · · · + (−2)|
= |(−2)(20)| = 40,
where there are 20 pairs (19, 21), (23, 25), (27, 29), . . . (95, 97) of values each summing to −2. Hence,
the answer is 40 = 040. ANSWER : 040. 2
Remark: To solve Example 6.3.1, it is not completely necessary to introduce the sophistication of
modular arithmetic given in the solution above. The solver could simply observe that A`−1 +A` = ±`
by noticing the pattern and then proceed to compute the sum in the last lines of the solution above.
Trigonometric Identities
Equally important in the knowledge of the six standard trigonometric functions given in (6.12) are the
numerous identities relating them. Constructing an exhaustive list of such identities is impossible,
but we can list some of the most frequently arising ones.
Looking once more at the side lengths of a right triangle, we recall that the Pythagorean Theorem
(see Theorem 8.2.4) gives us a relationship between the side lengths in a right triangle,
a2 + b2 = c2 .
Dividing through by c2 and using the formulas for sin x and cos x in Equation (6.13), we deduce the
trigonometric identity that is perhaps cited most frequently:
sin2 x + cos2 x = 1 (6.18)
for all real values x.

π
If we divide Equation (6.18) through by cos2 x, provided that cos x 6= 0 (i.e. x 6= + kπ for any
2
integer k), then we obtain the identity
tan2 x + 1 = sec2 x. (6.19)
Next, we present the double-angle formulas and half-angle formulas. These formulas can be deduced
as special cases of formulas we will present below for evaluating trigonometric functions of sums or
differences.
sin 2x = 2 sin x cos x (6.20)
cos 2x = cos2 x − sin2 x (6.21)
cos 2x = 2 cos2 x − 1 (6.22)

cos 2x = 1 − 2 sin2 x (6.23)
1 − cos 2x
sin2 x = (6.24)
2
1 + cos 2x
cos2 x = (6.25)
2
Equation (6.20) follows as a special case of Equation (6.28) below, while Equation (6.21) follows as
a special case of Equation (6.30). Then, Equations (6.22) and (6.23) follow from Equations (6.21)
and (6.18). Finally, Equations (6.24) and (6.25) both follow from easy algebraic rearrangements of
Equations (6.22) and (6.23).
Now let us focus on how we can use these trigonometric identities to solve problems in the AIME.
An excellent case in point is the following example.
Example 6.3.2. (2012 AIME-2, Problem #9) Let x and y be real numbers such that
sin x cos x 1
=3 and = . (6.26)
sin y cos y 2
The value of
sin 2x cos 2x
+ (6.27)
sin 2y cos 2y
p
can be expressed in the form , where p and q are relatively prime positive integers. Find p + q.
q
Solution: There are two terms to compute in the expression (6.27), the first of which is easy to
find by a simple application of Equation (6.20) along with the given information in (6.26):
sin 2x 2 sin x cos x sin x cos x 3

= = · = .
sin 2y 2 sin y cos y sin y cos y 2
To compute the second term in Equation (6.27), we must work a little harder. The two fractions in
(6.26) involve both x and y, and it would be very helpful to eliminate one of the variables. First,
rewrite
1
sin x = 3 sin y and cos x = cos y.
2
Now we can apply Equation (6.18) to conclude that
2
1 1
1 = sin2 x + cos2 x = (3 sin y)2 + cos y = 9 sin2 y + cos2 y.
2 4
We have now eliminated x. Replacing cos2 y with 1 − sin2 y, we have

1
1 = 9 sin2 y + (1 − sin2 y),
4
so that
3 35
= sin2 y.
4 4
Therefore,
3 27
sin2 y = and sin2 x = (3 sin y)2 = 9 sin2 y = .
35 35
Finally, we can use Equation (6.23) to write
54
cos 2x 1 − 2 sin2 x 1− 35 19
= 2 = 6 =− .
cos 2y 1 − 2 sin y 1− 35
29
Thus, returning to Equation (6.27), we compute that
sin 2x cos 2x 3 19 87 − 38 49
+ = − = = .
sin 2y cos 2y 2 29 58 58
Thus, we have p = 49 and q = 58, and p + q = 107. 2
Another frequently used collection of trigonometric identities involve the sum and difference formulas.
In fact, the identities that follow can actually be viewed as generalizations of the ones we studied in
(6.20) – (6.25). For instance, Equation (6.20) can be derived from Equation (6.28) below by setting
x = y. Once more, we bypass the proofs of these next identities, preferring instead to move towards
applications of these identities on AIME problems.
sin(x + y) = sin x cos y + cos x sin y (6.28)
sin(x − y) = sin x cos y − cos x sin y (6.29)
cos(x + y) = cos x cos y − sin x sin y (6.30)
cos(x − y) = cos x cos y + sin x sin y (6.31)
tan x + tan y
tan(x + y) = (6.32)
1 − tan x tan y
tan x − tan y
tan(x − y) = (6.33)
1 + tan x tan y
The two formulas (6.32) and (6.33) can be derived nicely from (6.28) – (6.31), in keeping with our
observation that information about all of the trigonometric functions can be related back to the sine
and cosine functions. Readers who would like some practice in the algebraic manipulation of these
sum and difference formulas are encouraged to carry out the verification of (6.32) and (6.33).
π
Note that if we substitute x = into Equations (6.29) and (6.31) and simplify, we deduce two more
2
useful relationships between the sine and cosine functions:
π π
sin − y = cos y and cos − y = sin y (6.34)
2 2
for all real numbers y. Similarly, if we substitute x = 0 into Equations (6.29) and (6.31) and simplify,
we have
sin(−y) = y and cos(y) = y (6.35)
for all real numbers y. The statements in Equation (6.35) simply state that the function f (y) = cos y
is an even function, while the function g(y) = sin y is an odd function. This is also clear from the
graphs of these two functions (Figure 6.3 above). We emphasize once more that the formulas in
Equations (6.34) and (6.35) can be extremely useful in many applications, and this includes AIME
problems. AIME participants should make sure to know these well.
The list of trigonometric identities provided above is by no means exhaustive, and it is possible to
use the ones listed in this section to derive many others. However, for most AIME problems, the list
provided here is sufficient, and related formulas can be derived on problems on an individual basis
as needed.
In the previous subsection, we reviewed the standard angles whose trigonometric values are important
to know (0, π/6, π/4, π/3 and π/2 radians). In some cases, however, AIME problems ask the
solver to compute a trigonometric value at a non-standard angle. There are two potentially fruitful
approaches to consider in such an instance.
AIME strategies for computing trigonometric values at non-standard angles:
(a): Find a way to express the non-standard angle as a sum, difference, or multiple of standard
angles and apply trigonometric identities such as (6.18) – (6.35) to determine the non-standard value
from the standard values, or
(b): Appeal directly to first principles by drawing right triangles with an acute angle equal to the
reference angle for the angle in question.
In our second solution to the next example, we take advantage of both of these strategies to solve a
problem involving an inverse trigonometric function. Recall that if y = f (x) denotes one of the six
standard trigonometric functions, then the function x = g(y) is an inverse for y = f (x) if f (g(y)) = y
and g(f (x)) = x for all x in the domain of f and for all y in the domain of g. Again, we will not take
space to discuss inverse trigonometric functions in detail in this text. Instead, we pass quickly to
examples from the AIME and leave the basic background material on inverse trigonometric functions
for one of the many standard texts on the subject.
Example 6.3.3. (1984 AIME, Problem #13) Find the value of
10 cot cot−1 3 + cot−1 7 + cot−1 13 + cot−1 21 .

Remark: We know that cot(cot−1 3) = 3, but we can by no means utilize this property in this
example. The basic problem is that none of the trigonometric functions, including the cotangent
function appearing here, respect addition. Thus, we cannot algebraically split the four terms ap-
pearing in the parentheses in order to evaluate the cotangent of each of them separately. The fact
that the sine, cosine, and tangent functions fail to respect addition is already clear from Equations
(6.28), (6.30), and (6.32).
Solution #1: Let
x = cot−1 3, y = cot−1 7, z = cot−1 13, w = cot−1 21.
This means that

cot x = 3, cot y = 7, cot z = 13, cot w = 21.
Therefore,
1 1 1 1
tan x = , tan y = , tan z = , tan w = . (6.36)
3 7 13 21
We are being asked to compute
10
10 cot(x + y + z + +w) = ,
tan(x + y + z + w)
and we can apply Equation (6.32) using A := x + y and B := z + w to find
tan(x + y + z + w) = tan(A + B)
tan A + tan B
= .
1 − tan A tan B
We have
1
tan x + tan y +1 10
1
tan A = tan(x + y) = = 3 1 71 = 21
20 =
1 − tan x tan y 1− 3 · 7 21
2
and
1
tan z + tan w + 1 34
34 1
tan B = tan(z + w) = = 13 1 211 = 273
272 = = .
1 − tan z tan w 1 − 13 · 21 273
272 8
Thus,
10 10 10(1 − tan A tan B)

10 cot(x + y + z + w) = = =
tan(x + y + z + w) tan(A + B) tan A + tan B
10(1 − 12 · 18 )
= 1 1
2 + 8
= 15 = 015.
ANSWER : 015. 2
Solution #2: Using x, y, z, and w as introduced in Solution #1 above, we can draw four right
triangles, one each possessing angle x, y, z, and w, as shown.
Figure 6.4: The four right triangles shown here can be drawn in view of the information given in
Equation (6.36).
From the pictures of these triangles in Figure 6.4, we can compute all six trigonometric functions at
x, y, z, and w. We want to compute
cos(x + y + z + w)
cot(cot−1 3 + cot−1 7 + cot−1 13 + cot−1 21) = cot(x + y + z + w) = .
sin(x + y + z + w)
In this particular case, it is going to be somewhat lengthy and tedious to compute cos(x + y + z + w),
so instead of repeating it all over again to compute sin(x + y + z + w), it will probably be easier to
use Equation (6.18) to rewrite sin(x + y + z + w) in terms of cos(x + y + z + w). Thus, the resulting
calculation we require is
cos(x + y + z + w)
cot(cot−1 3 + cot−1 7 + cot−1 13 + cot−1 21) = p .
1 − cos2 (x + y + z + w)
Using Equation (6.30), we can rewrite cos(x + y + z + w) in terms of the values of the cosine and
sine functions at x, y, z, and w. Therefore, before we proceed further, let us use Figure 6.4 above
to collect these eight values:
3 7 13 21
cos x = √ , cos y = √ , cos z = √ , cos w = √ ,
10 50 170 442
1 1 1 1
sin x = √ , sin y = √ , sin z = √ , sin w = √ .
10 50 170 442
Now, from Equation (6.28), we have
cos(x + y + z + w) = cos(x + y) cos(z + w) − sin(x + y) sin(z + w), (6.37)
so let us compute the required terms on the right side of Equation (6.37).
Therefore,
cos(x + y) = cos x cos y − sin x sin y
3 7 1 1 20 2
=√ ·√ −√ ·√ =√ =√ ,
10 50 10 50 500 5
cos(z + w) = cos z cos w − sin z sin w

13 21 1 1 272 8
=√ ·√ −√ ·√ =√ √ =√ ,
170 442 170 442 170 442 65
sin(x + y) = sin x cos y + cos x sin y

1 7 3 1 10 1
=√ ·√ +√ ·√ =√ √ =√ ,
10 50 10 50 10 50 5
and
sin(z + w) = sin z cos w + cos z sin w

1 21 13 1 1
=√ ·√ +√ ·√ =√ .
170 442 170 442 65
Plugging these results into Equation (6.37), we have

2 8 1 1 3
cos(x + y + z + w) = √ · √ − √ · √ = √ .
5 65 5 65 13
Thus, we have
√3
cos(x + y + z + w) 13 3
cot(cot−1 3 + cot−1 7 + cot−1 13 + cot−1 21) = p =q = .
1− cos2 (x + y + z + w) 1− 9 2
13
Multiplying by 10, we obtain the final answer: 10 · 3

2 = 15 = 015. 2
Remark: There is no particular reason that x and y must be paired together and that z and w
must be paired together in Equation (6.37). One can put the four variables into two pairs in any
desired fashion, and in fact, one could argue that the grouping
√ used√in the solution above involves
undue computations because the largest denominators, 170 and 442, get multiplied together.
The reader is invited to arrive at an easier solution by using a different set of pairings of x, y, z, and
w. ANSWER : 015. 2
Additional problems requiring the use of trigonometric identities will be given in the exercises. This
section has been mainly concerned with basic properties, particularly numerical data and functional
identities, of the trigonometric functions. The connection with geometry has been confined to the
basic definitions of these functions as they relate to the sides of a right triangle. However, it comes as
no surprise that trigonometry is intricately linked to geometry. Chapters 8 and 9 will focus squarely
on geometry problems in two and three dimensions, and we will see the trigonometric functions arise
again in some of those problems as well. For the time being, however, we want to continue with an
algebraic focus and conclude this chapter by considering a couple of examples that exhibit interplay
between the logarithmic functions and trigonometric functions.
6.4 Putting Logarithmic and Trigonometric Functions To-

gether
In the preceding two sections, we have considered several AIME problems involving logarithm and
trigonometric functions separately. There are also some examples that simultaneously involve both
types of functions.
Example 6.4.1. (2003 AIME, Problem #4) Given that
log10 sin x + log10 cos x = −1 (6.38)
and that
1
log10 (sin x + cos x) = (log10 n − 1), (6.39)
2
find n.
Solution: The left side of Equation (6.38) can be simplified using part (1) of Theorem 6.2.2, so we
write
−1 = log10 sin x + log10 cos x = log10 (sin x cos x),
so that Equation (6.2) implies that
1
sin x cos x = 10−1 = . (6.40)
10
Now let us turn our attention to Equation (6.39). Recall our “Warning” after Theorem 6.2.2,
which reminds us that there is no immediate property that allows us to simplify log10 (sin x + cos x).
However, we can make nice progress by computing
1 6
(sin x + cos x)2 = sin2 x + 2 sin x cos x + cos2 x = 1 + 2 · = ,
10 5
where we have used Equations (6.18) and (6.40). Moreover, rearranging Equation (6.39) and using
part (3) of Theorem 6.2.2, we observe that

2
6
log10 n − 1 = 2 log10 (sin x + cos x) = log10 (sin x + cos x) = log10 ,
5
or
6
log10 n = 1 + log10 .
5
Using part (2) of Theorem 6.2.2, we have

6 5n
1 = log10 n − log10 = log10 .
5 6
6.4. PUTTING LOGARITHMIC AND TRIGONOMETRIC FUNCTIONS TOGETHER 117
Therefore,
5n
= 10,
6
which implies that n = 12 = 012. ANSWER : 012. 2
Remark: As an alternate approach, observe that if we solve for n in Equation (6.39) and then use
Equation (6.40), we have
n = 102 log(sin x+cos x)+1
2
= 10log[(sin x+cos x) ]+1
= 10(sin x + cos x)2

1
= 10 1 + = 12.
5
Example 6.4.2. (1991 AIME, Problem #4) How many real numbers x satisfy the equation
1
log x = sin(5πx)?
5 2
Solution: We can rewrite the given equation as
log2 x = 5 sin(5πx),
or
32sin(5πx) = x.
This equation holds precisely for real numbers x such that the graphs of the functions f (x) =
32sin(5πx) and g(x) = x intersect. Since −1 ≤ sin(5πx) ≤ 1, the function f (x) takes values in the
range
1
≤ f (x) ≤ 32.
32
Hence, any intersection of f (x) with the line g(x) = x must occur in the interval [0, 32]. Moreover,
f (x) is periodic, with period 2π/5π = 2/5. Hence, on the interval [0, 32], f (x) makes exactly
32/(2/5) = 80 full oscillations. At x = 0, we have f (x) = f (0) = 1, and the function begins by
1
rising to y = 32 at x = . (See Figure 6.5.)
10
Figure 6.5: The graphs of the functions in this problem intersect at various points in the plane.
1 1
Thus, during the interval 0 ≤ x ≤ , f (x) 6= g(x). To the right of the point ( 10 , 32) (i.e. on the
10
1 1
interval , 32 ) on the graph of f (x), the function falls from y = 32 to y = exactly 80 times,
10 32
1
and the function rises from y = to y = 32 exactly 79 times. On each such rise or fall, the graph
32
of f (x) will intersect the graph of g(x) exactly once. Hence, the graphs cross one another exactly
80 + 79 = 159 times. ANSWER : 159. 2
6.5 Exercises
1. (1996 AIME, Problem #2) For each real number x, let bxc denote the greatest integer
that does not exceed x. For how many positive integers n is it true that n < 1000 and that
blog2 nc is a positive even integer?
2. (1995 AIME, Problem #2) Find the last three digits of the product of the positive roots
of √
1995xlog1995 x = x2 .
3. (1986 AIME, Problem #3) If tan x + tan y = 25 and cot x + cot y = 30, what is tan(x + y)?
4. (1994, Problem #4) Find the positive integer n for which
blog2 1c + blog2 2c + blog2 3c + · · · + blog2 nc = 1994.
(For real x, bxc is the greatest integer ≤ x.)
5. (2002 AIME, Problem #6) The solutions to the system of equations
log225 x + log64 y = 4
logx 225 − logy 64 = 1
are (x1 , y1 ) and (x2 , y2 ). Find log30 (x1 y1 x2 y2 ).
6. (1995 AIME, Problem #7) Given that (1 + sin t)(1 + cos t) = 5/4 and
m √
(1 − sin t)(1 − cos t) = − k,
n
where k, m, and n are positive integers with m and n relatively prime, find k + m + n.
7. (2008 AIME-2, Problem #8) Let a = π/2008. Find the smallest positive integer n such
that
2 cos(a) sin(a) + cos(4a) sin(2a) + cos(9a) sin(3a) + · · · + cos(n2 a) sin(na)

is an integer.
8. (2011 AIME, Problem #9) Suppose x is in the interval [0, π/2] and
3
log24 sin x (24 cos x) = .
2
Find 24 cot2 x.
22 m
9. (1991 AIME, Problem #9) Suppose that sec x + tan x = and that csc x + cot x = ,
7 n
m
where is in lowest terms. Find m + n.
n
6.5. EXERCISES 119
10. (2012 AIME, Problem #9) Let x, y, and z be positive real numbers that satisfy
2 logx (2y) = 2 log2x (4z) = log2x4 (8yz) 6= 0.

1
The value of xy 5 z can be expressed in the form p/q , where p and q are relatively prime
2
positive integers. Find p + q.
P44
cos n◦
11. (1997 AIME, Problem #11) Let x = Pn=1 44 . What is the greatest integer that does
◦
n=1 sin n
not exceed 100x?
12. (2009 AIME-2, Problem #11) For certain pairs (m, n) of positive integers with m ≥ n
there are exactly 50 distinct positive integers k such that |log m − log k| < log n. Find the
sum of all possible values of the product mn.
13. (2004 AIME, Problem #12) Let S be the set of ordered pairs (x, y) such that 0 < x ≤ 1,
0 < y ≤ 1, and blog2 ( x1 )c and blog5 ( y1 )c are both even. Given that the area of the graph of
S is m/n, where m and n are relatively prime positive integers find m + n. The notation bzc
denotes the greatest integer that is less than or equal to z.
14. (2007 AIME-2, Problem #12) The increasing geometric sequence x0 , x1 , x2 , . . . consists
entirely of integral powers of 3. Given that
7 7
!
X X
log3 (xn ) = 308 and 56 ≤ log3 xn ≤ 57,
n=0 n=0
find log3 (x14 ).
15. (2000 AIME-2, Problem #15) Find the least positive integer n such that
1 1 1 1
+ + ··· + = .
sin 45◦ sin 46◦ sin 47◦ sin 48◦ sin 133◦ sin 134◦ sin n◦
Chapter 7
Complex Numbers and

Polynomials
“I could never have imagined that my interest in mathematics would take me so far,
so fast. There is great satisfaction in knowing that every day I work, I am able to
solve problems that are interesting, exciting, and challenging.”
- M. Scott Elliott, Operations Research Analyst, Fedex Corporation
7.1 Introduction
Complex numbers play a crucial role in mathematics. Complex numbers make their first appearance
for many students when the student encounters the need to take the square root of a negative number.
For instance, consider the quadratic equation
x2 + x + 1 = 0. (7.1)
In using the quadratic formula to find the solutions for x, one obtains
√
−1 ± −3
x= . (7.2)
2
Despite the fact that Equation (7.1) has real (in fact,
√ integer) coefficients, the solutions given in
Equation (7.2) rely on some meaningful notion of −3. Mathematicians resolve this difficulty by
adjoining a new √symbol,√the “imaginary” number i, to the real numbers, with the property that
i2 = −1. Hence, −3 = 3i, so that the values of x obtained in (7.2) can be written
√
−1 ± 3i
x= .
2
121
122 CHAPTER 7. COMPLEX NUMBERS AND POLYNOMIALS
The set of complex numbers is the smallest collection of numbers that contains the real numbers
and the imaginary number i, and is closed with respect to operations of addition, subtraction,
multiplication, and division (by nonzero numbers). With this understanding, it can be shown that
the set of complex numbers consists precisely of all expressions of the form z = a + bi, where a and
b are real numbers. There are alternate forms in which complex numbers can be written that will
be discussed later. The form z = a + bi shown here is sometimes called the Cartesian form of z.
Complex numbers arise frequently in the study of polynomials. Of the many different types of
functions that arise in the AIME competitions, including exponential, logarithmic, trigonometric,
rational, and so on, perhaps the most common is the polynomial. Polynomials are very important in
many areas of mathematics and are a front-burner topic in the high school mathematics curriculum.
One is often interested in finding roots1 of polynomial equations, and those roots are frequently
complex numbers. Therefore, it is fitting that we study complex numbers and polynomials side by
side. In this chapter, we will start with a review of complex numbers, and then we will proceed to
see how they arise in the study of polynomial equations.
Let us briefly review the algebraic rules that apply to the complex numbers.
7.2 The Algebra of Complex Numbers
Every complex number z = a + bi (with a and b real numbers) can be associated with a point (a, b)
in the plane. In this context, the x-axis is generally referred to as the real axis, while the y-axis is
√ to as the imaginary axis. The modulus of z, written |z|, is the nonnegative real
generally referred
number |z| = a2 + b2 . The complex conjugate of z, written z, is the complex number z = a − bi.
Clearly, the complex number z is a real number if and only if z = z.
Given two complex numbers z1 = a + bi and z2 = c + di, we can perform the usual arithmetic
operations as recorded here:
Proposition 7.2.1. (Complex Numbers under Algebraic Operations) Let z1 = a + bi and

z2 = c + di be complex numbers, where a, b, c, and d are real. Then
• Addition: z1 + z2 = (a + c) + (b + d)i,
• Subtraction: z1 − z2 = (a − c) + (b − d)i,
• Multiplication: z1 z2 = (ac − bd) + (ad + cd)i,

z1 a + bi (a + bi)(c − di) (ac + bd) − (ad − bc)i
• Division: = = = .
z2 c + di c2 + d2 c2 + d2
Note as a special case of the multiplication operation, we have
z1 z1 = (a + bi)(a − bi) = a2 + b2 = |z|2 , (7.3)

1 Roots are sometimes referred to as zeros. By convention, the AMC competitions reserve the word “roots” to be
used in the context of solutions to polynomial equations, while “zeros” is reserved for more general types of functions.
In common practice, however, the terms are often used interchangeably.
7.2. THE ALGEBRA OF COMPLEX NUMBERS 123
which is always a nonnegative real number.

Already with this limited information, we can solve the following problem from the 2007 AIME.
Example 7.2.2. (2007 AIME, Problem #3) The complex number z is equal to 9 + bi, where b
is a positive real number and i2 = −1. Given that the imaginary parts of z 2 and z 3 are equal, find b.
Solution: It is natural to directly compute z 2 and z 3 in terms of b and then equate their imaginary
parts:
z 2 = (9 + bi)2 = (81 − b2 ) + 18bi and z 3 = (9 + bi)3 = (729 − 27b2 ) + (243b − b3 )i. (7.4)
Thus, we obtain
18b = 243b − b3 , (7.5)
or
b3 = 225b. (7.6)
2
Since b is a positive real number, we can divide both sides by b to obtain b = 225. Thus, since b is
assumed positive, we obtain b = 15 = 015. ANSWER : 015. 2
Remark: If one observes that it is not necessary to consider the real parts of z 2 and z 3 , then
there is no need to compute the full expansions of z 2 and z 3 carried out in (7.4). In particular,
it is unnecessary to compute 93 = 729. Instead, one can directly deduce (7.5). As an additional
alternative, one can compute
z 3 − z 2 = z 2 (z − 1) = (81 − b2 + 18bi)(8 + bi).
Setting the imaginary part of this equal to zero (since the imaginary parts of z 2 and z 3 are assumed
equal) immediately yields (7.6).
In Chapter 1, we studied systems of equations in the case where all unknowns were real numbers.
We can, of course, study systems in which the variables may assume complex values as well. The
2012 AIME supplies just such an example, and we conclude this section with it.
Example 7.2.3. (2012 AIME-2, Problem #8) The complex numbers z and w satisfy the system
20i
z+ =5+i
w
12i
w+ = −4 + 10i.
z
Find the smallest possible value of |zw|2 .
General Strategy Comment: We have a system with two equations and two unknowns z and w.
In some cases, such a system will have a unique solution for z and w. If so, it could be found, for
instance, by solving for z in the first equation and plugging the result into the second equation to
obtain w. However, if the reader tries to do this, the equation
12i
w+ 20i
= −4 + 10i
5+i− w
must be solved for w. Writing w = a + bi, if one tries to simplify the left side and then set the real
and imaginary components of each side equal, the resulting pair of equations becomes
a2 + b2 − 20 a2 + b2 − 4b
a + 4 = −12 and b − 10 = −60 ,
(5a − b)2 + (a + 5b − 20)2 (5a − b)2 + (a + 5b − 20)2
whose complexity far exceeds what we started with or what an AIME problem is trying to test in
its students. Not to mention the time and care required to reach this point! It is not even clear
whether or not the pair of equations above has a unique solution for a and b. Indeed, as the question
is phrased, it seems to suggest that the value of zw might not be unique.
Here is another idea: Looking at the leading terms of both equations on the left side, we might be
tempted to multiply the first equation through by w and the second equation through by z. This
yields
zw + 20i = (5 + i)w and zw + 12i = (−4 + 10i)z.
Then subtracting one from the other will eliminate the zw term:
8i = (5 + i)w − (−4 + 10i)z.
At this point, however, it becomes very unclear how to proceed to get information about zw,
especially since that term was just eliminated from the system! Either of the ideas mentioned briefly
here would be reasonable to consider. However, the take-away lesson from this is that sometimes
we should look more carefully for a trick that will have us on the fast track to a solution without
wading through excessive and tedious calculations. Sometimes the most obvious-looking approach
is not the best one to follow. Now let us take a look at an elegant solution that might not have been
the reader’s first thought, but with an open mind comes into focus.
Solution: We can multiply the two equations together on each side to obtain

20i 12i
z+ w+ = (5 + i)(−4 + 10i) = −30 + 46i.
w z
That is,
240
zw + 32i − = −30 + 46i.
zw
Hence,
240
zw − = −30 + 14i.
zw
If we now multiply through by zw, we see that zw satisfies the quadratic equation
x2 + (30 − 14i)x − 240 = 0.
According to the quadratic formula, the roots are

p
−(30 − 14i) ± (30 − 14i)2 + 960 p
x= = −15 + 7i ± (15 − 7i)2 + 240.
2
Now every complex number has two square roots. In this case, we seek a solution for real numbers
a and b such that
(a + bi)2 = (15 − 7i)2 + 240 = 416 − 210i.
7.3. THE GEOMETRY OF COMPLEX NUMBERS 125
Thus, we obtain
a2 − b2 = 416 and 2ab = −210.
Although a and b are not specified as integers, we expect that they are in order to ensure that
zw = (−15 + 7i) ± (a + bi) has |zw|2 equal to an integer in the range 000 – 999. Since a2 ≥ 416,
the smallest positive integer a that meets this restriction is a = 21. Then a2 = 212 = 441, and from
2ab = −210, we conclude that b = −5. Alternatively, we can have a = −21 and b = 5. Thus, we
have p
(15 − 7i)2 + 240 = ±(21 − 5i),
and so we have
zw = −15 + 7i ± (21 − 5i) ∈ {6 + 2i, −36 + 12i}.
Thus, the minimum possible value for |zw|2 is
|zw|2 = |6 + 2i|2 = 62 + 22 = 36 + 4 = 40 = 040.
Up to this point, we have considered complex numbers from a purely algebraic point of view. It is
also instructive and useful for applications to have a geometric command of the complex plane as
well. We are now ready to consider this point of view.
7.3 The Geometry of Complex Numbers
To each complex number z = a + bi (where a and b are real), we have seen how to naturally
associate with it a point (a, b) in the complex plane. Rendering complex numbers as points in the
plane opens the possibility of applying geometric notions associated with two-dimensional space to
complex numbers. For instance, the “distance” between two complex numbers z1 = a + bi and
z2 = c + di canpbe defined as the corresponding distance in the plane between the points (a, b) and
(c, d), which is (a − c)2 + (b − d)2 . Not coincidentally, this is precisely the modulus of z1 − z2 :
p
|z1 − z2 | = (a − c)2 + (b − d)2 . (7.7)
Another concept that is relevant here is that of vectors. Each point (a, b) in the complex √
plane can
be viewed as a vector whose magnitude is the distance from the origin to (a, b), namely a2 + b2 ,
and which points with the same orientation (i.e. direction) as the line segment directed from the
origin of the plane to the point (a, b).
Let us now consider an example from the AIME competition that uses the modulus concept in
Equation (7.7).
Example 7.3.1. (1999 AIME, Problem #9) A function f is defined on the complex numbers
by f (z) = (a + bi)z, where a and b are positive numbers. This function has the property that the
image of each point in the complex plane is equidistant from that point and the origin. Given that
|a + bi| = 8 and that b2 = m/n, where m and n are elatively prime positive integers, find m + n.
Figure 7.1: To each point (a, b), corresponding in the complex plane to z = a + bi, we can associate
a vector emanating from the origin and terminating at (a, b).
General Strategy Comment: Many times in this text we have and will face mathematical de-
scriptions that must be translated into mathematical equations. We saw this immediately at the
outset of this book, and in this example, we see it again. Before reading the solution below, can you
express the property of the given function as a mathematical equation?
Solution: The image (under f ) of any point z in the complex plane is simply f (z). The distance
from f (z) to the origin is simply |f (z)|. Moreover, the distance from a given point z to its image
f (z) is simply |f (z) − z|. Thus, the given property of the function can be expressed simply as:
|f (z) − z| = |f (z)|,
for all complex numbers z. We are invited to apply this property to any complex number z = c + di
that we choose. What point should we choose? It would make sense to start with a point that is
easy to evaluate. Since z = 0 turns out to give no useful information, let us try z = 1 instead: we
have f (1) = a + bi. We are given that
|(a + bi) − 1| = |a + bi|.
Therefore,
(a − 1)2 + b2 = a2 + b2 .
From this, we obtain −2a + 1 = 0. Therefore, a = 21 . Since we are given that a2 + b2 = 64, we find
that
1 255
b2 = 64 − = .
4 4
Since m = 255 and n = 4 are relatively prime, we have m + n = 255 + 4 = 259.
Remark: It is possible to obtain the answer by applying f to an arbitrary input z = c + di (for
any real values of c and d) and carrying through a similar calculation involving moduli of complex
numbers, but this is more algebraically tedious than necessary. ANSWER : 259. 2
Another benefit of associating complex numbers with points in the plane is the possibility of an
alternate description of complex numbers in terms of polar coordinates. That is, to each point
√ b
(a, b) 6= (0, 0) in the plane, we can associated a pair (r, θ), where r = a2 + b2 > 0 and tan θ =
a
π
when a 6= 0. If a = 0 and b > 0, we have θ = + 2πk for some integer k, and if a = 0 and b < 0,
2
3π
we have θ = + 2πk for some integer k. There is a unique angle θ with 0 ≤ θ < 2π (measured in
2
radians) associated with each such point (other than the origin). In terms of r and θ, we have
a = r cos θ and b = r sin θ.
Figure 7.2: A point (a, b), corresponding

√ in the complex plane to z = a + bi, can be described in
polar coordinates r and θ, where r = a2 + b2 is the distance from (a, b) to the origin and θ is the
angle with respect to the positive real axis.
Thus, we can write

z = a + bi = r cos θ + ir sin θ = r(cos θ + i sin θ). (7.8)
The expression appearing on the far right in Equation (7.8) is often referred to as the polar form
of the complex number z, and it has many practical uses, as we shall see. For starters, note from
Equation (7.8) that it is easy to find the modulus of a number written in this form:
p p
|z| = |r cos θ + ir sin θ| = (r cos θ)2 + (r sin θ)2 = r cos2 θ + sin2 θ = r.
Furthermore, if z1 = r1 (cos θ1 + i sin θ1 ) and z2 = r2 (cos θ2 + i sin θ2 ), then using the trigonometric
identities presented in Chapter 6, we can show easily that

z1 z2 = r1 r2 cos(θ1 + θ2 ) + i sin(θ1 + θ2 ) . (7.9)
Hence,
|z1 z2 | = r1 r2 = |z1 ||z2 |. (7.10)
Equations (7.9) and (7.10) provide a geometric alternative to the basic algebraic rules in Proposition
7.2.1 and will be useful in our examples to come.
One of the algebraic complications that arises in working with complex numbers written in the form
z = a + bi is that it becomes highly non-trivial to compute large powers or roots of z. For instance,
finding an explicit value for z 100 = (a + bi)100 would require a virtually impossible calculation with
the Binomial Theorem (Theorem 2.5.1) in general. Since we are going to encounter many AIME
problems that require us to compute powers or roots of complex numbers, we desire an alternative
method for doing such calculations. Fortunately, help is available in the form of the famous Euler
Formula in complex analysis. Euler’s Formula requires a bit of explanation since it involves evaluating
the function f (θ) = eiθ for a complex number θ. Of course, e ≈ 2.71828 and the indeterminate,
θ, appears here as an exponent. This implies that f is an exponential function. These functions
have an important place in the AIME competition, and these were studied alongside logarithmic
functions in Section 6.2. Our use of exponential functions in the present chapter is mainly to service
the application of Euler’s Formula.
So how do we actually evaluate f (θ) = eiθ , which involves raising the number e to (possibly) nonreal
exponent? In calculus, one encounters the concept of a power series representation for a function.
Without going into too much detail, it can be shown that for every real number x, we have
x2 x3 x4
ex = 1 + x + + + + ··· .
2! 3! 4!
For each fixed x, the expression on the right-hand side is a convergent infinite series, and the value
that the series converges to is ex . From here, we can choose to define for any complex number z the
following:
z2 z3 z4
ez := 1 + z + + + + ··· .
2! 3! 4!
Using calculus, it can be shown that the series defined here is convergent for all complex numbers
z. As a special case, we have the following result, known as Euler’s formula.
Theorem 7.3.2. (Euler’s Formula) For every complex number θ, we have
eiθ = cos θ + i sin θ. (7.11)
Perhaps you have seen a math T-shirt displaying the remarkably simple, elegant equation
eiπ = −1
that relates two of the most famous irrational numbers, π and e, and of course, the most famous
nonreal complex number, i. It merely results from substituting θ := π into (7.11). Euler’s Formula
is important not only in mathematics, but also in physics and engineering. In fact, the famous
physicist Richard Feynman called the equation “our jewel” and “one of the most remarkable, almost
astounding, formulas in all of mathematics.”2
Proofs of Euler’s Formula involve calculus, such as limits or power series expansions. We already
recorded the power series expansion for ez above. In like manner, power series expansions for sin z
and cos z are available, and from them, we can easily deduce Euler’s Formula. However, since
calculus is not required background for the AIME, we will not provide the details of such a proof
here.
The power of Euler’s Formula can be readily seen by combining Equations (7.8) and (7.11) to obtain
z = reiθ . (7.12)
Once a complex number has been rendered in the form of Equation (7.12), a calculation like
z 100 = (reiθ )100 = r100 (eiθ )100 = r100 ei(100θ) = r100 cos(100θ) + ir100 sin(100θ)
becomes easy. More generally, for any integer n, we have z n = rn ei(nθ) . Appealing once more to
Euler’s Formula, we can deduce de Moivre’s Theorem3 , named for this 17th and 18th century French
mathematician. We simply note that
(cos θ + i sin θ)n = (eiθ )n = e(iθ)n = ei(nθ) = cos(nθ) + i sin(nθ).
2 Feynman, Richard P. (1977). The Feynman Lectures on Physics, Vol. 1, Addision-Wesley.
3 Historically,
de Moivre’s Theorem was proven prior to Euler’s Formula. It is possible to prove de Moivre’s Theorem
independently, for instance, by using induction and trigonometric identities.
Theorem 7.3.3. (de Moivre’s Theorem) For every complex number θ and integer n, we have
(cos θ + i sin θ)n = cos(nθ) + i sin(nθ).
Using de Moivre’s Theorem or appealing directly to Euler’s Formula, we can solve problems similar
to our next one, which will serve as a nice warm-up for the AIME examples to come later in the
chapter.
Example 7.3.4. Find all complex numbers z satisfying z 10 = 1. (Complex numbers z such that
z k = 1 for a fixed integer k ≥ 2 are called kth roots of unity.)
Solution: Let us write z using polar coordinates as described above: z = reiθ , where r > 0 and
0 ≤ θ < 2π. Then z 10 = r10 e10iθ = 1, which implies that r = 1 and 10θ = 2πk for some integer k.
That is,
1
θ = πk.
5
Since the integers k = 0, 1, 2, 3, 4, 5, 6, 7, 8, and 9 result in ten distinct angles, none of which differ
by 2π, we obtain ten distinct values for z:
z = 1, z = eiπ/5 , z = e2iπ/5 , z = e3iπ/5 , z = e4iπ/5 ,
z = e5iπ/5 = −1, z = e6iπ/5 , z = e7iπ/5 , z = e8iπ/5 , z = e9iπ/5 .
Remark. Using Euler’s Formula (7.11), it is easy to translate the 10 complex numbers z listed
above into the standard a + bi form. For instance, the first three solutions for z listed above are
π π 2π 2π
z = 1, z = cos( ) + i sin( ), z = cos( ) + i sin( ).
5 5 5 5
2
Next, let us consider a few examples from the AIME competition. The first one emphasizes the
viewpoint that complex numbers can be viewed as vectors. It also shows that although the complex
numbers may be presented initially in Cartesian form, there may be advantages to looking at them
a different way.
Example 7.3.5. (2012 AIME, Problem #6) Let z = a + bi be the complex number with |z| = 5
and b > 0 such that the distance between (1 + 2i)z 3 and z 5 is maximized, and let z 4 = c + di. Find
c + d.
Solution: The complex numbers appearing in this question have all been rendered in Cartesian form,
but this is a good problem to study geometrically (i.e. polar form), since we are given information
regarding distances between points in the complex plane. Let us write z = 5eiθ , where 0 < θ < π
(since b > 0). Thus, z resides on a circle of radius 5 in the complex plane. We wish to maximize
|z 5 − (1 + 2i)z 3 | = |z 3 ||z 2 − (1 + 2i)| = 125|z 2 − (1 + 2i)|,
where we have used Equation (7.10). Therefore, we are really trying to maximize the expression
|z 2 − (1√+ 2i)|. Now√1 + 2i is a fixed point in the first quadrant of the complex plane on a circle of
radius 12 + 22 = 5, while z 2 sits somewhere on a circle of radius 25.
Figure 7.3: In order to maximize the distance between z 2 and 1 + 2i, the point z 2 should appear
across the origin (on the opposite side) from the position of 1 + 2i in the complex plane.
Here is the key concept to this problem: From Figure 7.3, we see that in order for z 2 and 1 + 2i to
be as far apart as possible, z 2 should sit so that its angle is 180◦ (or π radians) opposite to 1 + 2i. In
other words, z 2 should reside on a ray extending in the direction of the complex number (or vector)
pointing from the origin towards −(1 + 2i) = −1 − 2i. There is a unique point in the complex
plane that lies on this ray and on a circle of radius 52 = 25 centered at the origin. We can write
z 2 = t(−1 − 2i) for some t > 0. Since
√
25 = |z 2 | = |t(−1 − 2i)| = 5t,
we must have
25 √
t = √ = 5 5.
5
Thus, √
z 2 = −5 5(1 + 2i).
Hence,
z 4 = 125(1 + 2i)2 = 125(−3 + 4i) = −375 + 500i.
Hence, c = −375 and d = 500. The answer is c + d = −375 + 500 = 125. 2
Our next example, taken from the 1990 AIME competition, is a more intricate application of roots
of unity (see Example 7.3.4) that also involves a touch of number theory.
Example 7.3.6. (1990 AIME, Problem #10) The sets A = {z : z 18 = 1} and B = {w : w48 =
1} are both sets of complex roots of unity. The set C = {zw : z ∈ A and w ∈ B} is also a set of
complex roots of unity. How many distinct elements are in C?
Solution: The set A consists of the complex numbers z = eiθ such that 1 = z 18 = e18iθ , which
kπ
requires that 18θ = 2πk for some integer k. That is, θ = for some integer k. Therefore, we have
9
A = {eikπ/9 : k is an integer}.
Similarly, we have
B = {ei`π/24 : ` is an integer}.
Note that |A| = 18 and |B| = 48, because although there are infinitely many choices for k and `,
they generate only 18 (resp. 48) different reference angles for elements of A (resp. B) in the interval
[0, 2π). Therefore,
C = {zw : z ∈ A and w ∈ B} = {eiπ(k/9+`/24) : k and ` are integers}.
k `
We must determine how many different reference angles are possible for θ := + . We can write
9 24
8k+3`
eiπ(k/9+`/24) = eπi( 72 ) = e2πi( 8k+3`
144 ) .
Note that 8k + 3` is an integer, and the collection of all possible reference angles for elements of C
is
2π(8k + 3`)
: k and ` are integers, and 8k + 3` ∈ {0, 1, 2, 3, . . . , 143} .
144
(If 8k + 3` < 0 or 8k + 3` > 143, then we can replace 8k + 3` by its unique value (mod 144) in the set
{0, 1, 2, . . . , 143} to obtain a reference angle.) Thus, there are a maximum of 144 distinct reference
angles that could belong to elements of C. Since gcd(8, 3) = 1, we can write
8k0 + 3`0 = 1
for some integers k0 and `0 . In fact, many choices are possible, including (by inspection) k0 = −1
and `0 = 3, or k0 = 8 and `0 = −21. Hence, for every integer n = 0, 1, 2, . . . , 143, we have
n = (8k0 + 3`0 )n = 8(nk0 ) + 3(n`0 ).

n
Thus, we can make angles θ equal in value to any fraction of 2π units of the form that we wish,
144
2πn
with n = 0, 1, 2, . . . , 143 (again, for all n < 0 and n > 143, the angle θ = duplicates a reference
144
angle that has already been accounted for). Thus, we have 144 different angles that can be made,
and hence, 144 elements of C. ANSWER : 144. 2
Problems involving integer powers of a complex number are not limited to positive integers, as our
next example shows.
Example 7.3.7. (2000 AIME-2, Problem #9) Given that z is a complex number such that
1 1
z + = 2 cos 3◦ , find the least integer that is greater than z 2000 + 2000 .
z z
Solution: Since we are asked to compute large powers of the complex number z, we will write z
in polar form: z = reiθ , where r > 0 is real and −π < θ ≤ π (measured in radians). (The reason
for choosing the interval (−π, π] will become clear later, and is only a matter of convenience. Any
1 1
half-open interval of length 2π can be used.) Then = e−iθ . Converting 3◦ to π/60 radians4 , we
z r
are given that
π 1
2 cos =z+
60 z
1
= reiθ + e−iθ
r
1
= r(cos θ + i sin θ) + (cos θ − i sin θ)
r
1 1
= (r + ) cos θ + i(r − ) sin θ.
r r
θπ
4 Recall that there are 2π radians for every 360 degrees, so that any angle of θ◦ measures radians.
180
Therefore, since 2 cos(π/60) is real, equating real and imaginary parts yields
π 1
2 cos = (r + ) cos θ (7.13)
60 r
and
1
r− sin θ = 0. (7.14)
r
Therefore, from Equation (7.14) and the condition that θ ∈ (−π, π], one of three possibilities must
1
hold: (a) θ = 0, (b) θ = π, (c) r = .
r
π 1
Case 1: θ = 0. In this case, Equation (7.13) shows that 2 cos = r + , so that
60 r
π
r2 − 2r cos + 1 = 0.
60
The quadratic formula quickly shows that the only values of r that satisfy this equation are nonreal.
Since r is required to be real, we have reached a contradiction.
π 1
Case 2: θ = π. In this case, Equation (7.13) shows that 2 cos = −(r + ), which is impossible
60 r
since r cannot be negative.
1
Case 3: r = . This case must hold, since Cases 1 and 2 are both impossible. Since r > 0, we see
r
that r = 1. Equation (7.13) shows that
π
θ=± .
60
Had we used an interval for θ other than (−π, π], then a different pair of values for θ would need to
be considered. It is inconsequential, however, as the only information we need in order to complete
the calculation from here is the value of cos θ, which is the same regardless of which reference angle
for θ is used.
Therefore,
z = reiθ = e±i( 60 ) .
π
We conclude that

2000 1 ±i 100π ∓i 100π 100π 4π 1
z + =e 3 +e 3 = 2 cos = 2 cos =2 − = −1.
z 2000 3 3 2
The least integer greater than this is 0 = 000. ANSWER : 000. 2
Of course, any study of the geometry of complex numbers would be enhanced by having a firm
foundation in geometry itself. There are nice examples of AIME problems about complex numbers
in which pictures and geometrical concepts and shapes are essential. Therefore, we shall return to
this topic in Section 8.4 where we will explore exactly these types of problems with the relevant
geometric information in mind.
7.4. BASIC DEFINITIONS AND FACTS ABOUT POLYNOMIALS 133
For the time being, however, we wish to explore complex numbers in a different context. One of
the most common places that complex numbers arise in mathematics is as roots of polynomials,
even when those polynomials possess only real coefficients. Therefore, armed with the brief review
of complex numbers presented above, let us turn our attention in Section 7.4 to the theory of
polynomials, beginning with a brief review of the basic terminology. Then in Section 7.5, we will
explore AIME problems about polynomials with complex roots.
7.4 Basic Definitions and Facts about Polynomials
A polynomial in one variable x is a function of the form

p(x) = an xn + an−1 xn−1 + · · · + a1 x + a0 ,
where a0 , a1 , . . . , an are constants. In the AIME competition, the values of these constants are
assumed to be real numbers unless otherwise specified. The largest integer j such that aj 6= 0 is
called the degree of p(x), written deg p(x), and this quantity plays a crucial role in the theory of
polynomials. If r is any number such that p(r) = 0, we call r a root5 of p(x). We have the following
cornerstone result concerning roots of p(x).
Theorem 7.4.1. (Fundamental Theorem of Algebra) If p(x) is a polynomial of degree n, then
p(x) has exactly n roots.
Of course, some (or all) of the roots of p(x) may be repeated or complex, even if all of the constants
a0 , a1 , . . . , an are real. It is also well known that if the coefficients of the polynomial p(x) are all
real and z = a + bi is a complex root of p(x), where a and b are real numbers and b is nonzero, then
z = a − bi is also a complex root of p(x). That is, the nonreal roots of a polynomial p(x) come in
complex conjugate pairs: a ± bi. This is expressed formally in the next theorem.
Theorem 7.4.2. Consider a polynomial p(x) = an xn + an−1 xn−1 + · · · + a2 x2 + a1 x + a0 , where
a0 , a1 , . . . , an are real coefficients. If z is a complex number such that p(z) = 0, then p(z) = 0.
The basic justification behind Theorem 7.4.2 is that the complex conjugation operation respects
both addition and multiplication. This means that if z1 and z2 are complex numbers, then
z1 + z2 = z1 + z2 and z1 z2 = z1 z2 .
Since p(x) requires only the addition and multiplication of complex numbers, p(z) = p(z), from
which the conclusion follows. One nice consequence of Theorem 7.4.2 is that any polynomial of odd
degree must have at least one real root, since every polynomial must have an even number of nonreal
roots – they come in pairs of complex conjugates.
We can use the roots of a polynomial to manufacture a factorization of the polynomial. We observe
that a complex number r is a root of the polynomial p(x) if and only if p(x) = (x − r)q(x), where
q(x) is a polynomial with deg q(x) = deg p(x) − 1. The polynomial q(x) can be found by using long
division to divide p(x) by x − r. Therefore, by the Fundamental Theorem of Algebra, if
p(x) = an xn + an−1 xn−1 + · · · + a1 x + a0 (7.15)
5 Roots of a polynomial are sometimes alternatively referred to as zeros of the polynomial.
is a polynomial of degree n, it has n roots, say r1 , r2 , . . . , rn , and we can write
p(x) = an (x − r1 )(x − r2 ) . . . (x − rn ). (7.16)
In this way, the polynomial p(x) has been factored into a product of n linear (i.e. degree 1) factors.
In the case where the leading coefficient an is 1, we say that p(x) is a monic polynomial:
p(x) = xn + an−1 xn−1 + · · · + a1 x + a0 = (x − r1 )(x − r2 ) . . . (x − rn ). (7.17)
From the two expressions for p(x) in Equation (7.17), we can draw some further conclusions about
the roots of p(x). Note that since the roots of p(x) are unaffected by multiplying the polynomial
through by a constant, we may assume that p(x) is monic.
Proposition 7.4.3. Let p(x) be a monic polynomial of degree n with roots r1 , r2 , . . . , rn . Then
(a):

a0 , if n is even,
r1 r2 · · · rn =
−a0 , if n is odd.
(b):
X
ri = −an−1 .
1≤i≤n
(c):
X
ri rj = an−2 .
1≤i<j≤n
Part (a) of Proposition 7.4.3 follows from the fact that
a0 = p(0) = (−1)n r1 r2 . . . rn ,
while the second and third parts follow by comparing coefficients of xn−1 and xn−2 , respectively,
in the expressions appearing in (7.17). None of the results listed in Proposition 7.4.3 rely on p(x)
having real roots, and they may be applied just as well to polynomials with complex roots. A good
example of this is Problem #7 from the AIME 2010-2 competition, and it appears in the exercises
for this chapter.
Additional statements can also be made relating the roots of p(x) to the coefficients of xn−3 , xn−4 ,
and so on, but we will leave these for the reader to derive.
Let us now consider some examples from the AIME to illustrate some of these ideas.
Example 7.4.4. (2008 AIME-2, Problem #7) Let r, s, and t be the three roots of the equation
8x3 + 1001x + 2008 = 0.
Find (r + s)3 + (s + t)3 + (t + r)3 .

Solution: Let
p(x) = 8x3 + 1001x + 2008. (7.18)
Since r, s, and t are the roots of p(x), we can use Equation (7.16) to write
p(x) = 8(x − r)(x − s)(x − t) = 8x3 − 8(r + s + t)x2 + 8(rs + rt + st) − 8rst. (7.19)
Therefore, equating the coefficients in Equations (7.18) and (7.19), we find that
r + s + t = 0, (7.20)
1001
rs + rt + st = , (7.21)
8
and
2008
rst = − = −251. (7.22)
8
The solver should spend some moments to think carefully about how to proceed from here to compute
(r + s)3 + (s + t)3 + (t + r)3 .
On the one hand, Equations (7.20), (7.21), and (7.22) may provide enough information to solve the
system explicitly for r, s, and t. However, this appears to be a tedious task. On the other hand, we
could use these three equations to simplify the expression we are trying to find. In particular, by
using (7.20), we can write
(r + s)3 + (s + t)3 + (t + r)3 = (−t)3 + (−r)3 + (−s)3 = −(r3 + s3 + t3 ). (7.23)
We have replaced the original problem with a new problem, to determine r3 + s3 + t3 . But how can
we derive information about r3 + s3 + t3 from Equations (7.20) - (7.22)? The key observation is that
the expansion of (r + s + t)3 will include r3 + s3 + t3 as well as rst. While it may not yet be clear
exactly how this will be useful, let us try to expand this trinomial:
(r + s + t)3 = r3 + s3 + t3 + 3(r2 s + rs2 + r2 t + rt2 + s2 t + st2 ) + 6rst. (7.24)
We already know the value of r + s + t and rst. Therefore, the only part of Equation (7.24) standing
in the way of computing r3 + s3 + t3 is the expression r2 s + rs2 + r2 t + rt2 + s2 t + st2 . However, we
can work on this expression with factorization as follows:
r2 s + rs2 + r2 t + rt2 + s2 t + st2 = rs(r + s) + rt(r + t) + st(s + t)

= rs(−t) + rt(−s) + st(−r)
= −3rst.
Therefore, Equation (7.24) simplifies as follows:
03 = r3 + s3 + t3 + 3(−3rst) + 6rst = r3 + s3 + t3 − 3rst,
which implies that

r3 + s3 + t3 = 3rst = 3(−251) = −753.
Therefore,
(r + s)3 + (s + t)3 + (t + r)3 = −(r3 + s3 + t3 ) = 753.
ANSWER : 753. 2
Remark: This example illustrates an important idea for problem solvers. The entire solution may
not have been clear from the outset. This is no reason to give up. Instead, we recognized that by
computing (r + s + t)3 , there would be a lot of terms in the resulting expansion that are already
known or related to the terms we want to find. Thus, without seeing the end result in advance, we
push forward with that calculation, which gave us Equation (7.24) and a way to organize and use
the information that we had been previously given.
The next example has a great deal of similarity with the previous one. The reader is encouraged to
try it first on their own without reading the solution.
Example 7.4.5. (1996 AIME, Problem #5) The roots of
x3 + 3x2 + 4x − 11 = 0
are a, b, and c. The equation with roots a + b, a + c, and b + c is
x3 + rx2 + sx + t = 0.
Find t.
Solution: By Equation (7.16), we have

(x − a)(x − b)(x − c) = x3 + 3x2 + 4x − 11. (7.25)
Since
(x − a)(x − b)(x − c) = x3 − (a + b + c)x2 + (ab + ac + bc)x − abc, (7.26)
we can equate the coefficients of (7.25) and (7.26) to obtain
abc = 11, ab + ac + bc = 4, a + b + c = −3. (7.27)
By part (a) of Proposition 7.4.3, we know that
t = −(a + b)(a + c)(b + c).
Now for the crucial point – how best to compute t. Using the equations in (7.27), we can write
t = −(−3 − c)(−3 − b)(−3 − a)
= (a + 3)(b + 3)(c + 3)
= abc + 3(ab + ac + bc) + 9(a + b + c) + 27
= 11 + 3(4) + 9(−3) + 27
= 23 = 023.
ANSWER : 023. 2
The preceding example did not require us to determine the explicit values of the roots a, b, and
c. However, finding roots of polynomials is a common, central task in algebra at this level. The
determination of rational roots (i.e. roots that can be expressed as fractions of integers) can be
algorithmically carried out using the so-called Rational Roots Test in the case where the polynomial
has integer coefficients.
Theorem 7.4.6. (Rational Roots Test) Let p(x) = an xn + an−1 xn−1 + · · · + a1 x + a0 be a

polynomial with integer coefficients a0 , a1 , . . . , an . Every rational root of p(x) can be expressed in
the form r/s, where r is a divisor of a0 and s is a divisor of an .
For instance, in Example 5.3.3, we needed to search for roots of the equation
8x3 − 8x2 + 1 = 0.
According to the Rational Roots Test, the candidates for rational roots to this equation take the
form r/s, where r divides 1 and s divides 8. Thus, the list of candidates is
1 1 1
±1, ± , ± , and ± .
2 4 8
In that example, we discovered only one rational root, r = 12 . On problems such as this, the finite list
of candidates turns a seemingly impossible task of searching for rational roots into a finite algorithm.
Of course, for polynomials with non-rational roots, such an algorithm will not be fruitful in locating
those roots.
Of course, even when it is not possible to locate the roots of a polynomial, such a polynomial can still
admit a non-trivial factorization. The analogue in number theory is the fact that positive integers
often have many possible factorizations, even though according to the Fundamental Theorem of
Arithmetic there is only one prime factorization. For example, we could choose to factorize 260
as 260 = 26 · 10. As an illustration with polynomials, we could factor p(x) = x4 + 2x2 + 1 as
p(x) = (x2 + 1)2 , which in and of itself is not directly tied to knowing the roots of p(x).
In fact, it may at first be surprising to learn that much of the basic theory of numbers (e.g. divisibility,
factorizations, congruences, etc.) has an analogue in the setting of polynomials. One example is the
Division Algorithm, which states that we can always divide a nonzero integer a by a positive integer
d to obtain a unique quotient q and unique remainder r with 0 ≤ r < d such that
a = qd + r.
The condition that d divides a is precisely saying that the remainder r obtained in the Division
Algorithm is 0. We can do something similar for polynomials:
Proposition 7.4.7. (Division Algorithm) Suppose that p(x) and d(x) are nonzero polynomials
with deg d(x) ≥ 1. Then there are unique polynomials q(x) and r(x) with deg r(x) < deg d(x) such
that
p(x) = q(x)d(x) + r(x).
We say that d(x) is a factor of p(x) (or that d(x) divides p(x)) if r(x) = 0 in the Division Algorithm.
Here are a couple more important facts in this case:
Proposition 7.4.8. Suppose that d(x) divides p(x) and write p(x) = q(x)d(x) as in the Division
Algorithm. Then
(a): If t is a root of d(x), then t is also a root of p(x).
(b): If deg q(x) = k and deg d(x) = ` then deg p(x) = k + `.
To prove part (a), suppose that t is a root of d(x). Thus, d(t) = 0. Hence, p(t) = q(t)d(t) = q(t) · 0 =
0, so that t is a root of p(x) as well. Part (b) follows quickly by expanding the product
q(x)d(x) = (ak xk + ak−1 xk−1 + · · · + a0 )(b` x` + b`−1 x`−1 + · · · + b0 ).
Many nice applications of Propositions 7.4.7 and 7.4.8 can be found in the AIME competition.
However, before we consider some AIME examples, here is a nice prelude contained in the AMC 12
competition.
Example 7.4.9. (AMC 12 A 2005, Problem #24) Let P (x) = (x − 1)(x − 2)(x − 3). For
how many polynomials Q(x) does there exist a polynomial R(x) of degree 3 such that P (Q(x)) =
P (x) · R(x)?
Solution: The roots of P (x) are clearly x = 1, x = 2, and x = 3. By part (a) of Proposition 7.4.8,
the given equation P (Q(x)) = P (x) · R(x) implies that any root of P (x) must also be a root of
P (Q(x)). Hence,
P (Q(1)) = P (Q(2)) = P (Q(3)) = 0.
Thus, Q(1), Q(2), and Q(3) are roots of P (x), and thus,
Q(1), Q(2), and Q(3) must belong to the set {1, 2, 3}. (7.28)
Since deg(P (x)) = deg(R(x)) = 3, part (b) of Proposition 7.4.8 implies that P (Q(x)) has degree
6. Moreover, if deg(Q(x)) = n, then deg(P (Q(x))) = 3n. (To see this, observe that if Q(x) = axn
(with a 6= 0), then P (Q(x)) = (axn )3 + · · · = a3 x3n + · · · . Details are left to the reader.) From this
we see that Q(x) has degree 2.
If we write Q(x) = ax2 + bx + c (with a 6= 0), then the three values Q(1), Q(2), and Q(3) will
completely determine the values of a, b, and c. This can be seen either algebraically by using
Q(1), Q(2), and Q(3) to set up three equations that can be uniquely solved for a, b, and c, or
geometrically from the fact that the points (1, Q(1)), (2, Q(2)), and (3, Q(3)) in the plane determine
a unique polynomial of degree 2 or less. Therefore, according to (7.28) and the Multiplication
Principle (Theorem 2.2.1), we have 33 = 27 possible polynomials Q(x) of degree 2 or less. However,
any of the 27 choices that results in a = 0 must be ignored, since in that case Q(x) = bx + c is no
longer quadratic. The values of Q(1), Q(2), and Q(3) for which Q(x) is not quadratic (i.e., Q(x) is
linear of degree 1 or less) are summarized in the table below:
Q(1) Q(2) Q(3) Q(x)

1 1 1 Q(x) = 1
2 2 2 Q(x) = 2
3 3 3 Q(x) = 3
1 2 3 Q(x) = x
3 2 1 Q(x) = 4 − x
We conclude that among the 27 possible values of (Q(1), Q(2), Q(3)), there are 27 − 5 = 22 that
result in quadratic polynomials Q(x). Therefore, we have 22 possible quadratic polynomials Q(x).
ANSWER: 22. 2
Let us now conclude this section with a couple of examples from the AIME that demonstrate the
use of Propositions 7.4.7 and 7.4.8.
Example 7.4.10. (2010 AIME-2, Problem #6) Find the smallest positive integer n with the
property that the polynomial x4 − nx + 63 can be written as a product of two nonconstant polynomials
with integer coefficients.
Remark: We remind the reader here that if p(x) and q(x) are polynomials with real coefficients of
degree m and n, respectively, then p(x)q(x) is a polynomial of degree m + n. We will need this fact
in the solution below.
Solution: Since we must write the polynomial
p(x) = x4 − nx + 63 (7.29)
as a product of nonconstant polynomials, part (b) of Proposition 7.4.8 shows that we must either
express p(x) as a product of a polynomial of degree 1 and a polynomial of degree 3, or we must
express p(x) as a product of two quadratic polynomials. It is not clear at present which of these two
expressions can be achieved with the smallest integer n, so we will consider both cases:
Case 1: We can write p(x) as a product of a polynomial of degree 1 and a polynomial of

degree 3. Since the polynomial factors must contain integer coefficients, the degree 1 polynomial
must take the form x − a for some integer a. So we have
p(x) = (x − a)q(x), (7.30)
so that a is an integer root of p(x). It follows from Equations (7.29) and (7.30) that a must be a
divisor of 63. Thus, a = 1, 3, 7, 9, 21, or 63.
If a = 1, then
0 = p(1) = 64 − n,
so n = 64.
If a = 3, then
0 = p(3) = 81 − 3n + 63 = 144 − 3n,
so n = 48.
If a ≥ 7, then from
0 = p(a) = a4 − na + 63,
we solve for n to obtain
63
n = a3 + > 73 > 48.
a
Case 2: We can write p(x) as a product of two (monic) quadratic polynomials with
integer coefficients:
p(x) = (x2 + ax + b)(x2 + cx + d),
where a, b, c, d are integers. Expanding the right-hand side, we have
x4 − nx + 63 = x4 + (a + c)x3 + (b + d + ac)x2 + (ad + bc)x + bd.
Hence,
a + c = 0, b + d + ac = 0, ad + bc = −n, bd = 63.
This is a system of four equations and four (integer) unknowns. At first this may seem daunting,
but such calculations arise surprisingly often in algebra, and with some practice, they can be quite
tractable. In this instance, the trick is to reduce the number of variables in play by substituting
c = −a. The remaining equations become
b + d = a2 , a(b − d) = n, bd = 63.
Now since b and d are integers, we must have:
{b, d} = {1, 63} or {3, 21} or {7, 9}.
In the middle case, b + d = 24 is not a perfect square, contradicting b + d = a2 . In the first case,
b+d = 64, so a = ±8, from which we obtain n = a(b−d) = 8·62 = 496. In the third case, b+d = 16,
so a = ±4, from which we obtain n = 4 · 2 = 8.
Conclusion from Cases Analysis: The smallest value of n we have found in our analysis in either
case is n = 8, so this must be the answer. Indeed, we can factor p(x) as follows:
p(x) = (x2 + 4x + 9)(x2 − 4x + 7) = x4 − 8x + 63.
Thus, n = 8 = 008. ANSWER : 008.2

Example 7.4.11. (2003 AIME-2, Problem #9) Consider the polynomials
P (x) = x6 − x5 − x3 − x2 − x and Q(x) = x4 − x3 − x2 − 1.
Given that z1 , z2 , z3 , and z4 are roots of Q(x) = 0, find P (z1 ) + P (z2 ) + P (z3 ) + P (z4 ).
Solution: Using the Division Algorithm (Proposition 7.4.7), we divide P (x) by Q(x) via long
division to obtain
P (x) = (x2 + 1)Q(x) + (x2 − x + 1).
We are given that Q(zi ) = 0 for i = 1, 2, 3, 4, so that P (zi ) = zi2 − zi + 1 for each i = 1, 2, 3, 4.
Therefore,
P (z1 ) + P (z2 ) + P (z3 ) + P (z4 ) = z12 + z22 + z32 + z42 − z1 − z2 − z3 − z4 + 4.
Now we need to find

z1 + z2 + z3 + z4 and z12 + z22 + z32 + z42 .
By (7.16), we can write
Q(x) = (x − z1 )(x − z2 )(x − z3 )(x − z4 ),
and now we can use part (b) of Proposition 7.4.3 to compute
z1 + z2 + z3 + z4 = 1.
7.5. POLYNOMIALS WITH COMPLEX ROOTS 141
To find z12 + z22 + z32 + z42 , use part (c) of Proposition 7.4.3 as follows:
1 = (z1 + z2 + z3 + z4 )2
= (z12 + z22 + z32 + z42 ) + 2(z1 z2 + z1 z3 + z1 z4 + z2 z3 + z2 z4 + z3 z4 )
= (z12 + z22 + z32 + z42 ) + 2(−1).
Thus,
z12 + z22 + z32 + z42 = 3.
Hence,
P (z1 ) + P (z2 ) + P (z3 ) + P (z4 ) = (z12 + z22 + z32 + z42 ) − (z1 + z2 + z3 + z4 ) + 4 = 3 − 1 + 4 = 6 = 006.
ANSWER : 006.2
7.5 Polynomials with Complex Roots
The examples in the preceding section did not require us to consider complex-valued roots to the
polynomials in question. However, it would be a mistake to assume that this represents the majority
of problems related to roots of polynomials. With our study of complex numbers earlier in this
chapter, together with some good practice with polynomials, we are now well-equipped to tackle
problems about polynomials with complex roots in this section. Almost all of the background
material is already in place, so we will be able to pass quickly here to the problems. Recall that in
Section 7.4, we saw that the coefficients of a polynomial are related to the roots of the polynomial
(Proposition 7.4.3). We can use this basic information to solve the following problem.
Example 7.5.1. (1995 AIME, Problem #5) For certain real values of a, b, c, and d, the
equation x4 + ax3 + bx2 + cx + d = 0 has four nonreal roots. √The product of two of these roots is
13 + i and the sum of the other two roots is 3 + 4i, where i = −1. Find b.
Solution: According to Theorem 7.4.2, all nonreal roots of a polynomial equation always occur in
complex conjugate pairs. Therefore, we can write the four nonreal roots of the given equation as
r1 , r1 , r2 , and r2 . We are given that the product of two of the roots is 13 + i, but according to
Equation (7.3), r1 r1 and r2 r2 are real. Therefore, we must have either r1 r2 = 13 + i or r1 r2 = 13 + i.
We may as well assume without loss of generality that the former holds: r1 r2 = 13 + i. The given
information then implies that r1 + r2 = 3 + 4i. Now we have
x2 + ax3 + bx2 + cx + d = (x − r1 )(x − r2 )(x − r1 )(x − r2 )

= (x2 − (r1 + r2 )x + r1 r2 )(x2 − (r1 + r2 )x + r1 r2 ).
Since r1 r2 = 13 + i, we have
r1 r2 = 13 + i = 13 − i,
and since r1 + r2 = 3 + 4i, we have
r1 + r2 = 3 − 4i.
Therefore, we have
x2 + ax3 + bx2 + cx + d = (x2 − (3 − 4i)x + (13 + i))(x2 − (3 + 4i)x + (13 − i)).
Expanding this expression, the coefficient of x2 is
b = (13 − i) + (3 − 4i)(3 + 4i) + (13 + i) = 51 = 051.
ANSWER : 051. 2
It is not unusual for problems involving complex roots of polynomial equations to require one to
express the roots in polar form. That is, it is often necessary to write the complex root z in the
form z = reiθ . This is not surprising in view of the fact that polynomial equations involving z will
require the use of powers z k of z, and these powers are much simpler to compute when z has been
rendered in polar form.
Example 7.5.2. (1984 AIME, Problem #8) The equation
p(z) = z 6 + z 3 + 1 = 0
has a root reiθ with 90◦ < θ < 180◦ . Find θ.
General Strategy Comment: When dealing with polynomials of relatively high degree, it is
sometimes possible to invoke a change of variables to reduce the degree. Lower degree polynomials
are usually much easier to work with. For example, the roots of a polynomial of degree 2 are easily
obtained from the quadratic formula.
Solution: We observe by Theorem 7.4.6 that the only possible rational roots are z = 1 and z = −1.
Since p(1) = 3 6= 0 and p(−1) = 1 6= 0, we conclude that p(z) has no rational roots. Proceeding more
generally, let us attempt to reduce the degree of the polynomial (in view of the General Strategy
Comment above) by using a substitution. We make the change of variables y = z 3 , so that, in terms
of y, the equation becomes y 2 + y + 1 = 0, whose solutions are readily found with the quadratic
formula: √ √
−1 ± −3 −1 ± 3i
y= = . (7.31)
2 2
Figure 7.4: The roots of the quadratic equation y 2 + y + 1 = 0 can be expressed in both Cartesian
and polar form.
7.5. POLYNOMIALS WITH COMPLEX ROOTS 143
Using a polar description of y in the form y = reiθ , note that both roots in (7.31) have r = 1,
and the values of θ are θ = 2π 4π
3 or θ = 3 (measured in radians). Allowing for multiples of 2π, we
conclude that
1 2
y = e2π(k+ 3 ) and y = e2π(k+ 3 ) .
√
Now using z = 3 y, we have
k 1 k 2
z = e2π( 3 + 9 ) and z = e2π( 3 + 9 ) .
We must find k such that 2π( k3 + 91 ) or 2π( k3 + 29 ) lies between π/2 and π. This, in turn, requires
that k3 + 19 or k3 + 29 lies between 41 and 12 . Only with k = 1 in the former case does the result lie in
the specified range, so the angle, in radians, is
1 1 4 8π
θ = 2π( + ) = 2π · = .
3 9 9 9
In degrees6 , we have θ = 160. ANSWER : 160. 2
Now let us study some additional examples of polynomial equations whose roots are complex.
Example 7.5.3. (2001 AIME-2, Problem #14) There are 2n complex numbers that satisfy
both z 28 − z 8 − 1 = 0 and |z| = 1. These numbers have the form zm = cos θm + i sin θm , where
0 ≤ θ1 < θ2 < · · · < θ2n < 360 and angles are measured in degrees. Find the value of θ2 +θ4 +· · ·+θ2n .
Solution: The only real numbers that satisfy |z| = 1 are ±1, and since 1 and −1 do not satisfy
z 28 − z 8 − 1 = 0, each of the 2n solutions to z 28 − z 8 − 1 = 0 with |z| = 1 is nonreal. These nonreal
solutions come in n complex conjugate pairs. Hence, we know that 0 < θ1 < θ2 < · · · < θn < 180,
and once these values are found, the remaining values θn+1 , θn+2 , . . . , θ2n are easily derived, since
θn + θn+1 = 360, θn−1 + θn+2 = 360, ..., θ1 + θ2n = 360.
Now let us proceed to find the values θi for i = 1, 2, . . . , n. We can write a complex number z with
|z| = 1 in polar form as z = eiθ . Therefore,
z 28 − z 8 − 1 = e28iθ − e8iθ − 1 = (cos(28θ) − cos(8θ) − 1) + i (sin(28θ) − sin(8θ)) .
Hence, z satisfies z 28 − z 8 − 1 = 0 if and only if
cos(28θ) = cos(8θ) + 1 and sin(28θ) = sin(8θ). (7.32)
If we sum the squares of both sides of both equations and using the identity sin2 x + cos2 x = 1, we
derive that
1 = 2 + 2 cos(8θ).
1
That is, cos(8θ) = − . Therefore,
2
8θ = ±120 + 360t,
6 Recall once more that to convert an angle measure from radians to degrees, one must multiply the radian measure
180 π
by π
, and conversely, conversion from degrees to radians is accomplished by multiplying by 180 .
for some integer t. That is,

θ = ±15 + 45t,
for some integer t. The values of θ that take one of these forms in the interval 0 < θ < 180 are
θ = 15, 30, 60, 75, 105, 120, 150, and 165. These values of θ are therefore the only candidates for valid
solutions to the system (7.32), but it still must be determined which of them are indeed solutions.
Directly plugging each of the eight values of θ into (7.32) shows that the only valid choices for θ
with 0 < θ < 180 are θ = 15, 75, 105, and 165. Hence, n = 4, and we have
θ1 = 15, θ2 = 75, θ3 = 105, θ4 = 165, θ5 = 195, θ6 = 255, θ7 = 285, θ8 = 345.
Therefore,
θ2 + θ4 + θ6 + θ8 = 75 + 165 + 255 + 345 = 840.
ANSWER : 840. 2
Example 7.5.4. (1996 AIME, Problem #11) Let P be the product of those roots of z 6 + z 4 +
z 3 + z 2 + 1 = 0 that have positive imaginary part, and suppose that P = r(cos θ◦ + i sin θ◦ ), where
0 < r and 0 ≤ θ < 360. Find θ.
Solution: The key to starting this problem is to recognize that
p(z) = z 6 + z 4 + z 3 + z 2 + 1 (7.33)
is very similar to the form of a finite geometric series introduced in the last chapter. One might
hope to modify p(z) in such a way as to exploit this, but the tricky part is to see how best to modify
it. For instance, one option is to write
z7 − 1
p(z) + z 5 + z =
z−1
and obtain a common denominator:
z 7 − 1 − z(z 4 + 1)(z − 1)
p(z) = . (7.34)
z−1
Another option is to write

p(z) = (z 6 + z 4 + z 2 + 1) + z 3
and use the formula for a finite geometric series with common ratio z 2 on the portion in parentheses
to obtain
z8 − 1 z 8 − 1 + (z 2 − 1)z 3
p(z) = 2 + z3 = . (7.35)
z −1 z2 − 1
However, attempts to simplify either (7.34) or (7.35) by factoring in the numerator simply bring the
solver back to the original expression (7.33).
7.6. EXERCISES 145
Still, the strategy of utilizing a geometric series will be successful if the reader will persist just a bit
further and try a different manipulation as follows. We write
p(z) = z 6 + z 4 + z 3 + z 2 + 1
= (z 6 − z) + (z 4 + z 3 + z 2 + z + 1)
z5 − 1
= z(z 5 − 1) +
z−1
z(z − 1)(z 5 − 1) + (z 5 − 1)
=
z−1
(z − z + 1)(z 5 − 1)
2
= .
z−1
The advantage of this latter expression over (7.34) and (7.35) is that we can completely determine
all roots of the numerator; they consist of the two roots of the quadratic equation z 2 − z + 1 = 0
and the 5th roots of unity. Therefore, the six roots of p(z) = z 6 + z 4 + z 3 + z 2 + 1 = 0 consist of the
four solutions of z 5 − 1 = 0 other than z = 1 and the two solutions to z 2 − z + 1 = 0.
Following the model in Example 7.3.4, suppose that z = reiθ is a root of the equation z 5 − 1 = 0.
Then r5 e5iθ = 1. This implies that r = 1 and that 5θ = 2πk for some integer k. Therefore, θ = 2πk 5 ,
where k is an integer. If k = 0, we obtain θ = 0 and z = 1. To obtain four roots of z 5 − 1 = 0 besides
z = 1, we set k = 1, 2, 3, and 4. This gives us
z1 = eiθ1 , z2 = eiθ2 , z3 = eiθ3 , z4 = eiθ4 ,
where
2π 4π 6π 8π
θ1 = = 72◦ , θ2 = = 144◦ , θ3 = = 216◦ ,
θ4 = = 288◦ .
5 5 5 5
√
1±i 3
By the quadratic formula, the two roots of z 2 − z + 1 = 0 are z = , whose polar forms are
2
z5 = eiπ/3 and z6 = e5iπ/3 . Therefore, the two additional roots have polar forms with angle θ5 = 60◦
and θ6 = 300◦ , respectively. The values θ1 , θ2 , and θ5 are in the interval 0◦ < θ < 180◦ , and hence
have positive imaginary part. Therefore,
P = z1 z2 z5 = eiθ1 eiθ2 eiθ5 = ei(θ1 +θ2 +θ5 ) = cos 276◦ + i sin 276◦ ,
since 72 + 144 + 60 = 276. Therefore, θ = 276. ANSWER : 276. 2
7.6 Exercises
1. (2009 AIME, Problem #2) There is a complex number z with imaginary part 164 and a
positive integer n such that
z
= 4i.
z+n
Find n.
2. (1985 AIME, Problem #3) Find c if a, b, and c are positive integers which satisfy c =
(a + bi)3 − 107i, where i2 = −1.
3. (2001 AIME, Problem #3) Find the sum of all the roots, real and nonreal, of the equation
2001
x2001 + 12 − x = 0, given that there are no multiple roots.
4. (1993 AIME, Problem #5) Let P0 (x) = x3 + 313x2 − 77x − 8. For integers n ≥ 1, define
Pn (x) = Pn−1 (x − n). What is the coefficient of x in P20 (x).
5. (2005 AIME, Problem #6) Let P be the product of the nonreal roots of x4 −4x3 +6x2 −4x =
2005. Find bP c. (The notation bP c denotes the greatest integer that is less than or equal to
P .)
6. (2010 AIME-2, Problem #7) Let P (z) = z 3 + az 2 + bz + c, where a, b, and c are real.
There exists a complex number w such that the three roots of P (z) are w + 3i, w + 9i, and
2w − 4, where i2 = −1. Find |a + b + c|.
7. (2007 AIME, Problem #8) The polynomial P (x) is cubic. What is the largest value of k
for which the polynomials Q1 (x) = x2 + (k − 29)x − k and Q2 (x) = 2x2 + (2k − 43)x + k are
both factors of P (x)?
8. (2005 AIME-2, Problem #9) For how many positive integers n less than or equal to 1000
is
(sin t + i cos t)n = sin nt + i cos nt
true for all real t?
9. (1988 AIME, Problem #11) Let w1 , w2 , . . . , wn be complex numbers. A line L in the

complex plane is called a mean line for the points w1 , w2 , . . . , wn if L contains points (complex
numbers) z1 , z2 , . . . , zn such that
n
X
(zk − wk ) = 0.
k=1
For the numbers w1 = 32 + 170i, w2 = −7 + 64i, w3 = −9 + 200i, w4 = 1 + 27i, and

w5 = −14 + 43i there is a unique mean line with y-intercept 3. Find the slope of this mean
line.
10. (1998 AIME, Problem #13) If {a1 , a2 , a3 , . . . , an } is a set of real numbers, indexed so that
a1 < a2 < a3 < · · · < an , its complex power sum is defined to be a1 i + a2 i2 + a3 i3 + · · · + an in ,
where i2 = −1. Let Sn be the sum of the complex power sums of all nonempty subsets of
{1, 2, . . . , n}. Given that S8 = −176 − 64i and S9 = p + qi, where p and q are integers, find
|p| + |q|.
z+i
11. (2002 AIME, Problem #12) Let F (z) = for all complex numbers z 6= i, and let
z−i
1
zn = F (zn−1 ) for all positive integers n. Given that z0 = + i and z2002 = a + bi, where a
137
and b are real numbers, find a + b.
7.6. EXERCISES 147
12. (1989 AIME, Problem #14) Given a positive integer n, it can be shown that every complex
number of the form r + si, where r and s are integers, can be uniquely expressed in the base
−n + i using the integers 0, 1, 2, . . . , n2 as “digits”. That is, the equation
r + si = am (−n + i)m + am−1 (−n + i)m−1 + · · · + a1 (−n + i) + a0
is true for a unique choice of nonnegative integer m and digits a0 , a1 , . . . , am chosen from the
set {0, 1, 2, . . . , n2 }, with am 6= 0. We then write
r + si = (am am−1 . . . a1 a0 )−n+i
to denote the base −n + i expansion of r + si. There are only finitely many integers k + 0i
that have four-digit expansions
k = (a3 a2 a1 a0 )−3+i a3 6= 0.
Find the sum of all such k.
13. (1988 AIME, Problem #13) Find a if a and b are integers such that x2 − x − 1 is a factor
of ax17 + bx16 + 1.
14. (2005 AIME-2, Problem #13) Let P (x) be a polynomial with integer coefficients that
satisfies P (17) = 10 and P (24) = 17. Given that the equation P (n) = n + 3 has two distinct
integer solutions n1 and n2 , find the product n1 · n2 .
15. (2011 AIME, Problem #15) For some integer m, the polynomial x3 − 2011x + m has the
three integer roots a, b, and c. Find |a| + |b| + |c|.
Chapter 8
Plane Geometry
“Then, my noble friend, geometry will draw the soul towards truth, and create the
spirit of philosophy, and raise up that which is not unhappily allowed to fall down.”
- Plato, “The Republic”
8.1 Introduction
Geometry is a beautiful branch of mathematics rich in connections to mathematical history. In-

teresting solids, shapes, and geometrical patterns occupy the world we live in, and many of them
are extensions of what can be drawn on a two-dimensional plane. If you have ever visited an art
museum, you have seen mathematics, and specifically geometry, applied in fascinating ways to make
beautiful creations. A central problem in geometry is to study the geometric measures, which are
length, angle measure, area, and volume. Our focus on volume will be postponed until Chapter 9,
which is devoted to spatial geometry (in three-dimensional space). In the present chapter, we will
begin our study of geometry in the two dimensional plane with the others geometric measures.
Every AIME contains a healthy collection of geometry problems that involve such shapes as triangles,
parallelograms, circles, trapezoids, and others. The problems explore relationships between lengths,
angles, and areas in a variety of shapes. In this chapter, we study numerous useful results and
examples involving these shapes. The trigonometric functions we explored in Section 6.3 can often
be used to study these problems as well, as we shall see. The material in Chapter 7 is also relevant
here, since the complex number plane is just that, a plane that offers the prospect of a simultaneous
study of plane geometry and complex numbers to the mutual benefit of both. This has indeed been
done in the AIME competition (see Section 8.4).
The next two sections in this chapter will focus on triangles and circles, respectively. Before we
begin this study, however, let us take a few moments to review some of the basic information about
lines and angles that will be useful. It is expected that readers have been exposed to this material
through a text or course that emphasizes geometrical concepts. Therefore, this review will be brief
149
150 CHAPTER 8. PLANE GEOMETRY
so that we can pass quickly to examples from the AIME competition. Readers needing or wanting
a more comprehensive overview of fundamental concepts in geometry are urged to consult other
sources.
Measurement of angles is important in the study of triangles. One of the most important pieces of
knowledge about triangles, for instance, is the fact that the sum of the measures of all three angles
of a triangle must sum to 180◦ (or π radians). This is recorded in Theorem 8.2.1 below.
Angles are formed when two lines or line segments come together and meet at a single point. Figure
8.1 shows the situation of two lines `1 and `2 intersecting at a point in the plane, thus forming four
angles, numbered 1, 2, 3, and 4.
Figure 8.1: Two intersecting lines in the plane form adjacent pairs of supplementary angles.
The angles spanned by 1 and 2 comprise the entire half-plane on one side of the line `1 , and thus
their measures sum to 180◦ . Such angles, whose measures sum to 180◦ , are called supplementary
angles. In Figure 8.1, angles 2 and 3 are also supplementary, and so are angles 3 and 4.
By contrast, two angles whose measures sum to 90◦ are called complementary angles. Thus, for
instance, a triangle that contains complementary angles must be a right triangle1 , since two angles
have measures summing to 90◦ , and since the sum of all three measures in a triangle is 180◦ , the
third angle must measure 90◦ . (See Figure 8.2.)
Figure 8.2: Angles A and B in this right triangle are complementary.
Referring once more to Figure 8.1, we note that angles 1 and 3 have the same measure (this can be
deduced from the fact that 1 and 2 are supplementary and 2 and 3 are supplementary). Likewise,
angles 2 and 4 have the same measure. These opposite pairs are known as vertical angles, and
these will often be used in AIME problems.
Another important situation occurs when two parallel lines are cut by a third line known as a
transversal. This is illustrated in Figure 8.3. The angles 1 and 2 have the same measure, as
can be easily shown. They are called corresponding angles. Therefore, since both vertical and
1 Recall that a right triangle is any triangle that contains a 90◦ angle.
8.2. TRIANGLES 151
corresponding angles all have the same angle measure, we conclude that angles 2 and 3 have the
same measure. These angles are called alternate interior angles.
Figure 8.3: Two parallel lines cut by a transversal create corresponding angles and alternate interior
angles.
Let us summarize our observations in the following theorem.
Theorem 8.1.1. Consider two angles A and B. Then the measures of angle A and angle B are
the same if any of the following are true: A and B are vertical angles, A and B are corresponding
angles, or A and B are alternate interior angles.
Many of the introductory concepts regarding angle measures that we reviewed above are used fre-
quently in the study of triangles, including Theorem 8.1.1. We devote the next section to triangles.
Before we get there, let us make one final comment in this section about problem solving in geometry.
General Strategy Comment: As the reader has already seen in this introductory section, and
will continue to see throughout this chapter, it is often helpful if not essential that a diagram be
drawn in order to visualize what is going on and relationships between lines, angles, and so on.
8.2 Triangles
We will often denote a triangle with vertices A, B, and C by ∆ABC. As Figure 8.4 illustrates, it is
customary for the vertices of triangles to be denoted by capital letters with the sides opposite them
denoted with the corresponding lowercase letter. In addition, the measure of one of those angles,
denoted respectively by2 m∠A, m∠B, and m∠C, is the numerical value of the angle in question,
and could be measured in either degrees or radians.
Perhaps the first fact that one learns about angle measure in an arbitrary triangle ABC is the
following:
Theorem 8.2.1. In ∆ABC, we have
m∠A + m∠B + m∠C = 180◦ . (8.1)

2 Sometimes we simply write ∠A to denote both the angle and its measure.
Figure 8.4: Labelling a typical triangle.
Theorem 8.2.1 will be used heavily throughout our survey of geometry in the AIME. In fact, it can
be generalized to any convex n-gon as follows:
Theorem 8.2.2. The sum of the angle measures of all angles in a convex3 n-gon is 180(n − 2)
180(n − 2)
degrees. In particular, for a regular convex n-gon, each of its n angles measures degrees.
n
In the second statement of Theorem 8.2.2, recall that a regular convex n-gon is a convex n-gon
in which the angle measure at each vertex is the same. To understand Theorem 8.2.2, one simply
needs to subdivide the interior of the n-gon into triangles and apply Equation (8.1) to each triangle
therein. The reader can verify that, for n ≥ 3, there are n − 2 non-overlapping triangles that can
be drawn in the interior of the triangle. (Figure 8.5 illustrates this for n = 7.) Therefore, the total
measure (in degrees) of all angles within the n-gon is 180(n − 2).
Figure 8.5: We can subdivide a septagon (7-sided convex polygon) into five interior triangles.
Let us now consider a nice example from the AIME involving polygon angle measures. It also uses
information covered in Chapter 5 about arithmetic sequences, and thus provides a nice synthesis of
topics.
Example 8.2.3. (2011 AIME-2, Problem #3) The degree measures of the angles in a convex
18-sided polygon form an increasing arithmetic sequence with integer values. Find the degree measure
of the smallest angle.
Solution: Let the vertices of the convex 18-sided polygon be A1 , A2 , . . . , A18 as in Figure 8.6.
In order to apply Theorem 8.2.2, we draw additional line segments A1 Ai for each i = 3, 4, 5, . . . , 17,
thus forming 16 triangles, ∆A1 A2 A3 , ∆A1 A3 A4 , ∆A1 A4 A5 , . . . , ∆A1 A16 A17 , and ∆A1 A17 A18 , in
the interior of the convex 18-gon. The total sum of all of the measures of all angles of these triangles
is 16 · 180 = 2880, which must be the sum of the degree measures of the angles in the convex 18-gon.
3 A polygon is convex if, given any two points within the interior of the polygon, the line segment joining these
two points is also contained entirely within the interior of the polygon.
8.2. TRIANGLES 153
Figure 8.6: Subdividing the 18-gon in Example 8.2.3 into 16 triangles.
These degree measures can be written in increasing order via (5.1) as
a, a + k, a + 2k, ··· , a + 17k.
Thus, we have
a + (a + k) + (a + 2k) + · · · + (a + 17k) = 2880,
or
18a + (1 + 2 + 3 + · · · + 17)k = 2880.
Using part (1) of Theorem 5.4.2, this becomes
1
18a + (17)(18)(k) = 2880.
2
In other words,
18a + 153k = 2880, which implies that 2a + 17k = 320, (8.2)
where we have divided the first equation in (8.2) through by 9 to obtain the second equation. We
are seeking to determine the degree measure of the smallest angle, A1 , which we have denoted by
the integer a.
Equation (8.2) alone is not enough to specify the value of a. What other information do we have?
The key observation we need in addition to the work we have already done is that, since the 18-sided
polygon is convex, no angle measure can exceed 180◦ . Thus, in particular, the largest angle measure
cannot exceed 180◦ :
a + 17k < 180.
Together with Equation (8.2), this implies that a > 320 − 180 = 140. Equation (8.2) also implies
that k is even (since a is required to be an integer) and positive. Putting this together, we conclude
that k = 2. (The reader should check this.) Thus,
320 − 17 · 2
a= = 143.
2
ANSWER : 143. 2
Example 8.2.3 demonstrates quite plainly the importance of a firm foundation in triangles, because
any convex n-gon (with n ≥ 3) can be cut into a collection of interior triangles by drawing additional
segments on the n-gon. Thus, a mastery of the mathematical tools available in the study of triangles
will often facilitate the study of n-gons for n ≥ 4. We will see this again and again in the examples
and exercises in this chapter.
As we noted in Section 8.1, if two of the angles in ∆ABC are complementary, say ∠A and ∠B
are complementary, then using Equation (8.1), we see that m∠C = 90◦ , which implies that ∆ABC
is a right triangle. Speaking of right triangles, the most celebrated theorem of all involving right
triangles is the Pythagorean Theorem.
Theorem 8.2.4. (Pythagorean Theorem) If ∆ABC is a right triangle with legs of length a and
b and hypotenuse of length c, then a2 + b2 = c2 .
Figure 8.7: The Pythagorean Theorem states that the side lengths of the right triangle shown are
related via a2 + b2 = c2 .
Of course, with the Pythagorean Theorem, we can quickly derive the formula for the distance between
two points (x1 , y1 ) and (x2 , y2 ) in the plane:
Distance Formula: The distance between points P (x1 , y1 ) and Q(x2 , y2 ) in the plane is given by
p
D = (x1 − x2 )2 + (y1 − y2 )2 ≥ 0. (8.3)
Figure 8.8: The Pythagorean Theorem can be used to derive the formula for the distance between
points P (x1 , y1 ) and Q(x2 , y2 ) in the plane.
Not surprisingly, a great many geometry problems in the AIME competition utilize the Pythagorean
Theorem. Let us try a couple of examples of this now.
Example 8.2.5. (2011 AIME-2, Problem #2) On square ABCD, point E lies on side AD and
point F lies on side BC, so that BE = EF = F D = 30. Find the area of square ABCD.
8.2. TRIANGLES 155
Solution: As with many examples we are seeing in this chapter, a good starting point is a figure
illustrating the given information. We provide this in Figure 8.9.
Figure 8.9: Illustration of the problem in Example 8.2.5.
Since the area of square ABCD is simply (AB)2 , let us denote AB by x. Determining the value of
2
x becomes our main task. Since we also have AD = x, then ED = x by noting the positions of
3
the segments DF , F E, and EB in Figure 8.9. Thus,
2 1
AE = AD − ED = x − x = x.
3 3
Now we can apply the Pythagorean Theorem to ∆ABE:
2
2 1
x + x = 302 ,
3
or
10 2
x = 900.
9
Thus, the area of square ABCD is
9
x2 = · 900 = 810.
10
ANSWER : 810. 2
Remark: It is also worth noting the symmetry in Figure 8.9. We might, for instance, notice that
∆ABE has exactly 1/6 of the area of square ABCD, or that ∆BEF has exactly 1/3 of the area of
square ABCD. The interested reader may wish to attempt to discover a solution to this problem
that utlizes this symmetry, although we will not take the space here to do so.
Let us apply the Pythagorean Theorem in a slightly more involved AIME problem that also includes
a touch of number theory (Chapter 4).
Example 8.2.6. (2003 AIME, Problem #7) Point B is on AC with AB = 9 and BC = 21.
Point D is not on AC so that AD = CD, and AD and BD are integers. Let s be the sum of all
possible perimeters of ∆ACD. Find s.
General Strategy Comment: When faced with a point in the plane that lies off of a line that
exists elsewhere in the plane, it is often profitable to draw the segment from the stray point to the
nearest point on the line. This segment is always perpendicular to the line, and in this way, right
triangles can be constructed. We will use this strategy at the beginning of the solution to the present
example.
Solution: Let E denote the point along AC such that DE is perpendicular to AC. See Figure 8.10.
Since AD = CD, we have AE = CE = 15 and BE = 6. Let x = DE. Since D does not lie on AC,
x > 0. Now applying the Pythagorean Theorem (Theorem 8.2.4) to ∆ADE and ∆BDE, we find
that p p
AD = 152 + x2 = 225 + x2
and p p
BD = 62 + x2 = 36 + x2 .
Since AD and BD must both be integers, x must be an integer, 225 + x2 and 36 + x2 must both be
perfect squares, and they differ by 225 − 36 = 189. In other words, we are seeking positive integers
y and z such that y 2 − z 2 = 189 = 33 · 7. Thus, y > z, and (y − z)(y +√z) = 33 · 7. Both y − z and
y + z must be positive, and of course, y − z < y + z. Hence, y − z < 189 ≈ 14, so there are four
possible values for y − z, corresponding to the divisors of 189 that are less than 14:
Case 1: y − z = 1: In this case y + z = 189. Solving for y and z, we have y = 95 and z = 94. That
is, 952 = 225 + x2 and 942 = 36 + x2 . Thus, AD = 95 = CD. The perimeter of the triangle in this
case is 95 + 95 + 30 = 220.
Case 2: y − z = 3: In this case, y + z = 63. Solving for y and z, we have y = 33 and z = 30. That
case is 33 + 33 + 30 = 96.
case is 17 + 17 + 30 = 64.
is, 152 = 225 + x2 and 62 = 36 + x2 . In this case, x = 0, contrary to our observation that x > 0. So
we have no permissible triangle in this case.
8.2. TRIANGLES 157
Adding the perimeters of the triangles obtained in Cases 1,2, and 3, we have s = 220+96+64 = 380.
ANSWER : 380. 2
In the Pythagorean Theorem (Theorem 8.2.4), we have a direct relationship between the side lengths
of a right triangle. This naturally motivates us to ask the question: Are there any useful relationships
between the side lengths of any triangle? It turns out that there are indeed several. Let us begin
with the Triangle Inequality.
Theorem 8.2.7. (The Triangle Inequality) If ∆ABC is any triangle with side lengths a, b, and
c, then c ≤ a + b.
The Triangle Inequality states the rather plain fact that the length of any side of a triangle cannot
exceed the sum of the other two sides. This result is often explained by using vectors, and it is also
an illustration of the adage that “the shortest distance between two points is a straight line.” As we
mentioned, this fact is important to know and intuitively easy. Here is an example.
Example 8.2.8. (2006 AIME-2, Problem #2) The lengths of the sides of a triangle with positive
area are log10 12, log10 75, and log10 n, where n is a positive integer. Find the number of possible
values of n.
Solution: Note that this problem does not assume that the triangle is a right triangle. Therefore,
we cannot use the Pythagorean Theorem. We can only appeal to information related to the side
lengths of general triangles, and for these, the Triangle Inequality is relevant. Let us apply the
Triangle Inequality to the sides of the given triangle in each of the three ways that it can be applied:
log10 12 < log10 75 + log10 n, (8.4)
log10 75 < log10 12 + log10 n, (8.5)
and
log10 n < log10 12 + log10 75. (8.6)
Applying part (1) in Theorem 6.2.2 to each of these equations, we have
12 < 75n, 75 < 12n, n < 12 · 75 = 900.
Since n must be an integer, the latter two strict inequalities prove that 7 ≤ n ≤ 899, which results
in exactly 899 − 7 + 1 = 893 possible values of n. ANSWER : 893. 2
Continuing our focus on arbitrary triangles, let us turn our attention to angle relationships. In the
case of right triangles, in Chapter 6 we related the six standard trigonometric functions discussed
in Section 6.3 to the side lengths of a right triangle. For arbitrary triangles, two of the most useful
rules to remember are the Law of Sines and the Law of Cosines. We will not pause to prove these
results in this book, but both rules provide important relationships between the lengths of the sides
of a triangle and the angles in the triangle.
Proposition 8.2.9. (Law of Sines) Consider ∆ABC with the side opposite angle A of length a,
the side opposite angle B of length b, and the side opposite angle C of length c, as shown in Figure
8.11. Then
a b c
= = = 2R,
sin A sin B sin C
where R is the radius of the triangle’s circumcircle4 .
Figure 8.11: The Law of Sines and the Law of Cosines provide relationships between the side lengths
a, b, and c, and the measures of angles A, B, and C in any triangle.
Proposition 8.2.10. (Law of Cosines) Consider ∆ABC with the side opposite angle A of length
a, the side opposite angle B of length b, and the side opposite angle C of length c, as shown in Figure
8.11. Then
a2 = b2 + c2 − 2bc cos A,
b2 = a2 + c2 − 2ac cos B,
c2 = a2 + b2 − 2ab cos C.
If ∆ABC is a right triangle, say with right angle at A, then the first equation in the Law of Cosines
reduces to the Pythagorean Theorem: a2 = b2 + c2 .
Let us examine an AIME problem that can be solved using the Law of Cosines. One interesting
feature of this next example is that only a square and a circle are referenced in the problem, and
yet, we will see a heavy dependence on triangles in the solution.
Example 8.2.11. (2001 AIME-2, Problem #6) Square ABCD is inscribed in a circle. Square
EF GH has vertices E and F on CD and vertices G and H on the circle. The ratio of the area of
square EF GH to the area of square ABCD can be expressed as m/n where m and n are relatively
prime positive integers and m < n. Find 10n + m.
Remark: Without loss of generality, we may assume that square ABCD has side length 1. We can
let O denote the center of the circle, as shown in Figure 8.12.
What other point(s) on the figure are important to identify? Perhaps the midpoint of EF , or perhaps
the midpoint of GH? The most natural way to proceed from here depends on which way the solver
addresses this question. In the interest of completeness, we will present both options. The first one
will lead us to a natural use of the Law of Cosines (Proposition 8.2.10).
Solution #1: Let P denote the midpoint of CD, and form ∆OP H. This is depicted in Figure
8.13.
4 The circumcircle of a triangle is the unique circle in the plane of the triangle that passes through all of its vertices.
8.2. TRIANGLES 159
Figure 8.13: We add the point P , the midpoint of CD, to the previous figure, and consider ∆OP H.
By symmetry, observe that EP = F P . Let x = EP . Then square EF GH has side length 2x.
Therefore, the area of square EF GH is (2x)2 = 4x2 . Thus, we must find x. Let θ = m∠OP H
(measured in degrees) and consider ∆OP H. By the Law of Cosines, we have
OH 2 = OP 2 + P H 2 − 2(OP )(P H) cos θ.
Now OH is the radius of the circle, which has length

s
2 2
1 1 1
OH = + =√ .
2 2 2
Moreover,
1 p √ √
OP = and PH = x2 + (2x)2 = 5x2 = 5x.
2
Plugging these values into the Law of Cosines expression, we have
1 1 1 √
= + 5x2 − 2 · · 5 · x · cos θ
2 4 2
1 √
= + 5x2 − 5 · x · sin(90 − θ) (where we have applied Equation (6.34))
4
1 √
= + 5x2 + 5 · x · sin(θ − 90) (where we have applied Equation (6.35))
4
1 √ 2
= + 5x2 +
5·x· √ (where we have applied trigonometry to ∆EP H)
4 5
1
= 5x2 + 2x + .
4
That is,
1
5x2 + 2x − = 0. (8.7)
4
Using the quadratic formula, we have
−2 ± 3
x= ,
10
and since x must be positive (the length of any segment must be positive), we obtain
1
x= .
10
The area of square EF GH is therefore
4 1
(2x)2 = 4x2 = = .
100 25
Since the area of square ABCD is 1, the ratio of the area of square EF GH to the area of square
1
ABCD is . Thus, m = 1 and n = 25, and the answer is 10n + m = 251. ANSWER : 251. 2
25
Solution #2: Let Q denote the midpoint of GH, as shown in Figure 8.14.
Figure 8.14: We add the point Q, the midpoint of GH, to the figure, and consider ∆OGQ.
Then we have GQ = HQ. Let x = GQ. Then we have right triangle OGQ and can apply the
Pythagorean Theorem:
OG2 = OQ2 + GQ2 . (8.8)
1
Now OG is the radius of the circle, so as in Solution #1, we have OG = √ . We also have
2
1
OQ = + 2x and GQ = x.
2
Substituting into Equation (8.8), we obtain
2
1 1
= + 2x + x2 .
2 2
Once this is expanded and re-arranged algebraically, one arrives once more at Equation (8.7). From
this point, the remainder of the solution is the same as the one presented in Solution #1 above. 2
8.2. TRIANGLES 161
Remark: It may seem to the reader that Solution #2 is preferred over Solution #1, in that it
relies only on the more familiar Pythagorean Theorem, rather than the Law of Cosines. However,
it is always good to have flexible knowledge for the AIME competition. For instance, Solution
#2 depends on identifying the point Q, which may not be as obvious as the point P identified in
Solution #1. In summary, this is a nice problem that invites multiple approaches, and some solvers
may prefer Solution #1, while others may prefer Solution #2.
Example 8.2.12. (1989 AIME, Problem #10) Let a, b, c be the three sides of a triangle, and
let α, β, γ, respectively, be the angles opposite them. If a2 + b2 = 1989c2 , find
cot γ
.
cot α + cot β
Solution: We are confronted with a triangle with three known sides a, b, and c. We are given a
relationship between a2 , b2 , and c2 . The reader should therefore naturally pursue a strategy that
involves the Law of Cosines. Specifically, the Law of Cosines gives
c2 = a2 + b2 − 2ab cos γ.
Substituting 1989c2 = a2 + b2 and rearranging, we have 2ab cos γ = 1988c2 , or
ab cos γ = 994c2 . (8.9)
Using the Law of Sines, we have

a sin γ b sin γ
c= = . (8.10)
sin α sin β
If we replace c2 on the right hand side of Equation (8.9) with the product of the two expressions in
(8.10) for c, we will obtain factors of a and b that can be canceled from (8.9). Therefore,

a sin γ b sin γ
ab cos γ = 994c2 = 994 .
sin α sin β
Cancelling a and b from each side and rearranging the result, we have
sin α sin β cos γ = 994 sin2 γ.
From here, in keeping with one of the strategies we highlighted in Chapter 6 for handling trigono-
metric functions, we compute the desired expression
cot γ
cot α + cot β
by converting all quantities into terms of the sine and cosine functions and applying trigonometric
identities from Chapter 6 as needed:
cot γ cos γ
=
cot α + cot β sin γ cos α cos β
sin α + sin β
sin α sin β cos γ
=
sin γ(cos α sin β + sin α cos β)
994 sin2 γ
=
sin γ sin(α + β)
994 sin γ
=
sin(α + β)
994 sin γ
=
sin(180 − γ)
= 994,
since α + β + γ = 180 and sin γ = sin(180 − γ). Hence, the answer is 994. ANSWER : 994. 2
8.2.1 Congruent and Similar Triangles
Many AIME problems about triangles require the use of congruent or similar triangles. Two triangles
∆ABC and ∆A0 B 0 C 0 are congruent if there is a correspondence between the vertices of the two
triangles in such a way that the three sets of corresponding angles are all the same measure (such
angles are called congruent angles) and the three sets of corresponding sides are all the same length
(such segments are called congruent segments). Congruent triangles (or, in general, congruent
shapes) essentially look the same. One way to recognize this is to note that if we allow the triangles
to move as rigid objects (no bending or stretching), it is possible to place a triangle directly on top
of another triangle to which it is congruent, and it is a perfect match.
To test whether or not two triangles are congruent, it is not actually necessary to verify that all pairs
of sides and all pairs of angles are congruent. Indeed, we usually do not have all of this information,
but if we could conclude that two triangles are congruent from less information, then we could
conclude the additional information that we do not yet have. Any of the following properties is
sufficient to conclude that two triangles are congruent.
Theorem 8.2.13. Two triangles ∆ABC and ∆A0 B 0 C 0 are congruent if any of the following infor-
mation is known:
SSS congruence: All pairs of sides in the two triangles are congruent;
SAS congruence: Two pairs of sides and the angle between them5 in both triangles are congruent;
ASA congruence: Two pairs of angles and the side between them in both triangles are congruent.
Next, let us remind ourselves of the notion of similar triangles, another powerful concept in the
geometry problems on the AIME competition. Two triangles ∆ABC and ∆A0 B 0 C 0 are similar,
5 This is an important point; the congruent pair of angles must be between the two pairs of sides known to be
congruent.
8.2. TRIANGLES 163
sometimes written ∆ABC ∼ ∆A0 B 0 C 0 , if there is a corresponding between the vertices of the two
triangles, say A ↔ A0 , B ↔ B 0 , and C ↔ C 0 , in such a way that the three sets of corresponding
angles are all the same measure and the lengths of all of the corresponding sides of the two triangles
are proportional; that is,
AB AC BC
= 0 0 = 0 0. (8.11)
A0 B 0 AC BC
Proportionality represents a weakening of the condition given for congruent triangles, where the
corresponding sides of the two triangles were required to have the same length. Of course, if one
knows that m∠A = m∠A0 and m∠B = m∠B 0 , then it follows that m∠C = m∠C 0 also, and
therefore, the triangles are similar.
Let us consider a couple of examples.
Example 8.2.14. (1998 AIME, Problem #6) Let ABCD be a parallelogram. Extend DA
through A to a point P , and let P C meet AB at Q and DB at R. Given that P Q = 735 and
QR = 112, find RC.
Solution: This may be a case of a problem with too much information. There are so many similar
triangles in the problem that it can be hard to decide which ones to use. The reader may wish to
identify sets of similar triangles in Figure 8.15.
With the multitude of similar triangles available, there may be many avenues to the answer here.
Therefore, it is legitimate to ask: How do we decide which triangles to focus on? To address this,
note that since RC is the quantity we wish to find, we should choose triangles that involve RC.
Thus, for instance, we know that
∆RCB ∼ ∆RP D and ∆RCD ∼ ∆RQB. (8.12)
(The reader should verify this by using vertical angles and alternate interior angles arising from the
RB
parallel sides BC and AD, and the parallel sides AB and CD). We observe that the ratio can
RD
be exploited in both sets of similar triangles identified in Equation (8.12). In each of the similarity
RB
relations, we can therefore relate to RC. We have
RD
RC RB RQ
= = .
RP RD RC
Therefore,
RC 2 = RP · RQ = (735 + 112) · 112 = 847 · 112 = 24 · 72 · 112 ,
from which we deduce that
RC = 22 · 7 · 11 = 308.
ANSWER : 308. 2
General Strategy Comment: In Example 8.2.14 above, the problem begins with a simple paral-
lelogram ABCD. From there, additional points and lines are drawn onto the figure, as instructed in
the problem. Sometimes, however, it is fruitful to add additional points and lines to a figure, even
when not prompted to do so. Such is the case with our next example, where we can extend the
figure described by the problem in order to create similar triangles.
Example 8.2.15. (2001 AIME-2, Problem #13) In quadrilateral ABCD, ∠BAD ∼ = ∠ADC
and ∠ABD ∼ = ∠BCD, AB = 8, BD = 10, and BC = 6. The length CD may be written in the form
m/n, where m and n are relatively prime positive integers. Find m + n.
Solution: Extend segments AB and CD to an intersection point P , as shown in Figure 8.16.
Notice that ∆P AD is isosceles (the sides P A and P D are congruent), since m∠P AD = m∠P DA.
Recall that a triangle is called isosceles if (at least) two of its sides are congruent. Let us try to
identify a pair of similar triangles in Figure 8.16. Notice that ∠P is shared by several triangles,
including ∆P BC and ∆P DB. In fact, since ∠ABD ∼ = ∠BCD, the corresponding angles that are
supplementary to each of these are also congruent: ∠P BD ∼ = ∠P CB. Thus, the triangles ∆P BC
and ∆P DB are in fact similar, since they have two (and hence, three) congruent angles. Thus, the
ratios of corresponding sides must be the same:
BC PC PB
= = . (8.13)
BD PB PD
Since
BC 6 3
= = ,
BD 10 5
we can use (8.13) to obtain
3 3
PC = PB and PB = P D.
5 5
8.2. TRIANGLES 165
However, note that we can relate P D to P B in a different way:
PD = PA
= P B + BA
= P B + 8.
Thus,
3 3
PB = P D = (P B + 8),
5 5
which implies that P B = 12. Thus,
3 3 36
P A = P D = 20 and PC = P B = (12) = .
5 5 5
Thus,
36 64
CD = P D − P C = 20 − = .
5 5
Therefore, m = 64 and n = 5. Hence, m + n = 64 + 5 = 69 = 069. ANSWER : 069. 2
8.2.2 Area of a Triangle
There are several useful formulas giving the area of a triangle ∆ABC that arise frequently on AIME
problems. Let us enumerate a few of them here. (Note: The area of triangle ABC is sometimes
denoted by [ABC].)
• If ∆ABC has a base of length b and height h (where h is the length of the altitude extending
perpendicular to the base through the opposite vertex), then the area of triangle ABC is
1
[ABC] = bh. (8.14)
2
Figure 8.17: Determining the area of a triangle by using the length of its base and height.
a+b+c
• (Heron’s Formula) If ∆ABC has side lengths a, b, and c, then if s = denotes half
2
the perimeter of ∆ABC (sometimes dubbed the semiperimeter of the triangle), then the
area of ∆ABC is given by the formula
p
[ABC] = s(s − a)(s − b)(s − c). (8.15)
Figure 8.18: Determining the area of a triangle by using Heron’s Formula.
In particular, if the coordinates of the vertices A, B, and C are known, then Heron’s Formula
can be used to find the area of ∆ABC, since the distance formula can be used to determine
a, b, and c, if the coordinates of A, B, and C are known.
• (Area formula if one vertex rests at (0, 0) in the xy-plane) If ∆ABC has vertices (0, 0),
(a, b), and (c, d) then the area of ∆ABC is given by
1
[ABC] = |ad − bc|. (8.16)
2
This formula is derived using elementary theory of vectors.
Figure 8.19: Determining the area of a triangle positioned with coordinates in the xy-plane and one
vertex at the origin.
• Referring to ∆ABC in Figure 8.18, if we view b as the base length, then we can express the
height of the triangle as c sin A. Therefore, Equation (8.14) implies that
1
[ABC] = bc sin A.
2
Combining this formula with the equation
a
= 2R
sin A
in Proposition 8.2.9, we deduce that
abc
[ABC] = .
4R
An especially important class of triangles consists of the equilateral triangles. Of course, these are
triangles whose sides are all the same length, say a. In particular, every equilateral triangle is also
8.2. TRIANGLES 167
3
an isosceles triangle. Referring to Heron’s Formula, we see in this case that s = a, and thus, the
2
formula for the area of an equilateral triangle of side length a is
s
3 √
3 1 3 2
[ABC] = a a = a . (8.17)
2 2 4
Of course, since all angles in equilateral triangle ABC measure 60◦ , we could use basic trigonometry,
1
coupled with the formula [ABC] = bh above, to determine the same expression (8.17).
2
Let us next use some of these formulas to solve a couple of problems. The first one is actually from
the 2010 AMC exam and already provides a nice challenge.
Example 8.2.16. (AMC 12 A 2010, Problem #17) Equiangular hexagon ABCDEF has side
lengths AB = CD = EF = 1 and BC = DE = F A = r. The area of ∆ACE is 70% of the area of
the hexagon. What is the sum of all possible values of r?
Solution: According to Theorem 8.2.2 with n = 6, each angle of hexagon ABCDEF measure
180 · 4
= 120 degrees. It will be especially useful to draw a figure of the hexagon:
6
Our strategy will be to find the areas of both ∆ACE and hexagon ABCDEF in terms of r and
then relate them to one another from the given information.
Observe that ∆ABC, ∆CDE, and ∆EF A are all congruent, for instance, by using the ASA property.
Thus, AC = CE = EA. We can express this common length in terms of r by using the Law of
Cosines:
AC 2 = 1 + r2 − 2r cos(120◦ ) = 1 + r2 + r.
Thus, p
AC = CE = EA = 1 + r + r2 .
Thus, according to Equation (8.17), the area of (equilateral) triangle ACE is
√
3
[ACE] = (1 + r + r2 ).
4
Next, we will compute the area of the hexagon ABCDEF . Denoting this by [ABCDEF ], note that
[ABCDEF ] = [ACE] + [ABC] + [CDE] + [EF A]
√
3
= (1 + r + r2 ) + 3[ABC],
4
where we have used the fact that triangles ∆ABC, ∆CDE, and ∆EF A are all congruent in the
second step. Therefore, it is sufficient to find the area of ∆ABC. There are several ways to do this
that the reader may wish to explore. For instance, since all three side lengths of ∆ABC are known
(in terms of r), Heron’s Formula could be used. However, it is easier in this case to extend AB
outside the hexagon to a point P such that ∆AP C is a right triangle. (See Figure 8.21.)
Figure 8.21: To determine the area of ∆ABC, we extend AB outside of the hexagon in Figure 8.20
to a point P such that ∆AP C is a right triangle.
Notice that ∆BCP is a 30-60-90

√ triangle, and since the hypotenuse BC has length r, we conclude
1 3
that BP = r and CP = r. Thus, we can apply (8.14) to the triangles ∆AP C and ∆BP C to
2 2
obtain
[ABC] = [AP C] − [BP C]
√ ! √ ! √
1 r 3 1 r 3 3
= 1+ r − r = r.
2 2 2 2 2 2 4
Therefore, we have
√ √
3 2 3
[ABCDEF ] = (1 + r + r ) + 3[ABC] = (1 + r + r2 + 3r).
4 4
We are given that √

3
7 [ACE] (1 + r + r2 ) 1 + r + r2
= = √4 = .
10 [ABCDEF ] 3 2 1 + 4r + r2
4 (1 + 4r + r )
Next, we cross-multiply to obtain
10(1 + r + r2 ) = 7(1 + 4r + r2 ).
We rearrange and simplify this equation to
r2 − 6r + 1 = 0,
whose two roots are √

r = 3 ± 2 2.
Therefore, the sum of the possible values of r is 6. 2
Example 8.2.17. (2005 AIME, Problem #10) Triangle ABC lies in the Cartesian plane and
has area 70. The coordinates of B and C are (12, 19) and (23, 20), respectively, and the coordinates
of A are (p, q). The line containing the median to side BC has slope −5. Find the largest possible
value of p + q.
8.2. TRIANGLES 169
Solution: The line containing

the median to side BC must pass through the midpoint of the
35 39
segment BC, which is , . Figure 8.22 illustrates the problem.
2 2
Since we know the slope of this line and a point contained in the line, we can use the point-slope
form for the equation of this line:

39 35
y− = −5 x − .
2 2
That is,
y = −5x + 107.
Hence, the coordinates of A are (p, q) = (p, −5p + 107). Hence, we have determined the coordinates
of the vertices A, B, and C of ∆ABC, where A is given in terms of the unknown parameter p. Next
we wish to utilize the fact that the area of this triangle is 70. To do so, let us “translate” ABC so
that vertex B is moved to the origin. We must subtract 12 from the x-coordinates of all vertices
and subtract 19 from the y-coordinates of all points. Thus, we obtain
A0 = (p − 12, −5p + 88), B 0 = (0, 0), C 0 = (11, 1).
Using Equation (8.16), we find that the area is

1 1
[A0 B 0 C 0 ] = p − 12 − 11(−5p + 88) = 56p − 980.

2 2
Since this must be 70 (the area of the triangle remained unchanged when we translate it), we have

1
56p − 980 = 70.
2
That is, |56p − 980| = 140. Therefore, 56p − 980 = ±140, so that 56p = 1120 or 56p = 840. That is,
p = 20 or p = 15. Since p + q = p + (−5p + 107) = 107 − 4p, this expression is larger when p = 15:
p + q = 107 − 4 · 15 = 107 − 60 = 47 = 047.
ANSWER : 047. 2
Finally in this section, let us relate the area of triangles that we have just been discussing back to
similarity of triangles. Recall that if two triangles ∆ABC and ∆A0 B 0 C 0 are similar, then according
to Equation (8.11), the ratios of corresponding sides must be equal. The next result, which follows
easily from this, expresses the ratio of the areas of these triangles.
Theorem 8.2.18. Suppose that two triangles ∆ABC and ∆A0 B 0 C 0 are similar, say with side length
ratio
AB AC BC
= 0 0 = 0 0 = r,
A0 B 0 AC BC
then the ratio of the areas of the two triangles is
[ABC]
= r2 .
[A0 B 0 C 0 ]
Theorem 8.2.18 actually generalizes to any two similar polygons in the plane. We will have a nice
application of Theorem 8.2.18 in the next section in Example 8.3.5.
8.3 Circles
Any circle can be specified by giving its center (a, b) in the xy-plane and its radius r. The equation
of such a circle is (x − a)2 + (y − b)2 = r2 . In many applications, the physical center of the circle
is immaterial, and the only crucial piece of information regarding the circle is its radius r. For
instance, both the area and the circumference of a circle depend only upon knowledge of the radius.
Proposition 8.3.1. Let C be a circle of radius r. The area of C is πr2 and the circumference of C
is 2πr.
With this limited amount of information about circles, we can already solve the following problem
from the 2008 AIME competition.
Example 8.3.2. (2008 AIME, Problem #5) A right circular cone6 has base radius r and height
h. The cone lies on its side on a flat table. As the cone rolls on the surface of the table without
slipping, the point where the cone’s base meets the table traces a circular arc centered at the point
where the vertex touches the table. The cone first returns to its original position
√ on the table after
making 17 complete rotations. The value of h/r can be written in the form m n, where m and n
are positive integers and n is not divisible by the square of any prime. Find m + n.
Solution: Consider the circular arc traced by the cone on the table. Its radius is the slant height
of the cone.
√
Since the cone has radius r and 2 2
√ height h, its slant height is r + h . Therefore, the circular arc
2 2
traced by the cone has radius r + h . (See Figure 8.23.) On the other hand, the base of the cone
has circumference 2πr. Since the base of the cone makes 17 complete revolutions before making the
6 More information about cones can be found in the next chapter.
8.3. CIRCLES 171
circular arc into a complete circle, the circumference of the circular arc is 17(2πr) = 34πr. Therefore,
we have p
34πr = 2π r2 + h2 .
√ 2 2 2 2 2
That is, 17r√= r2 + h2 . Squaring both
√ √ sides, we have 289r = r + h or h = 288r . Hence, h =
288r = 12 2r. Therefore, h/r = 12 2. Hence, m + n = 12 + 2 = 14 = 014. ANSWER : 014. 2
The joint consideration of circles and triangles in AIME geometry problems is commonplace. Here
is another relatively easy example.
Example 8.3.3. (2004 AIME-2, Problem #1) A chord of a circle is perpendicular to a radius
at the midpoint of the radius. The ratio of the area of the larger of the two
√ regions into which the
aπ + b c
chord divides the circle to the smaller can be expressed in the form √ , where a, b, c, d, e, and
dπ − e f
f are positive integers, a and e are relatively prime, and neither c nor f is divisible by the square of
any prime. Find the remainder when the product abcdef is divided by 1000.
Solution: The situation is pictured in Figure 8.24, where we have assumed the circle has radius r.

r
Triangle OXY is composed of two identical 30 − 60 − 90 right triangles with legs of length and
√ 2
3r
. Therefore, the area of ∆OXY is
2
√
1 r √ 3 2
[OXY ] = ( 3r) = r .
2 2 4
1
Since ∠XOY = 120◦ = (360◦ ), the area of the sector the circle determined by segments OX,
3
OY , and the arc from X to Y on the circle is one-third of the area πr2 of the full circle, which is
1 2
πr . Now we can determine the area of the smaller region determined by the chord in question by
3
subtracting the area of ∆OXY from the area of this sector:
√
1 2 3 2
Area of smaller region = πr − r .
3 4
Since the larger region must occupy the remaining space in the full circle, we have
√ ! √
2 1 2 3 2 2 2 3 2
Area of larger region = πr − πr − r = πr + r .
3 4 3 4
Thus, the ratio of the larger region to the smaller region is
2 2
√
3 2
√
3 πr +
√4
r 8π + 3 3
Ratio = = √ ,
1
3 πr
2 − 3 2
4 r
4π − 3 3
where we multiplied all terms through by 12 in the last step in order to eliminate fractions. This is
done to render the answer into the form given in the question. Hence, a = 8, b = c = e = f = 3,
and d = 4. Therefore,
abcdef = 8 · 34 · 4 = 2592,
and the answer is 592. ANSWER : 592. 2
An important fact about angles within a circle is the following.
Proposition 8.3.4. Let C be a circle containing points A, B, C, and D, as shown in Figure 8.25.
Then m∠CAD = m∠CBD.
Figure 8.25: Illustration of Proposition 8.3.4.
The next example uses Proposition 8.3.4 and also illustrates a sophisticated use of both circles and
triangles to achieve a solution.
8.3. CIRCLES 173
Example 8.3.5. (2006 AIME-2, Problem #12) Equilateral ∆ABC is inscribed in a circle of
radius 2. Extend AB through B to a point D so that AD = 13, and extend AC through C to a
point E so that AE = 11. Through D, draw a line `1 parallel to AE, and through E, draw a line
`2 parallel to AD. Let F be the intersection of `1 and `2 . Let G be the point on the circle that is
collinear with A and F and distinct from A. Given that the area of ∆CBG can be expressed in
√
the form p q/r, where p, q, and r are positive integers, p and r are relatively prime, and q is not
divisible by the square of any prime, find p + q + r.
Solution: This problem requires a somewhat elaborate drawing, so one must be careful to draw it
correctly (see Figure 8.26).
Before proceeding further, let us briefly supply some guiding remarks about the overall strategy here.
With a well-drawn figure, one may suspect that ∆GBC is similar to ∆DF A. This does indeed turn
out to be the case, and we will explain this below. Therefore, if we can find the common ratio of
the side lengths of the two triangles ∆GBC and ∆DF A, then we will be able use Theorem 8.2.18
to find [GBC]. Readers seeking ownership of the solution are encouraged to stop here and try to
carry out this strategy on their own, before reading the lines below.
To see that ∆GBC ∼ ∆DF A, we will demonstrate that (at least) two corresponding pairs of angles
in these two triangles are congruent. Since ∠BAG and ∠BCG both subtend the arc BG on the
circle, they must have equal measure by Proposition 8.3.4. Likewise, since ∠CAG and ∠CBG both
subtend the arc GC on the circle, they must also have equal measure. However, sincee AE is
parallel to DF , we see that ∠GAC and ∠DF A are alternate interior angles. Thus, by Theorem
8.1.1, m∠GAC = m∠DF A. Therefore, since we have shown that two pairs of corresponding angles
in ∆GBC and ∆DF A are congruent, all three pairs must be, by Theorem 8.2.1. Hence, ∆GBC
and ∆DF A are similar.
Next, let us determine the lengths of the longest sides of these two triangles so that we can establish
the common ratio r of side lengths in these two triangles. From Figure 8.26, it is clear that the
longest side of ∆DF A is AF . Using the Law of Cosines (and the fact that7 m∠F DA = 120◦ ), we
have
1
AF 2 = AD2 + DF 2 − 2(AD)(DF ) cos 120◦ = 132 + 112 − 2 · 13 · 11 · (− ) = 169 + 121 + 143 = 433.
2
The longest side of ∆GBC is BC. If O denotes the center of the circle, then since ∆ABC is
equilateral, m∠BOC = 120◦ . Now either by appealing once more to the Law√ of Cosines, or by
creating 30-60-90 triangles within ∆OBC, the reader can verify that BC = 12. Hence, the ratio
of side lengths in ∆GBC to side lengths in ∆DF A is
√
BC 12
=√ ,
AF 433
and hence, Theorem 8.2.18 gives
√ !2
[CBG] 12 12
= √ = .
[DF A] 433 433
To compute [DF A], we can view the base of ∆DF A as side DF , which has length DF = 11. Then
the height of the triangle is
13 √
13 sin(∠EF D) = 13 sin 60◦ = 3.
2
Hence,
1 13 √ 143 √
[DF A] = · 11 · 3= 3.
2 2 4
We deduce that the area of ∆GBC is
143 √ 12 429 √
[GBC] = 3· = 3.
4 433 433
Thus, we have p = 429, q = 3, and r = 433. Therefore, p + q + r = 429 + 3 + 433 = 865.
ANSWER : 865. 2
In Example 8.3.5, a triangle is inscribed in a circle. Of course, it is not uncommon to reverse the
roles of the triangle and the circle. One can inscribe a circle inside of a triangle. Our final example
in this section does this. The solution will utilize an important result that arises in multiple AIME
problems.
Theorem 8.3.6. (Two Tangent Theorem) The distances from a vertex of a triangle and the two
points of tangency of an inscribed circle are equal. In particular, in Figure 8.27, we have AP = AQ.
We use Theorem 8.3.6 to solve our final example in this section.

7 This follows from the fact that AEF D is a parallelogram (whose angles sum to 360◦ ) with m∠A = m∠F = 60◦
and m∠D = m∠E.

8.3. CIRCLES 175
Figure 8.27: A circle is inscribed in ∆ABC. The distances from vertex A to points of tangency P
and Q are equal according to the Two Tangent Theorem: AP = AQ.
Example 8.3.7. (1999 AIME, Problem #12) The inscribed circle of triangle ABC is tangent
to AB at P , and its radius is 21. Given that AP = 23 and P B = 27, find the perimeter of the
triangle.
Solution: Let O denote the center of the inscribed circle, let Q be the point where the circle is
tangent to AC, and let R be the point where the circle is tangent to BC, as shown in Figure 8.28.
Let x = CQ. From the Two Tangent Theorem, we note that
AP = AQ = 23, BP = BR = 27, CQ = CR = x.
Therefore, the total perimeter of the triangle is
2s = 2 · (23 + 27 + x) = 100 + 2x.
Therefore, the semiperimeter of ∆ABC is
s = 50 + x.
We will compute [ABC], the area of ∆ABC, in two ways and equate the results. First, from Heron’s
formula (8.15), we have
p p
[ABC] = s(s − x − 23)(s − x − 27)(s − 23 − 27) = (50 + x)(27)(23)(x). (8.18)
Next, we observe that

[ABC] = [AOC] + [BOC] + [AOB]. (8.19)
The triangles ∆AOC, ∆BOC, and ∆AOB each have a height of 21, the radius of the circle, and a
base length that equals the length of one side of ∆ABC. Thus, using Equation (8.14) three times,
we have
1 1 1
[AOC] = (21)(23 + x), [BOC] = (21)(27 + x), [AOC] = (21)(23 + 27).
2 2 2
Thus, Equation (8.19) becomes
1
[ABC] = (21) [(23 + x) + (27 + x) + (23 + 27)] = 21 · (50 + x).
2
Matching this formula with Equation (8.18), we obtain
p
(50 + x)(27)(23)x = 21 · (50 + x).
Squaring both sides, we have
621x(50 + x) = 441(50 + x)2 .
From here, a short calculation yields
245
x= .
2
Thus, the total perimeter of the triangle is
2s = 100 + 2x = 100 + 245 = 345.
ANSWER: 345. 2
8.4 Geometrical Concepts in the Complex Plane
In Section 7.3, we studied complex numbers from a geometric perspective. The idea was that each
complex number z = a + bi √ can be identified with the point (a, b) in the Cartesian plane. By
measuring the distance r = a2 + b2 from (a, b) to the origin, as well as the angle θ that the vector
(a, b) makes with respect to the positive x-axis, we can rewrite z in polar form: z = reiθ . In the
AIME problems we explored in Section 7.3, we did not explicitly need many facts from geometry
beyond this basic set-up. However, some AIME problems do require facts that we have looked at
earlier here in this chapter. In this section, we will look at a couple of illustrations of this. Both of
these examples are somewhat intricate and require careful attention.
Example 8.4.1. (1992 AIME, Problem #10) Consider the region A in the complex plane that
z 40
consists of all points z such that both and have real and imaginary parts between 0 and 1,
40 z
inclusive. What is the integer that is nearest the area of A?
8.4. GEOMETRICAL CONCEPTS IN THE COMPLEX PLANE 177
Solution: Write z = a + bi, where a and b are real numbers. Then
z a b 40 40 40(a + bi) 40a 40b

= + i and = = = 2 + 2 i.
40 40 40 z a − bi a2 + b2 a + b2 a + b2
z 40
Therefore, in order for both and to have real and imaginary parts between 0 and 1, inclusive,
40 z
we must have
a
0≤ ≤ 1, (8.20)
40
b
0≤ ≤ 1, (8.21)
40
40a
0≤ 2 ≤ 1, (8.22)
a + b2
40b
0≤ ≤ 1. (8.23)
a2 + b2
In terms of the complex plane, the variable a represents a value with respect to the real axis, while
the variable b represents a value with respect to the imaginary axis. We must determine the area
of the region A consisting of points (a, b) that satisfy all four of the inequalities above. Equations
(8.20) and (8.21) dictate that the region A is contained inside of a square in the complex plane with
vertices P (0, 0), Q(40, 0), R(40, 40), and S(0, 40). This square has area 402 = 1600, and from it
we must subtract the area within this square consisting of pairs (a, b) that fail to satisfy Equations
(8.22) and (8.23). Figure 8.29 illustrates the situation.
We can rearrange Equation (8.22) as follows:
40a
0≤ ≤ 1 =⇒ 0 ≤ 40a ≤ a2 + b2 =⇒ a2 − 40a + b2 ≥ 0 =⇒ (a − 20)2 + b2 ≥ 202 .
a2+ b2
This latter equation defines the region outside of the circle centered at (20, 0) and with radius 20.
Therefore, the area of the portion of this circle that lies within the square P QRS must be deducted
from the area of the square. Exactly half of the area of the circle of radius 20 lies with the square,
so we must deduct 21 π(20)2 = 200π.
Likewise, Equation (8.23) defines the region lying outside of the circle centered at (0, 20) and with
radius 20. We can deduct another 200π from the area of the square to account for the portion of
this second circle that lies in the square P QRS. However, in so doing, we will have deducted the
area of the region B common to both circles twice.
We must therefore add back in the area of intersection between the two circles. Note that the two
circles intersect at (0, 0) and (20, 20), which can be seen in Figure 8.29 or by setting the equations
(a − 20)2 + b2 = 400 and a2 + (b − 20)2 = 400
equal to each other. The common region in question is bisected by the line y = x. Hence, we can
find the area of the common region above the line y = x and double it. The area of half of the
region B (common to both circles and lying above the line y = x) is the difference between the area
of one-fourth of the circle centered at (20, 0) and the right triangle with vertices (0, 0), (20, 0), and
(20, 20). That is, the
1 1
Area of one-half the region B = π(20)2 − (20)(20)
4 2
= 100π − 200.
Doubling this, we see that
Area of B = 200π − 400
must be added back in after deducting the areas of the two semicircles in order to compute the area
of A. Therefore, we conclude that
Area of A = 1600 − 2 · 200π + 200π − 400
= 1200 − 200π
= 200(6 − π).
Using the approximation π ≈ 3.14, we see that 200(6 − π) ≈ 200(2.86) = 572. ANSWER : 572. 2
Example 8.4.2. (2008 AIME-2, Problem #13) A regular hexagon with center at the origin
in the complex plane has opposite pairs of sides one unit apart. One pair of sides is parallel to the
imaginary axis. Let R be the region outside the hexagon, and let S = { z1 |z ∈ R}. Then the area of
√
S has the form aπ + b, where a and b are positive integers. Find a + b.
Solution: The first step is to draw the hexagon, already a non-trivial exercise. Figure 8.30 illustrates
the hexagon.
Since opposite sides are one unit apart and the hexagon has two parallel sides parallel to the imag-
inary axis, we know that the points A and B have x-coordinate 12 , and points D and E have
x-coordinate − 12 . Note also that all six triangles
∆OAB, ∆OBC, ∆OCD, ∆ODE, ∆OEF, ∆OF A
are equilateral. To see why, note that the regularity of the hexagon implies that each of the angles
at O in these triangles measures 60◦ . Thus, using alternate interior angles, we see for example that
m∠OAB = 60◦ . Likewise, all angles in all six of the triangles above measure 60◦ . Thus, the triangles
are equilateral, so that all of their side lengths must be equal. If we write the coordinates of A and
B as ( 21 , a) and ( 12 , −a) respectively, then since OA = AB, the distance formula (8.3) implies that
1
+ a2 = (2a)2 = 4a2 .
4
8.4. GEOMETRICAL CONCEPTS IN THE COMPLEX PLANE 179
Figure 8.30: The hexagon for Example 8.4.2.
1
Thus, a = √ . Thus,
2 3 s
2
1 1
OA = + a2 = √ .
2 3
1
Thus, by symmetry, all six vertices of the hexagon lie on the circle of radius √ centered at the
3
origin.
Let R0 denote the subset of R consisting of those points in the region outside the hexagon that lie
between the rays OA and OB (with −30◦ ≤ θ ≤ 30◦ ), indicated by the shaded region in Figure
8.30. That is,
1
R0 := {z = x + yi = reiθ : x ≥ and − 30◦ ≤ θ ≤ 30◦ }.
2
We will compute the area of
0 1 0
S = |z ∈ R ,
z
and then observe by symmetry that the area of S is six times the area of S 0 .
If we write a complex number z in polar form, z = reiθ , then
1 1 1
= iθ = e−iθ .
z re r
1
Therefore, the angle that makes with respect to the real axis in the complex plane goes from θ
z
to −θ under this transformation. Therefore, any point of R0 that lies outside the circle centered at
1 1 √
(0, 0) of radius √ is transformed via z 7→ to the interior of a circle of radius 3 centered at the
3 z
origin.
1
Now consider the points of R0 that lie inside the circle centered at (0, 0) of radius √ . (See Figure
3
1
8.30). To see how this region is transformed under z 7→ , we primarily need to consider the
z
1 1 1
transformation of the segment AB, which has equation x = , for − √ ≤ y ≤ √ . In polar
2 2 3 2 3
1 1
coordinates, x = becomes r cos θ = . The purpose for rewriting the equation in polar form is
2 2
1 1 1
that, under the transformation z 7→ , we have r cos θ 7→ cos(−θ) = cos θ. Thus, the polar form
z r r
of the transformed equation of the segment AB is
1 1
cos θ = ,
r 2
or
r = 2 cos θ.
Writing p x
r= x2 + y 2 and cos θ = ,
r
we can re-write this transformed equation in Cartesian form as
x2 + y 2 = 2x,
or
(x − 1)2 + y 2 = 1,
which is a circle of radius 1, centered at the point (1, 0). We need only consider the points in this
circle whose reference angle θ with respect to the positive x-axis satisfies −30◦ ≤ θ ≤ 30◦ . Therefore,
Figure 8.31 shows the region S 0 , whose area we now proceed to compute.
Figure 8.31: The region S 0 whose area we need to compute in Example 8.4.2.
As depicted in Figure 8.31, let X √ and Y denote the points where the circle of radius 1 centered at
O0 intersects the circle of radius 3 centered at the origin O. Consider ∆OO0 X. Using the fact that
√ !
√ 3 3
m∠O0 OX = 30◦ and OX = 3, the reader can confirm that X has coordinates X , . If we
2 2
view the base of ∆OO0 X as OO0 , then Equation (8.14) gives
√ √
0 1 3 3
[OO X] = · 1 · = .
2 2 4
√
0 3
By symmetry, [OO Y ] = . Moreover, using the distance formula (8.3), we find that O0 X = 1.
4
Thus, ∆OO0 X is isosceles, which implies that m∠OXO0 = 30◦ . Thus, according to Theorem 8.2.1,
we have m∠OO0 X = 120◦ . By symmetry, we find that m∠XOY = 120◦ , so that the area A of the
portion of the circle of radius 1 centered at O0 that lies to the right of the segments O0 X and O0 Y
is exactly one-third of the area of a circle of radius 1.
8.5. EXERCISES 181
Hence, the total area of S 0 is the sum of the one-third the area of the circle of radius 1 centered at
O0 and the two triangles [OO0 X] and [OO0 Y ], which is
[S 0 ] = A + [OO0 X] + [OO0 Y ]
√ √
π 3 3
= + +
3 √4 4
π 3
= + .
3 2
Hence, the area of S is √ √
[S] = 6[S 0 ] = 2π + 3 3 = 2π + 27.
Thus, a = 2 and b = 27, and we conclude that a + b = 2 + 27 = 29 = 029. ANSWER : 029. 2
8.5 Exercises
1. (2008 AIME, Problem #2) Square AIM E has sides of length 10 units. Isosceles triangle
GEM has base EM , and the area common to triangle GEM and square AIM E is 80 square
units. Find the length of the altitude to EM in ∆GEM .
2. (2007 AIME-2, Problem #3) Square ABCD has side length 13, and points E and F are
exterior to the square such that BE = DF = 5 and AE = CF = 12. Find EF 2 .
FIGURE from 2007 AIME-2, Problem 3 GOES HERE
3. (1999 AIME, Problem #4) The two squares shown share the same center O and have sides
of length 1. The length of AB is 43/99 and the area of octagon ABCDEF GH is m/n, where
m and n are relatively prime positive integers. Find m + n.
FIGURE from 1999 AIME, Problem 4 GOES HERE
4. (2006 AIME-2, Problem #6) Square ABCD has sides of length 1. Points E and F are
on BC and CD, respectively, so that ∆AEF is equilateral. A square with vertex B has sides
√ to those of ABCD and a vertex on AE. The length of a side of this smaller
that are parallel
a− b
square is , where a, b, and c are positive integers and b is not divisible by the square of
c
any prime. Find a + b + c.
5. (2008 AIME-2, Problem #5) In trapezoid ABCD with BC || AD, let BC = 1000 and
AD = 2008. Let ∠A = 37◦ , ∠D = 53◦ , and M and N be the midpoints of BC and AD,
respectively. Find the length M N .
6. (2005 AIME, Problem #7) In quadrilateral ABCD, BC = 8, CD = 12, AD = 10, and
√
m∠A = m∠B = 60◦ . Given that AB = p + q, where p and q are positive integers, find p + q.
7. (1988 AIME, Problem #7) In triangle ABC, tan ∠CAB = 22/7, and the altitude from A
divides BC into segments of length 3 and 17. What is the area of triangle ABC?
8. (2004 AIME, Problem #10) A circle of radius 1 is randomly placed in a 15-by-36 rectangle
ABCD so that the circle lies completely within the rectangle. Given that the probability that
the circle will not touch diagonal AC is m/n, where m and n are relatively prime positive
9. (2010 AIME-2, Problem #9) Let ABCDEF be a regular hexagon. Let G, H, I, J, K,
and L be the midpoints of sides AB, BC, CD, DE, EF , AF , respectively. The segments AH,
BI, CJ, DK, EL, and F G bound a smaller regular hexagon. Let the ratio of the area of the
m
smaller hexagon to the area of ABCDEF be expressed as a fraction where m and n are
n
10. (1986 AIME, Problem #9) In ∆ABC shown below, AB = 425, BC = 450 and CA = 510.
Moreover, P is an interior point chosen so that the segments DE, F G and HI are each of
length d, contain P , and are parallel to the sides AB, BC and CA, respectively. Find d.
11. (2007 AIME-2, Problem #9) Rectangle ABCD is given with AB = 63 and BC = 448.
Points E and F lie on AD and BC respectively, such that AE = CF = 84. The inscribed
circle of triangle BEF is tangent to EF at point P , and the inscribed circle of triangle DEF
is tangent to EF at point Q. Find P Q.
12. (2003 AIME-2, Problem #11) Triangle ABC is a right triangle with AC = 7, BC = 24,
and right angle at C. Point M is the midpoint of AB, and D is on the same side of √line AB as
m n
C so that AD = BD = 15. Given that the area of ∆CDM can be expressed as , where
p
m, n, and p are positive integers, m and p are relatively prime, and n is not divisible by the
square of any prime, find m + n + p.
13. (1990 AIME, Problem #12) A regular 12-gon is inscribed in a circle of radius 12. The√sum
of√the lengths
√ of all sides and diagonals of the 12-gon can be written in the form a + b 2 +
c 3 + d 6, where a, b, c, and d are positive integers. Find a + b + c + d.
14. (2000 AIME, Problem #14) In triangle ABC, it is given that angles B and C are congruent.
Points P and Q lie on AC and AB, respectively, so that AP = P Q = QB = BC. Angle ACB
is r times as large as angle AP Q, where r is a positive real number. Find the greatest integer
that does not exceed 1000r.
15. (2005 AIME-2, Problem #14) In ∆ABC, AB = 13, BC = 15, and CA = 14. Point D is
on BC with CD = 6. Point E is on BC such that ∠BAE ∼ = ∠CAD. Given that BE = p/q,
where p and q are relatively prime positive integers, find q.
Chapter 9
Spatial Geometry
“Where there is matter, there is geometry.”

- Johannes Kepler
9.1 Introduction
The previous chapter involved geometric notions confined to two dimensions. Now, we turn our
attention to the three-dimensional world of rectangular boxes, cylinders, cones, tetrahedra, pyramids,
spheres, and the like. Two of the quantities most frequently associated with three dimensional objects
are its surface area and, if it is closed, its volume. Throughout this chapter, in computing surface
areas, we note that we can use any of the area formulas from plane geometry that were provided in
Chapter 8.
One of the main challenges here can be to draw on paper figures that are insightful and accurate to
portray the important features of a three-dimensional problem. Very often, it is not important to
draw the entire three-dimensional figure, but just that portion of it crucial to solving the problem.
Moreover, many of the tools exploited in the preceding chapter (such as similarity, trigonometry,
and so on) continue to be relevant and useful here as well.
9.2 Rectangular Boxes
The rectangular box is arguably the simplest three-dimensional solid, with the cube being the easiest
special case within this class. The closed rectangular box in Figure 9.1 with side lengths a, b, and c
has volume V and surface area A given by the formulas
V = abc and A = 2ab + 2ac + 2bc. (9.1)
183
184 CHAPTER 9. SPATIAL GEOMETRY
Of course, if the box is not closed, then one of the sides of the box must be excluded from the
calculation of area A, and the volume V does not even make sense. As a special case, a cube is a
rectangular box whose side lengths are all the same, and in this case, the formulas in (9.1) reduce to
V = a3 and A = 6a2 . (9.2)
Figure 9.1: A closed rectangular box with side lengths a, b, and c.
Here is an example of an AIME problem that requires only the formula for the volume of a box
given in Equation (9.1).
Example 9.2.1. (2008 AIME-2, Problem #3) A block of cheese in the shape of a rectangular
solid measures 10 cm by 13 cm by 14 cm. Ten slices are cut from the cheese. Each slice has a width
of 1 cm and is cut parallel to one face of the cheese. What is the maximum possible volume in cubic
cm of the remaining block of cheese after ten slices have been cut off ?
Solution #1: With each slice that is cut off, one of the three dimensions of the block of remaining
cheese is reduced by 1. Therefore, after the ten slices are removed, the volume of the remaining
block has the form
V = (10 − x)(13 − y)(14 − z),
where x, y, and z are nonnegative integers such that
x + y + z = 10. (9.3)
The volume is maximized if the three dimensions of the box are as close to each other as possible.
Specifically, V is maximized if
10 − x = 13 − y = 14 − z.
For this, we need y = 3 + x and z = 4 + x. Substituting these expressions into Equation (9.3), we
obtain
10 = x + y + z = x + (3 + x) + (4 + x) = 7 + 3x,
from which it follows that
x = 1, y = 4, z = 5.
Thus, the maximum volume is
V = (10 − x)(13 − y)(14 − z) = 93 = 729.
ANSWER : 729. 2
Solution #2: As indicated in Solution #1 above, each slice removed from the cheese decreases one
of its dimensions by 1. The original cheese had dimensions summing to 10 + 13 + 14 = 37, so after
9.2. RECTANGULAR BOXES 185
10 slices are removed, the remaining block of cheese will have dimensions summing to 37 − 10 = 27.
To maximize the volume, we make all sides of the block the same length, 9 cm. The volume of a
cube of side length 9 is 93 = 729.
Now let us consider some additional examples.
Example 9.2.2. (1996 AIME, Problem #4) A wooden cube, whose edges are one centimeter
long, rests on a horizontal surface. Illuminated by a point source of light that is x centimeters directly
above an upper vertex, the cube casts a shadow on the horizontal surface. The area of the shadow,
which does not include the area beneath the cube, is 48 square centimeters. Find the greatest integer
that does not exceed 1000x.
Solution: Figure 9.2 shows the light sources at P above the vertex E of cube ABCDEF GH.
In Chapter 8, we discussed the notion of similar figures in the plane. Here, we have an example
of similar solids in space. In particular, pyramids P EF GH and P AQRS are similar figures, which
means that corresponding angles in the two pyramids are equal, and that the ratios of corresponding
sides of the two figures are all equal. Now P E = x and P A = 1 + x. By similarity, we have
PE PA
= ,
EF AQ
so that
P A · EF (1 + x) · 1
AQ = = .
PE x
2
1+x
Thus, the area of square AQRS is , so that the area of the shadow is therefore
x
2
1+x
Area of Shadow = − 1 = 48.
x
Hence,
2
1+x
= 49,
x
1+x 1 1
=7 or =6 or x = .
x x 6
Hence,
1000 2
1000x = = 166 ,
6 3
so the answer is 166. ANSWER : 166. 2
Example 9.2.3. (2002 AIME, Problem #11) Let ABCD and BCF G be two faces of a cube
with AB = 12. A beam of light emanates from vertex A and reflects off face BCF G at point P ,
which is 7 units from BG and 5 units from BC. The beam continues to be reflected off the faces of
the cube. The length of the
√ light path from the time it leaves point A until it next reaches a vertex
of the cube is given by m n, where m and n are integers and n is not divisible by the square of any
prime. Find m + n.
Solution: The beam alternates between hitting face ADEH and hitting the opposite face, BCF G
(see Figure 9.3).
An important observation is that, while it is possible for the light beam to impact other faces (such
as EF GH, CDEF , and so on), the distance travelled by the beam between√ impacting faces
√ ADEH
and BCF G is always identical, and it is the length of segment AP : 122 + 52 + 72 = 218. Since
gcd(5, 12) = 1 and gcd(7, 12) = 1, the beam will only reach a corner of the cube after every 12th
reflection. Hence, the
√total distance travelled by the beam after leaving point A and before impacting
another vertex is 12 218. Hence, m = 12 and n = 218, so that m + n = 230. ANSWER : 230. 2
9.3 Cylinders, Cones, and Spheres
A cylinder is formed when a closed, flat two-dimensional region (often called the base of the cylinder,
is stretched perpendicular to the region into a three-dimensional solid. A rectangular box is a cylinder
9.3. CYLINDERS, CONES, AND SPHERES 187
in which the base is a rectangle. More typically, a cylinder’s base is a circle (or portion thereof),
but in theory, the cylinder can have any shape.
Let us next give the formulas for volume V and surface area A of a (closed) cylinder of height h and
base of area B and perimeter p:
V = Bh and A = ph.
In particular, a right circular cylinder of radius r and height h has
V = πr2 h and A = 2πrh. (9.4)
In the case of a cone, the base is always assumed to be circular. If the top of the cone lies directly
above the center of the circular base, then the cone is sometimes called a right circular cone, and
there are well-known formulas for the volume A and surface area A for the closed right circular cone
with base radius r and height (distance from the top of the cone to the center of the base) h:
1 2 p
V = πr h and A = πr2 + πr h2 + r 2 . (9.5)
3
√
The quantity h2 + r2 is also the “slant height” of the cone, the distance from the top of the cone
to a point on the circular base. Its value results directly from the Pythagorean Theorem.
Finally, one of the most important objects in three-dimensional space is the sphere. A sphere consists
of a set of points S in space that are all equidistant from a given point P . The distance from P
to any point in S is called the radius of the sphere. The equation of a sphere of radius r that is
centered at the point (a, b, c) is most often expressed as
(x − a)2 + (y − b)2 + (z − c)2 = r2 . (9.6)
The volume and surface area formulas for a sphere are both expressed solely in terms of the radius
r of the sphere:
4
V = πr3 and A = 4πr2 . (9.7)
3
Now let us consider a few examples involving these shapes, beginning with an example involving a
right circular cylinder.
Example 9.3.1. (2003 AIME-2, Problem #5) A cylindrical log has diameter 12 inches. A
wedge is cut from the log by making two planar cuts that go entirely through the log. The first is
perpendicular to the axis of the cylinder, and the plane of the second cut forms a 45◦ angle with the
plane of the first cut. The intersection of these two planes has exactly one point in common with the
log. The number of cubic inches in the wedge can be expressed as nπ, where n is a positive integer.
Find n.
Solution: If the log is lying flat on a table, the first cut makes a vertical plane whose cross section in
the log is a circle. We may assume that the second cut shares only the point on this circle touching
the table. Because the angle of the second cut is 45◦ with the plane of the first cut, the length of
the wedge must equal the diameter of the cylinder (since the legs of a 45-45-90 triangle have the
same length). By symmetry, the volume of the wedge is exactly half of the volume of the cylinder of
height and diameter of length 12 inches (or radius r = 6 inches). Using Equation (9.4) the volume
of the whole cylinder is π · 62 · 12 = 432π. Hence, half of the cylinder has volume 216π. Therefore,
n = 216. ANSWER : 216. 2
Example 9.3.2. (2004 AIME, Problem #11) A solid in the shape of a right circular cone is
4 inches tall and its base has a 3-inch radius. The entire surface of the cone, including its base, is
painted. A plane parallel to the base of the cone divides the cone into two solids, a smaller cone-
shaped solid C and a frustrum-shaped solid F, in such a way that the ratio between the areas of the
painted surfaces of C and F and the ratio between the volumes of C and F are both equal to k. Given
that k = m/n, where m and n are relatively prime positive integers, find m + n.
Solution: Suppose that the plane parallel to the base of the cone is 4 − x units above the base, so
that the cone C has a height of x. If we denote the radius of C by r, we note that
4
x= r, (9.8)
3
since C is similar to the original right circular cone (and hence must maintain the same ratio of height
to radius). The idea now is to set up expressions representing the ratio of the painted surfaces of C
and F and the ratio of the volumes of C and F. To determine these quantities for F, we will subtract
the desired amount of the cone C from the original given cone. For instance, using subtraction and
Equation (9.5), we find that the volume of F is
1 2 1 4
Vol(F) = π3 · 4 − πr2 · x = π(12 − r3 ),
3 3 9
where we have used Equation (9.8) to eliminate x. Therefore, the ratio of volumes of C and F is
1 2 4 3
3 πr x 9r
k= = . (9.9)
π(12 − 49 r3 ) 12 − 49 r3
Next, we find the surface area of the slanted portion of F by subtracting the surface area of C from
the larger cone. Again, we appeal to Equation (9.5). The result is
p p 5
Painted Area of F = (9π + 3 32 + 42 π) − (πr2 + πr r2 + x2 ) = 24π − r2 π.
3
√
On the other hand, the painted portion of C has area πr r2 + x2 = 53 πr2 . Thus, the ratio of the
painted area of C to that of F is
5 2 5 2
3 πr 3r
k= 5 = . (9.10)
24π − 3 πr2 24 − 35 r2
Equating the expressions (9.9) and (9.10), we have
4 3 5 2
9r 3r
= .
12 − 49 r3 24 − 53 r2
Cross multiplying, we find that
5 2 4 4 5
r (12 − r3 ) = r3 (24 − r2 ),
3 9 9 3
9.4. TETRAHEDRA AND PYRAMIDS 189
and now we can simplify this to

32 3
20r2 = r .
3
We deduce that r = 15
8 . Hence, using either (9.9) or (9.10), we find that
4 3
9r 125
k= = .
12 − 49 r3 387
Therefore, m = 125 and n = 387. Hence, m + n = 125 + 387 = 512. ANSWER : 512. 2
Example 9.3.3. (AMC 12 B 2003, Problem #13) An ice cream cone consists of a sphere of
vanilla ice cream and a right circular cone that has the same diameter as the sphere. If the ice cream
melts, it will exactly fill the cone. Assume that the melted ice cream occupies 75% of the volume of
the frozen ice cream. What is the ratio of the cone’s height to its radius?
Solution: Let r denote the (common) radius of the cone and sphere. According to Equation (9.7),
the volume of the frozen ice cream is
4
Vfrozen = πr3 ,
3
so that the volume of the melted ice cream is
Vmelted = 0.75 · Vfrozen = πr3 .
Equating this to the volume of the cone with height, say h, we obtain
1 2
πr3 = πr h.
3
From this, we deduce that h = 3r. Thus, the ratio of the cone’s height to its radius is 3 : 1. 2
9.4 Tetrahedra and Pyramids
Like the cone, tetrahedra and pyramids can be oriented with a flat “base” and rising to a point
above the ground from all corners of the base. The base is a triangle in the case of a tetrahedron
and a rectangle (most often a square) in the case of a pyramid. To find the volume of each shape,
one simply multiplies one-third of the area of the base, A, times the height h:
1
Volume = Ah. (9.11)
3
While it is rare to find two nearly identical problems arising in the AIME competition, there is an
instance of topic repetition concerning the volume of a tetrahedron. We present one occurrence of
this problem here and leave the other as an exercise (see Problem 8).
Example 9.4.1. (1992 AIME, Problem #7) Faces ABC and BCD of tetrahedron ABCD meet
at an angle of 30◦ . The area of face ABC is 120, the area of face BCD is 80, and BC = 10. Find
the volume of the tetrahedron.
Solution: Orient the tetrahedron so that face ABC is regarded as the base. If we let h denote the
height of the tetrahedron (the distance from vertex D to the plane containing face ABC), then the
volume of the tetrahedron is
1
Volume = (Area of Base)(Height) = 40h.
3
To find h, first consider triangle BCD. It’s height a (distance from vertex D to the segment BC)
can be derived from its area:
1
Area of Triangle BCD = 80 = a(BC) = 5a.
2
Therefore, a = 16. Since this altitude in triangle BCD meets face ABC in an angle of 30◦ , we
see that h = a sin 30 = 21 a = 8. Thus, the volume of the tetrahedron is 40h = 40 · 8 = 320.
ANSWER : 320. 2
The next example illustrates the use of similar triangles in computing volumes of tetrahedra.
Example 9.4.2. (2003 AIME-2, Problem #4) In a regular tetrahedron, the centers of the four
faces are the vertices of a smaller tetrahedron. The ratio of the volume of the smaller tetrahedron to
that of the larger is m/n, where m and n are relatively prime positive integers. Find m + n.
Solution: Without loss of generality, assume that the sides of regular tetrahedron ABCD each
have length 1. See Figure 9.4.
Let O and P be the centers of faces DAB and ABC, respectively. Both DO and CP intersect
1
AB at its midpoint M . Trigonometry with 30-60-90 triangles quickly shows that M O = √ and
√ 2 3
3
MD = . Thus,
2
MO 1
= .
MD 3
Likewise, we have
MP 1
= .
MC 3
9.5. EXERCISES 191
Hence, triangles M OP and M DC are similar, and we conclude that

OP 1
= ,
DC 3
from which we deduce that OP = 13 . Therefore, the smaller tetrahedron is regular with side length
1
3 . Hence, since all sides of the smaller tetrahedron have length 31 that of the larger tetrahedron,
3
1 1
the ratio of the volumes of these two tetrahedra is then = . We conclude that m = 1 and
3 27
n = 27, so that m + n = 28 = 028. ANSWER : 028. 2
9.5 Exercises
1. (2002 AIME-2, Problem #2) Three vertices of a cube are P = (7, 12, 10), Q = (8, 8, 1),
and R = (11, 3, 9). What is the surface area of the cube?
2. (1985 AIME, Problem #2) When a right triangle is rotated about one leg, the volume of
the cone produced is 800π cm3 . When the triangle is rotated about the other leg, the volume of
the cone produced is 1920π cm3 . What is the length (in cm) of the hypotenuse of the triangle?
3. (2004 AIME-2, Problem #3) A solid rectangular block is formed by gluing together N
congruent 1-cm cubes face to face. When the block is viewed so that three of its faces are
visible, exactly 231 of the 1-cm cubes cannot be seen. Find the smallest possible value of N .
4. (2012 AIME-2, Problem #5) In the accompanying figure, the outer square S has side
length 40. A second square S 0 of side length 15 is constructed inside S with the same center
as S and with sides parallel to those of S. From each midpoint of a side of S, segments are
drawn to the two closest vertices of S 0 . The result is a four-pointed starlike figure inscribed
in S. The star figure is cut out and then folded to form a pyramid with base S 0 . Find the
volume of this pyramid.
INSERT FIGURE !!!!!
5. (2003 AIME, Problem #5) Consider the set of points that are inside or within one unit of
a rectangular parallelepiped (box) that measures 3 by 4 by 5 units. Given that the volume of
m + nπ
this set is , where m, n, and p are positive integers, and n and p are relatively prime,
p
find m + n + p.
6. (2013 AIME, Problem #7) A rectangular box has width 12 inches, length 16 inches, and
height m
n inches, where m and n are relatively prime positive integers. Three faces of the box
meet at a corner of the box. The center points of those three faces are the vertices of a triangle
with an area of 30 square inches. Find m + n.
7. (2000 AIME, Problem #8) A container in the shape of a right circular cone is 12 inches
tall and its base has a 5-inch radius. The liquid that is sealed inside is 9 inches deep when the
cone is held with its point down and its base horizontal. When the cone is held with its point
√
up and its base horizontal, the liquid is m − n 3 p inches deep, where m, n, and p are positive
integers and p is not divisible by the cube of any prime number. Find m + n + p.
8. (1984 AIME, Problem #9) In tetrahedron ABCD, edge AB has length 3 cm. The area
of face ABC is 15 cm2 and the area of face ABD is 12 cm2 . These two faces meet each other
at a 30◦ angle. Find the volume of the tetrahedron in cm3 .
9. (2005 AIME, Problem #9) Twenty-seven unit cubes are each painted orange on a set of
four faces so that the two unpainted faces share an edge. The 27 cubes are then randomly
arranged to form a 3 × 3 × 3 cube. Given that the probability that the entire surface of the
pa
larger cube is orange is b c , where p, q, and r are distinct primes and a, b, and c are positive
q r
integers, find a + b + c + p + q + r.
10. (2010 AIME, Problem #11) Let R be the region consisting of the set of points in the
coordinate plane that satisfy both |8 − x| + y ≤ 10 and 3y − x ≥ 15. When R is revolved
mπ
around the line whose equation is 3y − x = 15, the volume of the resulting solid is √ , where
n p
m, n, and p are positive integers, m and n are relatively prime, and p is not divisible by the
square of any prime. Find m + n + p.
11. 1998 AIME, Problem #10) Eight spheres of radius 100 are placed on a flat surface so that
each sphere is tangent to two others and their centers are the vertices of a regular octagon.
A ninth sphere is placed on the flat surface so √ that it is tangent to each of the other eight
spheres. The radius of this last sphere is a + b c, where a, b, and c are positive integers, and
c is not divisible by the square of any prime. Find a + b + c.
√
12. 2004 AIME-2, Problem #11) A right circular cone has a radius 600 and height 200 7.
A fly starts at a point on the surface of the cone whose distance from the vertex of the cone
is 125, and crawls along the surface of the√cone to a point on the exact opposite side of the
cone whose distance from the vertex is 375 2. Find the least distance that the fly could have
crawled.
13. (2005 AIME-2, Problem #10) Given that O is a regular octahedron, that C is the cube
whose vertices are the centers of the faces of O, and that the ratio of the volume of O to that
of C is m/n, where m and n are relatively prime positive integers, find m + n.
Chapter 10
Hints for the Exercises
We have all had the experience of being stumped by a math problem. This can happen for a variety
of reasons, but no matter what the reason, it can be a frustrating experience, especially for those
who hold themselves to exceptionally high standards. If being unable to solve a problem that one
feels they should be able to is a frustrating experience, then surely to read a succinct, clear solution
to such a problem is a deflating experience. We often wish that we had been given just a nudge
in the right direction, so that we could still claim ownership for the solution derived. This chapter
aims to provide just such a nudge, for each of the exercises contained in this text.
For each problem in the text, this chapter contains two types of hints. First, there is the “Hints
to Get Started” suggestion which is designed to just get the reader thinking about how to begin
solving the problem. Then, if that is not sufficient, or if the reader simply wants further guidance, a
section called “More Extensive Hints” provide more detailed strategies for solving the problems,
with less work for the reader to fill in to complete the solution. In reading the “more extensive
hints” for a particular problem, it is assumed that the reader has already read the corresponding
“hints to get started” for that problem.
Of course, many of the AIME problems contain multiple good solutions, and in the full solutions
in the final chapter, multiple solutions are sometimes given. However, the hints are usually only
oriented towards one particular solution, the one that the author feels is most likely to be easiest
for the reader to discover.
Again, to reiterate, readers who are not sure how to begin working on a problem in an exercise set
are encouraged to consult these hints and use them to take a fresh look at the problems on which
they are stuck. This will be a more effective way to learn and practice than simply reading the
solution provided in Chapter 11.
193
194 CHAPTER 10. HINTS FOR THE EXERCISES
10.1 Hints for Chapter 1
10.1.1 Hints to Get Started
1. Let g denote the number of girls at the party in the beginning, and let b denote the number
of boys at the party in the beginning. Therefore, total number of people at the party initially
is g + b. Now use the given information to write equations involving g and b.
2. Assume Rudolph bikes at a rate of r = 4k (miles per minute) for some constant k, and assume
Jennifer bikes at a rate of j = 3k (miles per minute). Compute Rudolph’s and Jennifer’s biking
times with no breaks and then add the amount of time for breaks. Be careful when counting
the number of breaks that each biker takes.
3. Write formulas A(t), B(t), and C(t) for the distance of Al, Bob, and Cy from the start of
walkway at time t, respectively. You will need to consider the possibilities for the order in
which the three walkers could be standing when one of the persons is halfway between the
other two.
4. If t denotes the total number of matches the player has played before the weekend and w
w
denotes the number of matches she has won before the weekend, what can you say about
t
w+3
and ?
t+4
5. The last two conditions in the newspaper story determine two equations relating the total
number of contestants at the festival, say x, to the total number of fish caught at the festival,
say f . Setting up two equations for these two unknowns and solving them will give the solution.
6. Add the five equations together.
7. Observe that Alpha’s success ratio was greater on the second day than the first day. Thus,
Beta’s largest overall success ratio is achieved when he answers the most questions on the
second day.
8. Start with an equation in which the base 10 number with ones digits a0 , tens digit a1 , and so
on is set equal to the base 7 number with the same digits. Together with the information that
0 ≤ ai ≤ 6 for each i, this equation will impose severe limitations on the values of a0 , a1 , and
so on.
9. One approach is to find the amount of time it takes to complete each quarter of the job
separately, given the loss of workers, as compared with the amount of time it would have taken
if each quarter had been completed on schedule. For simplicity, assume that, with the full
work force, each quarter of the work gets completed in exactly one hour. How long do the first
three-quarters of work take given the depleted work force? Use this to determine the number
of workers needed for the final quarter of work.
10. Let x denote the number of fish in the lake on May 1. Note that 60% of the 70 fish caught on
September 1 had been eligible for tagging on May 1.
10.1. HINTS FOR CHAPTER 1 195
11. This problem requires setting up several equations involving several unknowns. For instance,
try to write equations for the number of bananas each monkey ends up with (say M1 , M2 , and
M3 ) in terms of the number of bananas taken by each monkey from the pile (say x, y, and z,
respectively).
12. Re-express a3 and b3 in terms of c3 . Then the equation c3 − abc = 20 can be put in terms of
c alone. An algebraic manipulation then suffices to solve for c (and hence a and b).
13. This problem can be solved by “brute force” if one makes a table of showing Mary’s possible
scores as a function of the number of correct answers. Then one can simply look for the smallest
exam score that occurs only once in the table. Alternatively, one can derive inequalities
involving c and w that must hold in order for the value of c to be unique. From these
inequalities, the smallest possible score s corresponds to taking c as small as possible and w
as large as possible, subject to the inequalities.
14. Let v denote the speed of the escalator and let b denote Bob’s normal walking speed, in steps
per unit time. Then Bob’s effective speed on the escalator is b + v, while Al’s is 3b − v. Now
compute the amount of time Bob and Al each spend on the escalator as a function of b, v, and
n, where n is the number of steps on the escalator.
15. We can formulate equations relating the expressions in question by noting that
(axk + by k )(x + y) = axk+1 + by k+1 + xy(axk−1 + by k−1 ) (10.1)
for all integers k ≥ 1. Apply Equation (10.1) for k = 1, 2, 3, and 4 and substitute given
variables where possible.
10.1.2 More Extensive Hints
1. We are explicitly given that g = 0.6(g + b) and that g = 0.58(g + b + 20). Now solve for g + b,
the total number of people intially at the party. Now it is a routine matter to determine the
number of people who like to dance at the party, both before and after the 20 additional boys
arrive.
2. Verify that Rudolph’s total biking time is 50/r and Jennifer’s total biking time is 200/3r.
Rudolph takes 49 breaks while Jennifer takes 24 breaks prior to arriving at the 50-mile mark.
Now compute Rudolph’s total time and Jennifer’s total time and set them equal to each other
to solve for r.
3. Note that Bob will always be ahead of Cy on the walkway, so there are only three possible
orders that the three people could be arranged in. The formulas needed to proceed are A(t) =
6t, B(t) = 10(t − 2), and C(t) = 8(t − 4). Use these equations to show that two of the
three possible orders are impossible. For instance, if Bob is in front and Al is in back, then
B(t) − C(t) = C(t) − A(t) in order for Cy to be halfway between the other two. Show that
this gives a contradiction.
w+3
4. In terms of w and t, the given information translates to w = 2t and > 0.503. Now
t+4
derive an inequality for w and do not forget that w must be an integer.
f − 19 f − 108
5. The last two conditions given in the newspaper story dictate that = 6 and = 5.
x − 21 x−8
From these two equations, it is possible to solve for the unknowns x and f .
6. The result of adding the five equations together can be simplified to x1 +x2 +x3 +x4 +x5 = 31.
Subtract this equation from the appropriate equations from the original system in order to
find x4 and x5 .
7. Let us adopt the following notation. Suppose Beta answers k questions on the first day (and
thus 500 − k questions on the second day), with x correct answers on the first day and y
correct answers on the second day. Since Beta must answer at least one question correctly
each day but attains a percentage on the first day that is less than 160/300, the highest overall
percentage will be obtained when x = 1 and k = 2. Now compute the largest integer y such
that y/498 < 0.7.
8. For a 7-10 double, a0 + 10a1 + 100a2 + · · · = 2(a0 + 7a1 + 49a2 + · · · ). Thus, a0 + 4a1 =
2a2 + 314a3 + · · · . What does this condition tell you about a3 , a4 , and so on. Now choose
digits a0 , a1 , and a2 resulting in the largest possible 7-10 double.
9. The first quarter of work is completed in one hour, the second quarter in 10/9 hours, the third
quarter in 10/8 hours, leaving 4 − 1 − 10/9 − 10/8 hours for the final quarter of work to get
done. Use this to find the size of the work force required to complete the job on time. How
many additional workers must be hired to obtain this size?
10. Observe that 3/42 is the fraction of fish in the lake on May 1 that were tagged. Remember
that 60 fish were tagged on May 1. Set up an equality of ratios involving x and use it to
solve for x. Note, incidentally, that the deaths of some fish from the lake is immaterial to the
solution to this problem.
11. From the 3 : 2 : 1 radio given, we can write M1 = 21 (x + y + z), and so on for M2 and M3 .
Furthermore, the given information supplies the formulas M1 = 43 x + 38 y 11
24 z, and so on for M2
and M3 . Equate the expressions for M1 , M2 , and M3 and work algebraically towards solving
for x, y, and z.
12. After rewriting the equation c3 −abc = 20 in terms of c alone, we find a sixth-degree polynomial
for c. Use the quadratic formula to find two possible values for c3 (e.g. by replacing d = c3
in order to reduce the sixth-degree polynomial in c to a quadratic polynomial in d). Take the
maximum value of c3 in order to maximize a3 + b3 + c3 .
13. Observe that if w ≥ 4, then the score s is not uniquely obtained (why?). On the other hand,
if c + w ≤ 25, the score s is not uniquely obtained. Thus, we must have w ≤ 3 and c + w ≥ 26.
What choice of c and w results in the minimum possible score s?
14. Bob’s escalator time is n/(b + v) and Al’s escalator time is n/(3b − v).Be sure
you understand
n
why. Now use Equation (1.8) to obtain Bob’s total distance: 75 = b . Find a similar
b+v
formula using Al’s total distance. Using these two equations, it is possible now to algebraically
solve for n by first eliminating it from the two formulas for Bob and Al’s total distance.
15. Use the cases k = 2 and k = 3 in Equation (10.1) to find the values of xy and x + y. Then use
the case k = 4 to find the desired quantity.
1. Consider two cases. Case 1 consists of those plates that use two zeros, and Case 2 consists of
those plates that use less than two zeros. There are no plates common to both cases, so the
Sum Rule can be applied.
2. Consider two cases. Case 1 is that one man and one woman is chosen from each department,
and Case 2 is that two men are chosen from one department, two women are chosen from
another department, and one man and one woman are chosen from the third department.
3. Begin by counting the total number of line segments that can be constructed with endpoints at
vertices of P . Then subtract out the number of line segments that do not satisfy the definition
of a space diagonal.
4. Fix one particular meal to be correct. For the other two individuals who ordered the same
type of meal, consider the ways in which their meals could be incorrect and then proceed to
an analysis of cases. It might be helpful to draw a figure showing the nine positions arranged
in a row, using “B” to denote a beef meal, “C” to denote a chicken meal, and “F” to denote
a fish meal.
5. Consider two cases. Case 1 consists of numbers whose identical digit is 1, and Case 2 consists
of numbers whose identical digit is not 1. There are no numbers common to both cases, so the
Sum Rule can be applied.
6. Use the Subtraction Rule. Apply the stars and bars method with bars located wherever the
five selected natural numbers happen to be.
7. What is a6 ? Are there restrictions on which numbers can comprise the set {a1 , a2 , a3 , a4 , a5 }?
8. Try some examples first. They should reveal that it is more effective to create 4-digit parity-
monotonic numbers by starting with a4 , then selecting a3 , then selecting a2 , and finally,
selecting a1 .
9. The orientation of the coins and the color of the coins are independent, so we can multiply
the number of ways to establish each attribute of the coins in the stack. How many ways are
there to assign colors (four gold and four silver) to the coins? How many different orientation
arrangements of the coins are possible?
10. According to (i)-(iii), a complementary set must consist of three cards that are the same with
respect to 2 or fewer of the three attributes (shape, color, shade) and different with respect to
the remaining attributes. Count complementary sets in three cases according to whether the
three cards share 0, 1, or 2 of the attributes.
11. We can put the 15 seats in a straight line, with a Martian (M) in the first seat and Earthling
(E) in the last seat. The seating requirements dictate that the pattern must be
some M’s → some V’s → some E’s → some M’s → some V’s → some E’s → · · ·
Consider how many cycles through M → V → E are possible and count the number of seating
arrangements in each case.
12. Begin by writing four T H subsequences side by side, with space between each subsequence. One
HT subsequence must occur between each consecutive pair of T H subsequences. Therefore,
no additional T H or HT subsequences can occur. This determines in large part how the rest
of the sequence (which must have two HH and five T T subsequences) can be completed.
13. Call the two non-empty disjoint sets A and B. Each integer in {1, 2, 3, . . . , 10} has three
possible locations: placed in A, placed in B, or placed in neither set.
14. Call the numbers placed in the bottom row a1 , a2 , a3 , . . . , a10 . Start filling in the numbers in
the squares above it until a pattern is found. Use the Binomial Theorem.
15. The number of houses receiving mail must be 6, 7, 8, 9, or 10. Determine how many patterns
of mail are possible in each of these five cases. (All five cases need not be done separately, as
there are similarities that can be exploited between the cases.) Set up variables xi to represent
the number of houses between the ith and (i + 1)th house that receive mail. Then xi = 1 or
xi = 2 for each i. Additionally, the number of houses that do not receive mail before the first
house or after the last house must be 0, 1, or 2.
1. For Case 1, first choose two positions, among the five available, to place the zeros. Then
apply the Multiplication Principle. For Case 2, the Multiplication Principle can be applied
immediately.
2. For Case 1, there are two choices for each sex chosen in each department. For Case 2, begin
by choosing which department will have two male representatives, which department will have
two female representatives, and which department will have one of each.

26
3. The total number of line segments with vertices of P as endpoints is . Now subtract
2
the number of edges (since they join adjacent vertices) and all line segments joining all non-
adjacent vertices belonging to the same face.
4. For simplicity, assume the first person in the row receives the correct meal, say beef. Now the
second and third person in the row must receive the wrong meal, and there are two cases. Case
1 is the situation where the second and third person each receive the same (non-beef) meal,
and Case 2 is the situation where the second and third person receive different (non-beef)
meals. Now in each case, analyze the possible choices, and if necessary specify some choice
without loss of generality. Once you have a hypothetical assignment of meals to the first three
people in line, consider the next three people in line. With the first person (and correct meal)
specified, you should arrive at 6 ways to distribute the remaining meals in Case 1, and 18 ways
to distribute the remaining meals in Case 2.
5. For Case 1, first choose where the second 1 occurs in the number, then select the remaining
digits of the number. For Case 2, first choose which digit will be repeated, then choose the
remaining digit and positions. For both cases, apply the Multiplication Principle.

14
6. There are ways to choose five of the first 14 natural numbers. If we line up the
5
14 numbers we started with, the selection of five of them can be denoted by a string of 5
C’s (C is for “chosen”) and 9 N’s (N is for “not chosen”). If we view the C’s as “bars”,
and the N’s as “stars”, then we have five bars and nine stars. We are then trying to solve
x1 + x2 + x3 + x4 + x5 + x6 = 9 in nonnegative integers. What condition will ensure that no
two chosen numbers are consecutive?
7. Observe that a6 = 1 is required. Among the remaining 11 numbers, choose five of them to be
a1 , a2 , a3 , a4 , and a5 . Are there any additional choices to be made?
8. There are 10 choices for the rightmost digit a4 . Next, move to the left and consider the number
of choices for a3 . To determine this, it may be helpful to try some examples for a4 and see
how many choices you have for a3 . It turns out that the number of choices for a3 does not
depend on what choice was made for a4 . Now repeat this for a2 and then for a1 .
9. Looking at the stack from the top down, there are a set of coins face up followed by a set of
coins face down. It is actually possible to draw all of the stacks if desired.
10. Consider how many complementary three-card sets share no common attributes. In this case,
one card is a circle, one card is a square, and one card is a triangle. There are now 3! ways
to assign color to these shapes, and then 3! ways to assign shades to these shapes. Use the
Multiplication Principle to find the number of complementary sets of this type. Then proceed
to similar analyses for the other cases (one shared attribute and two shared attributes).
11. The factor of (5!)3 appearing in the question should be considered the number of ways of
arranging the five distinct Martians, five distinct Venusians, and five distinct Earthlings,
once their positions are chosen. Thus, N denotes the number of ways to select positions
for the type of committee member (M , V , or E) only. The pattern of some Martians fol-
lowed by some Venusians followed by some Earthlings can occur from one to five times. It
occurs one time in the pattern M M M M M V V V V V EEEEE and five times in the pattern
M V EM V EM V EM V EM V E. How many arrangements are possible with two cycles? three
cycles? four cycles?
12. There can be no H’s to the left of the four “TH” patterns, and no T ’s to the right of the four
“TH” patterns. Moreover, the three spaces between the four “TH” patterns must consist of a
sequence of H’s followed by a sequence of T ’s. Start by figuring the number of ways to achieve
two HH patterns. Then do a similar process to handle the T T patterns. Because the HH
and T T patterns are independent, the Multiplication Principle can be applied.
13. If we call the two non-empty, disjoint subsets A andB, then there are three possible choices
for the location of each integer in the set {1, 2, 3, . . . , 10}. This gives 310 outcomes for all
10 numbers. However, some of them must be removed because the sets A and B must be
non-empty. Use the Inclusion-Exclusion Principle. Also, note that the sets A and B are not
ordered.
14. The entries
in the
rth row down from the top of the array involve binomial coefficients of the
11 − r
form for s = 0, 1, 2, . . . , 11 − r. Now it is easy to see that the top entry in the
s
rectangular array involves coefficients that are mostly divisible by 3 regardless of the value of
ai . Isolate the terms whose binomial coefficients are not divisible by 3 and develop a list of
possible values of the corresponding ai .
15. If k houses receive mail (where 6 ≤ k ≤ 10), the line of 19 − k houses on Elm Street that
do not receive mail are split into k + 1 groups. Thus, we can consider the equation x1 +
x2 + · · · + xk+1 = 19 − k, where the given conditions require that xi ≤ 2 for each i (with
1 ≤ i ≤ k + 1) and x2 , x3 , . . . , xk ≥ 1. Use a change of variables to eliminate the requirement
that x2 , x3 , . . . , xk ≥ 1. For the requirement xi ≤ 2, consider each value of k in turn on a
case-by-case basis.
1. There are multiple ways to analyze this problem. For instance, we can let R denote a red
candy and B denote a blue candy. Terry and Mary draw a total of four candies, which forms
strings using R and B of length 4, for a total of 16 outcomes. Determine which of these 16
strings corresponds to Terry and Mary getting the same color combination, and then find the
probability that each of these strings occurs.

38
2. There are ways to choose two cards. Next count the number of pairs of cards in the
2
deck consisting of the same denominations.
3. In order for all three players to obtain tiles whose sum is odd, exactly one player must get
three odd tiles and the other two players must get one odd tile and two even ones. In how
many ways can this occur?
4. For each of the twelve gates, compute the fraction of gates that are within 400 feet of the original
gate. Now do the appropriate products and sums of probabilities to obtain the answer.
5. Let x denote the probability of obtaining exactly one head on one flip. Find the probability
of obtaining exactly one head on five flips and the probability of obtaining exactly two heads
on five flips, and set the results equal to each other.
6. We are essentially trying to find the number of sequences a ≤ b ≤ c ≤ d, where a, b, c, d ∈

{1, 2, 3, 4, 5, 6}. This can be done in cases, delineated according to the number of distinct
numbers in the set {a, b, c, d}.
7. If E1 denotes the event that there is an undefeated team and E2 denotes the event that there
is a winless team, we are trying to find p(E1 ∪ E2 ), which can be simplified by using Theorem
3.2.1. The choices involve which team is undefeated (or winless) along with outcomes of games
not involving this team.
8. There are eight possible outcomes of tossing three coins. Compute the probability of each of
these.
9. Assume that the six chosen numbers a1 , a2 , a3 , b1 , b2 , and b3 are 1, 2, 3, 4, 5, and 6 (the six
numbers are all distinct since they are chosen without replacement). One can then enumerate
all of the acceptable dimensions for the brick and the box.
10. Use the stars and bars method discussed in Chapter 2, with the positions of the heads denoting
the bars.
11. After the game between team A and team B, it is required that team A obtains at least as
many additional wins as team B. Enumerate all of the cases in which this occurs, along with
the number of different ways in which this could occur.
12. Make a table that shows the probability that the bug is at each vertex after the nth move,
starting with n = 0, 1, 2, . . . , 10.
13. Draw a graph whose horizontal axis is labelled with the number of shot attempts and the
vertical axis is labelled with the number of made shots. The player’s first ten shots trace
points along this graph (starting at (0, 0) and ending at (10, 4)) such that the ratio of made
shots to shot attempts stays below 0.4 and is equal to 0.4 after the tenth shot. In particular,
the player must miss the first two shots and make the last shot.
14. For no two teams to win the same number of games, it must be the case that
for each k =
40
0, 1, 2, . . . , 39, there is exactly one team that wins k games. A total of games occur;
2
how many of them meet the condition just described?
15. A “successful” string (in which a run of 5 heads is encountered before a run of two tails) either
begins with a head or with a tail. Consider these cases separately.
1. Of the sixteen possible (ordered) outcomes when Terry and Mary draw two candies each in
turn, exactly six of them result in Terry and Mary obtaining the same color combination.
These six outcomes are of two different types. Find the probabilities of each type.
2. For each of the nine denominations that still have four cards in the deck after two of the
cards were removed, there are six pairs that can be constructed. What about for the other
denomination?
3. The total number of ways that three people (in order) can select three tiles each can be
expressed as a product of combinations – this product is needed as the denominator of the
probability being computed in this problem. For the numerator, imagine that the first player
draws three odd tiles. In how many ways can this happen? Now determine the number of ways
that the second person can obtain one of the remaining odd tiles and two even tiles. Finally,
apply the Multiplication Principle.
4. Only four gates are within 400 feet of a gate at the end of the line, while more than four
gates are within 400 feet of a gate towards the middle of the line. It may help to write out a
table showing how many gates are within 400 feet of each gate. Your work can be reduced by
exploiting symmetry in this problem.

5
5. The probability of obtaining exactly k heads on five flips of this coin is p(x, k) = xk (1 −
k
x)5−k . Set p(x, 1) = p(x, 2) and solve for x.
6. Suppose the four rolls of the die yield k distinct outcomes (for some k ≤ 4). For each such
k, multiply the number of ways to choose which of the k outcomes occur by the number of
non-decreasing sequences of length 4 that have those k outcomes. Then apply the Sum Rule
to finish.
7. To compute |E1 |, for example, note that once an undefeated team is chosen (how many choices
here?), there are only six games remaining that do not involve this team. How many outcomes
can occur for this collection of six games? Multiply the choices that arise here. (Note that all
of these outcomes are equally likely.) Do a similar analysis to find |E2 | and |E1 ∩ E2 |.
8. Compute the probability of obtaining exactly k heads on three flips of the coin for each k =
0, 1, 2, 3. Then use the fact that Jackie and Phil perform these flips independently.
9. For the brick to be encloseable in the box, there is a condition on the two sets of dimensions
{a1 , a2 , a3 } and {b1 , b2 , b3 }. What is it? Try to enumerate all of the ways to successfully meet
this condition. You can simply use the numbers {1, 2, 3, 4, 5, 6} in doing this enumeration.
10. Use variables, say xi , to denote the number of tails between consecutive heads among the coin
flips. It will yield an equation of the form x1 + x2 + · · · + xh+1 = 10 − h, where h is the number
of heads. We must have x2 , x3 , . . . , xh ≥ 1 (why?). Now use stars and bars to solve.
11. The number

of
waysthat team A can have k additional wins while team B garners ` additional
5 5
wins is . Sum this value for each 0 ≤ ` ≤ k ≤ 5.
k `
12. Note that p0 = 1, q0 = 0, and r0 = 0. Thus, using the fact that pn = 12 qn−1 + 12 rn−1 for n ≥ 1
(and similar formulas for qn and rn ), one can simply compute p1 , q1 , r1 , p2 , q2 , r2 , and so on
until p10 is found.
(An alternative approach is to instead keep track of the number of clockwise and counter-
clockwise moves that occur in the 10-move process.)
13. From a graph of shots made versus shots attempted, observe that there are exactly 23 legal
paths from (0, 0) to (10, 4). What is the probability of the player’s path on this graph following
one of these 23 legal paths?
14. We are assigning each of 40 teams to a different number of wins in the interval 0–39. In how
many ways can this assignment be made?
15. Let pH denote the probability of obtaining a successful string that begins with H, and similarly
for pT . Now determine equations for pH in terms of pT and vice versa. Solve these equations
to find the values of pH and pT .
1. Write b = ak and c = a` for positive integers k and `. Observe that a must divide 100 and
enumerate cases according to the positive divisors of 100.
2. Write P using factorials and powers of 2.
3. Complete the square on the left-hand side, rearrange, and factor.
4. The right-hand side is a repeating decimal, hence can be written as a rational number. Cross
multiply to obtain an equality of integers and apply the Fundamental Theorem of Arithmetic.
5. Use Equation (4.3) for each factor k! in the product given (with k = 1, 2, 3, . . . , 100). This
enables us to express N as a double summation.
6. Start by computing T := p(1) + p(2) + p(3) + · · · + p(99). How is this sum related to p(1) +
p(2) + p(3) + · · · + p(999)?
7. Start by finding the values of a with 1 ≤ a ≤ 100 that generate a desired pair (a, b). Repeat
for each interval of 100 values of a.
8. Determine, in terms of x, y, and z, on how many of the 1000 steps that a particular switch
2x 3y 5z will be switched. Under what conditions on x, y, and z will this answer be a multiple
of four?
9. The first several sets S0 , S1 , S2 will contain several perfect squares. However, there comes a
point after which at most one perfect square can belong to the subsequent sets—what is that
point? And how many perfect squares are left to consider?
10. Write the prime factorization of 20042004 , the prime factorization of a typical divisor d, and a
condition on the prime factorization of d that is required in order for it to have exactly 2004
divisors.
11. The point labelled with 1993 is (1993)(1994)

2 −1 points clockwise around the circle from the point
labeled 1. Therefore, you need to find the smallest positive integer k such that k(k+1)2 −1 ≡
(1993)(1994)
2 − 1 (mod 2000).
12. Every number in the sum S can be written in the form xy, where x = 2k and y = 2` for a
suitable range of values of k and `.
13. Suppose the k consecutive positive integers are n + 1, n + 2, . . . , n + k. Write an expression for
311 in terms of n and k and appeal to the Fundamental Theorem of Arithmetic.
14. Under what condition is τ (n) odd? Therefore, for which n does S(n) switch parity (even to
odd, or odd to even)? Consider “streaks” of values S(i), S(i + 1), . . . , S(i + j) that are all odd
or all even.
15. When non-factorial tails arise, it is in moving from k! to (k + 1)! when (k + 1)! is divisible
by 25. Divisibility by 125, 625, or 3125 results in even more non-factorial tails arising in this
move. Ultimately, one needs to determine the integer x such that x! ends in 1992 zeros. Some
experimentation may help.
1. For each positive divisor a of 100, how many pairs (k, `) of integers satisfy k, ` ≥ 1 and
1 + k + ` = 100/a ? Now sum over all possible values of a.
200!
2. The number P can be written as 2100 ·100! . Now use Equation (4.3) to determine the number
of factors of 3 in 100! and 200!.
3. After completing the square, rearranging, and factoring on the left-hand side, observe that
both factors obtained have the same sign and the same parity (both are even, or both are
odd). Use the Fundamental Theorem of Arithmetic to derive two equations involving x and y
that can be solved for x and y.
4. If x = 0.d25d25 . . . , then d25 = 999x (show why). Write an integer equality involving n and x
and factor both sides of the equality. Note that any integer ending in “25” is divisible by 25,
and use the Fundamental Theorem of Arithmetic to find another two-digit integer divisor of
of d25.
5. Reverse the order of summation of the double sum obtained by applying Equation (4.3). Since
k ≤ 100, note that bk/5` c = 0 for all l ≥ 3. It only remains to compute bk/5c and bk/25c for
k = 1, 2, 3, . . . , 100.
6. Note that p(101) + p(102) + · · · + p(199) = T and p(201) + p(202) + · · · + p(299) = 2T , and
so on. Thus, one can readily use the value of T to compute the sum S. To compute T , it is
possible to apply a similar trick by starting with p(1) + · · · + p(9).
7. Find the possible values of a with 1 ≤ a ≤ 999 by considering the intervals 1 ≤ a ≤ 100,
then 101 ≤ a ≤ 200, then 201 ≤ a ≤ 300, and so on. The first interval needs to be handled
especially carefully, but the other intervals will all yield the same results. Applying symmetry
will shorten the amount of work involved.
8. The switch labelled 2x 3y 5z will be switched exactly (10 − x)(10 − y)(10 − z) times. (Why?)
To determine for how many (x, y, z) this is a multiple of four, do a case analysis based on how
many of x, y, z are odd.
9. There are 316 perfect squares lying in some set Si , since 3162 < 99999 but 3172 > 99999. Once
the difference between consecutive perfect squares n2 and (n + 1)2 exceeds 100 (what is the
smallest n for which this is the case?), then the numbers n2 , (n + 1)2 , . . . , each belong to a
different set Si .
10. Considering a divisor d = 2a 3b 167c of 20042004 , use Theorem 4.2.2 to generate an equation that
(a, b, c) must satisfy in order for d to have exactly 2004 positive divisors. Use combinatorics
to count the number of such d so generated.
11. Modulo 4000, we have k(k + 1) ≡ 1993 · 1994, so modulo 2000, we have k(k + 1) ≡ 42. Solving
this quadratic congruence equation and use the fact that 2000 = 24 · 53 .
12. Express S as a double summation using k and `. Factor this summation into two factors that
are each finite expressions that can be readily computed.
13. Write 2 · 311 = nk + k(k+1)

2 and use the Fundamental Theorem of Arithmetic to argue that k
must have the form k = 2i 3j where 0 ≤ i ≤ 1 and 0 ≤ j ≤ 11. Now consider these 24 values
of k to find the largest one that works.
14. Consecutive perfect squares n2 and (n + 1)2 differ by 2n + 1, so the odd and even “streaks” of
S(n) values will be of length 3, 5, 7, 9, and so on. There will be 22 odd streaks and 22 even
streaks (why?). How long are each of these streaks?
15. With a little experimentation, observe that there are 1973 zeros at the end of 7900!, while
there are 1998 zeros at the end of 8000!, so the value of x must be between 7900 and 8000.
Use experimentation to find x = 7980. Now use Equation (4.3) with p = 5 to find the number
of non-factorial tails less than 1992.
1. Write the five terms as p1 , p1 + k, p1 + 2k, p1 + 3k, and p1 + 4k. Try small values of k and
apply modular arithmetic.
2. Group terms together so as to create differences of squares.
3. Use the formula for an infinite geometric series for both the original series and the new series.
4. Compute a few additional terms by hand and look for a pattern.
5. Compute several of the early terms of the sequence in terms of a1 and a2 until a pattern
emerges.
6. Use a partial fractions decomposition of the form
1 A B
= + .
n2 − 4 n−2 n+2
7. Use the given information to write three equations involving three unknown quantities. Routine
algebraic manipulations can be used to solve for these quantities.
8. Since every term of the sequence can be written in the form
an 3n + an−1 3n−1 + an−2 3n−2 + · · · + a2 · 32 + a1 · 3 + a0 , (10.2)
where each coefficient a0 , a1 , a2 , . . . , an is either 0 or 1, there is a one-to-one, ordered corre-

spondence between the terms of the sequence at bit strings of zeros and ones.
9. Instead of computing a2 , a3 , a4 , and so on using the relation, try enumerating the terms in
reverse: am , am−1 , am−2 , and so on.
10. Begin by writing equations for x1 , x2 , etc. in terms of the other terms of the sequence. Now
sum both sides of all 100 equations.
11. A “brute force” approach in which the first 28 terms of the sequence are explicitly computed
(modulo 1000) will work. One can keep the numbers fairly small by exploiting negative numbers
(by using the modulus of 1000).
12. Let the first three terms of the sequence be labeled as a1 = 1, a2 = r, and a3 = r2 . Determine
additional terms a4 , a5 , a6 , a7 , etc. in terms of r. Then solve for r.
13. Start by determining the number of values of p for which b(p) = k for a fixed positive integer k.
Use this to rewrite the given summation for S as a different summation that can be evaluated
directly.
14. For 1 ≤ k ≤ 1995, observe that f (k) ∈ {1, 2, 3, 4, 5, 6, 7}. One needs to determine for how
many k we have f (k) equal to each one of these values.
15. Find an expression for x2k − x2k−1 and formulate a telescoping series.
1. The common difference k must be even (why?). Consider the cases k = 2 and k = 4 first
and show using modular arithmetic that, in both cases, it is impossible to obtain a sequence
consisting entirely of primes.
2. Simplify 992 − 982 , then 972 − 962 , then 952 − 942 , and so on. Group terms in the form
(4n + 3) + (4n + 2) − (4n + 1) − (4n) and simplify further.
3. The two infinite geometric series give rise to two equations and two unknowns. One of those
unknowns is the common ratio of the original series, which we can now solve for.
4. By using the given relation, compute x5 , x6 , x7 , . . . until you observe that the terms of the
sequence repeat cyclically. Use this observation to help you write x531 , x753 , and x975 in terms
of early terms within the sequence.
5. Note that the terms of the sequence repeat cyclically. What do the terms within each cycle
sum to? This will help you write the sum of the first 1492 terms and the sum of the first 1985
terms using only a1 and a2 . Then you can solve for a1 and a2 .
6. A telescoping series emerges from the partial fractions decomposition. The series thus reduces
to a small number of (uncancelled) terms to analyze.
7. The three equations can be written as 36 + k = a2 , 300 + k = (a + d)2 , and 596 + k = (a + 2d)2 ,
which has three unknowns. First eliminate k to reduce the problem to solving a system of two
equations for two unknowns (a and d).
8. The 100th largest term of this sequence can be found by writing 100 in base 2 and putting the
values (0 and 1) that arise into the corresponding coefficients of the expression (10.2).
9. The reverse enumeration of terms generates equations am−1 am−2 = 3 and am−2 am−3 = 6, and
more generally, am−k am−k−1 = 3k for positive integer k. Now substitute a suitable choice for
k.
10. Using S = x1 + x2 + · · · + x100 , derive S = 99S − 5050. Find S and use the 50th equation in
the array in order to find x50 .
11. A sophisticated solution rewrites the recurrence relation as ak+3 − ak+2 = ak+1 + ak . Now
sum both sides from k = 1 to k = n. On one side, telescoping behavior can be exploited.
12. The pattern that emerges from computing a4 , a5 , a6 , a7 , etc. is that for postive integers n,
a2n = (nr − n + 1)((n − 1)r − n + 2) and a2n+1 = (nr − n + 1)2 .
Now use given information to solve for r and then simplify these formulas for a2n and a2n+1 .
Finally, experiment with values of n until the desired result is obtained.
2007
X
13. Verify that b(p) = k for precisely 2k different values of p. Now write b(p) in the form
p=1
X
2k 2 and determine the appropriate range of values of k to sum.
√
14. We have f (k) = ` if and only if ` − 21 < 4 k < ` + 12 . For ` = 1, 2, 3, 4, 5, 6, 7, how many integers
k with 1 ≤ k ≤ 1995 satisfy the inequalities?
15. Square both sides of |xk | = |xk−1 + 3| and obtain a formula for x2k − x2k−1 . Sum this telescoping
expression from k = 1 to k = 2007 to obtain a formula for x1 + x2 + · · · + x2006 in terms of
x2007 . Now determine x2007 such that |x1 + x2 + · · · + x2006 | is minimized.
1. Observe that
blog2 nc = 2k if and only if 22k ≤ n < 22k+1 (10.3)
for each positive integer k.
2. Apply the base 1995 logarithm to each side of the given equation and apply the properties
given in Theorem 6.2.2.
1
3. Rewrite cot θ as and apply Equation (6.32).
tan θ
4. Make a table showing, for a given integer ` ≥ 0, the number of values of k such that blog2 kc = `.
Perform summation over k.
5. The terms on the left side of the second equation can be expressed in terms of the terms on
the left side of the first equation. Therefore, we can view this as a system of two equations
with two unknowns.
6. This problem mainly involves clever use of trigonometric identities. Start by multiplying out
the two products given in the question, and compare the results to one another.
7. The given expression is well-suited for an application of trigonometric formulas found in Section
6.3.
8. Eliminate the logarithms and roots and derive a cubic equation satisfied by sin x. Once the
value of sin x is determined, use the trigonometric identities to find cot x.
9. Use the trigonometric identity 1 + tan2 x = sec2 x.
10.
11. Write cos n◦ = sin(n + 45)◦ , and use one of the trigonometric identities in the chapter.
12. Use logarithm properties to rewrite the given strict inequality without logarithms. It is most
helpful to obtain an explicit inequality for k. From this inequality, can you write exactly what
the 50 integers that k can be are in terms of m and n?
13. Write out some half-open intervals of values of x and y on which the “even” requirement on
the logarithm floors are satisfied. Use a geometric series to sum the widths of these half-open
intervals in each direction.
14. Write xn = 3a+bn for suitable integers a and b. The given summations impose limitations on
the values of a and b. Apply logarithm properties to generate inequalities that force a unique
solution for a and b.
15. Multiply the given equation through by sin 1, and use the fact that
sin 1 = sin((x + 1) − x) = sin(x + 1) cos x − cos(x + 1) sin x.

1. Since 210 = 1024, observe that for n < 1000, we have blog2 nc < 10. Thus, the only positive
even values of k that must be considered in (10.3) are k = 2, 4, 6, and 8. Count the number of
values of n that occur in each interval of (10.3) for k = 2, 4, 6, and 8.
2. Taking the base 1995 logarithm of each side of the given equation and applying Theorem 6.2.2,
we obtain 21 + (log1995 x)2 = 2 log1995 x. Use the quadratic equation to find two positive roots
to the equation.
3. According to Equation (6.32), to evaluate tan(x+y), we must find tan x+tan y and tan x tan y.
The former value is given in the problem. The latter value can be found by rewriting cot x +
cot y = 30 using tan x and tan y.
m for precisely 2m values of k. Thus, the sum we are computing can be

4. Note that blog2 kc =X
viewed in the form m2m , with a few extra terms that must be added to the end of this
m
summation. Determine the range of values of m that must be summed.
5. Use the change-of-base formula (6.5) to rewrite the given system of equations in the form
a + b = 4 and a1 − 1b = 1, which can be routinely solved for a and b.
√
6. Write α := mn − k. Now add the two given equations to obtain a relationship between t and
α. Next, multiple the two given equations to obtain another relationship between t and α.
Now we have two equations, and two unknowns, t and α. Now solve for α.
7. Use (6.28) to expand sin(k 2 a + ka). One of the terms of this expansion has the right form
for the terms of the given expression, and one does not. Can you also use (6.28) to get the
unwanted term of this expansion to cancel? Once this is done, the given expression can be
re-expressed as two summations put together. Write out the first few terms of these two
summations – what do you notice?
8. Square both sides of the equation (24 sin x)3/2 = 24 cos x and obtain a cubic equation satisfied
by sin x. The Rational Roots Test can be used to find solutions for sin x. Then cot2 x can be
written solely in terms of sin2 x.
9. The trigonometric identity 1 + tan2 x = sec2 x together with the given equation sec x + tan x =
22
7 can be used to solve explicitly for sec x and tan x. From these values, csc x and cot x can
be found by uing the relationships between the different trigonometric functions.
10. We have xk = 4y 2 , (2x)k = 16z 2 , and (2x4 )k = 8yz. There is a relationship between the
right sides of these three equations; find it and then use it to determine the value of x (it does
not depend on k). Substitute this value of x into the three equations above, and now try to
determine xy 4 z from the three equations. It is not necessary to find explicit values of y and z.
√
11. Write sin(n + 45)◦ = 2
2 (sin n
◦
+ cos n◦ ), and use this to break the expression for x into two
terms.
12. Starting with the given inequality, Using properties of logarithms and performing exponenti-
ation, we find that n1 < m m
k < n. Isolating k, this becomes n < k < mn. Then mn − 51 ≤
m
n < mn − 50 (why?). Use n ≤ m to eliminate m and obtain an inequality that only involves
n. This will reduce the problem to a consideration of a small number of values of n.
13. Add up the widths of the intervals ( 12 , 1], ( 18 , 14 ], and so on, by using the formula for the sum of
an infinite geometric series given in Equation (5.12). Do a similar thing for intervals obtained
with respect to the variable y.
14. Note that log3 (xn ) = a + bn for all n. Thus, the given equation can be used to deduce
that 8a + 28b = 308. Similarly, the given inequality gives (check this!) 56 − a ≤ log3 (1 +
3b + 32b + · · · + 37b ) ≤ 57 − a. Now use the fact that log3 (x) is an increasing function and
37b < 1 + 3b + 32b + · · · + 37b < 8 · 37b to pinpoint more clearly the relationship between a and
b.
15. Using Equation (6.28), we can write
sin 1 sin((x + 1) − x)
=
sin x sin(x + 1) sin x sin(x + 1)
sin 1
and thus rewrite sin n by using the cotangent function. Finally, use the fact that cot theta +
cot(180 − θ) = 0 for all θ.
1. Substitute an expression for z into the given equation, cross multiply, and equate real and
imaginary parts of the resulting equation.
2. Consider the real and imaginary parts of (a + bi)3 − 107i, and use the fact that 107 is prime.
3. What is the degree of polynomial on the left-hand side? For each root r of the polynomial,
find another root to naturally pair with it. What is the sum of the two roots in each pair?
4. Begin by writing P20 (x) in terms of P19 (x), then replace P19 (x) in terms of P18 (x), then replace
P18 (x) in terms of P17 (x), and so on.
5. Use the fact that (x − 1)4 = x4 − 4x3 + 6x2 − 4x + 1.
6. Write w = u + vi, and use Proposition 7.4.3 to get expressions for a and b in terms of w. Now
use the fact that a and b are real to determine u and v.
7. Assume that P (x) is monic, and write
1
P (x) = (x − a)Q1 (x) = (x − b)Q2 (x), (10.4)
2
for constants a and b. Expand the latter two expressions and compare with P (x).
8. Manipulate the given equation into a form such that Euler’s Formula can be applied.
9. The mean line in question has an equation of the form y = mx + 3 and must contain five points
zk = xk + iyk (where k = 1, 2, 3, 4, 5).
10. What extra terms need to be added to S8 to obtain S9 ? These extra terms are expressible
using the terms comprising S8 as well, so S9 can be written in terms of S8 .
11. Compute the first few terms of {zn } and look for a pattern.
12. The task is to sum all integers k of the form
k = a3 (−3 + i)3 + a2 (−3 + i)2 + a1 (−3 + i) + a0 , (10.5)
where a0 , a1 , a2 , and a3 belong to the set {0, 1, 2, 3, 4, 5, 6, 7, 8, 9}. Consider the real and imag-
inary parts of any such k.
13. Use long division to divide ax17 + bx16 + 1 by x2 − x − 1. Determine conditions on a and b
necessary for the remainder to be zero.
14. Consider roots of the polynomial S(x) = P (x) − x − 3 and use divisibility.
15. First show that a, b, c 6= 0 and assume without loss of generality that |a| ≥ |b| ≥ |c| > 0 and
that a > 0. Factorize the cubic polynomial and equate coefficients. Use the resulting equations
to put limitations on the range of integer values for a and then test each value in the range to
find the one that works.
1. Write z = a + 164i. After substitution and cross multiplication in the given equation, we
obtain a + 164i = −656 + 4i(a + n). Now form two equations involving a and n by equating
the real and imaginary parts.
2. After expanding (a + bi)3 − 107i, obtain equations for the real and imaginary parts. One of
the equations obtained is b(3a2 − b2 ) = 107. Since 107 is prime, this gives two possible values
of b, and only one of these is actually possible.
3. Note that if r is a root of the polynomial on the left-hand side, then so is 1 − r. Since r 6= 1 − r
(why?), this produces pairs of roots to the polynomial. How many such pairs are there?
4. By writing P20 (x) in terms of P19 (x), then in terms of P18 (x), and so on, eventually obtain
P20 (x) = P0 (x − 210). Apply the Binomial Theorem to each term, keeping track of the
coefficient of x in each of these terms.
5. We need find the nonreal solutions of (x − 1)4 = 2006. Note that 2006 has two real fourth
roots and two nonreal fourth roots. What are they?
6. Use Proposition 7.4.3 to obtain a = 4 − 12i − 4w and b = 5w2 − 8w + 36wi − 48i − 27. The
imaginary parts of both of these expressions must be zero, from which we find w = 4 − 3i. This
enables us to write out all three roots of this polynomial and appeal once more to Proposition
7.4.3 to obtain the desired information.
7. Expand the expressions in Equation (10.4) to derive three equations relating a, b, and k by
comparing coefficients. Now solve for a, b, and k.
8. To apply Euler’s Formula, the imaginary number i should be multiplied by the sine function,
not the cosine function. To “correct” this, multiply both sides of the given equation through
by (−i)n . Now simplify and use Euler’s Formula.
9. Compute z1 + z2 + z3 + z4 + z5 , and use the result to find x1 + x2 + x3 + x4 + x5 and
y1 + y2 + y3 + y4 + y5 . How are xi and yi related?
10. To find S9 , it suffices to compute complex power sums for subsets of {1, 2, . . . , 9} that contain
“9”, since all other complex power sums are terms in S8 , whose value we already know.
The complex power sum of {a1 , a2 , . . . , ak , 9} is obtained from the complex power sum of
{a1 , a2 , . . . , ak } by adding 9ik+1 .
11. Observe that z3 = z0 .
12. Expanding Equation (10.5), we find that the imaginary part off the expression is 26a3 −6a2 +a1 ,
which must be 0. Use this to write a1 in terms of a2 and a3 . Thus, k can be expressed in
terms of a0 , a2 , and a3 . Now use the fact that these variables must all assume integer values
to reduce to an analysis of a small number of cases.
13. Upon dividing ax17 +bx16 +1 by x2 −x−1, the quotient will be a polynomial whose coefficients
are an integer linear combination of a and b with coefficients that follow the Fibonacci sequence.
It may be useful to produce a table to keep track of the coefficient of xk for each k ≤ 15. The
remainder, r(x) is a linear (degree 1) polynomial whose two coefficients are each expressed in
terms of a and b. This generates two equations for the unknowns a and b. Therefore, a and b
can be determined.
14. Both n1 and n2 are roots of S(x). Use Equation (7.16) to write an expression for S(x). Now
evaluate S(x) with each of the two formulas at x = 17 and x = 24. Finally, exploit knowledge
pertaining to divisibility of integers.
15. Using Equation (7.16) to write the polynomial as (x − a)(x − b)(x − c), we can use√Proposition
7.4.3 to write three equations involving a, b, and c. Use these to deduce that a > 2011. Also
use the fact that bc is maximized when b = c to show that a < 52. Now there are only finitely
many values of a to check, and only one of these is possible.
1. If GE crosses the square at B and GM crosses the square at C, then triangle GBC is similar
to triangle GEM .
2. Extend the segments AE and DF to an intersection point G. Now examine ∆DAG.
3. The given figure has eight congruent right triangles. How are the lengths of the legs of each
of these triangles related?
4. If we set BE = r, then since ∆AEF is equilateral, we use AE = EF to solve for r. Use similar
triangles to relate r to the side length of the smaller square.
5. Extend AB and CD to an intersection point E. Use similar triangles and the Law of Cosines
to find EM and EN .
6. Drop perpendiculars from both C and D to AB. This decomposes AB into three segments.
Find the lengths of each of the three segments. Alternatively, one could approach this problem
by extending segments AD and BC to an intersection point.
7. Split the given angle ∠CAB into the sum of two angles and use the formula for the tangent
of a sum of two angles, Equation (6.32).
8. The center of the circle must be at least one unit from the edge of the rectangle, and in order
that the circle does not intersect AC, it must be at least one unit from AC. The area that the
center can exist in consists of two congruent triangles. Find the area of these triangles.
9. Draw the figure and identify similar triangles. The segment AH is decomposed into three
sub-segments, one of which is the key length we must determine to answer this question. The
lengths of all segments can be found by utilizing the similar triangles.
10. Note that d = DP + EP . Use similar triangles to determine DP and EP in terms of d.
11. The distances from a vertex of a triangle and the two points of tangency of an inscribed circle
are equal (the Two Tangent Theorem, Theorem 8.3.6).
12. One approach is to place vertex C at (0, 0) and use the slopes of the lines in the figure to find
the coordinates of D and M .
13. The Law of Cosines can be used to determine the lengths of all sides and diagonals of the
figure.
14. Since several segments have the same length in this problem, it is possible to make isosceles
triangles with relative ease. (For instance, ∆QBP is isosceles.) Let α = m∠QBP and let
y = m∠P BC. Try to determine as many other angles in the figure as possible. Then apply
the Law of Sines.
15. Use Huron’s Formula to find [ABC], and then use Equation (8.14) to find [ABD] and [ACD].
It suffices to find [ABE].
1. Since ∆GBC is similar to ∆GEM , the ratio of lengths of corresponding sides. What is the area
of ∆ABE? Use this information to find AB, and by symmetry, CI. Now you can determine
BC, which has corresponding side EM in ∆GEM .
2. If we extend segments AE and DF to an intersection point G, then ∆DAG is actually con-

gruent to ∆CDF and ∆ABE. In particular, this enables one to determine AG and GD. Now
apply the Pythagorean Theorem.
3. Let J denote the vertex of one of the eight congruent right triangles, the one with AB as the
hypotenuse. Let AJ = x. Write BJ is terms of x and apply the Pythagorean Theorem on
∆ABJ. Write [ABJ] in terms of x, and then relate [ABCDEF GH] to [ABJ].
4. Let G lie along AB and H lie along AE such that the smaller square is BGHE, say with side
length s. Since ∆AGH and ∆ABE are similar (why?), we can set up a proportion relating
r = BE to s. Once we compute r, we will therefore know the value of s.
5. Note that ∆EBC and ∆EAD are similar triangles (why?). The enables us to set up a ratio
for EM/EN . Then M N = EN − EM . We can determine EN by applying the Law of Cosines
to ∆EN D. This will require us to determine the length ED, which can be done by observe
that ∠E is a right angle (why?).
6. Label the points E and F where the perpendiculars from C and D (respectively) meet AB.
Simple trigonometry can be used to compute AF and BE, so it only remains to find EF . To
do this, drop a perpendicular from C that meets DF at G to create rectangle CEF G. Now
EF = CG, and we can compute CG by identifying it as one leg of the right triangle CDG.
7. Let D denote the point where the altitude from A meets BC and set x := AD. To use Equation
(8.14) for the area of ∆ABC with base BC (of length 20), we only need to find x. To do this,
observe that tan(∠CAB) = tan(∠CAD + ∠DAB). Now both tan(∠CAD) and tan(∠DAB)
can be expressed in terms of x and Equation (6.32) can be applied.
8. Denote the vertices of one of the two (right) triangles whose area must be determined by E,
F , and G, such that the right angle is at G. To find [EF G], compute EG and F G. The slope
15 5
of the segment EF is − = − . Thus, ∆EF G is a 5-12-13 triangle. By extending the
36 12
segment EF to points where it intersects the rectangle ABCD, and by drawing perpendicular
segments from E and F to the nearest side of the rectangle, additional 5 − 12 − 13 triangles
can be formed.
9. Let X denote the intersection of AH and F G, and let Y denote the intersection of AH and BI.
If we assume that the sides the hexagon ABCDEF each have side length 1, then the ratio in
question that we must compute is simply XY 2 . Note that XY = AH − AX − Y H. Compute
AH by applying the Law of Cosines to ∆ABH. To find Y H, use the fact that ∆BHY and
∆AHB are similar triangles. The similarity of these triangles can also be used to compute
AX, since AX = BY .
10. Consider AC = AD + DG + GC. Now AD + GC = HP + P I = d by considering opposite

sides of parallelograms, so we can write AC = d + DG. Now, we have d = DP + EP , and it
is also possible to write DP and EP in terms of d. More precisely, we can relate DP to DG
by using similar triangles ∆DGP and ∆ACB. Putting this information together, the value of
DP can be obtained. A similar process will yield the value of EP .
11. Label the point where BE meets the inscribed circle of ∆BEF as G, and label the point
where BC meets the inscribed circle of ∆DEF as H. Then EG = EP (by the Two Tangent
Theorem), and several other similar relationships exist in the figure. Write P Q = F P − F Q
and replace F P and F Q with other quantities related by the Two Tangent Theorem.
12. Placing the figure on a coordinate plane, we have A(0, 7), B(24, 0), C(0, 0), M (x1 , y1 ), and
D(x2 , y2 ). Equation (8.16) now gives the formula for [CDM ]. It is easy to find x1 and y1 . To
find x2 and y2 , find the equation of the line that contains the segment M D.
13. Use the Law of Cosines to find AB, AC, AD, AE, and AF . It will be useful to verify that
√ √ √ √ √
√
q
6− 2 2+ 3 6+ 2
2− 3= and .
2 = 2
The last ingredient needed here is the number of segments of each of the six lengths computed.
14. Let θ = m∠(AP Q) and α = m∠(QBP ). Using isosceles triangle QAP , find a relationship
between θ and α. Focus on ∆BP C, where the Law of Sines can be applied.
15. View AB as the base of ∆ABE, and find [ABE] in terms of the ratio AE/AD. We can
determine this ratio by computing [ACE]/[ABD].
1. Find the distance between each pair of the points P , Q, and R. What does this information
reveal about the location of these points?
2. Revolution of the triangle along one of its legs generates a cone whose height is the length of
the leg of rotation and with radius equal to the length of the other leg.
3. If the dimensions of the block are a, b, and c, write an equation, in terms of a, b, and c, for
231, and use the prime factorization of 231.
4. Equation (9.11) gives the volume of a pyramid. The base area is determined from the square
S 0 . Note that the height of the pyramid is can be found as one of the legs of a right triangle
whose hypotenuse is the length of the altitude that meets a side of S 0 of any one of the four
triangles in the figure.
5. Break the set under consideration into various subsets, one of which is the parallelepiped itself.
Other subsets include portions of cylinders and spheres (of radius 1). Draw a picture to assist
in determining what fraction of these shapes (and how many of them) to include.
6. Determine the volume of the liquid in the container, and determine the volume of non-liquid
in the container. Once the cone is held with its point up, one can determine the height of the
cone of non-liquid resting above the liquid.
7. Treat side ABC as the base of the tetrahedron, and use the volume formula (9.11).
8. There are four different types of unit cubes: (1) unit cubes in the center of each face, (2) unit
cubes along the center of an edge between two faces, (3) unit cubes lying in a corner of the
larger cube, and (4) the one unit cube that is hidden in the center. Determine how many cubes
fall into each case, and find the probability of all exposed faces being orange for each type.
9. Draw a picture of the region R, and determine important points defining the boundary of R.
One of the edges of R is the line of revolution, and the resulting solid is conical. By computing
distances between points, determine the radius and height of the cone.
10. The centers of the eight spheres form an octagon – what is the edge length? Find the center
of the octagon and the radius of the ninth sphere that is placed in the center. The center of
the ninth sphere must lie somewhat above the plane of the octagon. Drawing some pictures
will help visualize this.
11. Imagine cutting the cone along a straight line running along the side of the cone, from the
point at the top to a point on the base, through the point where the fly starts. The surface of
the cone can then be laid flat as a sector of a circle. What is its radius and how large is the
sector?
12. Needs to be filled in!

Chapter 11
Solutions to Exercise Sets
11.1 Chapter 1 Solutions
1. (2008 AIME, Problem #1) Of the students attending a school party, 60% of the
students are girls, and 40% of the students like to dance. After these students are
joined by 20 more boy students, all of whom like to dance, the party is now 58%
girls. How many students now at the party like to dance?
Solution #1: Let g denote the number of girls initially at the party, and let b denote the
2
number of boys initially at the party. Note that b = g, since the ratio of boys to girls initially
3
at the party is 40:60. Since 58% of the party attendees are girls after 20 more boys arrive, we
have
5
g = 0.58(g + b + 20) = 0.58( g + 20)
3
29
= g + 11.6.
30
Thus,
2
g = 30(11.6) = 348 and b= (348) = 232.
3
(Observe that, up until now, we have made no use or reference to information given in the
problem about students who like to dance. We have isolated our focus on the easier task of
determining the number of boys and girls at the party.)
The number of students initially at the party who like to dance is thus 0.4(348 + 232) = 232.
After 20 more boys (who all like to dance) arrive, we have 232 + 20 = 252 students who like
to dance at the party. The answer is 252. 2
Remark: In Solution #1, we explicitly determined the number of girls and the number of
boys at the party. As our next solution shows, this information is actually unnecessary. The
217
218 CHAPTER 11. SOLUTIONS TO EXERCISE SETS
information required is only the value of g + b, the total number of persons initially at the
party.
Solution #2: Let g and b be as above in Solution #1. We are given that
g = 0.6(g + b) and g = 0.58(g + b + 20).
Setting these equations equal to each other, we have
0.6(g + b) = 0.58(g + b + 20).
That is, 0.02(g + b) = .58(20) = 11.6. Hence,
g + b = 50(11.6) = 580.
Therefore, 580 persons are at the party initially, and 40% of them like to dance. That is,
0.4(580) = 232 persons at the party initially like to dance. All 20 of the boys who arrive
afterwards like to dance, bringing the total number of persons that like to dance at the party
to 232 + 20 = 252. 2
2. (2008 AIME-2, Problem #2) Rudolph bikes at a constant rate and stops for a five-
minute break at the end of every mile. Jennifer bikes at a constant rate which is
three-quarters the rate that Rudolph bikes, but Jennifer takes a five-minute break
at the end of every two miles. Jennifer and Rudolph begin biking at the same
time and arrive at the 50-mile mark at exactly the same time. How many minutes
has it taken them?
Solution: Let us assume that Rudolph bikes at a rate of r miles per mile, while Jennifer bikes
at a rate of j miles per minute. Note that j = 43 r, since Jennifer’s biking rate is three-quarters
of Rudolph’s biking rate.
Let us begin by computing the amount of biking time (ignoring break time) required for each
of the bikers to reach 50 miles. We can compute the time from the formulas
Distance 50
Rudolph’s Biking Time = = ,
Rate r
and
Distance 50 200
Jennifer’s Biking Time = = = .
Rate j 3r
Now, Rudolph takes a total of 49 rest breaks prior to reaching the 50-mile mark (one five-
minute break after each mile from 1 to 49). Therefore, Rudolph’s total time, including breaks,
is
50 50
Rudolph’s Total Time = + 5 · 49 = + 245. (11.1)
r r
Now Jennifer takes only 24 rest breaks (one five-minute break after miles 2,4,6,. . . ,48), so
Jennifer’s total time, including breaks, is
200 200
Jennifer’s Total Time = + 5 · 24 = + 120. (11.2)
3r 3r
11.1. CHAPTER 1 SOLUTIONS 219
Since Rudolph and Jennifer arrive at the 50-mile mark simultaneously, we know from (11.1)
and (11.2) that
50 200
+ 245 = + 120.
r 3r
From this equation, we find that
50 50 2
125 = which implies that r= = .
3r 375 15
Hence, we can compute the time required for both bikers by plugging this value of r into either
(11.1) or (11.2). We use (11.1) here:
50
+ 245 = 375 + 245 = 620.
r
2
3. (2007 AIME, Problem #2) A 100 foot long moving walkway moves at a constant
rate of 6 feet per second. Al steps onto the start of the walkway and stands. Bob
steps onto the start of the walkway two seconds later and strolls forward along
the walkway at a constant rate of 4 feet per second. Two seconds after that, Cy
reaches the start of the walkway and walks briskly forward beside the walkway at
a constant rate of 8 feet per second. At a certain time, one of these three persons
is exactly halfway between the other two. At that time, find the distance in feet
between the start of the walkway and the middle person.
Solution: We can measure time t in seconds, starting at t = 0 when Al steps on the walkway.
According to Equation (1.8), after t seconds Al will be A(t) = 6t feet from the start of the
walkway. Similarly, adding Bob’s walking speed of 4 ft/sec to the walkway speed of 6 ft/sec
to get an effective speed of 10 ft/sec, Bob will be B(t) = (6 + 4)(t − 2) = 10(t − 2) feet from
the start of the walkway (for t ≥ 2, since Bob does not enter the walkway until 2 seconds after
Al). Finally, Cy will be C(t) = 8(t − 4) feet from the start of the walkway (for t ≥ 4, since Cy
does not beginning walking until 4 seconds after Al).
Notice that Bob starts moving on the walkway before Cy starts walking beside the walkway,
and Bob moves at a faster rate (10 ft/sec compared with Cy’s 8 ft/sec). Therefore, Bob will
always be ahead of Cy along the walkway. This gives us three possible orderings for Al, Bob,
and Cy at the moment that one of these three persons is exactly halfway between the other
two:
Order Case 1 Case 2 Case 3

First Bob Al Bob
Second Cy Bob Al
Third Al Cy Cy
In Case 1, Cy will be exactly halfway between Bob and Al if B(t) − C(t) = C(t) − A(t); that
is,
10(t − 2) − 8(t − 4) = 8(t − 4) − 6t.
This simplifies immediately to the absurd conclusion 12 = −32.
In Case 2, Bob will be exactly halfway between Al and Cy if A(t) − B(t) = B(t) − C(t); that
is,
6t − 10(t − 2) = 10(t − 2) − 8(t − 4).
This simplifies to −4t + 20 = 2t + 12, so that 6t = 8 or t = 34 . This is a contradiction, however,
since we must have t ≥ 4, since Cy does not begin walking until 4 seconds after Al.
The only remaining possibility, Case 3, must therefore hold. We set B(t) − A(t) = A(t) − C(t).
That is,
10(t − 2) − 6t = 6t − 8(t − 4).
This simplifies to 4t − 20 = −2t + 32. Therefore, 6t = 52 or t = 26
3 . In this case, Al is in the
middle, and the distance Al has travelled from the beginning of the walkway is
26 26
A( )=6· = 52 = 052.
3 3
2
4. (1992 AIME, Problem #3) A tennis player computes her win ratio by dividing the
number of matches she has won by the total number of matches she has played.
At the start of a weekend, her win ratio is exactly 0.500. During the weekend, she
plays four matches, winning three and losing one. At the end of the weekend, her
win ratio is greater that 0.503. What is the largest number of matches she could
have won before the weekend began?
Solution: If t denotes the total number of matches she has played before the weekend began
and w denotes the number of those she won, then since her win ratio is 0.500 before the
weekend, we know that t = 2w. After the weekend, we are given that
w+3 w+3
> 0.503, or > 0.503.
t+4 2w + 4
Cross-multiplication yields
w + 3 > 0.503(2w + 4) = 1.006w + 2.012.
Therefore,
0.988 988
0.006w < 0.988, so that w< = .
0.006 6
Since w must be a whole number, we conclude that

988
w≤ = 164.
6
The answer is 164. 2
5. (1993 AIME, Problem #3) The table below displays some of the results of last
summer’s Frostbite Falls Fishing Festival, showing how many contestants caught
n fish for various values of n.
n 0 1 2 3 ··· 13 14 15
number of contestants who caught n fish 9 5 7 23 ··· 5 2 1
In the newspaper story covering the event, it was reported that
• the winner caught 15 fish;

• those who caught 3 or more fish averaged 6 fish each;
• those who caught 12 or fewer fish averaged 5 fish each.
What was the total number of fish caught during the festival?
Solution: Let x denote the total number of contestants at the festival, and let f denote the
total number of fish caught at the festival. The last two conditions given in the newspaper
story dictate that
f −0·9−1·5−2·7 f − 13 · 5 − 14 · 2 − 15 · 1
=6 and = 5,
x−9−5−7 x−5−2−1
respectively. That is,
f − 19 f − 108
=6 and = 5.
x − 21 x−8
Hence, we have
f − 19 = 6(x − 21) and f − 108 = 5(x − 8). (11.3)
Then,
6(x − 21) + 19 = f = 5(x − 8) + 108.
We conclude that
6x − 107 = 5x + 68.
Hence, x = 175. Thus, using either formula in Equation (11.3), we can determine the total
number of fish f :
f = 5(x − 8) + 108 = 5(175 − 8) + 108 = 835 + 108 = 943.
6. (1986 AIME, Problem #4) Determine 3x4 + 2x5 if x1 , x2 , x3 , x4 , and x5 satisfy the
system of equations given below:
2x1 + x2 + x3 + x4 + x5 = 6
x1 + 2x2 + x3 + x4 + x5 = 12
x1 + x2 + 2x3 + x4 + x5 = 24
x1 + x2 + x3 + 2x4 + x5 = 48
x1 + x2 + x3 + x4 + 2x5 = 96.
Solution: Notice that if we add all five equations together, we obtain
6(x1 + x2 + x3 + x4 + x5 ) = 6 + 12 + 24 + 48 + 96 = 186.
Therefore,
x1 + x2 + x3 + x4 + x5 = 31. (11.4)
Subtracting (11.4) from the fourth equation in the original system gives x4 = 17 and subtract-
ing (11.4) from the fifth equation in the system gives x5 = 65. Therefore,
3x4 + 2x5 = 3 · 17 + 2 · 65 = 181.
2
7. (2004 AIME, Problem #5) Alpha and Beta both took part in a two-day problem-
solving competition. At the end of the second day, each had attempted questions
worth a total of 500 points. Alpha scored 160 points out of 300 points attempted
on the first day, and scored 140 out of 200 points attempted on the second day.
Beta, who did not attempt 300 points on the first day, had a positive integer score
on each of the two days, and Beta’s daily success ratio (points scored divided by
points attempted) on each day was less than Alpha’s on that day. Alpha’s two-day
success ratio was 300/500 = 3/5. The largest possible two-day success ratio that
Beta could have achieved is m/n, where m and n are relatively prime positive
integers. What is m + n?
Solution: Suppose that Beta attempts k points on the first day (where 1 ≤ k ≤ 499) and
earns x points. Then Beta attempts 500 − k points on the second day, and suppose Beta earns
y points on the second day. Since Beta’s success ratio was less on each day than Alpha’s, we
must have
x 160 8 y 140 7
< = and < = .
k 300 15 500 − k 200 10
We are seeking the largest possible two-day success ratio for Beta, which occurs when the
integer x + y is as large as possible. Now we have
8k 7(500 − k) 5
x+y < + = 350 − k.
15 10 6
Since k > 0, the largest possible integer value of x + y is 349. Thus, m = 349 and n = 500,
and we conclude that m + n = 349 + 500 = 849. 2
Some Intuition: Alpha’s success ratio was greater on the second day than the first day (since
140 160
> ). Therefore, Beta’s largest overall success ratio is achieved when he answers the
200 300
most possible questions on the second day (since he can achieve a higher percentage over the
greatest amount of problems). In order for Beta to achieve a percentage less than 160/300 on
the first day but with at least one correct answer, we must have k = 2 and x = 1. Then we
wish to compute the largest y such that
y y 140
= < = 0.7.
500 − k 498 200
This inequality gives y < 348.6, so the largest value of y is y = 348. Hence, Beta’s two day
success ratio is
1 + 348 349
= .
2 + 152 500
8. (2001 AIME, Problem #8) Call a positive integer N a 7-10 double if the digits of the
base-7 representation of N form a base-10 number that is twice N . For example,
51 is a 7 − 10 double because its base-7 representation is 102. What is the largest

7 − 10 double?
Solution: We can write the base-7 representation of N as
N = (· · · a3 a2 a1 a0 )7
= a0 + 7a1 + 49a2 + 343a3 + . . . ,
where a0 , a1 , a2 , a3 , . . . belong to the set {0, 1, 2, 3, 4, 5, 6}. In order for N to be a 7−10 double,
we have
a0 + 10a1 + 100a2 + 1000a3 + · · · = 2(a0 + 7a1 + 49a2 + 343a3 + . . . ). (11.5)
Rearranging Equation (11.5) so that all coefficients are positive, we have that
a0 + 4a1 = 2a2 + 314a3 + · · · + (10k − 2 · 7k )ak + · · · . (11.6)
Since 0 ≤ ai ≤ 6 for each i, the left side of (11.6) can be at most 30 (when a0 = a1 = 6), so
we see that
a3 = a4 = a5 = · · · = 0.
Hence,
a0 + 4a1 − 2a2 = 0.
The largest possible value of a2 is a2 = 6. From this, the largest possible value of a1 is a1 = 3.
Then a0 = 0. Hence, the largest 7-10 double is
N = a0 + 7a1 + 49a2 = 0 + 7 · 3 + 49 · 6 = 315.
2
9. (2004 AIME-2, Problem #5) In order to complete a large job, 1000 workers were
hired, just enough to complete the job on schedule. All the workers stayed on the
job while the first quarter of the work was done, so the first quarter of the work
was completed on schedule. Then 100 workers were laid off, so the second quarter
of the work was completed behind schedule. Then an additional 100 workers
were laid off, so the third quarter of the work was completed still further behind
schedule. Given that all workers work at the same rate, what is the minimum
number of additional workers, beyond the 800 workers still on the job at the end
of the third quarter, that must be hired after three-quarters of the work has been
completed so that the entire project can be completed on schedule or before?
Remark: The foundation of this problem involves the rate at which work is being done, not
the amount of time consumed while doing the work. Therefore, we can specify in advance the
total time to be any arbitrary value we like.
Solution: Assume for simplicity that the job is supposed to be completed in exactly 4 hours1 .
The first quarter of the work gets completed on schedule, which would be one hour. The
1 As stated in the “Remark” prior to the solution, we can assume any interval of time that we like for the large job
to be completed. We choose four hours here so that each quarter of the job, if completed on time, would take exactly
one hour.
second quarter of the work gets completed at 90% of the rate it was intended. Therefore, the
100
amount of time required for the second quarter of the work is hours. The third quarter
90
of the work gets completed at 80% of the rate it was intended. Therefore, the amount of time
100
required for the third quarter of the work is hours. This leaves
80
10 10 23
4−1− − =
9 8 36
hours to complete the rest of the work. Hence, during the last quarter, the work must be
completed at a rate of 36
23 of the usual rate. Hence, the number of workers must be at least

36
1000 · = 1566,
23
so we need at least
1566 − 800 = 766
additional workers during the fourth quarter. 2
10. (1990 AIME, Problem #6) A biologist wants to calculate the number of fish in a
lake. On May 1 she catches a random sample of 60 fish, tags them, and releases
them. On September 1 she catches a random sample of 70 fish and finds that 3
of them are tagged. To calculate the number of fish in the lake on May 1, she
assumes that 25% of these fish are no longer in the lake on September 1 (because
of death and emigrations), that 40% of the fish were not in the lake May 1 (because
of birth and immigrations), and that the number of untagged fish and tagged fish
in the September 1 sample are representative of the total population. What does
the biologist calculate for the number of fish in the lake on May 1?
Solution: We can assume that 60% of the fish caught on September 1 had been in the lake
on May 1 (since 40% of the fish are new to the lake after May 1). That is, we can assume
(0.6)(70) = 42 fish were eligible for tagging on May 1. Exactly three of these fish were caught
on September 1. We conclude that the fraction of fish tagged by the biologist on May 1 is
3 1
= . If x denotes the number of fish in the lake on May 1, we have
42 14
1 60
= .
14 x
Solving for x, we conclude that x = 840.
Remark: Beware of extraneous information that is irrelevant or misleading. In this problem,

the fact that 25% of the fish have died is unimportant, since we are only interested in live fish.
11. (2004 AIME-2, Problem #6) Three clever monkeys divide a pile of bananas. The
first monkey takes some bananas from the pile, keeps three-fourth of them, and
divides the rest equally between the other two. The second monkey takes some
bananas from the pile, keeps one-fourth of them, and divides the rest equally
between the other two. The third monkey takes the remaining bananas from the
pile, keeps one-twelfth of them and divides the rest equally between the other
two. Given that each monkey receives a whole number of bananas whenever the
bananas are divided, and the numbers of bananas the first, second, and third
monkeys have at the end of the process are in the ratio 3 : 2 : 1, what is the least
possible total for the number of bananas?
Solution: Let us use the following notation:
x = the number of bananas taken by the first monkey,

y = the number of bananas taken by the second monkey,
z = the number of bananas taken by the third monkey.
We wish to find the minimum possible value of x + y + z, which is the total number of bananas.
Let us also denote
M1 = the number of bananas the first monkey ends up with,
M2 = the number of bananas the second monkey ends up with,
M3 = the number of bananas the third monkey ends up with.
Since the ratio of the number of bananas the three monkeys have is 3 : 2 : 1, we deduce that
1 1 1
M1 = (x + y + z), M2 = (x + y + z), M3 = (x + y + z). (11.7)
2 3 6
This shows that M2 = 2M3 and M1 = 3M3 , as required. We also observe that
3 3 11
M1 = x + y + z,
4 8 24
since the first monkey keeps three-fourths of the bananas he takes, and he receives half of
the three-fourths of the bananas shared by the second monkey, and he receives half of the
eleven-twelfths of the bananas shared by the third monkey. Similarly, we obtain
3 3 11 1 1 11 1 3 1
M1 = x + y + z, M2 = x + y + z, M3 = x + y + z. (11.8)
4 8 24 8 4 24 8 8 12
Equating the expressions for M1 , M2 , and M3 found in Equations (11.7) and (11.8), we obtain
3 3 11 1
x + y + z = (x + y + z), (11.9)
4 8 24 2
1 1 11 1
x + y + z = (x + y + z), (11.10)
8 4 24 3
and
1 3 1 1
x + y + z = (x + y + z). (11.11)
8 8 12 6
Multiplying Equations (11.9) - (11.11) through by 24, we eliminate fractions:
18x + 9y + 11z = 12(x + y + z), (11.12)
3x + 6y + 11z = 8(x + y + z), (11.13)

3x + 9y + 2z = 4(x + y + z). (11.14)

From Equation (11.12), we find that
z = 6x − 3y, (11.15)
and substituting this into Equation (11.13) or (11.14), we have
3x + 6y + 11(6x − 3y) = 8(x + y + 6x − 3y)
or
69x − 27y = 56x − 16y
or
13x = 11y. (11.16)
Using Equations (11.15) and (11.16), we have that
26 51
x + y + z = x + y + (6x − 3y) = 7x − 2y = 7x − x= x.
11 11
Thus, the least possible number of bananas, x + y + z, occurs for the smallest possible value
of x. From Equation (11.16), observe that x must be a multiple of 11. However, since the
first monkey gives one-eighth of the x bananas it took to each of the other two monkeys, and
each monkey receives whole numbers of bananas, x must also be a multiple of 8. Putting this
together, we see that the minimum value of x is x = 11 · 8 = 88. Hence, the minimum possible
value of x + y + z is
51
x+y+z = x = 51 · 8 = 408.
11
2
12. (2010 AIME, Problem #9) Let (a, b, c) be a real solution of the system of equations
x3 − xyz = 2, y 3 − xyz = 6, z 3 − xyz = 20. (11.17)

m
The greatest possible value of a3 + b3 + c3 can be written in the form n, where m
and n are relatively prime positive integers. Find m + n.
Solution: From the system (11.17), observe that
a3 = c3 − 18 and b3 = c3 − 14.
Thus, we have
p p p
20 = c3 − c3 − 18 c3 − 14c = c3 − 3 (c3 − 18)(c3 − 14)c3 .
3 3
Rearranging this, we have
c3 (c3 − 18)(c3 − 14) = (c3 − 20)3 .
Expanding each side,
c9 − 32c6 + 252c3 = c9 − 60c6 + 1200c3 − 8000.

Simplifying this, we find that

28c6 − 948c3 + 8000 = 0,
or, upon dividing through by 4,
7c6 − 237c3 + 2000.
We can use the quadratic formula to find values of c3 :
√
3 237 ± 2372 − 56000 237 ± 13
c = = .
14 14
Since we are seeking the maximum possible value of a3 + b3 + c3 , we wish to take the largest
value of c3 , that is,
237 + 13 250 125
c3 = = = .
14 14 7
Thus,

125 375 151
a3 + b3 + c3 = (c3 − 18) + (c3 − 14) + c3 = 3c3 − 32 = 3 − 32 = − 32 = .
7 7 7
Hence, m = 151 and n = 7. Therefore, m + n = 158.
Remark: In this solution, we worked to express a and b in terms of c. This was a rather
arbitrary choice, and it would work equally well to express a and c in terms of b, or to
express b and c in terms of a. The interested reader is encouraged to work out these alternate
calculations. 2
13. (1984 AIME, Problem #10) Mary told John her score on the American High
School Mathematics Examination (AHSME), which was over 80. From this, John
was able to determine the number of problems Mary solved correctly. If Mary’s
score had been any lower, but still over 80, John could not have determined this.
What was Mary’s score? (Recall that the AHSME consists of 30 multiple-choice
problems and that one’s score, s, is computed by the formula s = 30 + 4c − w, where
c is the number of correct and w is the number of wrong answers; students are
not penalized for problems left unanswered.)
Solution #1: One “brute force” solution to this problem involves an examination of Mary’s
possible scores as a function of the number of correct answers c. In order for Mary to score
over 80, note that 13 ≤ c ≤ 30. Given such c, the possible scores for Mary on the AHSME
are, in descending order,
30 + 4c, 30 + 4c − 1, 30 + 4c − 2, ..., 30 + 4c − (30 − c) = 5c.
For each integer c with 13 ≤ c ≤ 30, let us form the set
Sc := {5c, 5c + 1, 5c + 2, . . . , 30 + 4c − 1, 30 + 4c}.
We are looking for the smallest number over 80 that belongs to a unique set Sc (for only one
value of c).
Number of Correct Answers Possible Scores

30 150
29 145, 146
28 140, 141, 142
27 135, 136, 137, 138
26 130, 131, 132, 133, 134
25 125, 126, 127, 128, 129, 130
24 120, 121, 122, 123, 124, 125, 126
23 115, 116, 117, 118, 119, 120, 121, 122
22 110, 111, 112, 113, 114, 115, 116, 117, 118
21 105, 106, 107, 108, 109, 110, 111, 112, 113, 114
.. ..
. .
As the table indicates, 119 is the smallest value that appears in the list of possible scores for
only one value of c. Hence, the answer is 119.
Solution #2: This problem can also be approached without “brute force” as follows. From
the formula
s = 30 + 4c − w,
we see that the value of s remains unchanged if c is increased by 1 and w is decreased by 4.
Thus, in order to obtain a unique value of c for a given score s, we see that
w ≤ 3. (11.18)
On the other hand, if c + w ≤ 25, then by increasing c by 1 and increasing w by 4, the same
score c is achieved. (This cannot be done if c + w ≥ 26 since it would require at least 31
questions.) Thus,
c + w ≥ 26. (11.19)
To satisfy Equations (11.18) and (11.19) with the smallest possible score s, we choose w as
large possible and c as small as possible: w = 3 and c = 23. This gives
s = 30 + 4c − w = 30 + 92 − 3 = 119.
14. (1987 AIME, Problem #10) Al walks down to the bottom of an escalator that is
moving up and he counts 150 steps. His friend, Bob, walks up to the top of the
escalator and counts 75 steps. If Al’s speed of walking (in steps per unit time) is
three times Bob’s speed, how many steps are visible on the escalator at any given
time? (Assume that this number is constant.)
Solution #1: Let n denote the number of steps visible at any one time on the escalator.
Suppose the speed of the escalator is given by v (measured in steps per unit time), and let
b denote Bob’s speed (measured in steps per unit time). Then Bob’s effective speed while
walking on the escalator is b + v, since he is walking in the same direction that the escalator
is moving. Meanwhile, Al’s effective speed while walking on the escalator is 3b − v, since he
is walking in the opposite direction to the escalator’s motion. Therefore, we can use Equation
(1.8), with the rates above and distance n, to determine the amount of time that Bob and Al
are on the escalator:
n n
Bob’s Escalator Time = and Al’s Escalator Time = .
b+v 3b − v
Thus, we can apply Equation (1.8) once more to obtain

n n
75 = b and 150 = 3b .
b+v 3b − v
Multiplying the former equation by 3, we have

n
225 = 3b .
b+v
Therefore,
225(b + v) = 3bn = 150(3b − v).
This simplifies to
5v = 3b.
Since
3bn
150 = ,
3b − v
we can simplify this to
5vn 5n
150 = = .
4v 4
Thus,
4
n = 150 = 120.
5
2
Solution #2: Imagine Al and Bob starting at opposite ends of the escalator at the same time.
The number n of steps visible on the escalator is precisely the number of steps separating the
two. We shall determine the number of steps each of them takes until they meet, and then we
will sum those numbers.
In the time that Bob takes 75 steps to reach the top of the escalator, observe that Al is capable
of taking 3 · 75 = 225 steps (since he moves at three times faster speed than Bob). However,
Al only needed 150 steps to reach the opposite end of the escalator, which means that Al can
walk down the escalator in 2/3 the time that it takes Bob to go up the escalator. Therefore,
at whatever point P the two meet, if the physical distance from P to Bob’s starting point
is x, then the distance from P to Al’s starting point is 32 x. Hence, the total length of the
escalator is 52 x, which means that Al and Bob have met 25 of the way from Bob’s starting point
to Al’s starting point. At this point, Bob has taken 2/5 of his steps on the escalator, while
Al has taken 3/5 of his steps on the escalator. So the total number of steps taken by the two
individuals is
2 3
(75) + (150) = 30 + 90 = 120.
5 5
2
15. (1990 AIME, Problem #15) Find ax5 + by 5 if the real numbers a, b, x, and y satisfy
the equations
ax + by = 3,
ax2 + by 2 = 7,
ax3 + by 3 = 16,
ax4 + by 4 = 42.
Solution: The key trick is to notice that
(ax + by)(x + y) = ax2 + by 2 + xy(a + b),

(ax2 + by 2 )(x + y) = ax3 + by 3 + xy(ax + by),
(ax3 + by 3 )(x + y) = ax4 + by 4 + xy(ax2 + by 2 ),
(ax4 + by 4 )(x + y) = ax5 + by 5 + xy(ax3 + by 3 ).
Set
C := x + y and D := xy.
Substituting into the four equations above, we find that
3C = 7 + D(a + b), 7C = 16 + 3D, 16C = 42 + 7D, 42C = (ax5 + by 5 ) + 16D. (11.20)
We see that the desired quantity ax5 + by 5 is attainable from knowledge of the values of C
and D. The middle two equations in (11.20),
7C = 16 + 3D and 16C = 42 + 7D,
can be routinely solved to find that C = −14 and D = −38.

Hence, we use the last equality in (11.20) to conclude our solution:
ax5 + by 5 = 42C − 16D = 42(−14) − 16(−38) = 20 = 020.
1. (2007 AIME-2, Problem #1) A mathematical organization is producing a set of

commemorative license plates. Each plate contains a sequence of five characters
chosen from the four letters in AIME and the four digits 2007. No character may
appear in a sequence more times than it appears among the four letters in AIME
or the four digits in 2007. A set of plates in which each possible sequence appears
N
exactly once contains N license plates. Find 10 .
Solution: To form a license plate that meets the constraints in the problem, we must use five
of the symbols A, I, M, E, 2, 0, 0, and 7. Note that the double occurrence of “0” implies that
it is allowed to occur up to two times in a license plate. Let us delineate (mutually exclusive)
cases according to how many 0’s occur in the license plate.

5
Case 1: Two zeros occur in a license plate. There are = 10 ways to choose two
2
of the five positions in the license plate in which to place the zeros. Three open spots remain.
The first can be filled with any of six symbols (A, I, M, E, 2, or 7), the second can be filled with
any of five remaining symbols, and the third can be filled with any of four remaining symbols.
Using the Multiplication Principle (Theorem 2.2.1), we have 10 · 6 · 5 · 4 = 1200 license plates
in this case.
Case 2: At most one zero occurs in a license plate. There are five slots to be filled
with distinct symbols from among A, I, M, E, 2, 0, and 7. The first slot can be filled with any
of seven symbols, the second slot can be filled with any of six remaining symbols, and so on.
Owing to the Multiplication Principle (Theorem 2.2.1), there are 7 · 6 · 5 · 4 · 3 = 2520 license
plates in this case.
Using the Sum Rule (Theorem 2.4.3), the number of license plates obtained in each of these
N
two cases gives a total of N = 1200 + 2520 = 3720 license plates. Therefore, = 372. 2
10
Remark: As stated in the text, one of the delicate decisions that one must make in using
cases to solve counting problems is how to delineate the cases most effectively. For instance,
in this problem one may well ask why Case 2 is not further expanded into two cases: (a) one
zero occurs in the licence plate and (b) no zeros occur in the license plate. Of course, (a) and
(b) are possibilities that apply to any of the characters, not just “0”. Therefore, it makes for
a cleaner solution to just treat “0” like every other character (i.e. it may or may not be used
once) than create unnecessarily refined cases. Still, it would not be out of the question to
envisage a decomposition of this problem into a different set of cases.
2. (2012 AIME-2, Problem #3) At a certain university, the division of mathemati-

cal sciences consists of the departments of mathematics, statistics, and computer
science. There are two male and two female professors in each department. A
committee of six professors is to contain three men and three women and must
also contain two professors from each of the three departments. Find the number
of possible committees that can be formed subject to these requirements.
Solution: We consider two (mutually exclusive) cases:
Case 1: One man and one woman from each department are chosen for the com-
mittee. In this case, there are two choices for each sex in each of the three departments,
and all six of these choices are independent from one another. Hence, by the Multiplication
Principle we have 26 = 64 committees of this type that can be formed.
Case 2: Two men are chosen from one department, two women from another de-
partment, and one man and one woman from the last department. There are three
choices for which department two men are chosen from, followed by two choices for which
department two women are chosen from (and the last department is then already determined).
Only in the last department must we make any further choices. We must choose which man
(two choices) and which woman (two choices) will be chosen for the committee. Thus, alto-
gether, we have (once more from the Multiplication Principle) 3 · 2 · 2 · 2 = 24 committees of

this type that can be formed.
Finally, we apply the Sum Rule to the two cases to arrive at the final answer: 64 + 24 = 88 =
088. 2
3. (2004 AIME, Problem #3) A convex polyhedron P has 26 vertices, 60 edges, and
36 faces, 24 of which are triangular, and 12 of which are quadrilaterals. A space
diagonal is a line segment connecting two non-adjacent vertices that do not belong
to the same face. How many space diagonals does P have?
Solution: Applying the Subtraction Rule, we will subtract the number of non-space
diagonals
26
from the total number of line segments joining vertices in P . There are = 325 line
2
segments connecting vertices of P , since each line segment is determined by an unordered
choice of two vertices in P . However, not all line segments qualify as space diagonals. In
particular, line segments connecting adjacent vertices, which simply form edges in P , are not
legitimate space diagonals. Thus, we must subtract the 60 edges of P from the total number
of line segments. We must also subtract line segments connecting non-adjacent vertices that
belong to the same face. Only vertices at the opposite corners of a quadrilateral face can be
non-adjacent and belong to the same face. There are 12 quadrilateral faces in P , and each
such face has two line segments connecting its opposite corners. Therefore, we must subtract
24 line segments of this type. Therefore, we have 325 − 60 − 24 = 241 space diagonals. 2
4. (2012 AIME, Problem #3) Nine people sit down for dinner where there are three
choices of meals. Three people order the beef meal, three order the chicken meal,
and three order the fish meal. The waiter serves the nine meals in random order.
Find the number of ways in which the waiter could serve the meal types to the
nine people so that exactly one person receives the type of meal ordered by that
person.
Solution: Let us denote a beef meal with B, a chicken meal with C, and a fish meal with
F. Imagine putting the nine people in a line such that the first (left) three ordered beef, the
middle three ordered chicken, and the last (right) three ordered fish.
(INSERT FIGURE !!!)
We will apply the Multiplication Principle by considering the subtasks required to obtain
exactly one correct meal. First, there are 9 ways to select one person to receive the correct
meal. Let us assume without loss of generality that the first person in the line receives the beef
meal, the correct type of meal. The next two people in the line (both of whom also ordered
the beef meal) must not receive beef. So there are two cases:
Case 1: The second and third person in line receive the same (non-beef ) type of
meal. There are two choices for what type of meal these two people receive (either chicken or
fish). Without loss of generality, assume they each receive chicken. So the pattern for the first
three people is BCC. So that none of the people who ordered fish receive the correct meal, the
middle three people must each receive fish. For the last three people, there are two beef meals
and one chicken meal to distribute. There are 3 ways to choose which of them receives the
chicken meal. So in this case, we have 2 · 3 = 6 ways to distribute the 8 incorrect meals.
Case 2: The second and third person in line receive different types of meals. The
pattern for the first three people must be either BCF or BFC. Hence, there are two choices.
Regardless of which pattern occurs, the middle three people must receive one beef and two fish
meals (they cannot receive chicken, and no fish meals can be leftover for the last three people).
There are 3 ways to decide which receives the beef meal. Finally, there is one beef meal and
two chicken meals to distribute to the last three people. Again, there are 3 ways to decide
which of them gets the beef meal. By the Multiplication Principle, there are 2 · 3 · 3 = 18 ways
to deliver the 8 incorrect meals.
Adding the results from Cases 1 and 2, we have 6 + 18 = 24 ways to distribute 8 incorrect
meals. There were 9 ways to choose one person to get the correct meal. Finally, we use the
Multiplication Principle to obtain 24 · 9 = 216 ways for the waiter to serve the meals with the
given requirements. ANSWER: 216. 2
5. (1983 AIME, Problem #10) The numbers 1447, 1005, and 1231 have something in
common: each is a 4-digit number beginning with 1 that has exactly two identical
digits. How many such numbers are there?
Solution: There are two types of numbers with four digits beginning with 1 that contain
exactly two identical digits. We will therefore use two cases.
Case 1: Numbers containing two 1’s. The first of the 1’s occurs in the left-most spot,
and there are three positions in which the second 1 can occur. In the two remaining positions
(whichever ones they are), we have nine choices for a digit in the first remaining slot and
then eight choices for a digit in the second remaining slot. By the Multiplication Principle
(Theorem 2.2.1), there are 3 · 9 · 8 = 216 numbers in this case.
Case 2: Numbers not containing two 1’s. In this case, there are ninechoices for which
3
digit occurs twice in the number (any digit other than 1). There are then = 3 choices
2
for which two slots the repeated digit occupies. Finally, there are eight digits remaining that
can fill the last open slot. Just like in Case 1, the Multiplication Principle (Theorem 2.2.1)
guarantees that there are 9 · 3 · 8 = 216 numbers in this case.
Computing the total sum of the numbers found in the two cases discussed above, we obtain
the answer: 216 + 216 = 432. 2
6. (2009 AIME-2, Problem #6) Let m be the number of five-element subsets that
can be chosen from the set of the first 14 natural numbers so that at least two of
the five numbers are consecutive. Find the remainder when m is divided by 1000.

14
Solution: There are a total of = 2002 ways to form a five-element subset of
5
{1, 2, 3, . . . , 14}. Let us next compute the number n of five-element subsets for which no two
numbers in the subset are consecutive. (Then the Subtraction Rule will give m = 2002 − n.)
We use the stars and bars method to compute n.
Consider a string of length 14 consisting of N ’s and C’s, where a C in the ith position means
that i was chosen for the five-element subset, and an N in the ith position means that i was
not chosen. For example, the string
N CN N N CN CN N N CN C
corresponds to the five element subset {2, 6, 8, 12, 14}. We can view the nine N ’s like stars and
the five C’s like bars. For each i = 1, 2, 3, 4, 5, 6, let xi denote the number of consecutive N ’s
prior to the ith occurrence of C (and x6 is the number of N ’s following the last C). Then we
must have
x1 + x2 + x3 + x4 + x5 + x6 = 9,
such that x2 , x3 , x4 , x5 ≥ 1 (since no two C’s in the string can be consecutive). Setting
yi = xi − 1 for i = 2, 3, 4, 5, we arrive at the equation
x1 + y2 + y3 + y4 + y5 + x6 = 5,
where each variable appearing here must have anonnegative

value. Hence, we have 5 stars
10
and 5 bars, and the number of solutions is n = = 252. Therefore, we conclude that
5
m = 2002 − 252 = 1750. The remainder when m is divided by 1000 is 750. 2
7. (2006 AIME-2, Problem #4) Let (a1 , a2 , a3 ,. . ., a12 ) be a permutation of (1, 2, 3,. . . ,12)
for which
a1 > a2 > a3 > a4 > a5 > a6 and a6 < a7 < a8 < a9 < a10 < a11 < a12 .
An example of such a permutation is (6, 5, 4, 3, 2, 1, 7, 8, 9, 10, 11, 12). Find the number
of such permutations.
Solution: We must have a6 = 1, since 1 cannot fit anywhere else in the strict inequalities
given. Next, observe that any five of the remaining 11 integers can take the values in the set
{a1 , a2 , a3 }, provided that the largest of them is a1 , next largest is a2 , and so on. Thus,
, a4 , a5
11
there are ) = 462 ways to select (and arrange) the numbers a1 , a2 , a3 , a4 , and a5 . The
5
remaining six numbers can then assume the values a7 , a8 , a9 , a10 , a11 , and a12 in one and only
one way so as to satisfy the inequality requirements. Thus, we have exactly 462 ways to build
such permutations.

11
Remark: It is also acceptable to choose the numbers a7 – a12 first. There are = 462
6
ways to do this. Then the remaining numbers can uniquely assume the values a1 – a5 . We
obtain the same final answer with this approach. The answer is 462. 2
8. (2007 AIME-2, Problem #6) An integer is called parity-monotonic if its decimal

representation a1 a2 a3 . . . ak satisfies ai < ai+1 if ai is odd, and ai > ai+1 if ai is even.
How many four-digit parity-monotonic integers are there?
Solution: This problem is trickier than it looks. A standard approach might suggest that we
determine the number of choices available in selecting each digit of a parity-monotonic integer
and then apply the Multiplication Principle (Theorem 2.2.1). If we do this in the normal way,
however, from left-most digit to right-most digit, the dependencies of the later digits on the
earlier digits makes the count extremely difficult. For instance, if the first digit is 1, then the
second digit can be any digit from 2 to 9, but if the first digit is 2, then the second digit must
be a 0 or 1. What this suggests, therefore, is that the number of choices available for selecting
the second, third, or fourth digits of the four-digit parity-monotonic integer depends highly on
the choice made for the previous digit. When such dependencies arise in the counting process,
the Multiplication Principle (Theorem 2.2.1) ceases to be valid.
The remedy that can be used to make the number of choices available for one digit independent
of preceding choices for other digits is to select the digits from right to left. There are 10 digits
that can be chosen for the right-most digit. It is easy to see that, regardless of what digit is
chosen, the next digit to the left always has four choices. For instance, if the right-most digit
is 0, then the next digit to the left can be 2,4,6, or 8. The table below shows that, given any
digit in the parity-monotonic integer, the next digit to the left can be any one of exactly four
digits:
Right-most digit Options for digit to its left

0 2, 4, 6, 8
1 2, 4, 6, 8
2 1, 4, 6, 8
3 1, 4, 6, 8
4 1, 3, 6, 8
5 1, 3, 6, 8
6 1, 3, 5, 8
7 1, 3, 5, 8
8 1, 3, 5, 7
9 1, 3, 5, 7
Using this reasoning, we see that there are 10 choices for the right-most digit, four choices for
the next digit to the left (independent of the previous choice), then four choices for the next
digit to the left, and finally, four choices for the left-most digit. Since all choices are made
independently, we can apply the Multiplication Principle (Theorem 2.2.1) to obtain a total of
10 · 4 · 4 · 4 = 640 parity-monotonic integers. 2
9. (2005 AIME, Problem #5) Robert has 4 indistinguishable gold coins and 4 in-
distinguishable silver coins. Each coin has an engraving of a face on one side,
but not on the other. He wants to stack the eight coins on a table into a single
stack so that no two adjacent coins are face to face. Find the number of possible
distinguishable arrangements of the 8 coins.
Solution: Let us assume each coin’s face is on the “up” side of the coin. We can easily
enumerate the possible configurations (without reference to which coins are gold and which
are silver) of the coins in the stack that avoid having two adjacent coins face to face. We note
that “down” refers to a coin that is face down in the stack, and “up” refers to a coin that is
face up in the stack. Here are the possible stacks:
down up up up up up up up up
down down up up up up up up up
down down down up up up up up up
down down down down up up up up up
down down down down down up up up up
down down down down down down up up up
down down down down down down down up up
down down down down down down down down up

8
For each of the nine stacks shown above, we have = 70 ways to assign four of the eight
4
coins in the stack to be gold (the remaining four coins in the stack are then silver by default).
Observe that the color (gold or silver) of each coin is independent of the coin’s orientation (up
or down), and in fact, it is fine to first choose the color of each coin in the stack, and then
orient the face (up or down) aftewards. Therefore, we can use the Multiplication Principle
to determine the number of ways of completing the two independent tasks we have described
(face configuration and color selection) and obtain the final answer: 9 · 70 = 630.
Remark: Of course, it is not necessary to enumerate the stacks of coins as we have done
above. In general, for n coins, there will be n + 1 allowable stacks, indexed by the position
within the stack, if any, where the coin orientation switches. 2
10. (1997 AIME, Problem #10) Every card in a deck has a picture of one shape –
circle, square, or triangle, which is painted in one of three colors – red, blue, or
green. Furthermore, each color is applied in one of three shades – light, medium,
or dark. The deck has 27 cards, with every shape-color-shade combination rep-
resented. A set of three cards from the deck is called complementary if all of the
following statements are true:
(i) Either each of the three cards has a different shape or all three of the cards
have the same shape.
(ii) Either each of the three cards has a different color or all three of the cards
have the same color.
(iii) Either each of the three cards has a different shade or all three of the cards
have the same shade.
How many different complementary three-card sets are there?

Solution #1: There are several viable solutions to this problem, and we present two of them
here. In the first solution, we count complementary three-card sets by enumerating mutually
exclusive cases according to how many of the three characteristics given in (i), (ii), and (iii)
are the same and how many are different in the complementary sets.
Case 1: The three cards meet (i)-(iii) by being different with respect to all three
attributes (shape, color, and shade). In this case, one card pictures a circle, one card
pictures a square, and one card pictures a triangle. (No choices have been made yet.) Now
each color must occur on exactly one of the three cards. There are 3! = 6 ways to assign
colors to the three shapes occurring on the three cards. Likewise, since each shade must occur
on exactly one of the three cards, there are also 3! = 6 ways to assign shades to the three
shapes occurring on the three cards. Note that the color assignments and shade assignments
are made independently from one another. Hence, by the Multiplication Principle (Theorem
2.2.1), there are 3! · 3! = 36 complementary sets in this case.
Case 2: The three cards meet (i)-(iii) by being different with respect to exactly
two of the three attributes (shape, color, and shade). There are three choices for which
of the three attributes the three cards share in common. Without loss of generality, suppose
all three cards picture the same shape. There are three choices for what the common shape
in the set is. The three cards will therefore picture the same shape in three different colors.
(There are no choices made while colors are assigned.) Now there are 3! = 6 ways to assign
shades to the three colors occurring on the three cards. Multiplying the number of choices
described here, we find 3 · 3 · 3! = 54 complementary sets in this case.
Case 3: The three cards meet (i)-(iii) by being different with respect to exactly
one of the three attributes (shape, color, and shade). There are three choices for
which two of the three attributes the three cards share in common. Without loss of generality,
suppose all three cards picture the same shape and color. There are three choices for what the
common shape in the set is, and there are three choices for what the common color in the set
is. Finally, each card is assigned a different shade. (There are no choices made while shades
are assigned.) The number of choices enumerated in this case, obtained by the Multiplication
Principle (Theorem 2.2.1), is 3 · 3 · 3 = 27.
Note that it is impossible for the three cards to be different in none of the attributes; that
would imply that all three cards were painted identically. Since the deck only contains one
card of each shape-color-shade combination, no two cards can be identical. Therefore, only
Cases 1–3 can occur.
Summing the results in each of the three cases above gives the answer: 36 + 54 + 27 = 117. 2
Solution #2: Given any two cards in the deck, we can explore the number of ways to add a
third card so that a complementary set is formed. If the two chosen cards have the same shape,
then the third card must have the same shape also. On the other hand, if the two chosen cards
have different shpaes, then the third card must have a different shape also. Thus, the shape of
the third card is determined already from the two given shapes. The same goes for color and
shade. We conclude that any two given cards in the deck determine a unique complementary
three-card set. However, the same complementary three-card
set would be generated from
27
any two of its three members. Thus, while there are ) ways to choose two cards, the
2
complementary three-card set so determined arises three different times
depending on which
27
two of its three cards are chosen. In other words, the number counts each set three
2
times. Hence, our final answer is
1 27
= 117.
3 2
2
Remark: This problem is not well-suited to the Subtraction Rule, since failure of three cards
to form a complementary set can happen in multiple ways, as any one (or more) of conditions
(i), (ii), and/or (iii) could fail.
11. (2009 AIME, Problem #10) The Annual Interplanetary Mathematics Examina-
tion (AIME) is written by a committee of five Martians, five Venusians, and five
Earthlings. At meetings, committee members sit at a round table with chairs
numbered from 1 to 15 in clockwise order. Committee rules state that a Martian
must occupy chair 1 and an Earthling must occupy chair 15. Furthermore, no
Earthling can sit immediately to the left of a Martian, no Martian can sit imme-
diately to the left of a Venusian, and no Venusian can sit immediately to the left
of an Earthling. The number of possible seating arrangements for the committee

is N (5!)3 . Find N .
Solution: Let us begin by assuming that all Martians look identical, all Venusians look
identical, and all Earthlings look identical. We denote a Martian by “M”, a Venusian by “V”,
and an Earthling by “E”. Figure 28(a) shows one example of a permissible seating arrangement
for the Martians, Venusians, and Earthlings, and Figure 28(b) shows this same arrangement
placed in a straight line, beginning with Chair 1 at the far left and ending with Chair 15 at the
far right. We can analyze the problem most effectively with the seats arranged in a straight
line.
FIGURE 28 GOES HERE
The key observation is that the number of groupings of Martians (i.e. sets of consecutively
seated Martians) must be the same as the number of groupings of Venusians and the number
of groupings of Earthlings. For instance, there is only one way to have one grouping of each:
M M M M M V V V V V EEEEE
At the opposite extreme, there is only one way to have five groupings of each (where each
grouping in this case consists of a single individual):
M V EM V EM V EM V EM V E
However, we can also have two, three, or four groupings of each:

Case 1: Two groupings of each. Let m1 denote the number of Martians in the first
grouping, and let m2 denote the number of Martians in the second grouping. We have m1 +
m2 = 5 such that m1 , m2 > 0. There are four valid solutions:
(m1 = 4, m2 = 1), (m1 = 3, m2 = 2), (m1 = 2, m2 = 3), and (m1 = 1, m2 = 4).
Independent of this, we have the same four solutions for the Venusians and for the Earthlings.
Thus, there are 43 = 64 valid ways to seat the committee with two groupings of each.
Case 2: Three groupings of each. Along with the notation from Case 1 above, we introduce
m3 to denote the number of Martians in the third grouping. We have
m1 + m2 + m3 = 5 such that m1 , m2 , m3 > 0.
There are six valid solutions:
(m1 = 3, m2 = 1, m3 = 1), (m1 = 1, m2 = 3, m3 = 1), (m1 = 1, m2 = 1, m3 = 3),
(m1 = 2, m2 = 2, m3 = 1), (m1 = 2, m2 = 1, m3 = 2), (m1 = 1, m2 = 2, m3 = 2).

Independent of this, we have the same six solutions for the Venusians and for the Earthlings.
Thus, there are 63 = 216 valid ways to seat the committee with three groupings of each.
Case 3: Four groupings of each. Along with the notation from Case 2 above, we introduce
m4 to denote the number of Martians in the fourth grouping. We have
m1 + m2 + m3 + m4 = 5 such that m1 , m2 , m3 , m4 > 0.
There are four valid solutions:
(m1 = 2, m2 = 1, m3 = 1, m4 = 1), (m1 = 1, m2 = 2, m3 = 1, m4 = 1),
(m1 = 1, m2 = 1, m3 = 2, m4 = 1), (m1 = 1, m2 = 1, m3 = 1, m4 = 2).

Independent of this, we have the same four solutions for the Venusians and for the Earthlings.
Thus, there are 43 = 64 valid ways to seat the committee with three groupings of each.
A summary of the number of ways to seat the committee as a function of the number of
groupings is listed in the table below:
Number of Groupings Number of Seating Arrangements

1 1
2 64
3 216
4 64
5 1
Summing the results, we find that the number of permissible seating arrangements, under the
standing assumption that all individuals from the same planet look identical, is
1 + 64 + 216 + 64 + 1 = 346.
The fact that the individuals from the same planet are actually distinguishable implies that
for each arrangement we have counted above, we have (5!)3 arrangements that involve distin-
guishable individuals (5! for each planet’s members). Thus, N = 346. 2
12. (1986 AIME, Problem #13) In a sequence of coin tosses one can keep a record
of the number of instances when a tail is immediately followed by a head, a head
is immediately followed by a head, etc. We denote these by T H, HH, etc. For
example, in the sequence HHT T HHHHT HHT T T T of 15 coin tosses we observe
that there are five HH, three HT , two T H, and four T T subsequences. How many
different sequences of 15 coin tosses will contain exactly two HH, three HT , four
T H, and five T T subsequences?
Solution: Notice that long strings of consecutive H’s or T ’s supply multiple adjacent HH or
T T subsequences, whereas HT or T H subsequences occur uniquely when the coin toss outcome
changes. Thus, we know that we get a head followed by a tail (HT string) exactly three times,
and we get a tail followed by a head (T H string) exactly four times. So let us begin by putting
down four T H subsquences on our paper2 , with space between them to (possibly) fill with
additional characters:
TH TH TH TH
2 One can also begin with three HT subsequences and follow a similar approach.
In each of the three spaces between these four T H subsequences, a HT subsequence must occur.
Therefore, since we are constructing sequences with three HT subsequences, there can be no
other HT subsequences. Hence, no H can occur to the left of the four subsequences shown,
and no T can occur to the right of the four subsequences shown. Moreover, any characters
occupying any of the three spaces between the four subsequences must consist of a sequence of
(possibly no) H characters followed by a sequence of (possibly no) T characters. To produce
two HH subsequences, two H characters must be introduced within the spaces between the
three T H subsequences or at the end. There are four spaces, so there are four ways to put
4
both H characters in the same space, and = 6 ways to place a single H in two of the
2
four spaces. This is a total of 10 ways to place the H characters. Finally, there are four spaces
in which five T characters can be placed (at the beginning, or in the spaces between the four
T H subsequences), each of which creates a T T subsequence. This can be viewed as a problem
in stars and bars (see Section 2.3.2), with each T to be added playing the role of a star ∗ and
the characters between the four playing the role of the bars |. Hence, we have five stars
spaces
8
and three bars, for a total of = 56 ways to position the T characters. Placing the two
5
H characters and the five T characters are independent tasks, with 10 ways to place the H’s
and 56 ways to place the T ’s. Therefore, we have a total of 10 · 56 = 560 sequences of the
desired form. 2
13. (2002 AIME-2, Problem #9) Let S be the set {1, 2, 3, . . . , 10}. Let n be the number
of sets of two non-empty disjoint subsets of S. (Disjoint sets are defined as sets that
have no common elements.) Find the remainder obtained when n is divided by
1000.
Solution: We consider the process of building two non-empty disjoint subsets A and B of S.
(Note that the two non-empty sets in this question are not labeled, so we will need to account
for that later.) Each integer 1, 2, . . . , 10 will either be placed in A, placed in B, or placed in
neither set. Thus, we have three independent choices for each integer, leading to 310 outcomes,
but some of these leave A or B (or both) empty and must be subtracted. If we denote by X
those outcomes for which A = ∅ and by Y those outcomes for which B = ∅, then we need to
determine |X ∪ Y |. Note that |X| = 210 (each of the ten integers is either placed in B or in
neither set), and similarly, |Y | = 210 . Finally, |X ∩ Y | = 1 (why?). Thus,
|X ∪ Y | = |X| + |Y | − |X ∩ Y | = 210 + 210 − 1 = 2047.
Using the Subtraction Rule, we conclude that there are
310 − 2047 = 59049 − 2047 = 57002
ways to create two non-empty labeled subsets A and B from S. However, the configuration
with A = {1, 2, 4} and B = {3, 9} is the same as A = {3, 9} and B = {1, 2, 4}. We have
double-counted each of the sets of two non-empty disjoint subsets of S. Thus,
1
n= (57002) = 28501.
2
Dividing by 1000 and taking the remainder, the answer is 501. 2
14. (2007 AIME-2, Problem #13) A triangular array of squares has one square in the
first row, two in the second, and, in general, k squares in the kth row for 1 ≤ k ≤ 11.
With the exception of the bottom row, each square rests on two squares in the
row immediately below, as illustrated in the figure. In each square of the eleventh
row, a 0 or a 1 is placed. Numbers are then placed into the other squares, with
the entry for each square being the sum of the entries in the two squares below it.
For how many initial distributions of 0’s and 1’s in the bottom row is the number
in the top square a multiple of 3?
Solution: Let the entries in the bottom row of the triangular array be a1 , a2 , a3 , a4 , a5 , a6 ,
a7 , a8 , a9 , a10 , and a11 . (See Figure !!!!!!.)
FIGURE 29 GOES HERE – FIX VARIABLE NAMES!!!!!!!!!!!
The entries in the row above that take the form ai + ai+1 (for i = 1, 2, . . . , 10), the entries in
the row above that take the form ai + 2ai+1 + ai+2 (for i = 1, 2, . . . , 9), the entries in the row
above that take the form ai + 3ai+1 + 3ai+2 + ai+3 (for i = 1, 2, . . . , 8), the entries in the row
above that take the form ai + 4ai+1 + 6ai+2 + 4ai+3 + ai+4 (for i = 1, 2, . . . , 7), and so on.
The coefficients of these variables are precisely the binomial coefficients. More precisely, the
11 − r
coefficients of the entries occurring in the rth row are the binomial coefficients for
s
s = 0, 1, 2, . . . , 11 − r. Thus, the entry at the top of the triangular array is

10 10 10 10 10 10
a1 + a2 + a3 + · · · + a9 + a10 + a11 .
0 1 2 8 9 10
It is easy to see that all of the coefficients

10 10 10 10 10 10 10
, , , , , ,
2 3 4 5 6 7 8
are multiples of 3. Therefore, regardless of the values chosen for a3 , a4 , a5 , a6 , a7 , a8 , and a9 ,
the top square of the array is a multiple of 3 if and only if

10 10 10 10
a1 + a2 + a10 + a11
0 1 9 10
is a multiple of three (i.e. a1 + 10a2 + 10a10 + a11 is a multiple 3). For those that have studied
number theory notation Chapter 4, we can write this condition in a mathematical equation as
a+b+j+k ≡0 (mod 3)
There are five ways in which this can occur when a, b, j, and k are either 0 or 1, as the table
below summarizes:
a b j k a+b+j+k
0 0 0 0 0
1 1 1 0 3
1 1 0 1 3
1 0 1 1 3
0 1 1 1 3
We conclude from the Multiplication Principle that the total number of initial distributions of
0’s and 1’s such that the top square is a multiple of 3 is
5 · 27 = 640.
15. (2001 AIME, Problem #14) A mail carrier delivers mail to the nineteen houses
on the east side of Elm Street. The carrier notices that no two adjacent houses
ever get mail on the same day, but that there are never more than two houses
in a row that get no mail on the same day. How many different patterns of mail
delivery are possible?
Solution: Let k denote the number of houses that receive mail. Since no two adjacent houses
can get mail, k ≤ 10. On the other hand, since there are never more than two houses in a row
that get no mail, k ≥ 6. We will count the number of patterns of mail delivery in each case
k = 6, 7, 8, 9, and 10 and sum the results.
The k houses that receive mail separate the remaining houses of Elm Street into k + 1 groups.
Let us say that there are xi houses in the ith group for i = 1, 2, . . . , k + 1. None of the houses
in these k + 1 groups receive mail, so we have
x1 + x2 + · · · + xk+1 = 19 − k. (11.21)
Note that xi ≤ 2 for each i (since there are never more than two houses in a row that get no
mail). However, x2 , x3 , . . . , xk ≥ 1 (since no two adjacent houses both receive mail). Therefore,
for i = 2, 3, . . . , k, we set yi = xi − 1. Setting y1 = x1 and yk+1 = xk+1 , then Equation (11.21)
becomes
y1 + y2 + y3 + · · · + yk + yk+1 = 19 − k − (k − 1) = 20 − 2k.
Now each yi is a nonnegative integer, and y1 , yk+1 ≤ 2 while yi ≤ 1 for i = 2, 3, . . . , k. We
consider the following cases:
Case 1: y1 = yk+1 = 2. Then we have
y2 + y3 + · · · + yk = 20 − 2k − 4 = 16 − 2k.
Out of the k − 1 variables on the left side of the equation, we 16 − 2k of them to

must choose
k−1
assume the value 1 (the others must be 0). Thus, we have solutions. Note that
16 − 2k
this case only applies for 6 ≤ k ≤ 8.
Case 2: y1 = 2, yk+1 6= 2. Then we have
y2 + y3 + · · · + yk+1 = 20 − 2k − 2 = 18 − 2k.
18 − 2k of them to
Out of the k variables on the left side of the equation, wemust choose
k
assume the value 1 (the others must be 0). Thus, we have solutions. Note that
18 − 2k
this case only applies for 6 ≤ k ≤ 9.
Case 3: y1 6= 2, yk+1 = 2. This yields the same number of solutions as Case 2, by symmetry.
Case 4: y1 6= 2, yk+1 6= 2. Then we have
y1 + y2 + · · · + yk+1 = 20 − 2k.
Out of the k + 1 variables on the left side of the equation, we 20 − 2k of them to

must choose
k+1
assume the value 1 (the others must be 0). Thus, we have solutions. This case
20 − 2k
applies for 7 ≤ k ≤ 10.
Adding the results for all cases (with the values of k specified and doubling the values in Case
2), we arrive at the answer:

5 6 7 6 7 8 9
+ + +2 + + +
4 2 0 6 4 2 0

8 9 10 11
+ + + + = 351.
6 4 2 0
2
REVIEW FIGURE 30 WHICH WAS LISTED TO GO HERE.
1. (2004 AIME-2, Problem #2) A jar has 10 red candies and 10 blue candies. Terry
picks two candies at random, then Mary picks two of the remaining candies at
random. Given that the probability that they get the same color combination,
irrespective of order, is m/n, where m and n are relatively prime positive integers,
find m + n.
Solution: We will use R to denote a red candy, and B to denote a blue candy. Since four
candies are drawn in total by Terry and Mary, we can view the outcomes as a string of four
letters (the first two corresponding to Terry’s chosen candies and the last two corresponding
to Mary’s chosen candies). For example, the string RBRR would indicate that Terry got one
candy of each color and Mary got two red candies. There are 16 possible strings. The 16
strings can be enumerated via a tree diagram (see Figure 31).
FIGURE 31 GOES HERE !!!!!!!!!
Of course, we are not concerned with all outcomes listed on this tree, only the the six high-
lighted ones:
{RRRR, BBBB, RBRB, RBBR, BRRB, BRBR}. (11.22)
The first two outcomes in (11.22) are each of the same type (and occur with equal probabilities):
10 9 8 7
· · · .
20 19 18 17
To see this, note that on the first pick, 10 of the 20 candies are red. Assuming a red one is
picked first, on the second pick, there are 9 of 19 candies that are red, and so on.
The last four outcomes in (11.22) are each of the same type (and occur with equal probabilities):
10 10 9 9
· · · .
20 19 18 17
To see this, consider the case RBRB. There is a 10 20 chance of getting a red candy on the first
pick. On the second pick, there are 10 blue candies and 9 red candies in the bag, thus giving
a 10
19 probability of obtaining a blue candy on the second pick. On the third pick, there are
9
9 candies of each color, thus a 18 chance of obtaining a red candy. Finally, on the last pick,
9
there are 9 blue candies and 8 red candies in the bad, thus giving a 17 probability of obtaining
a blue candy on the last pick.
Summing the probabilities of each of the six outcomes, we find that the probability that Terry
and Mary get the same color combination is

10 9 8 7 10 10 9 9 118
2 · · · +4 · · · = .
20 19 18 17 20 19 18 17 323
Hence, m = 118 and n = 323, so that m + n = 118 + 323 = 441. 2
Remark: The tree diagram in Figure 31 is not necessary to solve the problem. Rather, it
is helpful as a book-keeping device to assure the solver that all necessary cases have been
considered. On larger problems (more outcomes, more tasks, etc.), the tree diagram becomes
impractical. Even in this case, the solver may prefer only to write the proabilities on branches
associated with the six desired outcomes, as it will take more time to write out the entire tree
as shown in Figure 31. However, in situations where many different questions are being asked
about the same experiment, it may save time in the long run to simply draw the expanded
tree and make many different references to it for many different questions.
2. (2000 AIME-2, Problem #3) A deck of forty cards consists of four 1’s, four 2’s,
. . . , and four 10’s. A matching pair (two cards with the same number) is removed
from the deck. Given that these cards are not returned to the deck, let m/n be
the probability that two randomly selected cards also form a pair, where m and n
Solution: The sample space S consists of all possible (unordered)
selections of a pair of cards
38
from the 38 cards in the deck. Therefore |S| = = 703. On the other hand, the
2
event E consists of all pairs of cards consisting of two cards of the same
value. From the nine
4
denominations for which there are four cards in the deck, we have = 6 ways to choose
2
a pair from that denomination. On the other hand, there is only one pair available in the
denomination that has been reduced to two cards. This gives us a total of |E| = 9 · 6 + 1 = 55.
Therefore, the probability that two randomly selected cards form a pair is
|E| 55
p(E) = = .
|S| 703
Therefore, we have m = 55 and n = 703. Therefore m + n = 758. 2
3. (1998 AIME, Problem #4) Nine tiles are numbered 1, 2, 3, . . . , 9, respectively. Each
of three players randomly selects and keeps three of the tiles, and sums those three
values. The probability that all three players obtain an odd sum is m/n, where m
and n are relatively prime positive integers. Find m + n.
Solution: The sample space S consists of all possible ways that three players (say, A, B, and
C) can select three tiles at random. Then

9 6 3
|S| = = 84 · 20 · 1 = 1680,
3 3 3
since the first player may choose any three of the nine tiles, the second player may choose any
three of the remaining six tiles, and the third player must take the last three tiles.
Next we compute the size of the event E in which all three players have tiles summing to an
odd number. In order for this to happen, exactly one player must draw three odd tiles, and
each of the other two players must draw one odd tile apiece. There are 3 choices for which
player draws three
odd tiles. That player must then select three of the five odd-numbered
5
tiles. There are = 10 ways for that player to select those tiles. The second player must
3
4
choose one of the remaining odd tiles (2 choices) and two of the four even tiles ( =6
2
choices). Finally, the last player must take the remaining three tiles. Using the Multiplication
Principle, we conclude that
|E| = 3 · 10 · 2 · 6 = 360.
Therefore, the probability that all three players obtain an odd sum is
|E| 360 3
p(E) = = = .
|S| 1680 14
We conclude that m = 3 and n = 14, so the answer is m + n = 3 + 14 = 17 = 017. 2
4. (2010 AIME-2, Problem #4) Dave arrives at an airport which has twelve gates
arranged in a straight line with exactly 100 feet between adjacent gates. His
departure gate is assigned at random. After waiting at that gate, Dave is told
the departure gate has been changed to a different gate, again at random. Let
m
the probability that Dave walks 400 feet or less to the new gate be a fraction ,
n
where m and n are relatively prime positive integers. Find m + n.
Solution: The likelihood that Dave needs to walk 400 feet or less to the second gate depends
on where the first gate is. The following table summarizes the number of gates that are within
400 feet of the first gate for all possible locations of the first gate:
Original Gate Number ki of Gates within

i 400 Feet ofGate i
1 4 (gates 2,3,4,5)
2 5 (gates 1,3,4,5,6)
3 6 (gates 1,2,4,5,6,7)
4 7 (gates 1,2,3,5,6,7,8)
5 8 (gates 1,2,3,4,6,7,8,9)
6 8 (gates 2,3,4,5,7,8,9,10)
7 8 (gates 3,4,5,6,8,9,10,11)
8 8 (gates 4,5,6,7,9,10,11,12)
9 7 (gates 5,6,7,8,10,11,12)
10 6 (gates 6,7,8,9,11,12)
11 5 (gates 7,8,9,10,12)
12 4 (gates 8,9,10,11)
While this table need not be generated in order to solve this problem, some readers may gain
confidence in their work by having an easy reference to exploit while doing the calculations to
come below.
If we let Ei denote the event that Dave begins at gate i and then walks 400 feet or less to the
new gate (for each i = 1, 2, . . . , 12), then from the table above, we have
1 ki
p(Ei ) = ·
12 11
for each i = 1, 2, . . . , 12.
The events Ei are all mutually disjoint since Dave cannot start from more than one gate, and
therefore, part (4) of Theorem 3.2.1 can be applied:
p(E1 ∪ E2 ∪ · · · ∪ E12 ) = p(E1 ) + p(E2 ) + · · · + p(E12 )

1 4 1 5 1 6 1 4
= · + · + · + ··· + ·
12 11 12 11 12 11 12 11
1 4 5 6 7 8
= 2· +2· +2· +2· +4·
12 11 11 11 11 11
8 + 10 + 12 + 14 + 32 76 19
= = = .
132 132 33
Therefore m = 19 and n = 33. We conclude that m + n = 19 + 33 = 52 = 052. 2
Remark: The summation at the end of this solution can be reduced by half if we exploit
symmetry by recognizing that p(Ei ) = p(E13−i ) for each i = 1, 2, . . . , 12.
5. (1989 AIME, Problem #5) When a certain biased coin is flipped 5 times, the
probability of getting heads exactly once is not equal to 0 and is the same as that
of getting heads exactly twice. Let i/j, in lowest terms, be the probability that
the coin comes up heads exactly 3 times out of 5. Find i + j.
Solution: Assume that the probability of obtaining a head is x. Then the probability of
obtaining a tail is 1 − x.
The probability of obtaining exactly one head on five flips is 5x(1 − x)4 .
The factor of 5 occurs here because the head can occur on any of the five flips.

5
The probability of obtaining exactly two heads on five flips is x2 (1 − x)3 ,
2

5
where the factor of = 10 arises from the fact that the two heads can occur on any two
2
of the five flips of the coin. We are given in the problem that the two probabilities computed
above are equal:
5x(1 − x)4 = 10x2 (1 − x)3 . (11.23)
Since the probability of getting heads is not 0, we know that x 6= 0 and 1 − x 6= 0. Hence, we
may cancel common factors from both sides of Equation (11.23) to obtain
1 − x = 2x.
Therefore, x = 31 . The probability of obtaining exactly three heads in five flips is

5 1 4 40
x3 (1 − x)2 = 10 · · = .
3 27 9 243
Therefore, we have i = 40 and j = 243, so that the answer is i + j = 40 + 243 = 283. 2

6. (2001 AIME, Problem #6) A fair die is rolled four times. The probability that
each of the final three rolls is at least as large as the roll preceding it may be
expressed in the form m/n, where m and n are relatively prime positive integers.
Find m + n.
Solution#1: The sample space S consists of all ordered four-tuples (a, b, c, d), where a, b, c, d
belong to the set {1, 2, 3, 4, 5, 6}. Therefore, |S| = 64 . The event E is the subset of S con-
sisting of all ordered four-tuples (a, b, c, d) such that a ≤ b ≤ c ≤ d and a, b, c, d belong to
{1, 2, 3, 4, 5, 6}. Let us examine several cases enumerated by the number of distinct members
that the set {a, b, c, d} contains:

6
Case 1: a, b, c, and d are all distinct: In this case, there are = 15 such four-tuples.
4
Case 2: {a, b, c, d} contains exactly three distinct members: This requires that one of
6
the numbers occurs twice in the four-tuple (a, b, c, d). There are = 20 ways to choose
3
three numbers, and then 3 ways to decide which of those three numbers occurs twice in the
four tuple. By the Multiplication Principle (Theorem 2.2.1), this gives a total of 20 · 3 = 60
four tuples consisting of three distinct numbers.

6
Case 3: {a, b, c, d} contains exactly two distinct members: There are = 15
2
ways to choose two distinct numbers. If the two distinct numbers are a and b with a < b,
then we can form three different four-tuples (a, a, a, b), (a, a, b, b), or (a, b, b, b). Hence, by the
Multiplication Principle (Theorem 2.2.1), we have 15 · 3 = 45 four-tuples in this case.
Case 4: All four numbers a, b, c, and d are identical: There are six such four-tuples:
{(1, 1, 1, 1), (2, 2, 2, 2), (3, 3, 3, 3), (4, 4, 4, 4), (5, 5, 5, 5), (6, 6, 6, 6)}.
Adding the number of four-tuples in each of the cases above, we have |E| = 15 + 60 + 45 + 6 =
126. Therefore, the probability that each of the final three rolls is at least as large as the roll
preceding it is
|E| 126 21 7
= 4 = = .
|S| 6 216 72
Thus, m = 7 and n = 72, so that m + n = 7 + 72 = 79 = 079. 2
Solution #2: When a fair die is rolled four times, there is a unique way to list the values that
occured on the four rolls in weakly increasing (i.e. non-decreasing) order. Thus, the weakly
increasing sequences of length 4 produced by the four rolls of the die are in correspondence
with the different lists of four numbers (including repetitions) from {1, 2, 3, 4, 5, 6}. The latter
number can be computed as the number of solutions in nonnegative integers to
x1 + x2 + x3 + x4 + x5 + x6 = 4,
where xi denotes the number of times that the number i was rolled. This can be readily
9
computed using the stars and bars method with 4 stars and 5 bars: = 126. Thus, out
5
4
of the 6 sequences of four integers chosen from {1, 2, 3, 4, 5, 6}, exactly 126 of them are in
126 7
weakly increasing order. Thus, the probability in question is 4 = , and we obtain m = 7
6 72
and n = 72. Hence, m + n = 79 = 079. 2
7. (1996 AIME, Problem #6) In a five-team tournament, each team plays one game
with every other team. Each team has a 50% chance of winning any game it plays.
m
(There are no ties.) Let be the probability that the tournament will produce
n
neither an undefeated team nor a winless team, where m and n are relatively prime
integers. Find m + n.

5
Solution: In this tournament, there are a total of = 10 games played. Keeping track
2
of the outcome of each game results in a sample space S consisting of 210 outcomes of the
tournament. So we have |S| = 210 = 1024.
Let E1 be the event that there is an undefeated team, and let E2 be the event that there is a
winless team. We wish to compute p(E1 ∪ E2 ) = 1 − p(E1 ∪ E2 ). We need to determine |E1 |,
|E2 |, and |E1 ∩E2 |. To compute |E1 |, we have 5 choices for which team is undefeated. Selection
of this team determines the outcome of all four games played by that team, leaving only six
games left for which two outcomes are possible. Thus, |E1 | = 5 · 26 = 320. By symmetry, we
also have |E2 | = 5 · 26 = 320. For |E1 ∩ E2 |, there are 5 choices for which team is undefeated,
followed by 4 choices for which team is winless. Only the three games involving the other three
teams are in doubt after this choice is made. Thus, |E1 ∩ E2 | = 5 · 4 · 23 = 160. We are asked
to compute
p(E1 ∪ E2 ) = 1 − p(E1 ∪ E2 )

= 1 − p(E1 ) + p(E2 ) − p(E1 ∩ E2 )
= 1 − p(E1 ) − p(E2 ) + p(E1 ∩ E2 )

|E1 | |E2 | |E1 ∩ E2 |
=1− − +
|S| |S| |S|
320 320 160 15 17
=1− − + =1− = .
1024 1024 1024 32 32
We conclude that m = 17 and n = 32, so that m + n = 17 + 32 = 49 = 049. 2
8. (2010 AIME, Problem #4) Jackie and Phil have two fair coins and a third coin
4
that comes up heads with probability . Jackie flips the three coins, and then Phil
7
m
flips the three coins. Let be the probability that Jackie gets the same number
n
of heads as Phil, where m and n are relatively prime positive integers. Find m + n.
Solution: We can compute the likelihood of each of the eight possible outcomes from flipping
the three coins. We assume without loss of generality that the biased coin is tossed third.
Outcome Outcome of Tossing Three Coins Probability of this Outcome

Number (Third Coin is Biased)
1 HHH 4/28
2 HHT 3/28
3 HTH 4/28
4 THH 4/28
5 HTT 3/28
6 THT 3/28
7 TTH 4/28
8 TTT 3/28
For each i = 0, 1, 2, 3, let Ei denote the event that exactly i heads are obtained when each coin is
flipped once. Note that E3 corresponds to Outcome 1 in the table above, while E2 corresponds
to Outcomes 2,3, and 4, E1 corresponds to Outcomes 5,6, and 7, and E0 corresponds to
Outcome 8. We have
4
p(E3 ) =
28
3 4 4 11
p(E2 ) = + + =
28 28 28 28
3 3 4 10
p(E1 ) = + + =
28 28 28 28
3
p(E0 ) =
28
Since the results of Jackie’s experiment of flipping three coins and Phil’s experiment of flipping
three coins are independent, we conclude from part 5 of Theorem 3.2.1 that the probability
that Jackie and Phil each obtain i heads in their experiment is p(Ei )2 . Thus, the probability
that Jackie and Phil obtain the same number of heads is
m
= p(E0 )2 + p(E1 )2 + p(E2 )2 + p(E3 )2
n
2 2 2 2
4 11 10 3
= + + +
28 28 28 28
42 + 112 + 102 + 32 246 123
= = = .
282 784 392
Thus, m = 123 and n = 392, so that m + n = 515. 2
9. (1993 AIME, Problem #7) Three numbers, a1 , a2 , a3 , are drawn randomly and
without replacement from the set {1, 2, 3, . . . , 1000}. Three other numbers, b1 , b2 , b3 ,
are then drawn randomly and without replacement from the remaining set of
997 numbers. Let p be the probability that, after a suitable rotation, a brick of
dimensions a1 × a2 × a3 can be enclosed in a box of dimensions b1 × b2 × b3 , with the
sides of the brick parallel to the sides of the box. If p is written as a fraction in
lowest terms, what is the sum of the numerator and denominator?
Solution: In this experiment, six distinct
numbers have been drawn in two groups of three.
6
Given six numbers, there are = 20 ways to put them into two groups of three,
3
{a1 , a2 , a3 } and {b1 , b2 , b3 }. It remains to determine how many of these 20 arrangements
satisfy the requirement of the problem. For simplicity, we may assume that the six chosen
numbers are 1, 2, 3, 4, 5, and 6. Then we can simply enumerate those groupings that satisfy
the requirement of the problem:
Brick Dimensions Box Dimensions

1,2,3 4,5,6
1,2,4 3,5,6
1,2,5 3,4,6
1,3,4 2,5,6
1,3,5 2,4,6
Hence, of the 20 ways to place six numbers into two groups of three, five of them lead to
5 1
acceptable brick and box dimension. Therefore, we have p = = . The sum of the
20 4
numerator and denominator here is 1 + 4 = 5 = 005.
Remark: As the solution shows, it is immaterial how many numbers belong to the initial set,
in this case 1000, as long as there are at least six numbers. 2
10. (1990 AIME, Problem #9) A fair coin is to be tossed 10 times. Let i/j, in lowest
terms, be the probability that heads never occur on consecutive tosses. Find i + j.
Solution #1: The experiment of performing ten tosses of the fair coin results in a sample
space S consisting of strings of “H” (heads) and “T” (tails) of length 10. There are 210 such
strings, all of them equally likely. We have |S| = 210 = 1024. We must now determine the
number of such strings that do not contain consecutive heads. Such strings will comprise the
set E. A string with no consecutive heads can have at most five heads (six heads would require
at least five tails to separate each pair of neighboring heads, resulting in a string longer than
length 10). Let h denote the number of heads. We must have 0 ≤ h ≤ 5, and if t denotes the
number of tails, we have t = 10 − h. Placing h heads in a string creates h + 1 spaces (including
a space before the first “H” and a space after the last “H”). Let xi (i = 1, 2, . . . , h + 1) denote
the number of tails in the ith space. Then we have
x1 + x2 + · · · + xh+1 = 10 − h
with x2 , x3 , . . . , xh ≥ 1 in order to guarantee that no two heads are consecutive. If we set

y1 = x1 , yh+1 = xh+1 and yi = xi − 1 for each i = 2, 3, . . . , h, then we obtain the new equation
y1 + y2 + · · · + yh+1 = 11 − 2h,
to be solved for nonnegative integers y1 , y2 , . . . , yh+1 . The method of stars and bars (Section
2.3.2) can be used to solve
this. In this case, we have 11 − 2h stars and h bars, so the total
11 − h
number of solutions is . Summing this expression for h = 0, 1, 2, 3, 4, 5, we find
h
that

11 10 9 8 7 6
+ + + + + = 1 + 10 + 36 + 56 + 35 + 6 = 144
0 1 2 3 4 5
of the strings in S do not have consecutive heads. Thus, |E| = 144, and the probability that
heads never occur on consecutive tosses is
|E| 144 9
= = .
|S| 1024 64
Therefore, we conclude that i = 9, j = 64, and so the answer is i + j = 9 + 64 = 73 = 073. 2

Solution #2: This solution is based on some experience with recurrence relations for se-
quences, and in particular, the Fibonacci sequence (see Section 5.4). We let En denote the set
of sequences of heads (H) and tails (T) of length n that do not contain two consecutive heads.
Such sequences can be split into two (mutually exclusive) types: (a) those that begin with T,
and (b) those that begin with H. For strings of type (a), there are |En−1 | strings of length n
that do not contain two consecutive heads. For strings of type (b), the H must be followed by
T (to avoid two consecutive heads), and there are |En−2 | ways to complete the string of length
n such that we have no two consecutive heads. By the Sum Rule, we have
|En | = |En−1 | + |En−2 |. (11.24)
Since |E1 | = 2 and |E2 | = 3, we leave it to the reader to apply (11.24) repeatedly to find |E3 |,
|E4 |, |E5 |, and so on to find |E10 | = 144. Thus, the probability that heads never occur on
consecutive tosses is
|E10 | 144 9
= = .
210 1024 64
Therefore, we conclude that i = 9, j = 64, and so the answer is i + j = 9 + 64 = 73 = 073. 2
11. (2006 AIME-2, Problem #10) Seven teams play a soccer tournament in which
each team plays every other team exactly once. No ties occur, each team has a
50% chance of winning each game it plays, and the outcomes of the games are
independent. In each game, the winner is awarded 1 point and the loser gets 0
points. The total points are accumulated to decide the ranks of the teams. In the
first game of the tournament, team A beats team B. The probability that team A
finishes with more points than team B is m/n, where m and n are relatively prime
positive integers. Find m + n.
Solution #1: Following the game between team A and team B, both of these teams have
five more games to play. There are 210 different outcomes for these 10 games. In order for
team A to finish with more points than team B, team A must win at least as many games
5
as team B in the five remaining games each team has to play. Note that there are
k
equally-likely ways for either team to win exactly k games. Since the wins of team A and the
wins of team B in their remaining games are independent, we can apply the Multiplication
Principle (Theorem 2.2.1) to the separate results for each team. So if team A has k additional
wins andteam
B has ` additional wins, then the number of ways that this can occur is given
5 5
by . The table below shows the various ways that team A can win at least as
k `
many games as team B along with the number of ways that each such outcome can occur:
k ` Ways for this to occur

0 0 1
1 0 5
1 1 25
2 0 10
2 1 50
2 2 100
3 0 10
3 1 50
3 2 100
3 3 100
4 0 5
4 1 25
4 2 50
4 3 50
4 4 25
5 0 1
5 1 5
5 2 10
5 3 10
5 4 5
5 5 1
The sum of the results in the last column is 638. Hence, among the 210 = 1024 possible results
in the last five games played by team A and team B, team A finishes with more points than
team B in 638 of them. Hence, the probability that team A finishes with more points than
team B is
638 319
= .
1024 512
Hence, m = 319 and n = 512, and we conclude that m + n = 831. 2
Solution #2: This solution uses the symmetry that there is the same probability that team
A wins more of its remaining games than team B as there is that team B wins more of its
remaining games than team A.
The probability that team A finishes with more points than team B is precisely the same as
the probability that team A wins at least as many of its remaining games as team B. Let p
be the probability that team A and team B win exactly the same number of their remaining
games. Thus, using the symmetry alluded to above, the probability that team A wins more of
its remaining games than team B is 21 (1 − p). We have
2 2 2 2 2 2
5 5 5 5 5 5
+ + + + +
0 1 2 3 4 5 252
p= = .
1024 1024
Hence, by symmetry, the probability that team A wins more of its remaining games than team
B is
1 1 772 386
(1 − p) = · = .
2 2 1024 1024
Thus, the probability that team A wins at least as many games as team B is
1 252 + 386 638 319
p + (1 − p) = = = .
2 1024 1024 512
Hence, m = 319 and n = 512, and we conclude that m + n = 831. 2
12. (2003 AIME-2, Problem #13) A bug starts at a vertex of an equilateral triangle.
On each move, it randomly selects one of the two vertices where it is not currently
located, and crawls along a side of the triangle to that vertex. Given that the
probability that the bug moves to its starting vertex on its tenth move is m/n,
where m and n are relatively prime positive integers, find m + n.
Solution #1: At each move, the bug either chooses to move clockwise (denoted by A) or
counter-clockwise (denoted by B) around the triangle. Thus, the 10 moves of the bug can be
represented as a string of A’s and B’s of length 10. There are 210 such strings, and we can
establish a sample space for this problem consisting of these 210 strings.
Let nA denote the number of A’s in one of these strings, and let nB be the number of B’s in
one of these strings. The key observation is that, in order for the bug to return to the starting
vertex after 10 moves, nA − nB must be a multiple of 3. The only possible values of nA and
nB that satisfy this requirement as well as nA + nB = 10 are
nA nB
2 8
5 5
8 2
Thus, we must sum the number of strings of length 10 that contain 2, 5, or 8 A’s:

10 10 10
+ + = 45 + 252 + 45 = 342.
2 5 8
342
Thus, the probability that the bug returns to its starting position on the tenth move is 10 =
2
171
. Hence, we have m = 171 and n = 512, so that m + n = 171 + 512 = 683. 2
512
Solution #2: Denote the vertices of the equilateral triangle by P , Q, and R, and assume
that the bug starts at vertex P . Let pn (resp. qn , rn ) denote the probability that bug is at
vertex P (resp. Q, R) after the nth move. Using the fact that for all n ≥ 1,
1 1
pn = qn−1 + rn−1 ,
2 2
1 1
qn = pn−1 + rn−1 ,
2 2
1 1
rn = pn−1 + qn−1 ,
2 2
we can easily fill in the table below:
Move # pn qn rn
0 1 0 0
1 0 1/2 1/2
2 1/2 1/4 1/4
3 1/4 3/8 3/8
4 3/8 5/16 5/16
5 5/16 11/32 11/32
6 11/32 21/64 21/64
7 21/64 43/128 43/128
8 43/128 85/256 85/256
9 85/256 171/512 171/512
10 171/512
Hence, we have m = 171 and n = 512, so that m + n = 171 + 512 = 683. 2

13. (2002 AIME-2, Problem #12) A basketball player has a constant probability of
.4 of making any given shot, independent of previous shots. Let an be the ratio
of shots made to shots attempted after n shots. The probability that a10 = .4 and
an ≤ .4 for all n such that 1 ≤ n ≤ 9 is given to be pa q b r/(sc ), where p, q, r, and s are
primes, and a, b, and c are positive integers. Find (p + q + r + s)(a + b + c).
Solution: We can draw a graph showing the number of made shots as a function of the
number of shots taken. One example is illustrated in Figure 32.
FIGURE 32 GOES HERE

Since a10 = .4, the point (10, 4) must be on the graph, and no point lying above the line
y = .4x (where x denotes the number of shots taken and y denotes the number of shots made)
can appear on the graph. In particular, the player must make the tenth shot, so that (9, 3)
occurs on the graph. Also, the player must miss the first two shots, so that (1, 0) and (2, 0)
must occur on the graph. We can view the shot history of the basketball player, consisting of
a collection of 11 points, beginning at (0, 0) and ending at (10, 4), as a path consisting of four
moves in the “northeast” direction and six moves in the “east” direction, as shown in Figure
32. The numbers labelling the vertices of the graph indicate the number of legal paths from
(0, 0) to the given point. We see that there are 23 legal paths from (0, 0) to (10, 4). Each of
these 23 paths occurs with probability (.6)6 (.4)4 , since each of the six “east” moves occurs with
probability .6 and each of the four “northeast” moves occurs with probability .4. Therefore,
the probability that the basketball player’s shot history obeys the conditions set forth in the
problem is
23 · 66 · 44 214 · 36 · 23 24 · 36 · 23
23(.6)6 (.4)4 = 10
= 10 10
= .
10 2 5 510
Therefore, we have p = 2, q = 3, r = 23, s = 5, a = 4, b = 6, and c = 10. The answer to the
problem is thus
(p + q + r + s)(a + b + c) = (2 + 3 + 23 + 5)(4 + 6 + 10) = 33 · 20 = 660.
14. (1999 AIME, Problem #13) Forty teams play a tournament in which every team
plays every other team exactly once. No ties occur, and each team has a 50%
chance of winning any game it plays. The probability that no two teams win
the same number of games is m/n, where m and n are relatively prime positive
integers. Find log2 n.
Solution: Each team plays exactly 39 games, so that the number of wins any given team earns
must be a member of the set {0, 1, 2, . . . , 39}. In order for no two teams to win the same number
of games, there must be exactly one team that wins k games, for each k = 0, 1, 2, . . . , 39. Let
p be the probability that team 1 wins 39 games, team 2 wins 38 games, team 3 wins 37 games,
and so on (with team 40 winning no games). In order for this outcome to occur, it must be
the case that team x beats team y if andonly if x < y. The likelihood that team x beats
40
team y is 1/2, and since there are = 780 unordered pairs {x, y}, we conclude that
2
1
p = 780 . There are 40! ways to shuffle the 40 teams so that one of them wins 40 games,
2
one of them wins 39 games, and so on. (See Section 2.3.1.) Therefore, the probability that
40!
no two teams win the same number of games is 780 . This fraction is not in lowest terms,
2
however. We must determine the number of factors of 2 in the number of 40!. Problems of
this nature are discussed in detail in Chapter 4. (See, for instance, Examples 4.2.4 and 4.2.6.)
There are 20 even numbers in {1, 2, . . . , 40}, 10 multiples of 4, 5 multiples of 8, 2 multiples
of 16 and 1 multiple of 32. Thus, the factorization of 40! into a product of primes consists
40!
of 20 + 10 + 5 + 2 + 1 = 38 factors of 2. Hence, when 780 is reduced to lowest terms, the
2
denominator, n, is n = 2780−38 = 2742 . Thus, log2 n = 742. 2
15. (1995 AIME, Problem #15) Let p be the probability that, in the process of repeat-
edly flipping a fair coin, one will encounter a run of 5 heads before one encounters
a run of 2 tails. Given that p can be written in the form m/n, where m and n are
relatively prime positive integers, find m + n.
Solution: We use H to denote the occurrence of a head and T to denote the occurrence of a
tail. Let us call a string successful if it is a sequence of heads and tails in which a run of 5 heads
is encountered before a run of 2 tails. Let pH denote the probability of obtaining a successful
string that begins with H, and let pT denote the probability of obtaining a successful string
that begins with T . Any successful string beginning with T must consist of a T followed by a
successful string that begins with an H. Therefore, it follows that
1
pT = pH . (11.25)
2
Conversely, any successful string that begins with H must begin with exactly one of the
following patterns:
HHHHH HHHHT HHHT HHT HT.
The first string occurs with probability 1/32, while the remaining strings can be used to cast
pH in terms of pT . Using a generalization of part 4 of Theorem 3.2.1 to five events, we have
1 1 1 1 1 1 1 1 15
pH = + pT + pT + pT + pT = + (1 + 2 + 4 + 8)pT = + pT . (11.26)
32 16 8 4 2 32 16 32 16
Solving Equations (11.25) and (11.26) simultaneously, we find that
1 1
pH = and pT = .
17 34
Thus, the probability of obtaining five consecutive heads before two consecutive tails is
1 1 3
p = pH + pT = + = .
17 34 34
Therefore, m = 3 and n = 34, and so we conclude that m + n = 3 + 34 = 37 = 037. 2
1. (2007 AIME-2, Problem #2) Find the number of ordered triples (a, b, c) where a, b,
and c are positive integers, a is a factor of b, a is a factor of c, and a + b + c = 100.
Solution: Since a is a factor of b and c, we can write b = ak and c = a` for positive integers
k and `. Since b, c > 0, we must have k, ` > 0. Now we have
100 = a + b + c = a + ak + a` = a(1 + k + `), (11.27)
which implies that a must be a divisor of 100. We can consider each divisor a of 100 in turn,
from which (11.27) gives us
100
k+`= − 1.
a
100
Then since k, ` ≥ 1, there are − 2 choices for (k, `). Each pair (k, `) corresponds in one-to-
a
one fashion to the distinct choices for (b, c) that then form the triple (a, b, c). We summarize
all of the results in a table.
a k+` Number of choices for (k, `)
1 99 98
2 49 48
4 24 23
5 19 18
10 9 8
20 4 3
25 3 2
50 1 0
100 0 0
Summing the results in the last column of the table and observing that each such pair (k, `)
corresponds to a distinct triple (a, b, c) such that a + b + c = 100, we find the total number of
triples desired in the problem:
98 + 48 + 23 + 18 + 8 + 3 + 2 + 0 + 0 = 200.
2
2. (2006 AIME-2, Problem #3) Let P be the product of the first 100 positive odd
integers. Find the largest integer k such that P is divisible by 3k .
Solution #1: We note that
200! 200!
P = 1 · 3 · 5 · · · 197 · 199 = = 100 ,
2 · 4 · 6 · · · 198 · 200 2 100!
so the number of factors of 3 in P can be computed as the number of factors of 3 in 200! minus
the number of factors of 3 in 100!. We can use Equation (4.3). The number of factors of 3 in
200! is
200 200 200 200
+ + + = 66 + 22 + 7 + 2 = 97,
3 9 27 81
and the number of factors of 3 in 100! is

100 100 100 100
+ + + = 33 + 11 + 3 + 1 = 48.
3 9 27 81
Thus, we conclude that k = 97 − 48 = 49 = 049. 2
Solution #2: We have

P = 1 · 3 · 5 · · · 197 · 199.
There are exactly 33 numbers in this product that contain (at least) one factor of 3, namely:
3, 9, 15, 21, . . . 183, 189, and 195. Of these, all multiples of 9 actually contain (at least) two
factors of 3, namely: 9, 27, 45, 63, 81, 99, 117, 135, 153, 171, and 189. Of these, the multiples of
27 actually contain (at least) three factors of 3, namely: 27, 81, 135 and 189. Finally, there
is just one multiple of 81, which is 81 itself, that contains four factors of 3. The table below
summarizes our observations:
Odd integers between 1 and 200 Number of Entries

` with at least ` factors of 3 in Previous Column
1 3, 9, 15, · · · , 189, 195 33
2 9, 27, 45, 63, 81, 99, 117, 135, 153, 171, 189 11
3 27, 81, 135, 189 4
4 81 1
Summing the values in the last column of the table, we conclude that k = 33 + 11 + 4 + 1 =
49 = 049. 2
3. (2008 AIME, Problem #4) There exist unique positive integers x and y that satisfy
the equation x2 + 84x + 2008 = y 2 . Find x + y.
Solution: Completing the square, we find that
(x + 42)2 + 244 = y 2 .
Therefore,
y 2 − (x + 42)2 = 244.
Factoring the difference of squares, we have
(y − (x + 42)) (y + (x + 42)) = 244 = 4 · 61. (11.28)
From (11.28), observe that y − x − 42 and y + x + 42 must both be positive, since y + x + 42 > 0
(owing to the fact that x, y > 0) and both factors must have the same sign since their product
is positive. Also note that
(y + x + 42) − (y − x − 42) = 2x + 84
is even, so y − x − 42 and y + x + 42 must both be even3 . Appealing again to Equation

(11.28), we see that the only way to decompose 244 as a product of two even positive integers
is 244 = 2 · 122. Thus, we must have
y − x − 42 = 2 and y + x + 42 = 122.
From these two equations, we solve quickly for x and y to obtain x = 18 and y = 62. Hence,
x + y = 18 + 62 = 80 = 080. 2
Remark: Even if the reader had not noticed that y − x − 42 and y + x + 42 must both be even,
one can still arrive at the answer by considering a few other possible factorizations of 244 into
two factors and deducing a contradiction in all cases except y −x−42 = 2 and y +x+42 = 122.
4. (1989 AIME, Problem #3) Suppose n is a positive integer and d is a single digit
in base 10. Find n if
n
= 0.d25d25d25 . . . .
810
3 They cannot both be odd, since their product is even.

Solution: The right-hand side is a repeating decimal and can be represented by the fraction
d25
. (To see this, let x = 0.d25d25 . . . and note that d25 = 1000x − x = 999x.) Therefore,
999
n d25
= .
810 999
Cross multiplying, we have
999 · n = 810 · d25.
Factoring 999 and 810 as products of primes, we have 33 · 37 · n = 2 · 34 · 5 · (d25), or, after
cancelling common factors,
37n = 2 · 3 · 5 · (d25). (11.29)
Since any number that ends in “25” must be a multiple of 25, d25 is a multiple of 25. Moreover,
from Equation (11.29), d25 is a multiple of 37 as well (since a factor of 37 must occur in the
expression on the right-hand side, by the Fundamental Theorem of Arithmetic). But the only
three-digit number that is a multiple of both 25 and 37 is 25 · 37 = 925. Hence, d = 9, and
thus, from Equation (11.29), 37n = 2 · 3 · 5 · 25 · 37, and n = 2 · 3 · 5 · 25 = 750. 2
Remark: Even from the fact that d25 is a multiple of 37 alone (discarding the fact that d25
is a multiple of 25), we can deduce that d = 9, because d25 must equal 37 times some number
ending in 5. After trying 37 · 5 and 37 · 15, neither of which ends in “25”, we must conclude
that 37 · 25 = d25.
5. (2006 AIME, Problem #4) Let N be the number of consecutive 0’s at the right end
of the decimal representation of the product 1!2!3!4! · · · 99!100!. Find the remainder
when N is divided by 1000.
Solution: The number of consecutive 0’s at the right end of the decimal representation of the
given product is precisely the number of factors of 10 in this product. Each factor of 10 is
comprised of one factor of 2 and one factor of 5. In any number k!, the number of factors of
5 in its prime factorization cannot exceed the number of factors of 2. Therefore, the number
of factors of 10 in the given product is precisely the number of factors of 5. We can sum the
number of factors of 5 occurring in k! for each k with 1 ≤ k ≤ 100.
According to Equation (4.3), the number of factors of 5 occurring in k! is given by
∞
X k
.
5`
`=1
Hence, the number of consecutive 0’s at the right end of the decimal representation of the
product 1!2!3!4! · · · 99!100! is
∞
100 X 2 X
100 100 100
X k X k X k X k
N= = = + ,
5` 5` 5 25
k=1 `=1 `=1 k=1 k=1 k=1

k
where we have reversed the order of summation and noted that for ` ≥ 3, = 0 since
5`
k ≤ 100. The (nonzero) terms in the first summation on the right-hand side are as follows,
five terms each of 1,2, · · · , 19, and then one term (for k = 100) of 20. The (nonzero) terms of
the second summation on the right-hand side are: 25 terms each of 1,2, and 3, and then one
term (for k = 100) of 4. Thus, we have
N = [5 (1 + 2 + 3 + · · · + 19) + 20] + [25 (1 + 2 + 3) + 4]

(19)(20)
= 5· + 20 + [25 · 6 + 4]
2
= 950 + 20 + 150 + 4
= 1124.
Therefore, the remainder when N is divided by 1000 is 124. 2
6. 1994 AIME, Problem #5) Given a positive integer n, let p(n) be the product of
the nonzero digits of n. (If n has only one digit, then p(n) is equal to that digit.)
Let
S = p(1) + p(2) + p(3) + · · · + p(999).
What is the largest prime factor of S?
Solution #1: This is a nice example of a problem where it can be beneficial to look at a
smaller example first. We might, for instance, consider
T = p(1) + p(2) + p(3) + · · · + p(99).
Now of course,
p(1) + p(2) + · · · + p(9) = 1 + 2 + · · · + 9 = 45.
Thus,
p(10) + p(11) + p(12) + · · · + p(19) = p(10) + 45 = 46,

p(20) + p(21) + p(22) + · · · + p(29) = 2 · 46,
p(30) + p(31) + p(32) + · · · + p(39) = 3 · 46,
..
.
p(90) + p(91) + p(92) + · · · + p(99) = 9 · 46.
Thus,
T = 45 + 46(1 + 2 + 3 + · · · + 9) = 45 + 46 · 45 = 45 · 47.
How can this help us determine S? The key is the next observation:
T = p(1) + p(2) + · · · + p(99)

T = p(101) + p(102) + · · · + p(199)
2T = p(201) + p(202) + · · · + p(299)
..
.
9T = p(901) + p(902) + · · · + p(999).
All terms of S with the exception of p(100), p(200), · · · , p(900) are listed in this array. There-
fore,
S = p(100) + p(200) + · · · + p(900) + T + T + 2T + · · · + 9T
= (1 + 2 + · · · + 9) + T (1 + 1 + 2 + · · · + 9)
= 45 + 46T
= 45 + 45 · 46 · 47
= 45(1 + 46 · 47)
= 45 · 2163
= 45 · 21 · 103.
Clearly, 103 is the largest prime factor of S. 2
Solution #2: As the problem is stated, the number of digits that are multiplied in computing
p(n) depends on how many of the digits of n are zero. However, one way to skirt this depen-
dence is to institute a rule that, anytime a zero is encountered as a digit in n, we multiply by 1
(which means we essentially “do nothing” to the value of p(n)). In so doing, we can make the
computation of each integer p(n) a product of three digits. Hence, if we compute the product
(1 + 1 + 2 + · · · + 8 + 9)(1 + 1 + 2 + · · · + 8 + 9)(1 + 1 + 2 + · · · + 8 + 9),
where the first “1” in each factor is reserved for the case that the corresponding digit is zero,
the distribution of terms gives all of the terms of S and one extra term (corresponding to
choosing the first “1” in all three factors). Thus, using the well-known formula for a difference
of cubes4 , we have
S = (1 + 1 + 2 + · · · + 8 + 9)3 − 1 = 463 − 13 = (46 − 1)(462 + 46 + 1) = 45 · 2163 = 45 · 21 · 103,
and the largest prime factor of S is 103. 2
7. (2006 AIME-2, Problem #7) Find the number of ordered pairs of positive integers
(a, b) such that a + b = 1000 and neither a nor b has a zero digit.
Solution #1: We will examine the possible values of a with 1 ≤ a ≤ 999 by considering
intervals of 100 values of a. We summarize the results for a ≤ 500 in the table below:
Range of values for a Values of a such that a and b have no zero digits # of pairs
1 ≤ a ≤ 100 1, 2, . . . , 9, 11, 12, . . . , 19, . . . , 81, 82, . . . , 89 81
101 ≤ a ≤ 200 111, 112, . . . , 119, . . . , 181, 182, . . . , 189 72
201 ≤ a ≤ 300 211, 212, . . . , 219, . . . , 281, 282, . . . , 289 72
301 ≤ a ≤ 400 311, 312, . . . , 319, . . . , 381, 382, . . . , 389 72
401 ≤ a ≤ 500 411, 412, . . . , 419, . . . , 481, 482, . . . , 489 72
By symmetry, the results for a > 500 will consist of the same number of pairs (a, b). Therefore,
the total number of pairs is
2(81 + 72 + 72 + 72 + 72) = 2 · 369 = 738.

4 This formula is a3 − b3 = (a − b)(a2 + ab + b2 ).
Solution #2: Assume at first that both a and b are three-digit numbers and write a and b
each as three-digit numbers, say
a = rst and b = xyz,
where none of the digits r, s, t, x, y, z are zero. Since a + b = 1000, we must have
t + z = 10, s + y = 9, and r + x = 9.
There are 9 choices for the value of the nonzero digit t (from which z is determined), then 8
choices for the value of the nonzero digit s (from which y is determined), and then 8 choices
for the value of r (from which x is determined). By the Multiplication Principle, this gives us
a total of 9 · 8 · 8 = 576 pairs (a, b).
However, this only accounts for pairs of three-digit numbers. It is possible that a or5 b has
only one or two digits. Let us assume for the moment that a has only two digits, say a = st,
where neither s nor t is zero, and b is as above. Again, there are 9 choices for the value of t
(from which z is determined), then 8 choices for s (from which y is determined), and in this
case, x = 9. By the Multiplication Principle, we therefore have 9 · 8 = 72 pairs (a, b) of this
type. Moreover, there are 9 pairs (a, b) where a is a single digit (1–9). So we have a total of
72 + 9 = 81 pairs in which a has two or fewer digits. By symmetry, we will also have 81 such
pairs in the case where b has two or fewer digits.
Adding all of the results above, we have 576 + 81 + 81 = 738 pairs. 2
Remark: Solution #2 is really a counting problem via three cases: (a) both numbers a and b
are three-digit numbers, (b) one of the numbers a or b (but not both) has only two digits, and
(c) one of the numbers a or b (but not both) has only one digit. We obtained 576 pairs in case
(a), 144 pairs in case (b), and 18 pairs in case (c). Then the answer is 576 + 144 + 18 = 738. 2
8. (1999 AIME, Problem #7) There is a set of 1000 switches, each of which has four
positions, called A, B, C, and D. When the position of any switch changes, it is
only from A to B, from B to C, from C to D, or from D to A. Initially each switch
is in position A. The switches are labeled with the 1000 different integers 2x 3y 5z ,
where x, y, and z take on the values 0, 1, . . . , 9. At step i of a 1000-step process,
the ith switch is advanced one step, and so are all the other switches whose labels
divide the label on the ith switch. After step 1000, has been completed, how many
switches will be in position A?
Solution: Observe that a switch is in position A if and only if the number of times it has been
advanced is a multiple of 4. Therefore, we begin by determining the number of times that a
given switch 2x 3y 5z is advanced. This switch is advanced a number of times that is precisely
equal to the number of switches whose label 2a 3b 5c has 2x 3y 5z as a divisor. This requires
x ≤ a ≤ 9, y ≤ b ≤ 9, and z ≤ c ≤ 9.
There are 10−x values of a, 10−y values of b, and 10−z values of c that satisfy these inequalities.
Therefore, the number of switches whose label has 2x 3y 5z as a divisor is (10−x)(10−y)(10−z).
5 This is clearly an exclusive “or”, since a and b must sum to 1000.
Therefore, we must determine for how many triples (x, y, z) we have (10 − x)(10 − y)(10 − z)
divisible by 4. Let us consider four cases:
Case 1: x, y, z are all odd. In this case (10 − x)(10 − y)(10 − z) is odd, hence never divisible
by 4. Thus, we have no triples to count in this case.
Case 2: Exactly two of x, y, z are odd: There are three choices for which one member of
{x, y, z} is even. Let us say x is even. Then y and z are odd, and 10 − y and 10 − z are both
odd, so that (10 − x)(10 − y)(10 − z) will be divisible by 4 if and only if 10 − x is divisible by 4,
if and only if x = 2 or x = 6. Therefore, there are two choices for the value of x. Since y and
z are odd, they each can independently assume a value from the set {1, 3, 5, 7, 9}. So each of
these variables has a value that can be chosen in one of five ways. Multiplying the choices we
have enumerated, there are 3 · 2 · 5 · 5 = 150 triples (x, y, z) that satisfy the requirement that
(10 − x)(10 − y)(10 − z) is divisible by 4.
Case 3: Exactly one of x, y, z is odd: There are three choices for which one member of
{x, y, z} is odd. Without loss of generality, let us say x is odd. It can be chosen to be any
value from the set {1, 3, 5, 7, 9}. Since both y and z are even, 10 − y and 10 − z will both be
even, and hence, (10 − x)(10 − y)(10 − z) is guaranteed to be divisible by 4. Thus, each of the
5 · 5 = 25 ways of assigning even values to y and z will produce a triple (x, y, z) that satisfies
the requirement that (10 − x)(10 − y)(10 − z) is divisible by 4. Multiplying the choices we have
enumerated here, there are 3 · 5 · 25 = 375 triples (x, y, z) to count in this case.
Case 4: None of x, y, z are odd. In this case, x, y, z are all even, and regardless of their
values, we will have (10 − x)(10 − y)(10 − z) divisible by 4. There are 53 = 125 such triples to
count in this case.
Summing the results in these four cases, we find that there are 150 + 375 + 125 = 650 triples
(x, y, z) such that (10 − x)(10 − y)(10 − z) is divisible by 4. Therefore, we have 650 switches
that finish the 1000-step process in position A. 2
Remark: Instead of counting triples (x, y, z) such that (10−x)(10−y)(10−z) is divisible by 4,

the reader might prefer to attack the complementary problem and then apply the Subtraction
Rule. That is absolutely appropriate, and the interested reader is encouraged to pursue this
alternate strategy to solve the problem. 2
9. (2008 AIME, Problem #7) Let Si be the set of all integers n such that 100i ≤ n <
100(i + 1). For example, S4 is the set {400, 401, 402, . . . , 499}. How many of the sets
S0 , S1 , S2 , . . . , S999 do not contain a perfect square?
Solution: The largest member of the last set, S999 , is 99, 999. Let us begin by determining
the number of perfect squares √ that belong to√ some set S0 , S1 , . √ . . , S999 . This requires us
to compute
√ (approximately) 100,000 = 100 10. Since 3.1 < 10 < 3.2, we know that
310 < 100,000 < 320. Direct computation shows that 3162 = 99, 856 and 3172 = 100, 489.
Therefore, 316 perfect squares (12 , 22 , . . . , 3162 ) belong to some set S0 , S1 , . . . , S999 . However,
some of these perfect squares (such as, for instance, 102 , 112 , 122 , 132 , 142 all belong to S1 )
belong to the same set Si .
Observe that the difference between two consecutive squares is (n + 1)2 − n2 = 2n + 1. For
n ≥ 50, (n + 1)2 − n2 > 100, and since |Si | = 100 for each i, we see that 502 , 512 , . . . , 3162
each belong to a different set Si . On the other hand, for n < 50, the difference between
two consecutive squares is less than 100. Therefore, since 2400 < 492 < 2500, each of the sets
S1 , S2 , . . . , S24 must contain a perfect square. In the remaining 975 sets S25 , S26 , . . . , S999 , each
of the perfect squares 502 , 512 , . . . , 3162 must occur in distinct sets. There are 316−50+1 = 267
perfect squares here. Hence, the number of sets among the 975 sets S25 , S26 , . . . , S999 that do
not contain a perfect square is 975 − 267 = 708. 2
10. (2004 AIME-2, Problem #8) How many positive integer divisors of 20042004 are
divisible by exactly 2004 positive divisors?
Solution: Since the prime factorization of 2004 is 2004 = 22 · 3 · 167, we have 20042004 =
24008 · 32004 · 1672004 . Every divisor d of 20042004 therefore takes the form d = 2a 3b 167c , where
0 ≤ a ≤ 4008, 0 ≤ b ≤ 2004, and 0 ≤ c ≤ 2004. (11.30)
According to Theorem 4.2.2, the number of divisors of such a number d is
τ (d) = (a + 1)(b + 1)(c + 1).
Hence, we must determine the number of solutions (a, b, c) to
(a + 1)(b + 1)(c + 1) = 2004 = 22 · 3 · 167 (11.31)
subject to the conditions in (11.30). In accordance with the Fundamental Theorem of Arith-
metic, we are determining the number of ways to assign the prime factors on the right side of
Equation (11.31) to the three factors a + 1, b + 1, and c + 1 on the left side. There are six ways
to distribute the two factors of 2 among the three factors on the left side (three ways to assign
them both to the same factor and three ways to assign one each to two of the factors). Next,
there are (independently) three ways to distribute the factor of 3 to one of the factors on the
left side of (11.31), and there are (independently) three ways to distribute the factor of 167 to
one of the factors on the left side of (11.31). By the Multiplication Principle (Theorem 2.2.1),
we have a total of 6 · 3 · 3 = 54 ways to assign the prime factors of 2004 to a + 1, b + 1, and
c + 1. Hence, the answer is 54 = 054. 2
11. (1993 AIME, Problem #9) Two thousand points are given on a circle. Label
one of the points 1. From this point, count 2 points in the clockwise direction
and label this point 2. From the point labeled 2, count 3 points in the clockwise
direction and label this point 3. See Figure. Continue this process until the labels
1, 2, 3, . . . , 1993 are all used. Some of the points on the circle will have more than
one label and some points will not have a label. What is the smallest integer that
labels the same point as 1993?
FIGURE from 1993 AIME, Problem 9 GOES HERE
Solution #1: First observe that the point with label k (with k ≥ 2) occurs
k(k + 1)
2 + 3 + · · · + k = (1 + 2 + · · · + k) − 1 = −1 (mod 2000)
2
points clockwise around the circle from the point labeled 1. In particular, the point labeled
1993 is
(1993)(1994)
−1 (mod 2000)
2
points clockwise around the circle from the point labeled 1. Therefore, we are seeking the
smallest positive integer k such that
k(k + 1) (1993)(1994)
−1≡ −1 (mod 2000),
2 2
or
k(k + 1) (1993)(1994)
≡ (mod 2000),
2 2
or
k(k + 1) ≡ 1993 · 1994 (mod 4000). (11.32)
Equation (11.32) implies, in particular, that
k(k + 1) ≡ 1993 · 1994 ≡ (−7)(−6) ≡ 42 (mod 2000),
since any equation that holds modulo n also holds modulo any divisor of n (see part 5 of
Proposition 4.4.1). Thus, we can solve
k(k + 1) ≡ 42 (mod 2000) (11.33)
and then find the smallest such solution k that satisfies Equation (11.32). We can rewrite
Equation (11.33) as
k 2 + k − 42 ≡ (k + 7)(k − 6) ≡ 0 (mod 2000).
Note that 2000 = 24 · 53 , and since k + 7 and k − 6 differ by 13, only one of them can be even
and only one of them can contain factors of 5. Thus, we have two possible cases:
Case 1: k + 7 is a multiple of 125 and k − 6 is a multiple of 16. The smallest positive
value of k such that k + 7 is divisible by 125 is k = 118, and in this case, k − 6 is a multiple
of 16 (since 112 = 16 · 7).
Case 2: k + 7 is a multiple of 16 and k − 6 is a multiple of 125. Clearly, no positive
value of k less than 118 will meet both of these requirements.
Now it is a simple matter to plug k = 118 into Equation (11.32), or one of the equations above
it, to verify that it is indeed a solution to our problem. 2
Solution #2: We can observe that
1993 · 1994 ≡ 2042 (mod 4000).
Thus, referring to Equation (11.32), we are seeking the smallest positive integer k such that
k(k + 1) ≡ 2042 (mod 4000).
Since 4000 = 25 · 53 = 32 · 125, this congruence equation is equivalent to the system
k(k + 1) ≡ 2042 ≡ 26 (mod 32) and k(k + 1) ≡ 2042 ≡ 42 (mod 125),

since divisibility by 4000 is equivalent to divisibility by both 32 and 125. The trick now is to
rewrite each of the congruences so that the resulting quadratic equation in k can be factored.
In the case of the first equation, for instance, we replace 26 with 26 + 64 = 90, since 90 ≡ 26
(mod 32):
k 2 + k − 90 ≡ 0 (mod 32) and k 2 + k − 42 ≡ 0 (mod 125).
Factoring, we obtain
(k + 10)(k − 9) ≡ 0 (mod 32) and (k + 7)(k − 6) ≡ 0 (mod 125). (11.34)
Consider the latter congruence, which states that 125 divides (k + 7)(k − 6). It is impossible
for both k + 7 and k − 6 to be divisible by 5 (since they differ by 13), so we conclude that either
k + 7 is divisible by 125 or k − 6 is divisible by 125. Some of the positive values of k satisfying
this condition (in increasing order) are
k = 6, 118, 131, 243, 256, 368, . . . . (11.35)
Applying similar reasoning to the congruence (k + 10)(k − 9) ≡ 0 (mod 32), we find that either
k + 10 is divisible by 32 or k − 9 is divisible by 32. Some of the positive values of k satisfying
this condition (in increasing order) are
k = 9, 22, 41, 54, 73, 86, 105, 118, 137, . . . . (11.36)
The solution to this problem is the smallest positive integer k that satisfies both of the con-
gruences in Equation (11.34). Comparing the lists in (11.35) and (11.36), we conclude that
k = 118 is the answer.
Remark: In solving systems of congruences with moduli that are pairwise relatively prime
(such as in this case, where we have moduli 32 and 125), it is also possible to appeal to the
Chinese Remainder Theorem. However, this requires the student to know this theorem as well
as the somewhat complicated formula for determining the solution to the system. In this case,
this process would result in some large numbers and cumbersome calculations, so we have
opted for the direct solution presented above. 2
12. (2000 AIME, Problem #11) Let S be the sum of all numbers of the form a/b,
where a and b are relatively prime positive divisors of 1000. What is the greatest
integer that does not exceed S/10?
Solution: We have 1000 = 23 · 53 . Since the numbers to be summed have the form a/b, where
a and b are relatively prime, each such number can be expressed as a product xy, where x = 2k
for some integer k with −3 ≤ k ≤ 3, and y = 5` for some integer ` with −3 ≤ ` ≤ 3. Thus, we
can view S as a sum of 49 terms (7 choices each for k and `). There are a variety of ways to
organize these terms. We provide one example here:

3
X 3
X
S= 2k 5`
k=−3 `=−3
3
! 3
!
X X
k `
= 2 5
k=−3 `=−3

1 1 1 1 1 1
= + + +1+2+4+8 + + + 1 + 5 + 25 + 125
8 4 2 125 25 5
127 19531
= ·
8 125
= 2480.437
Hence, the greatest integer that does not exceed S/10 is 248. 2
13. (1987 AIME, Problem #11) Find the largest possible value of k for which 311 is
expressible as the sum of k consecutive positive integers.
Solution: Suppose that 311 is the sum of k consecutive positive integers, say6
311 = (n + 1) + (n + 2) + (n + 3) + · · · + (n + k)
k(k + 1)
= nk + ,
2
where we have used the formula in Theorem 5.4.2 for the sum of the first k positive integers.
Rearranging this expression, we can write
2 · 311 = 2nk + k(k + 1) = k(2n + k + 1). (11.37)
Thus, k divides 2 · 311 . Therefore, we must have k = 2i 3j for some integers i and j with
0 ≤ i ≤ 1 and 0 ≤ j ≤ 11. There are only 24 ways to assign values of i and j subject to
these restrictions, and hence, there are only 24 possible values of k. They can each be checked
directly, and it will go even faster once one realizes that many of these values of j can be
almost immediately eliminated. For instance, when i = 1, Equation (11.37) becomes
2 · 311 = 2 · 3j (2n + 2 · 3j + 1),
so that
311−j = 2n + 2 · 3j + 1. (11.38)
Note that 11 − j ≥ j, for if 11 − j < j, then the right side of (11.38) would exceed the left side.
Thus, we must have j ≤ 5. Indeed, if j = 5, we obtain
36 = 729 = 2n + 2 · 35 + 1 = 2n + 486 + 1,
which can be solved for the positive integer n. Thus, k = 2 · 35 = 486 works.
6 If one calls these positive integers n, n + 1, . . . , n + k − 1, an essentially equivalent argument to what follows can
be used.
On the other hand, if i = 0, then Equation (11.37) reads
2 · 311 = 3j (2n + 3j + 1),
or
2 · 311−j = 2n + 3j + 1.
If j ≥ 6, the right-hand side once more exceeds the left-hand side, a contradiction. Thus,
j ≤ 5 in this case as well, but the value of k obtained is less than 486 found earlier. Therefore,
k = 486 is the solution to the problem. 2
14. (2005 AIME, Problem #12) For positive integers n, let τ (n) denote the number of
positive integer divisors of n, including 1 and n. For example τ (1) = 1 and τ (6) = 4.
Define S(n) by
S(n) = τ (1) + τ (2) + · · · + τ (n).
Let a denote the number of positive integers n ≤ 2005 with S(n) odd, and let b
denote the number of positive integers n ≤ 2005 with S(n) even. Find |a − b|.
Solution: A couple of basic observations will make this problem quite tractable:
(a) τ (n) is odd if and only if n is a perfect square.

Reason: For each divisor d of n, n/d is another divisor of n. Provided that d 6= n/d,
therefore, we can view d and n/d as a pair of divisors of n. Hence, if n is not a perfect
square, then all divisors of n belong to a pair, so that the number of divisors of n is even.
However, if d = n/d for some integer d, then n = d2 is a perfect square and {d, n/d} does
not form a pair of divisors of n. Pairing up all of the other divisors of n, we end up with
an odd number of divisors of n in this case.
(b) Based on the first observation, we see that, for n ≥ 2, the parity (i.e. even or odd
property) of S(n − 1) and S(n) are opposite precisely when n is a perfect square.
(c) Consecutive perfect squares n2 and (n + 1)2 differ by
(n + 1)2 − n2 = 2n + 1.
If we enumerate the first few values of S(n), we see that S(1), S(2), S(3) are odd, then
S(4), S(5), S(6), S(7), S(8) are even, then S(9), S(10), S(11), S(12), S(13), S(14), S(15) are odd,
and so on. From this pattern, observe that for all odd n, the streak
{S(n2 ), S(n2 + 1), . . . , S((n + 1)2 − 1)} (11.39)
consists solely of odd integers, while for all even n, the streak (11.39) consists solely of even
integers. Note that 442 < 2005 < 452 , so that there are going to be 22 “odd streaks” and 22
“even streaks” in the list {S(1), S(2), . . . , S(2004), S(2005)}. To determine a, we note that the
first odd streak has length 3, the second one has length 7, the third one has length 11, and so
on. Thus, summing the first 22 positive integers of the form 4k + 3 (where k an integer with
0 ≤ k ≤ 21), we have
a = 3 + 7 + 11 + 15 + 19 + · · · + 75 + 79 + 83 + 87
= 990.
Since S(n) is either even or odd for every n ≥ 1, we have a + b = 2005, and we conclude that
b = 2005 − 990 = 1015. Therefore, |a − b| = 25 = 025.
Caution: If one elects instead to add the lengths of the 22 even “streaks”, be careful to note that
the last such “streak” is cut short: S(442 ), S(442 + 1), . . . , S(2005). Since 2005 − 442 + 1 = 70,
the last term to be summed is only 70, not 89 as it otherwise would be. 2
15. (1992 AIME, Problem #15) Define a positive integer n to be a factorial tail if
there is some positive integer m such that the decimal representation of m! ends
with exactly n zeros. How many positive integers less than 1992 are not factorial
tails?
Solution: Reviewing from Example 4.2.4, the number of zeros at the end of any positive
integer N is the number of factors of 5 in the prime factorization of N . Clearly, the number
of factors of 5 in the number k! is a (weakly) increasing function of the positive integer k.
Key Observation: The number (k + 1)! will contain r more factors of 5 than k! for some integer
r ≥ 1 if and only if k + 1 = a · 5r , where a is a positive integer with gcd(a, 5) = 1.
For instance, if k + 1 is a multiple of 25, for example, then (k + 1)! will contain (at least) two
more factors of 5 in its prime factorization than k does, and hence, the decimal representation
of (k + 1)! ends in (at least) two more zeros than the decimal representation of k! does. For
instance, 74! ends in 15 zeros, while 75! ends in 17 zeros, so that 16 is not a factorial tail.
Similarly, if k + 1 is a multiple of 125, then (k + 1)! will contain (at least) three more factors
of 5 in its prime factorization than k does, and hence, there will be (at least) two non-factorial
tails associated with the move from k! to (k + 1)!. Extending this one step further, we see that
if k + 1 is a multiple of 625, then (k + 1)! will contain (at least) four more factors of 5 in its
prime factorization than k does, and hence, there will be (at least) three non-factorial tails
associated with the move from k! to (k + 1)!. And so on.
Our strategy, then, is to find, if possible, an integer x such that x! ends in 1992 zeros. Then we
can determine the number of multiples of 25, 125, 625, and so on, less than x. We will be able
to use this information to determine the number of non-factorial tails. To do this, we seek an
algorithm for determining the number of zeros at the end of x! for every (positive) integer x.
The answer is that each (positive) multiple of 5 that does not exceed x contributes one factor
of 5 to x!, each multiple of 25 that does not exceed x contributes one additional factor of 5 to
x!, each multiple of 125 that does not exceed x contributes yet another additional factor of 5
to x!, and so on. Therefore, we can test x by summing the quotients obtained by dividing x
by all powers of 5. This sum is the number of zeros at the end of x!.
1992
Since 100! ends in 24 zeros (see Example 4.2.4) and = 83, it is reasonable to expect that
24
8300! ends in approximately 1992 zeros. The problem is that there are no multiples of 125,
625, 3125, etc. between 1 and 100, but there are such multiples between 1 and 8300. As they
are encountered, they engender extra factors of 5. Therefore, we must be more careful. We can
safely assert that x < 8300. If we apply the algorithm described in the preceding paragraph to
the values of x listed in the table below, we obtain the number of zeros at the end of x! given
in the last column (by summing the values in the previous five columns):
x Factors Factors Factors of Factors Factors Number of zeros

of 5 of 25 of 125 of 625 of 3125 at the end of x!
7500 1500 300 60 12 2 1874
7600 1520 304 60 14 2 1898
7700 1540 308 61 12 2 1923
7800 1560 312 62 12 2 1948
7900 1580 316 63 12 2 1973
8000 1600 320 64 12 2 1998
From the table, we see that in order for x to have 1992 zeros at the end of its decimal
representation, we must have 7900 < x < 8000. A little more experimentation gives the exact
answer: x = 7980 (actually, any integer x with 7980 ≤ x ≤ 7984 works).

7980
Therefore, since = 319, there are 319 multiples of 25 between 1 and 7980, thus 319
25
non-factorial tails.
However,
each multiple of 53 = 125 contributes an additional non-factorial
7980
tail. There are = 63 multiples of 125 between 1 and 7980, hence 63 additional non-
125
factorial tails.
However,
each multiple of 54 = 625 contributes yet another non-factorial tail.
7980
There are = 12 such additional non-factorial tails. Finally, each multiple of 55 = 3125
625
7980
contributes yet another non-factorial tail. There are = 2 such additional non-factorial
3125
tails. Summing the number of non-factorial tails, we achieve the answer:
319 + 63 + 12 + 2 = 396.
1. (1999 AIME, Problem #1) Find the smallest prime that is the fifth term of an
increasing arithmetic sequence, all four preceding terms also being prime.
Solution: We are seeking to find a sequence of five prime numbers p1 < p2 < p3 < p4 < p5
with a common difference between consecutive primes in the sequence. First observe that each
prime in the sequence must be odd. Otherwise, if p1 = 2, then since p2 must be odd, the
common difference p2 − p1 would be odd, which would imply that p3 is again even, contrary
to the fact that 2 is the only even prime. Since each prime in the sequence is odd, note that
the common difference k is even.
If k = 2, the sequence reads
p1 , p1 + 2, p1 + 4, p1 + 6, p1 + 8. (11.40)
Modulo 3, every congruence class7 0, 1, and 2 must appear in the sequence (11.40) since p1 ,
p1 + 2, and p1 + 4 are mutually distinct modulo 3. Hence, at least one of the terms of (11.40) is
7 That is, remainder upon division by 3.
a multiple of 3 (hence, since prime, equal to 3). This requires p1 = 3, but then p4 = p1 + 6 = 9
is not prime, a contradiction. Hence, k > 2.
If k = 4, it is easy to see by considering the terms

p1 , p1 + 4, p1 + 8, p1 + 12, p1 + 16
modulo 5 that one of these five numbers must be divisible by 5 (hence, since prime, equal to
5). This requires p1 = 5. However, if p1 = 5, then we have p2 = p1 + 4 = 9 is not prime.
Therefore, k > 4.
Next, we try k = 6. We quickly see that it is possible to create an arithmetic sequence of five
primes with common difference 6:
5, 11, 17, 23, 29.
Therefore, since the problem asks for the fifth term of this sequence, we see that the answer is
29 = 029. 2
2. (2008 AIME-2, Problem #1) Let N = 1002 + 992 − 982 − 972 + 962 + · · · + 42 + 32 − 22 − 12 ,
where the additions and subtractions alternate in pairs. Find the remainder when
N is divided by 1000.
Solution #1: We can group every other term and use a difference of squares as follows:
N = (1002 − 982 + 962 − 942 + · · · + 42 − 22 ) + (992 − 972 + 952 − 932 + · · · + 32 − 12 )
= (100 − 98)(100 + 98) + (96 − 94)(96 + 94) + · · · + (4 − 2)(4 + 2)+
+ (99 − 97)(99 + 97) + (95 − 93)(95 + 93) + · · · + (3 − 1)(3 + 1)

= 2 (100 + 98 + 96 + · · · + 4 + 2) + (99 + 97 + 95 + · · · + 5 + 3 + 1)
100
X
=2 k
k=1
(100)(101)
=2
2
= 10100,
where we have used part 1 of Theorem 5.4.2. The remainder when N is divided by 1000 is
100. 2
Solution #2: This follows a similar approach to Solution #1, but this time, we apply the
difference of squares factorization to each pair of consecutive terms, with the exception of 1002 .
Hence, we obtain
N = 1002 + (992 − 982 ) − (972 − 962 ) + (952 − 942 ) − · · ·
= 1002 + (99 − 98)(99 + 98) − (97 − 96)(97 + 96) + (95 − 94)(95 + 94) − · · ·
= 1002 + (99 + 98 − 97 − 96) + (95 + 94 − 93 − 92) + · · · + (3 + 2 − 1 − 0)
= 1002 + 4 · 25
= 1002 + 100
= 10100,
where we note that each grouping of four terms (4n + 3) + (4n + 2) − (4n + 1) − 4n sums to
4 and has a multiple of 4 as its smallest (rightmost) term. Therefore, since we have 25 such
groupings of four terms among the first 100 nonnegative integers, these groupings collectively
sum to 4 · 25 = 100, as shown in the calculation. Thus, N = 1002 + 100 = 10100 and the
answer is 100. 2
3. (2005 AIME-2, Problem #3) An infinite geometric series has sum 2005. A new
series, obtained by squaring each term of the original series, has sum 10 times the
sum of the original series. The common ratio of the original series is m/n, where
m and n are relatively prime positive integers. Find m + n.
Solution: Assume the original series has first term a and common ratio r. In order for it to
sum to a finite value, we know that |r| < 1. According to Equation (5.12), the sum of this
series is
a
= 2005. (11.41)
1−r
The new series has first term a2 , common ratio r2 , and sums to 10 times that of (11.41):
a2
= 20050. (11.42)
1 − r2
Using Equation (11.41), we can eliminate a by substituting a = 2005(1 − r) into Equation
(11.42) to obtain
20052 (1 − r)2
= 20050.
1 − r2
This can be simplified to
2005(1 − r)2 = 10(1 − r2 ) = 10(1 − r)(1 + r).
Therefore,
2005(1 − r) = 10(1 + r),
or
2015r = 1995.
Thus,
1995 399
r=
= .
2015 403
Since 399 and 403 are relatively prime8 , we conclude that m = 399 and n = 403. Thus,
m + n = 399 + 403 = 802. 2
4. (2001 AIME-2, Problem #3) Given that
x1 = 211,
x2 = 375,
x3 = 420,
x4 = 523, and
xn = xn−1 − xn−2 + xn−3 − xn−4 when n ≥ 5,
8 Any common positive divisor would have to be at most 403 − 399 = 4, and only 1 satisfies this condition.
find the value of x531 + x753 + x975 .

Solution: Obviously it is not practical to compute x531 , x753 , x975 directly by applying the
recurrence relation over and over. However, in accord with one of our recommended strategies
for handling sequences and series, we should suspect that a pattern might emerge by computing
a few additional terms. Indeed, we find that
x5 = 267, x6 = −211, x7 = −375, x8 = −420, x9 = −523, x10 = −267,
x11 = 211, x12 = 375, x13 = 420, x14 = 523, x15 = 267, . . .
From these data, it should be apparent that xn = −xn−5 for all n ≥ 6. This can be verified
rigorously by substituting xn−1 = xn−2 −xn−3 +xn−4 −xn−5 into the given recurrence relation
to obtain
xn = xn−1 − xn−2 + xn−3 − xn−4
= (xn−2 − xn−3 + xn−4 − xn−5 ) − xn−2 + xn−3 − xn−4
= −xn−5 ,
which holds for all n ≥ 6. Thus, xn = xn−10 for all n ≥ 11, and thus, x531 = x1 , x753 = x3 ,
and x975 = x5 . We conclude that
x531 + x753 + x975 = x1 + x3 + x5

= 211 + 420 + 267
= 898.
2
5. (1985 AIME, Problem #5) A sequence of integers a1 , a2 , a3 , . . . is chosen so that

an = an−1 − an−2 for each n ≥ 3. What is the sum of the first 2001 terms of this
sequence if the sum of the first 1492 terms is 1985, and the sum of the first 1985
terms is 1492?
Solution: It may be unclear how to begin, in which case it may well be profitable to start by
writing out the first several terms. Of course, in this case, we are not given any of the initial
values of the sequence. Still, the relation an = an−1 − an−2 will enable us to write as many
terms as we want in terms of a1 and a2 . Let us begin by doing this and summarazing the
results in a table:
n an
1 a1
2 a2
3 a2 − a1
4 −a1
5 −a2
6 a1 − a2
7 a1
8 a2
9 a2 − a1
.. ..
. .
Observe that a7 = a1 , a8 = a2 , a9 = a3 , and so on. In fact, the sequence repeats itself

cyclically after six terms. Therefore, we should compute
a1 + a2 + a3 + a4 + a5 + a6 = a1 + a2 + (a2 − a1 ) + (−a1 ) + (−a2 ) + (a1 − a2 ) = 0.
Now, since
2001 = 333 · 6 + 3,
the sum of the first 2001 terms is the same as
2001
X
an = a1 + a2 + a3 = a1 + a2 + (a2 − a1 ) = 2a2 . (11.43)
n=1
Next we will use the given information to determine the value of a2 .

Since the sum of the first 1492 terms is 1985 and 1492 = 248 · 6 + 4, we know that
1985 = a1 + a2 + a3 + a4 = 2a2 − a1 , (11.44)
and since the sum of the first 1985 terms is 1492 and 1985 = 330 · 6 + 5, we know that
1492 = a1 + a2 + a3 + a4 + a5 = a2 − a1 . (11.45)
Subtracting Equation (11.45) from Equation (11.44), we conclude that a2 = 493. Hence, using
Equation (11.43), we get
2001
X
an = 2a2 = 2 · 493 = 986.
n=1
2
10000
X 1
6. (2002 AIME-2, Problem #6) Find the integer that is closest to 1000 .
n=3
n2 − 4
Solution: The trick is to rewrite the terms of the series so that a telescoping behavior on the
terms can be employed. To achieve this, observe that for all n ≥ 3, we can write
1 1 A B
= = + , (11.46)
n2 − 4 (n − 2)(n + 2) n−2 n+2
where A and B must satisfy the requirement
A(n + 2) + B(n − 2) = 1
that arises from cross multiplication in (11.46). Therefore, for all n ≥ 3, we have
(A + B)n + (2A − 2B) = 1,
which implies that

A+B =0 and 2A − 2B = 1.
1
Solving this system of two equations for the two unknowns A and B, we find that A = and
4
1
B = − . Therefore, we have
4
10000 10000
X
X 1 1 1
1000 = 1000 −
n=3
n2 − 4 n=3
4(n − 2) 4(n + 2)
10000
X 1
1
= 250 −
n=3
n−2 n+2
10000 10000
!
X 1 X 1
= 250 −
n=3
n−2 n=3
n+2

1 1 1 1 1 1 1
= 250 1 + + + · · · + − − − − ··· −
2 3 9998 5 6 7 10002

1 1 1 1 1 1 1
= 250 1 + + + − − − −
2 3 4 9999 10000 10001 10002

25 1 1 1 1
= 250 − − − −
12 9999 10000 10001 10002

1 1 1 1
= 520.83 − 250 + + + .
9999 10000 10001 10002
In the last expression, observe that

1 1 1 1
250 + + +
9999 10000 10001 10002
1
amounts to a value of approximately 10 ,so even after subtracting it, the value of this last
10000
X 1
expression is still greater than 250.5. Hence, the closest integer to 1000 2−4
is 251. 2
n=3
n
7. (1989 AIME, Problem #7) If the integer k is added to each of the numbers 36,
300, and 596, one obtains the squares of three consecutive terms of an arithmetic
sequence. Find k.
Solution: We can write
36 + k = a2 , 300 + k = (a + d)2 , 596 + k = (a + 2d)2 ,
where k, a, and d are unknowns. Subtracting the first equation from the second, we obtain
2ad + d2 = 264, (11.47)
and subtracting the first equation from the third, we obtain
4ad + 4d2 = 560. (11.48)
Multiplying (11.47) by 2 and subtracting from (11.48), we arrive at the equation 2d2 = 32.
Therefore, d = ±4. Substituting this into Equation (11.47), we find that ±8a = 248, so that
a = ±31. Therefore, 36 + k = a2 = 961. Therefore, k = 961 − 36 = 925. 2
8. (1986 AIME, Problem #7) The increasing sequence 1, 3, 4, 9, 10, 12, 13, . . . consists
of those positive integers which are powers of 3 or sums of distinct powers of 3.
Find the 100th term of this sequence (where 1 is the 1st term, 3 is the 2nd term,
and so on).
Solution: Every term in this sequence can be uniquely represented in the form
an 3n + an−1 3n−1 + an−2 3n−2 + · · · + a2 · 32 + a1 · 3 + a0 , (11.49)
where each coefficient a0 , a1 , a2 , . . . , an is either 0 or 1. Therefore, the terms of this sequence

are in one-to-one correspondence with strings of zeros and ones. That is, we can make the
identification
an 3n + an−1 3n−1 + an−2 3n−2 + · · · + a2 · 32 + a1 · 3 + a0 −→ (an an−1 an−2 . . . a2 a1 a0 )2 ,
where the latter expression is a sequence of digits in base 2. Furthermore, this correspondence
identified here is order-preserving. This means that the 100th largest number that is expressible
in the form (11.49) with each ai ∈ {0, 1} will correspond precisely under our identification to the
100th largest positive number (which, of course, is 100) written in base 2. Now if we write the
number 100 in base 2, we get 100 = (1100100)2 (since 100 = 26 + 25 + 22 ). This base 2 number
corresponds via the identification above to the number 36 + 35 + 32 = 729 + 243 + 9 = 981,
which is therefore the 100th term of the sequence, as requested. 2
9. (2005 AIME-2, Problem #11) Let m be a positive integer, and let a0 , a1 , . . . , am be

a sequence of real numbers such that a0 = 37, a1 = 72, am = 0, and
3
ak+1 = ak−1 −
ak
for k = 1, 2, . . . , m − 1. Find m.
Solution: We have seen other examples of problems where enumeration of the first several
terms was useful for finding a pattern, and it is not unreasonable to search for such a pattern
here. However, in this case, no reasonable pattern will be forthcoming. (This should not
surprise us, as Problem 11 is sufficiently deep within the examination so as to require something
a bit more creative.)
Rather, the clever idea on this problem is to try an enumeration of terms process “in reverse”,
beginning with am = 0, and working backwards to compute am−1 , am−2 , and so on.
Since am = 0, we see that
3
am−2 = ,
am−1
or
am−1 am−2 = 3. (11.50)
Next,
3
am−1 = am−3 − ,
am−2
so that
am−1 am−2 = am−2 am−3 − 3. (11.51)
Therefore, from (11.50) and (11.51), we have
am−2 am−3 = 6. (11.52)
Next,
3
am−2 = am−4 − ,
am−3
so that
am−2 am−3 = am−3 am−4 − 3,
which implies that
am−3 am−4 = 9. (11.53)
Continuing the pattern emerging from Equations (11.50), (11.52), and (11.53), we see that
am−k am−k−1 = 3k (11.54)
for all positive integers k. Now we can substitute any value of k in Equation (11.54) for which
we will be able to evaluate am−k and am−k−1 . Since a0 and a1 are given, the correct choice is
k = m − 1, from which we obtain
a1 a0 = 3(m − 1).
Thus,
a1 a0 37 · 72
m−1= = = 888.
3 3
Therefore, m = 889. 2
10. (2000 AIME, Problem #10) A sequence of numbers x1 , x2 , . . . , x100 has the property
that, for every integer k between 1 and 100, inclusive, the number xk is k less than
the sum of the other 99 numbers. Given that x50 = m/n, where m and n are
relatively prime positive integers, find m + n.
Solution: We have
x1 = x2 + x3 + · · · + x100 − 1,
x2 = x1 + x3 + · · · + x100 − 2,
x3 = x1 + x2 + · · · + x100 − 3,
..
.
x50 = x1 + x2 + · · · + x49 + x51 + · · · + x100 − 50
..
.
x100 = x1 + x2 + · · · + x99 − 100.
The reader likely senses the relevance of summing the first 100 positive integers. Indeed, each
of these integers is subtracted, one per equation, at the right side of the array above. Thus, it
makes sense to sum both sides of all 100 of these equations. For convenience, we will denote
S = x1 + x2 + x3 + · · · + x100 .
As we sum on the right side, each xi will be summed exactly 99 times (xi is missing only from
the ith equation in the array). Thus, we have
100 · 101
S = 99S − (1 + 2 + 3 + · · · + 100) = 99S − = 99S − 5050.
2
5050 2525
Therefore, S = = . To obtain the value of x50 from this, examine the 50th equation
98 49
in the array above:
x50 = x1 + x2 + · · · + x49 + x51 + · · · + x100 − 50 = S − x50 − 50,
where we have added and subtracted x50 on the right side. Hence,
2x50 = S − 50,
from which we have

1 1 2525 25 101 25 3 75
x50 = (S − 50) = − 50 = −2 = · = .
2 2 49 2 49 2 49 98
Therefore, m = 75 and n = 98, so that m + n = 173. 2
11. (2006 AIME-2, Problem #11) A sequence is defined as follows: a1 = a2 = a3 = 1,

and, for all positive integers n, an+3 = an+2 + an+1 + an . Given that a28 = 6090307,
28
X
a29 = 11201821, and a30 = 20603361, find the remainder when ak is divided by
k=1
1000.
Solution #1: By using some clever manipulation with the recurrence relation
an+3 = an+2 + an+1 + an ,
it is possible to derive a clever solution to this problem (see below), but it is more straight-
forward to simply compute the summation (modulo 1000) by brute force. In some cases, it is
useful to replace positive numbers with negative numbers (modulo 1000) in order to keep the
numbers that arise small. This strategy is utilized in filling in the table below:
n
X
n an (mod 1000) ak (mod 1000)
k=1
1 1 1
2 1 2
3 1 3
4 3 6
5 5 11
6 9 20
7 17 37
8 31 68
9 57 125
10 105 230
11 193 423
12 355 778
13 -347 431
14 201 632
15 209 -159
16 63 -96
17 473 377
18 -255 122
19 281 403
20 499 902
21 -475 427
22 305 732
23 329 61
24 159 220
25 -207 13
26 281 294
27 233 527
28 307 834
Therefore, the answer is 834. 2
Solution #2: For the reader interested in a more clever solution, observe that
ak+3 = ak+2 + ak+1 + ak ,
for all positive integers k, which can be rewritten as
ak+3 − ak+2 = ak+1 + ak .
If we sum both sides of this equation for k = 1 to n and use the fact that the left side is
telescoping, we have
n
X n
X n
X n
X
an+3 − a3 = (ak+3 − ak+2 ) = (ak+1 + ak ) = ak+1 + ak .
k=1 k=1 k=1 k=1
At this point, we can simply substitute n = 27 to obtain
27
X
a30 − a3 = (a2 + a3 + · · · + 28) + (a1 + a2 + · · · + a27 ) = a1 + 2 ak + a28 .
k=2
Hence,
27
X 1 1
= a2 + a3 + · · · + a27 = (a30 − a3 − a1 − a28 ) = (20603361 − 1 − 1 − 6090307) = 7256526.
2 2
k=2
Thus,
28
X 27
X
ak = a1 + ak + a28 = 1 + 7256526 + 6090307 = 13346834,
k=1 k=2
and the remainder upon division by 1000 is 834. 2
12. (2004 AIME-2, Problem #9) A sequence of positive integers with a1 = 1 and
a9 + a10 = 646 is formed so that the first three terms are in geometric progression,
the second, third, and fourth terms are in arithmetic progression, and, in general,
for all n ≥ 1, the terms a2n−1 , a2n , a2n+1 are in geometric progression, and the terms
a2n , a2n+1 , a2n+2 are in arithmetic progression. Let an be the greatest term in this
sequence that is less than 1000. Find n + an .
Solution: The first three terms of the sequence take the form 1, r, and r2 , for some positive
integer r. All subsequent terms of the sequence can be expressed in terms of r. Therefore, if
we can determine a9 and a10 , the relation a9 + a10 = 646 will provide an equation from which
we can solve for r. Now a3 − a2 = r2 − r, and since a2 , a3 , a4 forms an arithmetic progression,
we have
a4 = a3 + (a3 − a2 ) = r2 + (r2 − r) = r(2r − 1).
Next,
a4 r(2r − 1) 2r − 1
= = ,
a3 r2 r
and since a3 , a4 , a5 forms a geometric progression, we have

a4
a5 = a4 · = (2r − 1)2 .
a3
Since a4 , a5 , a6 forms an arithmetic progression with common difference a5 −a4 = (2r−1)(r−1),

we obtain
a6 = a5 + (a5 − a4 ) = (2r − 1)2 + (2r − 1)(r − 1) = (2r − 1)(3r − 2).
Continuing the analysis in this way, we can build the following table:
n an
1 1
2 r
3 r2
4 r(2r − 1)
5 (2r − 1)2
6 (2r − 1)(3r − 2)
7 (3r − 2)2
8 (3r − 2)(4r − 3)
9 (4r − 3)2
10 (4r − 3)(5r − 4)
.. ..
. .
More generally, one can prove using a simultaneous induction argument (we omit this here)
that for n ≥ 1, we have

a2n = (nr − n + 1)((n − 1)r − n + 2)
(11.55)
a2n+1 = (nr − n + 1)2 .
Therefore, we have
646 = a9 + a10
= (4r − 3)2 + (4r − 3)(5r − 4)
= (4r − 3)(9r − 7).
This can be rewritten as the quadratic equation
36r2 − 55r − 625 = 0.
By the quadratic equation, the roots of this are

√
55 ± 93025 55 ± 305
r= = .
72 72
Since r = a2 must be a positive integer, we conclude that
55 + 305 360
r= = = 5.
72 72
Using this, can simplify Equation (11.55):

a2n = (4n + 1)(4n − 3)
(11.56)
a2n+1 = (4n + 1)2 ,
for all n ≥ 1.
We wish to determine the largest value of n such that an < 1000. Using the second √ line of
Equation (11.56), we can get a close estimate by setting (4n + 1)2 = 1000. Now, 31 < 1000 <
32, so we have 4n + 1 ≈ 31, or n ≈ 7.5. Plugging n = 7 or n = 8 into Equation (11.56), we
find that a14 = 29 · 25 = 725, a15 = 292 = 841, a16 = 33 · 29 = 957, and a17 = 332 = 1089.
Hence, a16 is the largest term less than 1000. We conclude that n + an = 16 + 957 = 973. 2
13. (2007 AIME, Problem #11) For each positive integer p, let b(p) denote the unique
√ 1
positive integer k such that |k − p| < . For example, b(6) = 2 and b(23) = 5. If
2
2007
X
S= b(p), find the remainder when S is divided by 1000.
p=1
Solution: It is possible to find a pattern by simply computing b(p) for the first several positive
integers p. The following table (or a portion of it) can be generated fairly quickly:
p b(p) p b(p)
1 1 17 4
2 1 18 4
3 2 19 4
4 2 20 4
5 2 21 5
6 2 22 5
7 3 23 5
8 3 24 5
9 3 25 5
10 3 26 5
11 3 27 5
12 3 28 5
13 4 29 5
14 4 30 5
15 4 31 6
16 4 32 6
Since b(p) = 1 for two values of p, b(p) = 2 for four values of p, b(p) = 3 for six values of p, and
so on, the conjecture based on this table is that the value of b(p) equals some number k for
exactly 2k values of p. Of course, independent of the table, we can also proceed more formally
to see that this is true.
Observe that
√ 1
b(p) = k ⇐⇒ |k − p| <
2
1 √ 1
⇐⇒ − < k − p <
2 2
1 √ 1
⇐⇒ − < p − k <
2 2
1 √ 1
⇐⇒ k − < p < k +
2 2
2 2
1 1
⇐⇒ k − <p< k+ .
2 2
2 2
1 1
Since k− and k+ are non-integers that differ by
2 2
2 2
1 1
k+ − k− = 2k,
2 2
we conclude that b(p) = k for exactly 2k values of p. Thus, for k = 1, 2, . . . , m, the total
2007
X
contribution to the sum b(p) is
p=1
2007 m m
X X X m(m + 1)(2m + 1) m(m + 1)(2m + 1)
b(p) = (2k)k = 2 k2 = 2 · = ,
p=1
6 3
k=1 k=1
where we have used the second equation in Theorem 5.4.2. The m values of k giving rise to
this sum correspond to 2 + 4 + 6 + · · · + 2m values of p in the sum defining S. Therefore, we
are only interested in choosing a value of m such that
2 + 4 + 6 + · · · + 2m ≤ 2007.
But by part 1 of Theorem 5.4.2, we have
2 + 4 + 6 + · · · + 2m = 2(1 + 2 + · · · + m) = m(m + 1).
The largest positive integer m with m(m + 1) ≤ 2007 can be experimentally found to be
m = 44. In fact, 44 · 45 = 1980. Thus,
1980 44
X X 44 · 45 · 89
b(p) = 2k 2 = = 58740.
p=1
3
k=1
For p = 1981, 1982, . . . , 2007, which is 27 values of p, we have b(p) = 45. Thus,
2007
X 1980
X 2007
X
S= b(p) = b(p) + b(p) = 58740 + 27 · 45 = 58740 + 1215 = 59955.
p=1 p=1 p=1981
When S is divided by 1000, the remainder is 955. 2
Remark: This solution illustrates a powerful technique for evaluating series in the case when
many of the terms are identical to one another. Namely, we can sometimes reformulate the
Xm
summation an from the perspective of the output terms an rather than the term number
n=1
n for the series. In this example, we considered all of the possible values of an (in this case,
an ∈ {1, 2, 3, . . . , 44, 45}) and then proceeded to determine how many terms of the original
2007-term series had each of these 45 values. Then we re-cast the series of 2007 in terms of a
different series with just 45 terms that could be conveniently calculated.
1995
√ X 1
14. (1995 AIME, Problem #13) Let f (n) be the integer closest to 4
n. Find .
f (k)
k=1
√
4
√
4
Solution #1: Note that f is a (weakly) increasing function, and since 1995 < 2401 = 7,
we know that the value of f (k) belongs to the set {1, 2, 3, 4, 5, 6, 7} for each k = 1, 2, . . . , 1995.
For each i in {1, 2, 3, 4, 5, 6, 7}, we wish to determine how many values of k satisfy f (k) = i.
√ 3 34 81
Note that f (k) = 1 provided that 4 k < . That is, k < 4 = . Hence, for k = 1, 2, 3, 4, 5,
2 2 16
we have f (k) = 1.
Next,
3 √
4 5 81 54 625
f (k) = 2 if and only if < k < if and only if <k< 4 = .
2 2 16 2 16
Hence, for k = 6, 7, . . . , 38, 39, we have f (k) = 2.
Next,
5 √4 7 625 74 2401
2 2 16 2 16
Next,
7 √
4 9 2401 94 6561
2 2 16 2 16
Next,
9 √
4 11 6561 114 14641
f (k) = 5 if and only if < k< if and only if <k< 4 = .
2 2 16 2 16
Next,
11 √
4 13 14641 134 28561
f (k) = 6 if and only if < k< if and only if <k< 4 = .
2 2 16 2 16
Finally, for k = 1786, 1787, . . . , 1995, we have f (k) = 7. We summarize these findings in the
table below:
k f (k) Number of k values in this range

1,2,. . . ,5 1 5
6,7,. . . ,39 2 34
40,41,. . . ,150 3 111
151,152,. . . ,410 4 260
411,412,. . . ,915 5 505
916,917,. . . ,1785 6 870
1786,1787,. . . ,1995 7 210
Thus,
1995 5 39 150 410 915 1785 1995
X 1 X 1 X1 X 1 X 1 X 1 X 1 X 1
= + + + + + +
f (k) 1 2 3 4 5 6 7
k=1 k=1 k=6 k=40 k=151 k=411 k=916 k=1786
1 1 1 1 1 1
= 5 · 1 + 34 · + 111 · + 260 · + 505 · + 870 · + 210 ·
2 3 4 5 6 7
= 5 + 17 + 37 + 65 + 101 + 145 + 30
= 400.
Remark: This solution illustrates a powerful technique for evaluating series in the case when
many of the terms are identical to one another. Namely, we can sometimes reformulate the
Xm
summation an from the perspective of the output terms an rather than the term number
n=1
n for the series. In this example, we considered all of the possible values of 1/f (k), which
are 1, 1/2, 1/3, 1/4, 1/5, 1/6, and 1/7. We then determined how many of the 1995 terms of
the original series assume each of these values. Finally, we sum the seven terms obtained by
taking each of the seven different values of 1/f (k) and multiplying them respectively by the
number of terms from the original series that assume each of those values. Let us flush out
this approach a bit more rigorously in the next solution to this problem.
1995
√
4
√
4 1X
Solution #2: Since 1995 < 2401 = 7, we know that every term of the series
f (k)
k=1
assumes one of the seven values 1, 1/2, 1/3, 1/4, 1/5, 1/6 or 1/7. It only remains to determine
how many of the 1995 terms assume each of the values. The reader should verify that f (k) = `
if and only if
4 4
1 1
`− <k < `+ .
2 2
The number of values of k that satisfy these inequalities is9
4 4
1 1
`+ − `− = 4`3 + `.
2 2
Thus, we have:
` Number of values
of k with f (k) = `
1 5
2 34
3 111
4 260
5 505
6 870
7 1379
The total number of values of k (as indicated by the right column) exceeds 1995 here, so we
must truncate the number of values of k for which f (k) = 7 so that the series has exactly 1995
terms. Since 5 + 34 + 111 + 260 + 505 + 870 = 1785, we have room for 1995 − 1785 = 210
values of k with f (k) = 7. Thus, we conclude that
1995
X 1 1 1 1 1 1 1
= 5 · 1 + 34 · + 111 · + 260 · + 505 · + 870 · + 210 · = 400.
f (k) 2 3 4 5 6 7
k=1
2
9 Compute this by expanding both fourth powers and subtracting.
15. (2006 AIME, Problem #15) Given that a sequence satisfies x0 = 0 and |xk | =
|xk−1 +3| for all integers k ≥ 1, find the minimum possible value of |x1 +x2 +· · ·+x2006 |.
Solution: It can be tedious to do substantial algebraic manipulations with equations involving
absolute value, so in this problem, it might be useful to eliminate absolute values by taking
the square of those expressions. For instance, since |xk | = |xk−1 + 3| for all integers k ≥ 1, we
have
x2k = (xk−1 + 3)2 = x2k−1 + 6xk−1 + 9.
Therefore,
x2k − x2k−1 = 6xk−1 + 9,
and so10
2007
X 2007
X
(x2k − x2k−1 ) = (6xk−1 + 9).
k=1 k=1
The sum on the left-hand side is telescoping, so it is easy to evaluate:
2007
X
x2k − x2k−1 = x22007 .
k=1
At the same time, we can simplify the right-hand side to obtain

x22007 = 6(x0 + x1 + x2 + · · · + x2006 ) + 9 · 2007.
Therefore, since x0 = 0,
x22007 − 9 · 2007
x1 + x2 + · · · + x2006 = .
6
Hence, |x1 + x2 + · · · + x2006 | assumes a minimum value when x22007 is as close to 9 · 2007 as
possible. From the fact that x0 = 0 and |xk | = |xk−1 + 3| for each k, we observe that xk is a
multiple of 3 for all k. Therefore, we can write x2007 = 3y for some integer y. Thus,
|x22007 − 9 · 2007| = |9y 2 − 9 · 2007| = 9|y 2 − 2007|,
√
which is minimized by finding the integer y closest to 2007, which is the integer y = 45. (It
is also acceptable to choose y = −45. The solution and answer are unaffected by this.) Thus,
1352 − 9 · 2007

x2007 = 3y = 135. Hence, the minimum value of |x1 + x2 + · · · + x2006 | is =
6
162
= 27 = 027. 2
6
1. (1996 AIME, Problem #2) For each real number x, let bxc denote the greatest
integer that does not exceed x. For how many positive integers n is it true that
n < 1000 and that blog2 nc is a positive even integer?
10 Note that we sum k from 1 to 2007 because, on the right-hand side, we wish to have the sum carry from x to
1
x2006 so as to correspond to the given expression |x1 + x2 + · · · + x2006 | given in the statement of the problem.
Solution: Note that

blog2 nc = 2 if and only if 4 ≤ n < 8,
and n must exceed 1000 in order for blog2 nc ≥ 10. Therefore, we just need to sum the number
of values of n arising from each of the four inequalities exhibited above to find the total:
4 + 16 + 64 + 256 = 340. 2
2. (1995 AIME, Problem #2) Find the last three digits of the product of the positive
roots of √
1995xlog1995 x = x2 .
Solution: The given equation in some ways resembles the formulas in Equation (6.3), but it
is not quite right. Nevertheless, it might be useful to bring the exponent log1995 x down. To
do this, let us apply the base 1995 logarithm to each side of the given equation (since part 3
of Theorem 6.2.2 can be used to bring powers down in front) to obtain
√
log1995 ( 1995xlog1995 x ) = log1995 (x2 ).
Using part (1) in Theorem 6.2.2, we have

√
log1995 1995 + log1995 (xlog1995 x ) = log1995 (x2 ).
Now applying part (3) in Theorem 6.2.2, we obtain
1
+ (log1995 x)2 = 2log1995 x.
2
1
Setting y := log1995 x, we can rewrite the preceding equation more simply as + y 2 = 2y,
2
which can be rewritten as
1
y 2 − 2y + = 0.
2
Using the quadratic equation to solve for y, we obtain
√ √
2± 2 2
y= =1± .
2 2
Since y = log1995 x, we have 1995y = x. Therefore, we obtain two solutions for x:
√ √
2 2
x = 19951+ 2 and x = 19951− 2 .
The product of these two roots is

√ √
2 2
19951+ 2 19951− 2 = 19952 .
It is not necessary to fully compute the number 19952 , since the problem only asks for the last
three digits, so a partial computation is enough to get the last three digits: 025. Alternatively,
for those readers that have reviewed Chapter 4 and are comfortable with congruence arithmetic,
we can simply note that
19952 ≡ (−5)2 ≡ 25 (mod 1000) ,
so the last three digits are 025. 2
3. (1986 AIME, Problem #3) If tan x+tan y = 25 and cot x+cot y = 30, what is tan(x+y)?
Solution: Provided one is aware of Equation (6.32), this problem is fairly straightforward.
There are really no other alternative approaches, unless one simply derives Equation (6.32)
from Equations (6.28) and (6.30). Ultimately, all solutions to this problem run through Equa-
tion (6.32):
tan x + tan y
tan(x + y) = .
1 − tan x tan y
We are already given tan x + tan y, and to find tan x tan y, we can use the given fact that
cot x + cot y = 30 as follows:
1 1 tan x + tan y 25
30 = cot x + cot y = + = = ,
tan x tan y tan x tan y tan x tan y
so that
25 5
tan x tan y = = .
30 6
Thus,
tan x + tan y 25 25
tan(x + y) = = 5 = 1 = 150.
1 − tan x tan y 1− 6 6
2
Remark: We have commented in the text that converting all trigonometric functions into
sine and cosine functions is a common strategy for solving problems in trigonometry. This
example shows, however, that this is not a hard and fast rule that always applies. Here, it was
most profitable to describe all relevant quantities in terms of tangent and cotangent functions,
and converting to sines and cosines would only complicate the solution.
4. (1994, Problem #4) Find the positive integer n for which
blog2 1c + blog2 2c + blog2 3c + · · · + blog2 nc = 1994. (11.57)
(For real x, bxc is the greatest integer ≤ x.)
Solution: This problem can best be solved by constructing a table such as this one:
blog2 kc Values of k Number of Values of k

0 1 1
1 2, 3 2
2 4, 5, 6, 7 4
3 8, . . . , 15 8
4 16, . . . , 31 16
5 32, . . . , 63 32
.. .. ..
. . .
For those who prefer not to construct an ad hoc table, we can simply note that for every
nonnegative integer m,
blog2 kc = m if and only if 2m ≤ k ≤ 2m+1 − 1,
which holds for precisely 2m values of k. Thus, if we group together terms for which blog2 kc
takes a common value and then sum them, we get an expression of the form
X
m2m = 0 · 1 + 1 · 2 + 2 · 4 + 3 · 8 + 4 · 16 + 5 · 32 + . . . ,
m
where m represents a nonnegative integer value and 2m is the number of terms being summed
that have value m. We can list the values of m2m in a table for ease of reference:
m m2m
0 0
1 2
2 8
3 24
4 64
5 160
6 384
7 896
8 2064
We wish to sum terms on the left side of Equation (11.57) until we reach 1994, and by adding
the values in the second column of the table above, the largest value of ` such that
`
X
m2m ≤ 1994
m=0
is m = 7:
7
X
m2m = 1538.
m=0
7
X
This sum is achieved by adding the first 2m = 28 − 1 = 255 terms of the left-hand side
m=0
of Equation (11.57). Since 1994 − 1538 = 456, we need to add enough additional terms, each
456
equal to blog2 kc = 8, to sum to 456. Since = 57, we must add 57 more terms, each of
8
value 8, to the 255 terms whose values were each 7 or less. Thus, the total number of terms in
the sum on the left side of Equation (11.57) is n = 255 + 57 = 312. 2
5. (2002 AIME, Problem #6) The solutions to the system of equations
log225 x + log64 y = 4
logx 225 − logy 64 = 1
are (x1 , y1 ) and (x2 , y2 ). Find log30 (x1 y1 x2 y2 ).

Solution: We notice that expressions of the form loga b and logb a appear simultaneously in
this system of equations, so we recognize the potential to use the logarithm change-of-base
formula (6.6). For notational convenience, we set
a := log225 x and b := log64 y.
Then we apply Equation (6.6) to obtain

1 1
= logx 225 and = logy 64.
a b
Therefore, the given linear system can be expressed as
1 1
a+b=4 and − = 1. (11.58)
a b
Since b = 4 − a, the second equation in (11.58) becomes
1 1
− = 1.
a 4−a
Hence,
4 − 2a = a(4 − a),
so that
a2 − 6a + 4 = 0.
Using the quadratic formula, we find that
√
a=3± 5.
Therefore, √ √
b = 4 − a = 4 − (3 ± 5) = 1 ∓ 5.
Hence, the solutions for x and y are
√ √
x = 225a = 2253± 5
and y = 64b = 641∓ 5
.
Therefore, we have
√ √ √ √
x1 y1 x2 y2 = 2253+ 5
641− 5
2253− 5
641+ 5
= 2256 · 642 = 212 · 312 · 512 = 3012 .
We conclude that
log30 (x1 y1 x2 y2 ) = log30 (3012 ) = 12 = 012.
2
6. (1995 AIME, Problem #7) Given that
(1 + sin t)(1 + cos t) = 5/4 (11.59)
and
m √
(1 − sin t)(1 − cos t) = − k, (11.60)
n
where k, m, and n are positive integers with m and n relatively prime, find k+m+n.
Solution: We notice that if the left sides of Equations (11.59) and (11.60) are each expanded,
there will be much in common. Before we do this, let us set
m √
α := − k
n
for notational convenience. Expanding Equations (11.59) and (11.60), we have
5
1 + sin t + cos t + sin t cos t = and 1 − sin t − cos t + sin t cos t = α.
4
Adding these two equations, we find that
5
2 + 2 sin t cos t = + α. (11.61)
4
Equation (11.61) supplies one equation relating the unknown α that we seek in terms of t. If
we can manufacture a second such equation, we will be in a good position to solve for α. A
key observation now is to observe by Equation (6.18) that
(1 + sin t)(1 − sin t) = 1 − sin2 t = cos2 t,
and likewise,
(1 + cos t)(1 − cos t) = 1 − cos2 t = sin2 t.
Motivated by this observation, let us multiply Equations (11.59) and (11.60):
5
(1 + sin t)(1 + cos t)(1 − sin t)(1 − cos t) = α
4
5
(1 − sin2 t)(1 − cos2 t) = α
4
2 2 5
cos t sin t = α.
4
Thus, √
5α
sin t cos t = ,
2
which we can substitute into Equation (11.61) to obtain
√ 5
2+ 5α = + α.
4
Hence,
2
3
5α = α− ,
4
from which we obtain the quadratic equation
13 9
α2 − α+ = 0.
2 16
Using the quadratic formula, we find that
√
q
13 169 9
2 ± 4 − 4 13 160 13 √
α= = ± = ± 10.
2 4 4 4
Thus, we have m = 13, n = 4, and k = 10. The answer is k + m + n = 10 + 13 + 4 = 27 = 027.
2
7. (2008 AIME-2, Problem #8) Let a = π/2008. Find the smallest positive integer n
such that
2 cos(a) sin(a) + cos(4a) sin(2a) + cos(9a) sin(3a) + · · · + cos(n2 a) sin(na)

(11.62)
is an integer.
Solution: The terms appearing in (11.62) consist of products of cosine and sine functions, so
we would be wise to consider using Equation (6.28), or perhaps Equation (6.29). For instance,
sin(k 2 a + ka) = cos(k 2 a) sin(ka) + cos(ka) sin(k 2 a). (11.63)
The first term on the right-hand side of (11.63) resembles the terms of the given expression,
but the second term does not. However, we can get the second term to cancel by similarly
using
sin(−k 2 a + ka) = cos(−k 2 a) sin(ka) + cos(ka) sin(−k 2 a) (11.64)
and the fact that cos(−x) = cos x and sin(−x) = − sin x for all real numbers x (see Equation
(6.35)). Thus, Equation (11.64) becomes
sin(−k 2 a + ka) = cos(k 2 a) sin(ka) − cos(ka) sin(k 2 a). (11.65)
Adding Equations (11.63) and (11.65), we obtain
sin(k 2 a + ka) + sin(−k 2 a + ka) = 2 cos(k 2 a) sin(ka). (11.66)
The right-hand side of (11.66) is exactly a typical term of the expression (11.62). Summing k
from k = 1 to k = n, we see that the expression (11.62) can be written as
n
X n
X
sin(k 2 a + ka) + sin(−k 2 a + ka) = (sin((k + 1)ka) + sin((1 − k)ka))
k=1 k=1
Xn
= (sin((k + 1)ka) − sin((k − 1)ka)) ,
k=1
where we have used the sine formula from (6.35) in the last step.
The latter sum is actually telescoping:
n
X
(sin((k + 1)ka) − sin((k − 1)ka)) = [sin(2a) − sin(0)] + [sin(6a) − sin(2a)] + · · ·
k=1
= sin(n2 a + na) − sin(0) = sin(n2 a + na).
Hence, we are seeking the smallest positive integer n such that

2
(n + n)π
sin
2008
mπ
is an integer. Since y(x) = sin x assumes an integer value if and only if x has the form 2 for
some integer m, we set
(n2 + n)π mπ
=
2008 2
and find that
n2 + n = 1004m.
Thus, we must find the smallest positive integer n such that n2 + n = n(n + 1) is a multiple of
1004 = 4 · 251. Since 251 is prime, n(n + 1) must contain a factor of 251. The smallest value
of n for which this can be true is n = 250, but in this case, n(n + 1) will not be divisible by
4. The next smallest value of n for which n(n + 1) contains a factor of 251 is n = 251. Since
n + 1 = 252 is a multiple of 4, we can say that (251)(252) is a multiple of 1004, and we are
done by choosing n = 251. 2
8. (2011 AIME, Problem #9) Suppose x is in the interval [0, π/2] and
3
log24 sin x (24 cos x) = . (11.67)
2
Find 24 cot2 x.
Remark: This problem utilizes the Rational Roots test, which is not covered until Chapter
7 in the text. Readers unfamiliar with this test may wish to study this result prior to reading
the solution below.
Solution: Equation (11.67) implies that
(24 sin x)3/2 = 24 cos x.
By squaring both sides, we obtain
(24 sin x)3 = (24 cos x)2 ,
which can be simplified to

24 sin3 x = cos2 x = 1 − sin2 x.
Thus, sin x is a solution to the equation
24y 3 + y 2 − 1 = 0. (11.68)
Since we know that the answer to this problem, 24 cot2 x, is an integer, it is likely that sin x is
a rational number, and we can use the Rational Roots Test (Theorem 7.4.6) to examine the
rational number candidates for the roots of Equation (11.68), which are:
1 1 1 1 1 1 1
y= ± , ± , ± , ± , ± , ± , ± , ±1.
24 12 8 6 4 3 2
A relatively quick trial-and-error process with these candidates leads to y = 13 . Therefore, we

conclude that
1
sin x = .
3
Thus,
24 cos2 x 24(1 − sin2 x) 24(1 − 19 )
24 cot2 x = 2 = 2 = 1 = 24 · 8 = 312.
sin x sin x 9
2
22 m
9. (1991 AIME, Problem #9) Suppose that sec x+tan x = and that csc x+cot x = ,
7 n
m
where is in lowest terms. Find m + n.
n
Solution: There are several ways to exploit trigonometric identities to solve this problem.
Lesser known identities, such as half-angle formulas for y(x) = tan x, can be used to provide
a relatively direct solution. However, we will instead use the trigonometric identity involving
sec x and tan x that is most likely to come to mind here, Equation (6.19):
1 + tan2 x = sec2 x.
Since we are given another relationship between sec x and tan x in the problem, we should be
able to make good headway with a substitution, for instance
22
tan x = − sec x.
7
Thus, we have
2
2 2 22 484 44
sec x = 1 + tan x = 1 + − sec x =1+ − sec x + sec2 x,
7 49 7
and so after cancelling sec2 x from both sides and rearranging, we obtain
44 484 533
sec x = 1 + = ,
7 49 49
and therefore,
533 7 533
sec x = · = .
49 44 308
Now that this value has been determined, we can set to work computing the other trigono-
metric functions of x that are needed for this problem. Unfortunately, it will be difficult from
a practical and computation standpoint to compute csc x from sec x. (Why?) Therefore, al-
ternatively we can start over with the same trigonometric identity and same approach again,
this time solving for tan x: We have
2
22 484 44
1 + tan2 x = sec2 x = − tan x = − tan x + tan2 x,
7 49 7
so that after cancelling and rearranging these terms,
44 484 435
tan x = −1= .
7 49 49
Hence,
435 7 435
tan x = · = .
49 44 308
Therefore,
308 308 533 533
cot x = and csc x = cot x · sec x = · = .
435 435 308 435
Thus,
533 308 841 29
csc x + cot x =
+ = = .
435 435 435 15
Hence, m = 29 and n = 15, so that m + n = 29 + 15 = 44 = 044. 2
10. (2012 AIME, Problem #9) Let x, y, and z be positive real numbers that satisfy
2 logx (2y) = 2 log2x (4z) = log2x4 (8yz) 6= 0. (11.69)
1
The value of xy 5 z can be expressed in the form p/q , where p and q are relatively
2
prime positive integers. Find p + q.
Solution: Until we find x, y, and z, we do not know what the nonzero value expressed three
ways in Equation (11.69) is. For the time being, let us call this value k. (Note that while the
value of k may help us to solve this problem, its value is not necessary to compute, as we will
see.) We can formulate three equations involving k by using the three equal expressions for k
in Equation (11.69) along with part (3) of Theorem 6.2.2:
xk = (2y)2 = 4y 2 , (11.70)
(2x)k = (4z)2 = 16z 2 , (11.71)

and
(2x4 )k = 8yz. (11.72)
Comparing these equations, we notice that if we multiply the right-hand sides of Equations
(11.70) and (11.71), we will get 64y 2 z 2 = (8yz)2 , the square of the right side of Equation
(11.72). Thus, the corresponding results on the left-hand sides of these three equations must
obey the same relationship:
(xk )((2x)k ) = (2x4 )2k ,
or
2k x2k = 22k x8k .
Simplifying to get powers of 2 on one side and powers of x on the other side, we obtain
2−k = x6k ,
and now taking the kth root of both sides, we have

r
1 1
= x6 , which implies that x = = 2−1/6 .
6
2 2
Substituting this value of x into Equations (11.70) – (11.72), we arrive at three new equations:
2−k/6 = 4y 2 , (11.73)
25k/6 = 16z 2 , (11.74)

and
2k/3 = 8yz. (11.75)
Now if we square Equation (11.73) and then multiply that result with Equation (11.75), the
left hand side will become 1. Alternatively, we notice that this operation will give us the
expression y 4 z on the right-hand side, which is the majority of the expression xy 4 z whose
value we are trying to obtain. In any case, the result is
1 = (4y 2 )2 (8yz) = 27 y 4 z.
Thus,
y 4 z = 2−7 ,
and so
xy 4 z = 2−1/6 2−7 = 2−43/6 .
Therefore, p = 43 and q = 6, so that
p + q = 43 + 6 = 49 = 049.
Remark: The interested reader may wish to determine the value of k. The answer is k = 6.
11. (1997 AIME, Problem #11) Let
44
X
cos n◦
n=1
x= 44
.
X
◦
sin n
n=1
What is the greatest integer that does not exceed 100x?

Solution: Using the first formula in Equation (6.34) to replace cos n◦ with sin(90 − n)◦ , we
can write
X44 X89 X44
cos n◦ sin n◦ sin(n + 45)◦
n=1 n=46 n=1
x = 44
= 44
= 44
. (11.76)
X X X
◦ ◦ ◦
sin n sin n sin n
n=1 n=1 n=1
Using Equation (6.28), we can write

√
◦ 2
sin(n + 45) = (sin n◦ + cos n◦ ),
2
and substituting this into the calculation (11.76) above, we obtain

√ 44 √ 44 44
2X 2 X X
(sin n◦ + cos n◦ ) sin n◦ + cos n◦ √
2 n=1 2 n=1 n=1 2
x = 44
= 44
= (1 + x).
X X 2
sin n◦ sin n◦
n=1 n=1
√
Hence, multiplying throughout by 2, we obtain
√
2x = 1 + x,
or √
( 2 − 1)x = 1.
Hence, √
1 1 2+1 √
x= √ =√ ·√ = 2 + 1 ≈ 2.414.
2−1 2−1 2+1
Thus, 100x ≈ 241.4, and the greatest integer that does not exceed this is 241. 2
√
Remark: Note that it is important to keep enough of the digits of 2 past the decimal point
to be able to determine between what two integers the number 100x falls.
12. (2009 AIME-2, Problem #11) For certain pairs (m, n) of positive integers with
m ≥ n there are exactly 50 distinct positive integers k such that |log m − log k| <
log n. Find the sum of all possible values of the product mn.
Solution: Since the inequality in question involves a difference of logarithms, we apply part
(2) of Theorem 6.2.2 to rewrite this as

log m < log n.

k
That is, m
−log n < log < log n.
k
Recall that the notation log x refers to the base 10 logarithm of x, so exponentiating all three
quantities in this relation with the function f (x) = 10x , we obtain
10−log < 10log( k ) < 10log n .

m
n
That is,
1 m
< < n, (11.77)
n k
where part (3) of Theorem 6.2.2 has been used in the first step. Since we want to determine
the conditions on (m, n) for which there are exactly 50 solutions for k, it will be helpful for
us to get an explicit inequality for k. To do this, we can take the reciprocal of the inequality
chain in (11.77) to get
1 k
< < n,
n m
and multiplying through by m, we conclude that

m
< k < mn. (11.78)
n
In order for there to be exactly 50 integers k that satisfy (11.78), we observe that these integer
values are mn − 1, mn − 2, mn − 3, . . . , mn − 50. Thus,
m
mn − 51 ≤ < mn − 50.
n
In order to isolate the integer m, we can multiply through by n and rearrange the results to
obtain
50n 51n
<m≤ 2 . (11.79)
n2 − 1 n −1
Since we are also given that n ≤ m, we deduce from the right-hand inequality in (11.79) that
51n
n< .
n2 − 1
√
Rearranging this, we obtain n2 < 52. Thus, we need only consider n with 1 ≤ n ≤ 7. First
observe that n = 1 leads to no solutions; this can be seen, for example, from (11.78).
For n = 2, (11.79) becomes
100 102
<m≤ = 34,
3 3
so the only integer value that works is m = 34.
150 153
<m≤ , which implies that 18.75 < m ≤ 19.125
8 8
so the only integer value that works is m = 19.
200 204
<m≤ , which implies that 13.3 < m ≤ 13.6,
15 15
so no integer values of m work in this case.
250 255
24 24
300 306
35 35

350 357
<m≤ , which implies that 7.2 < m ≤, 7.4375
48 48
Thus, the possible values of (m, n) are (34, 2) and (19, 3). The sum of the possible values of
mn is therefore 34 · 2 + 19 · 3 = 68 + 57 = 125. 2
13. (2004 AIME, Problem #12) Let S be the set of ordered pairs (x, y) such that
0 < x ≤ 1, 0 < y ≤ 1, and blog2 ( x1 )c and blog5 ( y1 )c are both even. Given that the area
of the graph of S is m/n, where m and n are relatively prime positive integers find
m + n. The notation bzc denotes the greatest integer that is less than or equal to
z.

1 1
Solution: Note that for 0 < x ≤ 1, we have that log2 is even if and only if belongs
x x
to one of the half-open
intervals
[1, 2), [4,
8), [16, 32), . . . . Therefore, x must belong to one
1 1 1 1 1
of the intervals ,1 , , , , , . . . . The general interval in this list has the form
2 8 4 32 16
1 1
, , where k = 0, 1, 2, . . . . The width of this interval is
22k+1 22k
1 1 1
− 2k+1 = 2k+1 .
22k 2 2
Summing over all values of k and using Equation (5.12), the total width of the collection of
intervals for the values of x given is
∞ ∞ k
X 1 1X 1 1 1 1 4 2
= = · 1 = · = .
22k+1 2 4 2 1− 4
2 3 3
k=0 k=0

1
Next we perform exactly the same analysis for the y coordinate. We find that log5 is
y
1
even if and only if belongs to one of the intervals [1, 5), [25, 125), [625, 3125), . . . . Therefore, y
y
1 1 1 1 1
must belong to one of the intervals ,1 , , , , , . . . . The general interval
5 125 25 3125 625
1 1
in this list has the form , , where k = 0, 1, 2, . . . . The width of this interval is
52k+1 52k
1 1 4
2k
− 2k+1 = 2k+1 .
5 5 5
Summing over all values of k and applying Equation (5.12) once more, we find that the total
width of the collection of intervals for the values of y given is
∞ ∞ k
X 4 4X 1 4 1 4 25 5
= = · 1 = 5 · 24 = 6 .
52k+1 5 25 5 1 − 25
k=0 k=0
To find the total area of S, we simply multiply the one-dimensional widths in the x and y
directions to obtain
2 5 10 5
Area of S = · = = .
3 6 18 9
14. (2007 AIME-2, Problem #12) The increasing geometric sequence x0 , x1 , x2 , . . . con-
sists entirely of integral powers of 3. Given that
7 7
!
X X
log3 (xn ) = 308 and 56 ≤ log3 xn ≤ 57,
n=0 n=0
find log3 (x14 ).

Solution: Let x0 = 3a for some integer a, and suppose the geometric sequence {xn } has
common ratio 3b , where b is an integer. Then xn = 3a+bn for each positive integer n, and
log3 (xn ) = a + bn. (11.80)
Thus,
7
X
308 = log3 (xn )
n=0
= 8a + (1 + 2 + · · · + 7)b = 8a + 28b.
Dividing through by 4, we obtain the relation
2a + 7b = 77. (11.81)
Next, we have
7
X
xn = 3a + 3a+b + 3a+2b + · · · + 3a+7b = 3a (1 + 3b + 32b + · · · + 37b ). (11.82)
n=0
Hence from Equation (11.82) and part (1) of Theorem 6.2.2, we have
7
!
X
a b 2b 7b
log3 xn = log3 3 (1 + 3 + 3 + · · · + 3 )
n=0
= a + log3 (1 + 3b + 32b + · · · + 37b ).
Hence, we have
56 − a ≤ log3 (1 + 3b + 32b + · · · + 37b ) ≤ 57 − a. (11.83)
Moreover, since log3 (x) is an increasing function of x,
7b = log3 (37b ) < log3 (1+3b +32b +· · ·+37b ) < log3 (8·37b ) = log3 8+log3 (37b ) < 2+7b. (11.84)
Putting the inequalities in (11.83) and (11.84) together, we find that
56 − a < 7b + 2 and 7b < 57 − a.

Therefore, 7b < 57 − a < 7b + 3, which implies that either
57 − a = 7b + 1 or 57 − a = 7b + 2.
In the latter case, we have a + 7b = 55, and together with (11.81), this implies that a = 22.
However, this in turn implies (again from (11.81)) that 7b = 33, contrary to the assumption
that b is an integer. Therefore, the former case, 57 − a = 7b + 1, must hold. Thus, a + 7b = 56.
Again appealing to (11.81), we find that a = 21, which in turn implies that 7b = 35. Thus,
b = 5. Therefore, from Equation (11.80), we conclude that
log3 (x14 ) = a + 14b = 21 + 70 = 91 = 091.
2
15. (2000 AIME-2, Problem #15) Find the least positive integer n such that
1 1 1 1
+ + ··· + = . (11.85)
sin 45◦ sin 46◦ ◦
sin 47 sin 48◦ ◦
sin 133 sin 134◦ sin n◦
Solution: First observe that
sin 1 = sin((x + 1) − x) = sin(x + 1) cos x − cos(x + 1) sin x,
by Equation (6.29). Hence, from Equation (11.85),

sin 1 sin 1 sin 1 sin 1
= + + ··· +
sin n sin 45 sin 46 sin 47 sin 48 sin 133 sin 134
sin[(45 + 1) − 45] sin[(47 + 1) − 47] sin[(133 + 1) − 133]
= + + ··· +
sin 45 sin 46 sin 47 sin 48 sin 133 sin 134
sin 46 cos 45 − cos 46 sin 45 sin 134 cos 133 − cos 134 sin 133
= + ··· +
sin 45 sin 46 sin 133 sin 134
= (cot 45 − cot 46) + (cot 47 − cot 48) + · · · + (cot 133 − cot 134)
= cot 45 − (cot 46 + cot 134) + (cot 47 + cot 133) − · · · + (cot 89 + cot 91) − cot 90
= cot 45
= 1.
Hence, sin n◦ = sin 1◦ , and we conclude that n = 1 = 001. 2
1. (2009 AIME, Problem #2) There is a complex number z with imaginary part 164
and a positive integer n such that
z
= 4i.
z+n
Find n.
Solution: Write z = a + 164i, for some real number a. We have
a + 164i = z
= 4i(z + n)
= 4i(a + n + 164i)
= −656 + 4i(a + n).
Equating the real and imaginary parts on both sides, we conclude that
a = −656 and 164 = 4(a + n).
Hence, a + n = 41, and

n = 41 − a = 41 − (−656) = 697.
2
2. (1985 AIME, Problem #3) Find c if a, b, and c are positive integers which satisfy
c = (a + bi)3 − 107i, where i2 = −1.
Solution: Note that
(a + bi)3 − 107i = (a3 − 3ab2 ) + (3a2 b − b3 − 107)i.
Since this must equal the positive integer c, we conclude that
a3 − 3ab2 = c and 3a2 b − b3 − 107 = 0
from the fact that two complex numbers are equal if and only if their real parts are the same
and their imaginary parts are the same. Rearranging the latter equation, we have
b(3a2 − b2 ) = 107.
Since 107 is prime and b is a positive integer, we must have either b = 1 or b = 107.
Case 1: b = 107. In this case, 3a2 − b2 = 1, so that 3a2 = b2 + 1. This requires that b2 + 1 is
a multiple of 3. However,
1072 + 1 ≡ (−1)2 + 1 ≡ 2 (mod 3),
so 1072 + 1 is not a multiple of 3. Thus, this case is impossible.

Case 2: b = 1. In this case, 3a2 − b2 = 107, so that
108
a2 = = 36.
3
Since a is positive, we conclude that a = 6 and
c = a3 − 3ab2 = 63 − 18 = 216 − 18 = 198.
2
3. (2001 AIME, Problem #3) Find the sum of all the roots, real and nonreal, of the
2001
equation x2001 + 12 − x = 0, given that there are no multiple roots.
2001
Solution #1: The degree of the polynomial p(x) = x2001 + 21 − x is 2000, and therefore,
by the Fundamental Theorem of Algebra (Theorem 7.4.1), p(x) has 2000 (not necessarily
distinct) roots. Observe that
1
r is a root of p(x) if and only if − r is a root of p(x),
2
since p(r) = p( 21 − r). Since p( 14 ) > 0, r = 41 is not a root of p(x). Therefore, we can group the
roots of p(x) into pairs of two distinct roots consisting of r and 12 − r. The sum of the roots
comprising each pair is r + ( 21 − r) = 12 . Since there are a total of 2000 roots, we have 1000
pairs. Hence, the sum of all the roots of p(x) is 21 · 1000 = 500. 2
Solution #2: We are being asked to determine the sum of the roots of a polynomial, so
part (b) of Proposition 7.4.3 appears to be germane. In this case, observe that the degree of
2001
p(x) = x2001 + 12 − x is 2000. Part (b) of Proposition 7.4.3 requires the polynomial to
be monic. For p(x), the coefficient of x2000 is (by the Binomial Theorem)

2001 1 2001
(−1)2000 = .
1 2 2
Hence, the polynomial
2
q(x) := p(x)
2001
is monic of degree 2000 and has the same roots as p(x). Thus, according to part (b) of
Proposition 7.4.3, we must find −a1999 , where a1999 is the coefficient of x1999 in q(x). According
to the Binomial Theorem, this coefficient is
2
2 2001 1
a1999 = · (−1)1999 = −500,
2001 2 2
and so the sum of the roots of q(x), and hence of p(x), is −a1999 = 500. 2
4. (1993 AIME, Problem #5) Let P0 (x) = x3 + 313x2 − 77x − 8. For integers n ≥ 1,
define Pn (x) = Pn−1 (x − n). What is the coefficient of x in P20 (x).
Solution: We are given a recursively defined sequence of polynomials, so it is natural to back-
substitute the relation Pn (x) = Pn−1 (x − n) into itself repeatedly, starting with n = 20. This
gives us
P20 (x) = P19 (x − 20)

= P18 ((x − 20) − 19)
= P17 (((x − 20) − 19) − 18)
..
.
= P0 (x − (20 + 19 + 18 + · · · + 2 + 1))
= P0 (x − 210)
= (x − 210)3 + 313(x − 210)2 − 77(x − 210) − 8.
To find the coefficient of x, we can apply the Binomial Theorem to each term in this expression.
The coefficient of x in (x − 210)3 is 3(−210)2 , the cofficient of x in (x − 210)2 is −2 · 210, and
the cofficient of x in 77(x − 210) is 77. Thus, the coefficient of x in P20 (x) is
3(−210)2 − 2 · 313 · 210 − 77 = 210(630 − 626) − 77 = 840 − 77 = 763.
Remark: Note that another way to begin is to examine P1 (x), P2 (x), P3 (x), and so on, until
a pattern emerges. This was a common problem solving theme in Chapter 5 in working with
sequences. We have
P1 (x) = P0 (x − 1) = (x − 1)3 + 313(x − 1)2 − 77(x − 1) − 8,
P2 (x) = P1 (x − 2) = (x − 3)3 + 313(x − 3)2 − 77(x − 3) − 8,

P3 (x) = P2 (x − 3) = (x − 6)3 + 313(x − 6)2 − 77(x − 6) − 8,
P4 (x) = P3 (x − 4) = (x − 10)3 + 313(x − 10)2 − 77(x − 10) − 8,
and so on.
The pattern reveals that for each positive integer k, we have
3 2
k(k + 1) k(k + 1) k(k + 1)
Pk (x) = Pk−1 (x − k) = x − + 313 x − − 77 x − − 8.
2 2 2
Therefore,
P20 (x) = (x − 210)3 + 313(x − 210)2 − 77(x − 210) − 8,
which is the expression we obtained previously for P20 (x).
5. (2005 AIME, Problem #6) Let P be the product of the nonreal roots of
x4 − 4x3 + 6x2 − 4x = 2005.
Find bP c. (The notation bP c denotes the greatest integer that is less than or equal
to P .)
Solution: The coefficients of the polynomial p(x) = x4 − 4x 3 2
+6x − 4x are 1, −4, 6, and −4.
4
These numbers happen to be the binomial coefficients ± for k = 0, 1, 2, 3. These are
k
4
precisely the coefficients we obtain from (x − 1) , but with the constant term missing from
p(x). In fact
p(x) + 1 = x4 − 4x3 + 6x2 − 4x + 1 = (x − 1)4 .
Thus, we need to look for the nonreal solutions of
(x − 1)4 = 2006.
√ y = x −√
If we set 1, we have y 4 = 2006. There are four values of y that satisfy this√equation:
y = ± 4 2006,
√ ±i 4
2006. Hence, the four corresponding values of x are x = 1 ± 4 2006 and
4
x = 1 ± i 2006. Hence,
√4
√
4
√
P = (1 + i 2006)(1 − i 2006) = 1 + 2006.
√ √ √ √
We need to find b 2006c. Since 40 = 1600 < 2006 < 2500 = 50, we can compute 412 ,
422 , and so on. We eventually find that
√ √ √
44 = 1936 < 2006 < 2025 = 45.
Thus, we have √
45 < 1 + 2006 < 46,
so that bP c = 045. 2
6. (2010 AIME-2, Problem #7) Let P (z) = z 3 + az 2 + bz + c, where a, b, and c are real.
There exists a complex number w such that the three roots of P (z) are w + 3i,
w + 9i, and 2w − 4, where i2 = −1. Find |a + b + c|.
Solution: Write w = u + vi, where u and v are real. From Proposition 7.4.3, we have
a = − [(w + 3i) + (w + 9i) + (2w − 4)] = 4 − 12i − 4w (11.86)
and
b = (w +3i)(w +9i)+(w +3i)(2w −4)+(w +9i)(2w −4) = 5w2 −8w +36wi−48i−27. (11.87)
Now we will exploit the fact that a and b are real. From Equation (11.86), we see that v = −3.
Now substituting w = u − 3i into Equation (11.87), we obtain
b = 5(u − 3i)2 − 8(u − 3i) + 36(u − 3i)i − 48i − 27,
whose imaginary part is
−30u + 24 + 36u − 48 = 6u − 24 = 0.
Hence, u = 4. Therefore, we have

w = 4 − 3i.
Hence, the three roots of P (z) are r1 = 4, r2 = 4+6i, and r3 = 4−6i. Returning to Proposition
7.4.3 with this information, we have
a = −(r1 + r2 + r3 ) = −12,
b = r1 r2 + r1 r3 + r2 r3 = 16 + 24i + 16 − 24i + 52 = 84,

and
c = −(r1 r2 r3 ) = −4(52) = −208.
Thus,
|a + b + c| = | − 12 + 84 − 208| = 136.
2
7. (2007 AIME, Problem #8) The polynomial P (x) is cubic. What is the largest value
of k for which the polynomials Q1 (x) = x2 +(k−29)x−k and Q2 (x) = 2x2 +(2k−43)x+k
are both factors of P (x)?
Solution #1: When computing factors or roots of polynomials, constant factors can always
be safely inserted or removed without changing the factors or roots. For this reason, by
multiplying P (x) through by the reciprocal of the its leading coefficient, we may assume that
P (x) is monic. This assumption does not affect the divisibility of P (x) by Q1 (x) or Q2 (x).
Since P (x) is cubic and Q1 (x) and Q2 (x) are both quadratic factors of P (x), degree consider-
ations imply that there must exist constants a and b such that
1
P (x) = (x − a)Q1 (x) and P (x) = (x − b)Q2 (x). (11.88)
2
The factor of 1/2 in front of the second formula is required since P (x) is monic and Q2 (x) has
leading coefficient 2. Inserting the formulas for Q1 (x) and Q2 (x) given in the problem and
expanding both expressions in (11.88), we obtain
P (x) = (x − a)Q1 (x) = x3 + (k − 29 − a)x2 − (k + ak − 29a)x + ak
and
1 1 1
P (x) = (x − b)Q2 (x) = x3 + (2k − 43 − 2b)x2 + (k + 43b − 2bk)x − bk.
2 2 2
Equating coefficients for the two expressions for P (x) just obtained, we have
1
k − 29 − a = (2k − 43 − 2b), (11.89)
2
1
−(k + ak − 29a) = (k + 43b − 2bk), (11.90)
2
b = −2a. (11.91)
Substituting Equation (11.91) into Equation (11.89) and simplifying (note that k cancels in
(11.89)), we obtain a = −2.5. Therefore, using (11.91), we obtain b = 5. Substituting these
values for a and b into (11.89) or (11.90), a routine calculation allows us to solve for k. The
result is k = 030. 2
Solution #2: Let a, b, c denote the roots of P (x). Since Q1 (x) and Q2 (x) divide P (x), every
root of Q1 (x) and Q2 (x) must also be a root of P (x). Since Q1 (x) and Q2 (x) are not multiples
of one another, they cannot have an identical list of (two) roots. Thus, let us assume that
the roots of Q1 (x) are a and b, while we assume that the roots of Q2 (x) are a and c. (We
must assume that Q1 (x) and Q2 (x) do share one common root since there are only three roots
of P (x), and two of them must be roots of Q1 (x) and two of them must be roots of Q2 (x).)
Hence, using Equation (7.16), we have
Q1 (x) = (x − a)(x − b) and Q2 (x) = 2(x − a)(x − c).
Expanding these equations and using the formulas for Q1 (x) and Q2 (x) given, we conclude
that
x2 − (a + b)x + ab = x2 + (k − 29)x − k (11.92)
and
2x2 − 2(a + c)x + 2ac = 2x2 + (2k − 43)x + k. (11.93)
Equating coefficients in Equations (11.92) and (11.93), we have
a + b = 29 − k, (11.94)
ab = −k, (11.95)
2(a + c) = 43 − 2k, (11.96)
and
2ac = k. (11.97)
From Equations (11.95) and (11.97), we see that either a = 0 or
b = −2c. (11.98)
If a = 0, then Equation (11.95) implies that k = 0. We will see momentarily, however, that in
the case when (11.98) holds, we obtain a larger value of k. Substituting (11.98) into (11.94),
we have
a − 2c = 29 − k. (11.99)
Adding Equation (11.99) to Equation (11.96) yields 3a = 72 − 3k, or
a = 24 − k.
Substituting this into Equation (11.94) and simplifying, we find that b = 5. Hence, from
(11.95), we see that
k = −5a = −5(24 − k) = 5k − 120,
from which it quickly follows that k = 30 = 030. 2
Remark: The set of algebraic steps we took to solve Equations (11.94) – (11.97) in Solution
#2 above are certainly not unique. Readers are encouraged to pursue their own line of solution
and compare with what we have recorded here.
8. (2005 AIME-2, Problem #9) For how many positive integers n less than or equal
to 1000 is
true for all real t?
Solution: The expressions on both sides of the given equation are reminiscent of Euler’s
Formula eiθ = cos θ + i sin θ (see Equation (7.11) for further discussion). In the expression
sin t + i cos t appearing on the left side of the equation given in the problem, the imaginary
element i is associated with the cosine function, rather than the sine function dictated in Euler’s
Formula. But we can “correct” this phenomenon by multiplying through by ±i: specifically,
observe that
(−i)(sin t + i cos t) = cos t − i sin t = e−it .
Therefore, let us multiply both sides of
through by (−i)n :
(−i)n (sin t + i cos t)n = (−i)n (sin nt + i cos nt)

(cos t − i sin t)n = (−i)n (sin nt + i cos nt)
= (−i)n−1 (cos nt − i sin nt).
Therefore, we have (e−it )n = (−i)n−1 e−int . Since e−int 6= 0 for all n and for all t, we conclude
that (−i)n−1 = 1. This equation holds if and only if n − 1 is a multiple of 4. Therefore,
among the first 1000 positive integers, exactly 250 of them (n = 1, 5, 9, . . . , 997) will satisfy
(−i)n−1 = 1. Therefore, the answer is 250. 2
9. (1988 AIME, Problem #11) Let w1 , w2 , . . . , wn be complex numbers. A line L in

the complex plane is called a mean line for the points w1 , w2 , . . . , wn if L contains
points (complex numbers) z1 , z2 , . . . , zn such that
n
X
(zk − wk ) = 0.
k=1
For the numbers w1 = 32 + 170i, w2 = −7 + 64i, w3 = −9 + 200i, w4 = 1 + 27i, and

w5 = −14 + 43i there is a unique mean line with y-intercept 3. Find the slope of
this mean line.
Solution: The mean line for L must contain five points
z1 = x1 + y1 i, z2 = x2 + y2 i, z3 = x3 + y3 i, z4 = x4 + y4 i, and z5 = x5 + y5 i
such that
5
X 5
X
zk = wk = (32 + 170i) + (−7 + 64i) + (−9 + 200i) + (1 + 27i) + (−14 + 43i) = 3 + 504i.
k=1 k=1
The mean line with y-intercept 3 and slope m has equation y = mx + 3 in the complex
plane. We are seeking to determine the value of m. This implies that yi = mxi + 3 for each
i = 1, 2, 3, 4, 5. We conclude that
3 + 504i = z1 + z2 + z3 + z4 + z5
= (x1 + x2 + x3 + x4 + x5 ) + (y1 + y2 + y3 + y4 + y5 )i
= (x1 + x2 + x3 + x4 + x5 ) + [m(x1 + x2 + x3 + x4 + x5 ) + 15]i.
Equating the real and imaginary parts of both sides of this expression, we obtain
x1 + x2 + x3 + x4 + x5 = 3 and m(x1 + x2 + x3 + x4 + x5 ) + 15 = 504.
Therefore, 3m + 15 = 504, which implies that 3m = 489, or m = 163. Since m denotes the
slope of the mean line, we conclude that the answer is 163. 2
10. (1998 AIME, Problem #13) If {a1 , a2 , a3 , . . . , an } is a set of real numbers, indexed
so that a1 < a2 < a3 < · · · < an , its complex power sum is defined to be a1 i + a2 i2 +
a3 i3 + · · · + an in , where i2 = −1. Let Sn be the sum of the complex power sums of all
nonempty subsets of {1, 2, . . . , n}. Given that S8 = −176 − 64i and S9 = p + qi, where
p and q are integers, find |p| + |q|.
Solution: The key idea here is to look for a relationship between the sum S8 , which is given,
and the sum S9 , which is sought. We see that the sum S9 includes all of the terms that
collectively sum to S8 , and in fact as we will explain below, the terms of S8 each occur twice
in the sum S9 . Let us now proceed to the solution: There are a total of 29 − 1 = 511 non-
empty subsets of {1, 2, . . . , 9}, and hence, there are 511 terms in the sum S9 . However, any
term that corresponds to a subset not containing “9” actually belongs to the sum S8 . We are
already given the value of S8 , so to find S9 , it suffices to compute the sum of the complex
power sums for subsets of {1, 2, . . . , 9} that contain “9”. Of course, the complex power sum
of {a1 , a2 , . . . , ak , 9} is the same as the complex power sum of {a1 , a2 , . . . , ak } plus 9ik+1 .
Therefore, all of the terms arising in the sum S8 appear again in the sum S9 , with one extra
term, 9ik+1 added on. Hence, the value of S8 is summed twice in the computation of S9 . In
8
addition, for each k = 0, 1, 2, . . . , 8, there are terms whose value is 9ik+1 which must
k
be added. Thus, we have
8
X 8
S9 = 2S8 + · 9ik+1
k
k=0

8 8 8 8 8 8 8
= 2S8 + 9 i − −i + +i − −i + +i
1 2 3 4 5 6 7
= 2(−176 − 64i) + 9 (i − 8 − 28i + 56 + 70i − 56 − 28i + 8 + i)
= 2(−176 − 64i) + 9(16i)
= 2(−176 − 64i) + 144i
= −352 + 16i.
Therefore, p = −352 and q = 16. Therefore, the answer is |p| + |q| = 352 + 16 = 368. 2
8
X 8
Remark: Observe that to compute ·9ik+1 , we can appeal to the Binomial Theorem
k
k=0
if we recognize that
8 8
X 8 k+1
X 8
· 9i = 9i ik
k k
k=0 k=0
= 9i(1 + i)8
√
= 9i( 2eiπ/4 )8
= 9i(16e2πi )
= 144i,
where we have used the polar form of z = 1 + i in order to compute z 8 quickly.
z+i
11. (2002 AIME, Problem #12) Let F (z) = for all complex numbers z 6= i, and
z−i
1
let zn = F (zn−1 ) for all positive integers n. Given that z0 = + i and z2002 = a + bi,
137
where a and b are real numbers, find a + b.
Solution: We have
z1 = F (z0 ), z2 = F (F (z0 ), z3 = F (F (F (z0 ))), ....
In effect, we are looking at a sequence of complex numbers. To compute z2002 , we must
theoretically compose the function F with itself 2002 times, starting with the given input
value z0 . Since this is obviously not practical, we suspect that after only a few applications of
F , a pattern develops that will enable us to quickly deduce the value z2002 . We have
z0 + i
z1 = F (z0 ) = ,
z0 − i
z0 +i
+i

z0 + i z0 + i + i(z0 − i)
z2 = F (z1 ) = F = zz00 −i
+i
=
z0 − i z0 −i − i z0 + i − i(z0 − i)
z0 +i+i(z0 −i)
z +i−i(z −i) + i

z0 + i + i(z0 − i)
z3 = F (z2 ) = F = z0 +i+i(z0 −i)
z0 + i − i(z0 − i) 0 0
−i
z0 +i−i(z0 −i)
z0 + i + i(z0 − i) + i(z0 + i) + z0 − i 2z0 + 2iz0
= = = z0 .
z0 + i + i(z0 − i) − i(z0 + i) − z0 + i 2 + 2i
This calculation shows that F (F (F (z))) = z for all z 6= i. Hence, z0 = z3 = z6 = · · · = z2001 .
Therefore,
1
z0 + i + 2i
z2002 = F (z2001 ) = F (z0 ) = = 137 1 = 1 + 274i.
z0 − i 137
Thus, a = 1 and b = 274. Hence, we conclude that the answer to this problem is a + b =
1 + 274 = 275. 2
12. (1989 AIME, Problem #14) Given a positive integer n, it can be shown that every
complex number of the form r + si, where r and s are integers, can be uniquely
expressed in the base −n + i using the integers 0, 1, 2, . . . , n2 as “digits”. That is,
the equation
r + si = am (−n + i)m + am−1 (−n + i)m−1 + · · · + a1 (−n + i) + a0
is true for a unique choice of nonnegative integer m and digits a0 , a1 , . . . , am chosen
from the set {0, 1, 2, . . . , n2 }, with am 6= 0. We then write
r + si = (am am−1 . . . a1 a0 )−n+i
to denote the base −n + i expansion of r + si. There are only finitely many integers
k + 0i that have four-digit expansions
k = (a3 a2 a1 a0 )−3+i a3 6= 0.
Find the sum of all such k.
Solution: Although the question is rather lengthy and perhaps intimidating upon first read,
what it is saying is really quite straightforward. We are seeking to add up all integers k that
can be written in the form
k = a3 (−3 + i)3 + a2 (−3 + i)2 + a1 (−3 + i) + a0
= a3 (−18 + 26i) + a2 (8 − 6i) + a1 (−3 + i) + a0
= (−18a3 + 8a2 − 3a1 + a0 ) + (26a3 − 6a2 + a1 )i.
Note that a0 , a1 , a2 , and a3 are chosen from the set S = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9}. Since k is an
integer, we must have 26a3 − 6a2 + a1 = 0. That is,
a1 = 6a2 − 26a3 . (11.100)
Hence, we can write

k = −18a3 + 8a2 − 3a1 + a0
= −18a3 + 8a2 − 3(6a2 − 26a3 ) + a0
= 60a3 − 10a2 + a0 .
Using (11.100), we see that a3 ≤ 2, for if a3 ≥ 3, it is impossible for both a1 and a2 to belong
to S. Since we are given that a3 6= 0, we have only two possibilities: a3 = 2 or a3 = 1.
Case 1: a3 = 2. Here, a1 = 6a2 − 52. Thus, since we must have a1 ≥ 0, we must have a2 = 9.
Then a1 = 6a2 − 26a3 = 54 − 52 = 2. Since we have no constraints on a0 , we obtain ten
different values of k here, which can be computed using k = 60a3 − 10a2 + a0 = 30 + a0 . The
integers obtained are 30, 31, 32, 33, 34, 35, 36, 37, 38, and 39.
Case 2: a3 = 1. In order for a1 to belong to S, (11.100) requires that a2 = 5. Since we

again have no constraints on a0 , once more we garner ten different values of k here as well,
which can be computed using k = 60a3 − 10a2 + a0 = 10 + a0 . The integers obtained are
10, 11, 12, 13, 14, 15, 16, 17, 18, and 19.
The sum of the values obtained in the above two cases are
(30+31+32+33+34+35+36+37+38+39)+(10+11+12+13+14+15+16+17+18+19) = 490.
13. (1988 AIME, Problem #13) Find a if a and b are integers such that x2 − x − 1 is a
factor of ax17 + bx16 + 1.
Solution #1: We can perform long division of p(x) = ax17 + bx16 + 1 by d(x) = x2 − x − 1
and determine the values of a and b required to obtain a quotient q(x) that has a remainder
r(x) = 0. (See the Division Algorithm, Proposition 7.4.7.) The result is
q(x) = ax15 + (a + b)x14 + (2a + b)x13 + (3a + 2b)x12 + · · · + (987a + 510b)
and
r(x) = (1597a + 987b)x + (987a + 610b + 1). (11.101)
Note that the coefficients of a and b appearing in the terms of both q(x) and r(x) consist
of Fibonacci numbers, with the coefficient of a being one Fibonacci number larger than the
coefficient of b. The table below summarizes these coefficients; compare with the list in (5.16).
k Coefficient of xk in q(x)
15 a
14 a+b
13 2a + b
12 3a + 2b
11 5a + 3b
10 8a + 5b
9 13a + 8b
8 21a + 13b
7 34a + 21b
6 55a + 34b
5 89a + 55b
4 144a + 89b
3 233a + 144b
2 377a + 233b
1 610a + 377b
0 987a + 510b
We see that from Equation (11.101) that r(x) = 0 if and only if

1597a + 987b = 0 and 987a + 610b = −1.
We can eliminate b from this system of two equations and two unknowns by multiplying the
first equation by 610, the second equation by −987, and adding the results:
(610)(1597) − 9872 a = 987.

Since (610)(1597) − 9872 = 1, we conclude that a = 987. 2
Solution #2: For x2 − x − 1 to be a factor of ax17 + bx16 + 1, note by part (a) of Proposition
7.4.8 that any roots of x2 − x − 1 must also be roots of ax17 + bx16 + 1. But according to the
quadratic formula, the solutions of x2 − x − 1 = 0 are precisely
√ √
1+ 5 1− 5
x0 = and 1 − x0 = .
2 2
Thus, we must have
ax17 16
0 + bx0 + 1 = 0 and a(1 − x0 )17 + b(1 − x0 )16 + 1 = 0. (11.102)
√
1± 5
Since these equations involve powers of , we are reminded of the formula (5.17), which
2
tells us that the nth Fibonacci number is
1
fn = √ (xn0 − (1 − x0 )n ).
5
This solution hopes to take advantage of this formula. To √ do so, we should subtract the second
equation in (11.102) from the first, and then divide by 5. The result is

1
√ a(x17 0 − (1 − x 0 )17
) + b(x 16
0 − (1 − x 0 ) 16
) = 0.
5
That is,
af17 + bf16 = 0.
Referring to the list (5.16), we have f16 = 987 and f17 = 1597, so that
1597a + 987b = 0. (11.103)
One easy solution to this for (a, b) is (987, −1597), where we chose a > 0 since we know that
the value of a, requested in this question, is an AIME answer in the (nonnegative) range 000
– 999. Moreover, we observe from Equation (11.103) that 987 divides 1597a. However, we can
quickly check that 1597 shares no common factors with 987 (i.e. gcd(1597, 987) = 1), so we
conclude from Proposition 4.3.1 that 987 must divide a. Thus, since we know that a ≤ 999,
we must a = 987. 2
Remark: Solution #2 makes use of the fact that the answers to AIME problems are always
integers in the interval 000 – 999. It is important to realize that AIME questions are never
written with the intention that you must depend on this knowledge (see, for example, Solution
#1). Still, as Solution #2 demonstrates, it can be absolutely useful for an AIME contestant
to take advantage of the AIME answer requirements!
14. (2005 AIME-2, Problem #13) Let P (x) be a polynomial with integer coefficients
that satisfies P (17) = 10 and P (24) = 17. Given that the equation P (n) = n + 3 has
two distinct integer solutions n1 and n2 , find the product n1 · n2 .
Solution: We can turn the integer solutions n1 and n2 into roots of a polynomial by defining
S(x) := P (x) − x − 3.
Then S(n1 ) = S(n2 ) = 0, so by using Equation (7.16), we can write
S(x) = Q(x)(x − n1 )(x − n2 ),
where Q(x) is a polynomial. Now
S(17) = P (17) − 20 = −10 and S(24) = P (24) − 27 = −10.
Hence,
Q(17)(17 − n1 )(17 − n2 ) = Q(24)(24 − n1 )(24 − n2 ) = −10,
which implies that all of the following are divisors of 10:
n1 − 17, n2 − 17, n1 − 24, n2 − 24.
Let us focus first on n1 . The two numbers n1 − 17 and n1 − 24 differ by 7 and are both
divisors of 10, which implies that they belong to the set {−10, −5, −2, −1, 1, 2, 5, 10}. The
only pairs of elements of this set that differ by exactly 7 are: (a) −5, 2, and (b) −2, 5. Since
n1 −17 > n1 −24, we must have n1 −17 ∈ {2, 5}. Hence, n1 ∈ {19, 22}. By the same reasoning,
we also have n2 ∈ {19, 22}. Since n1 6= n2 , we conclude that {n1 , n2 } = {19, 22}, so that
n1 · n2 = 19 · 22 = 418.
2
15. (2011 AIME, Problem #15) For some integer m, the polynomial x3 − 2011x + m has
the three integer roots a, b, and c. Find |a| + |b| + |c|.
Solution: From Equation (7.16), we can write
x3 − 2011x + m = (x − a)(x − b)(x − c)

= x3 − (a + b + c)x2 + (ab + ac + bc)x − abc,

a + b + c = 0, (11.104)
ab + ac + bc = −2011, (11.105)
and
abc = −m. (11.106)
While we are not told what m is, we can draw a useful conclusion about it from the fact
observe that m 6= 0, for otherwise
that the polynomial has only integer roots. Namely, we √
x3 −2011x+m = x(x2 −2011) has two non-integer roots ± 2011. Since m 6= 0, then Equation
(11.106) implies that a, b, c 6= 0.
At this point, we have no additional information about a, b, and c. To get further, it will be
helpful to make some assumptions. For instance, since we possess exactly the same information
about each of a, b, and c, we may assume without loss of generality that
|a| ≥ |b| ≥ |c| > 0. (11.107)
We will assume here that a > 0. It is easy to see that in the case a < 0, we simply need to
reverse all of the signs in the discussion below. In the end, we need to compute |a| + |b| + |c|,
which means the actual signs of a, b, and c are immaterial.
We will try to establish a finite range of (positive) values that a can assume, and then we will
test each one. It follows from Equations (11.104) and (11.107) that b < 0 and c < 0. Now we
can use Equation (11.105) to write
−2011 = ab + ac + bc = a(b + c) + bc = −a2 + bc,
which implies that

a2 = bc + 2011.
Since bc > 0, we conclude that √
a> 2011 > 44.
On the other hand, we can find an upper bound on a by placing a maximum value on bc. Note
a
that b + c = −a and bc will attain a maximum value when b = c; that is, b = c = − . Thus,
2
a 2
a2 = bc + 2011 < − + 2011,
2
which implies that
3 2
a < 2011
4
or
8044
a2 < ≈ 2681.3 < 2704 = 522 .
3
Thus,
a < 52.
Thus, we have a ∈ {45, 46, 47, 48, 49, 50, 51}. Using
bc = a2 − 2011,
we can make the following table:
a bc Prime Factorization of bc b+c Conclusion

45 14 2·7 −45 Impossible
46 105 3·5·7 −46 Impossible
47 198 2 · 32 · 11 −47 Impossible
48 293 293 −48 Impossible
49 390 2 · 3 · 5 · 13 −49 b = −39, c = −10
50 489 3 · 163 −50 Impossible
51 590 2 · 5 · 59 −51 Impossible
There is only one solution for a, b, and c that satisfies the conditions. We conclude that
|a| + |b| + |c| = 49 + 39 + 10 = 98 = 098. 2
1. (2008 AIME, Problem #2) Square AIM E has sides of length 10 units. Isosceles
triangle GEM has base EM , and the area common to triangle GEM and square
AIM E is 80 square units. Find the length of the altitude to EM in ∆GEM .
Solution: Suppose that GE crosses the square at B and GM crosses the square at C, as
shown in Figure 34. Let R denote the midpoint of BC, and let S denote the midpoint of EM .
Note that R lies along the altitude GS whose length we are asked to compute. We summaraize
this in Figure 34.
FIGURE 34 GOES HERE
Note that ∆BGR is similar ∆EGS, since both triangles share angle ∠BGR and both triangles
contain a right angle. Therefore, using Equation (8.11), we have
BR GR GR
= = . (11.108)
ES GS GR + 10
1
We have ES = EM = 5, and we can easily compute BR as follows. Since the area of square
2
AIM E is 100 square units, the sum of the areas of triangles ∆ABE and ∆ICM must be 20
square units. Since these triangles are congruent, each has area 10 square units. Therefore,
using Equation (8.14) for the area of a triangle, we deduce that AB = CI = 2. Therefore,
BC = AI − AB − CI = 10 − 2 − 2 = 6,
1
so that BR = BC = 3.
2
Hence, from Equation (11.108), we find that
3 GR
= ,
5 GR + 10
so that, after cross-multiplying and simplifying, we have
3(GR + 10) = 5GR.
Thus, GR = 15. We conclude that
GS = GR + 10 = 15 + 10 = 25 = 025.
2. (2007 AIME-2, Problem #3) Square ABCD has side length 13, and points E and
F are exterior to the square such that BE = DF = 5 and AE = CF = 12. Find EF 2 .
Solution: We can extend AE and DF and let G denote the intersection point of the extended
segments, as shown in Figure 35.
FIGURE 35 GOES HERE
From the sketch, we suspect that ∠G is a right angle, but this can also be carefully justified
as follows. The triangles ∆AEB and ∆CF D are right (5-12-13) triangles. Since ∠EAB
is complementary to both ∠EBA and ∠GAD, we see that angles ∠EBA and ∠GAD are
congruent. Similarly, ∠CDF is complementary to both ∠DCF and ∠ADG. Therefore, ∠DCF
and ∠ADG are also congruent angles. Thus, two (and hence, all three) of the angles in ∆DGA
are congruent to the angles in ∆AEB and ∆CF D. Thus, we must have that ∠G is a right
angle.
Since the hypotenuse of right triangle DGA has length AD = 13, which is the same as the
hypotenuses of right triangles ∆AEB and ∆CF D, we conclude that all three of these triangles
are actually congruent. Hence, AG = 5 and GD = 12. Thus, we have
EG = EA + AG = 12 + 5 = 17 and GF = F D + DF = 12 + 5 = 17.
Finally, we can apply the Pythagorean Theorem to ∆EGF to conclude that
EF 2 = EG2 + GF 2 = 172 + 172 = 578.
Remark: Of course, not all of the solution presented here must be formally considered to
obtain the correct answer. Still, mathematicians insist on a rigorous solution, which is what
we have provided. 2
3. (1999 AIME, Problem #4) The two squares shown share the same center O
and have sides of length 1. The length of AB is 43/99 and the area of octagon
ABCDEF GH is m/n, where m and n are relatively prime positive integers. Find
m + n.
Solution #1: Consider the right triangle in Figure 36, whose hypotenuse is AB.
FIGURE 36 GOES HERE
Let J denote the vertex at the right angle. The perimeter of ∆AJB is 1, since the three sides
of ∆AJB have lengths that match the lengths of the three segments comprising one side of
43 56
either square. Therefore, AJ + BJ = 1 − = . Let
99 99
56
AJ = x and BJ = − x.
99
By the Pythagorean Theorem, we have

2 2
56 43
x2 + −x = .
99 99
Expanding this and rearranging the expression, we obtain

2 2
112 56 43
2x2 − x+ − = 0.
99 99 99
On the other hand, using Equation (8.14), we have that the area of ∆ABJ is

1 56
[ABJ] = x −x .
2 99
We view octagon ABCDEF GH as a square of side 1 with four triangles congruent to ∆ABJ
removed. That is, the area of octagon ABCDEF GH is
[ABCDEF GH] = 1 − 4[ABJ]

56
= 1 − 2x −x
99

112 2
=1− x − 2x
99
" 2 #
2
56 43
=1− −
99 99

56 43 56 43
=1− − +
99 99 99 99
13
=1−
99
86
= .
99
Therefore, we have m = 86 and n = 99, so that m + n = 185. 2
Solution #2: The octagon can be decomposed into eight triangles, each of which is congruent
43
to ∆OAB by symmetry. Taking AB as the base of ∆OAB, then this triangle has base and
99
1
height . Hence,
2
1 1 43
[OAB] = · · .
2 2 99
Hence,
43 86
[ABCDEF GH] = 8[OAB] = 2 · = .
99 99
Therefore, we have m = 86 and n = 99, so that m + n = 185. 2
4. (2006 AIME-2, Problem #6) Square ABCD has sides of length 1. Points E and
F are on BC and CD, respectively, so that ∆AEF is equilateral. A square with
vertex B has sides that are parallel to those of √ ABCD and a vertex on AE. The
a− b
length of a side of this smaller square is , where a, b, and c are positive
c
integers and b is not divisible by the square of any prime. Find a + b + c.
Solution: We will employ a use of similar triangles. Refer to Figure 37 below. We have
labelled BE = DF = r, and thus, F C = EC = 1 − r. Furthermore, we denote the corners of
the smaller square as shown in Figure 37. In particular, we set BG = GH = s, which is the
side length of the smaller square.
FIGURE 37 GOES HERE

Since ∆AGH and ∆ABE are both right triangles that share an acute angle, they are similar.
Thus,
AG AB 1−s 1 1 1
= or = or =1+ . (11.109)
GH BE s r s r
Our goal is to solve for s, but using Equation (11.109), it will suffice to determine r. For this,
note that since ∆AEF is equilateral, we have AE = EF . Since
p p
AE = 1 + r2 and EF = 2(1 − r)2 ,
we obtain
1 + r2 = 2(1 − r)2 .
This simplifies to the quadratic equation
r2 − 4r + 1 = 0,
√
quadratic formula supplies the roots r = 2 ± 3. However, since r < 1, we must have
and the √
r = 2 − 3. Now, from Equation (11.109), we deduce that
√
1 1 3− 3
=1+ √ = √ .
s 2− 3 2− 3
Therefore, √ √
2− 3 3− 3
s= √ = .
3− 3 6
Therefore, we have a = 3, b = 3, and c = 6. Thus, a + b + c = 3 + 3 + 6 = 12 = 012. 2
Remark: A slight modification to the solution above is to view Figure 37 as sitting in the
xy-plane, with the point A at the origin. The equation of the line containing segment AE is
1
y= x,
r
1
since the slope of this line is . The vertex H lies along this line and has coordinates (s, 1 − s).
r
Thus,
1 s
1 − s = (s) = ,
r r
from which we can quickly deduce Equation (11.109) and proceed with the solution above.
5. (2008 AIME-2, Problem #5) In trapezoid ABCD with BC || AD, let BC = 1000 and
AD = 2008. Let ∠A = 37◦ , ∠D = 53◦ , and M and N be the midpoints of BC and AD,
respectively. Find the length M N .
Solution: Extend AB and CD so that they intersect, say at point E. (See Figure 38.)
FIGURE 38 GOES HERE

By virtue of sides BC being parallel to AD, angles ∠EBC and ∠EAD are corresponding
angles, and hence have equal measure by Theorem 8.1.1. The same can be said for angles
∠ECB and ∠EDA. Thus, ∆EBC and ∆EAD are similar triangles. Therefore, we have
EM 1000 EC
= = . (11.110)
EN 2008 ED
Moreover, we observe by Theorem 8.2.1 that ∠E is a right angle. Thus, ∆EBC and ∆EAD
are right triangles.
Consider triangle ∆EN D. We have N D = 1004 (half the length of segment AD) and, applying
trigonometry to the right triangle EAD, we have ED = 2008 cos 53◦ . Therefore, we can use
the Law of Cosines to determine EN :
EN 2 = N D2 + ED2 − 2(N D)(ED)(cos 53◦ )
= 10042 + (2008 cos 53◦ )2 − 2(1004)(2008 cos 53◦ )(cos 53◦ )
= 10042 .
Hence, EN = 1004. From the ratio established in (11.110), we deduce that EM = 500. Hence,
1000
M N = EN − EM = EN − (EN ) = 1004 − 500 = 504.
2008
2
6. (2005 AIME, Problem #7) In quadrilateral ABCD, BC = 8, CD = 12, AD = 10, and
√
m∠A = m∠B = 60◦ . Given that AB = p + q, where p and q are positive integers,
find p + q.
Solution #1: In Figure 39, we have dropped perpendiculars from both C and D to the
segment AB. Label the point where the perpendicular from C meets AB as E, and label the
point where the perpendicular from D meets AB as F .
FIGURE 39 GOES HERE
Then we have
AB = AF + EF + BE. (11.111)
Using right triangle BCE, since BC = 8 and m∠B = 60◦ , we have
√
BE = 4 and CE = 4 3.
Similarly, using right triangle ADF , since AD = 10 and m∠A = 60◦ , we have
√
AF = 5 and DF = 5 3.
Thus, to complete the calculation of AB in Equation (11.111), we need to compute EF . To do

this, let us draw a perpendicular segment from C that meets DF at G.This creates a rectangle,
√ √
CEF G, and we note that EF = CG. Since CE = 4 3, DF = 5 3, and CE is parallel to
DF , we have √
DG = DF − CE = 3.
In ∆CGD, the Pythagorean Theorem gives
CG2 = CD2 − DG2 = 144 − 3 = 141.
Thus, √
EF = CG = 141.
√
Thus,
√ substituting the values into Equation (11.111), we conclude that AB = 5 + 4 + 141 =
9 + 141, so that p = 9 and q = 141. We have p + q = 150. 2
Solution #2: As with Solution #1, this route to the answer also involves drawing additional
points on the given figure. The strategy here is to note that since angles A and B have the
same measure, we can build an isosceles triangle by extending the sides AD and BC to a
common point E. This is shown in Figure ??????
INSERT FIGURE !!!!!!!!
In fact, since the sum of the angle measures in a triangle is 180◦ , we have m∠E = 60◦ , so that
∆EAB is actually equilateral. Let us denote the side length of ∆EAB by x. Then
DE = x − 10 and CE = x − 8.
We can solve for x by applying the Law of Cosines to ∆EDC using angle ∠E:
CD2 = DE 2 + CE 2 − 2(DE)(CE)(cos 60◦ )

= (x − 10)2 + (x − 8)2 − 2(x − 10)(x − 8)(cos 60◦ )
= x2 − 18x + 84.
Since CD = 12, this can be reduced to
x2 − 18x − 60 = 0.
√
√ the quadratic formula, we find that√x = 9 ± 141. Since x > 0, we must have x =
With
9 + 141, so we conclude that AB = 9 + 141, as in the first solution. 2
7. (1988 AIME, Problem #7) In triangle ABC, tan ∠CAB = 22/7, and the altitude
from A divides BC into segments of length 3 and 17. What is the area of triangle
ABC?
Solution: Let D be the point on segment BC at which the altitude from A meets BC, and
let x = AD > 0. (See Figure ?????)
INSERT FIGURE !!!!!!!!!!!!!!!!!!!!!!!!!!!!
22
We are given tan ∠CAB = , and Figure ??? shows that ∠CAB can be decomposed into
7
two angles:
m∠CAB = m∠CAD + m∠DAB.
Therefore, we should consider using the identity
tan α + tan β
tan(α + β) = .
1 − tan α tan β
We have
22
= tan ∠CAB
7
= tan(∠CAD + ∠DAB)
tan ∠CAD + tan ∠DAB
=
1 − tan ∠CAD · tan ∠DAB
17
+3
= x 17 x 3
1− x · x
20x
= 2 .
x − 51
Cross multiplying, we have
22(x2 − 51) = 140x,
or 22x2 − 140x − 1122 = 0. This quadratic equation can be easily solved for x to obtain x = 11.
(The other root of this quadratic equation is negative.) Thus, ∆ABC has height x = 11 and
base 17 + 3 = 20, so that according to Equation (8.14), the area of triangle ABC is
1
[ABC] = (17 + 3)x = 10x = 110.
2
2
8. (2004 AIME, Problem #10) A circle of radius 1 is randomly placed in a 15-by-36
rectangle ABCD so that the circle lies completely within the rectangle. Given that
the probability that the circle will not touch diagonal AC is m/n, where m and n
are relatively prime positive integers, find m + n.
Solution: The center of a circle of radius 1 that is entirely contained in the rectangle ABCD
must fall within a 13-by-34 rectangle that excludes a strip of width 1 from all edges of the
rectangle. In order for the circle to avoid touching AC, the center of the circle must be at least
one unit away from AC. As Figure 40 shows, the possible locations of the center of the circle
so that the circle does not intersect AC forms two congruent right triangles, one of which is
∆EF G.
FIGURE 40 GOES HERE

We wish to determine the area of ∆EF G. We must compute the lengths of the legs of ∆EF G.
From E, draw a perpendicular to meet AD at H. The slope of EF matches the slope of
15 5
AC, which is − = − . Therefore, if we extend EF to the point I where it meets AD,
36 12
5
then ∆IHE is a 5-12-13 right triangle. Since HE = 1, this implies that HI = . We can
12
also draw a perpendicular from I to a point J on AC, thereby forming another 5-12-13 right
13
triangle, ∆AIJ. Since IJ = 1, we have AI = . Hence,
12
5 13 18 3
AH = + = = ,
12 12 12 2
and thus, the height of ∆EF G is
3 25
EG = 15 − −1= .
2 2
Next, we determine the length GF . Draw a perpendicular from F to a point K on CD. Extend
EF to a point L on CD and consider the right triangle F KL. Since the hypotenuse F L has
5 12
slope − and F K = 1, we deduce that KL = . Finally, draw a perpendicular from L to a
12 5
point M on AC to form a 5-12-13 right triangle LM C. Since LM = 1, the hypotenuse of this
13
5-12-13 triangle has length LC = . Hence,
5
12 13
KC = + = 5,
5 5
and therefore,
GF = 36 − 5 − 1 = 30.
Hence, according to Equation (8.14), the area of triangle EF G is
1 25 375
[EF G] = · · 30 = .
2 2 2
By symmetry, there is a second triangle of equal area above AC in which the center of the
circle of radius 1 can be located. So the total area that the center of the circle can be placed
375
so that the circle does not touch AC is 2 · = 375. If we ignore the restriction about
2
not touching diagonal AC, the total area in which the center of the circle can be placed is
13 · 34 = 442. Note that m = 375 and n = 442 are relatively prime. Hence, the answer is
m + n = 375 + 442 = 817. 2
9. (2010 AIME-2, Problem #9) Let ABCDEF be a regular hexagon. Let G, H, I, J,
K, and L be the midpoints of sides AB, BC, CD, DE, EF , AF , respectively. The
segments AH, BI, CJ, DK, EL, and F G bound a smaller regular hexagon. Let the
ratio of the area of the smaller hexagon to the area of ABCDEF be expressed as
m
a fraction where m and n are relatively prime positive integers. Find m + n.
n
Solution #1: Denote the intersection of segments AH and F G by X, and denote the inter-
section of segments AH and BI by Y . (See Figure 41.)
FIGURE 41 GOES HERE
We may as well assume that the side lengths of ABCDEF is 1. By a generalization of Theorem
8.2.18, the ratio of the areas of the two hexagons is
2
m XY
= = XY 2 .
n AB
Therefore, we focus on determining XY . First observe that
XY = AH − AX − Y H. (11.112)
We will determine all three values on the right-hand side of (11.112).

We begin with AH. By Theorem 8.2.2, each interior angle of a regular hexagon measure 120◦ .
Thus, in ∆ABH, we know two side lengths (AB = 1 and BH = 1/2) and the angle between
them (m∠ABH = 120◦ ), so that we can apply the Law of Cosines:
2
2 2 2 ◦ 2 21 1 7 2
AH = AB + BH − 2(AB)(BH) cos 120 = AB + BH + (AB)(BH) = 1 + + = .
2 2 4
Hence, √
7
AH = . (11.113)
2
Next, we determine Y H. Observe that ∠XY I is an interior angle of the smaller regular hexagon
GHIJKL, so again we have m∠XY I = 120◦ . Now using vertical angles (see Theorem 8.1.1),
we have m∠BY H = 120◦ . Since ∆ABH and ∆BY H both share ∠Y HB, and additionally
they both contain a 120◦ angle, they are similar. That is, ∆ABH is similar to ∆BY H. Hence,
YH BH 1/2 1
= =√ =√ .
BH AH 7/2 7
Hence,
BH 1
YH = √ = √ . (11.114)
7 2 7
Finally, to find AX, observe that AX = BY by symmetry. Using the similarity of the same
triangles once more, we use the fact that AB = 2BH to conclude that
1
BY = 2Y H = √ .
7
Thus,
1
AX = BY = √ . (11.115)
7
Substituting the values obtained in (11.113), (11.114), and (11.115) into Equation (11.112),
we find that
√
7 1 1 7−2−1 2
XY = AH − AX − Y H = −√ − √ = √ =√ .
2 7 2 7 2 7 7
Hence,
2
2 4
XY 2 =√ = ,
7 7
and thus, m = 4, n = 7, and the answer is m + n = 11 = 011. 2
Solution #2: We can use a coordinates-based approach. Place the outer hexagon so that
its center, O, coincides with the origin point (0, 0) of the coordinate plane. (See Figure
???????????)
INSERT FIGURE !!!!!!!!!!!!!!!!!!!!!!!!!!!!
As in Solution #1, assume that all sides of the hexagon have length 1. Then the coordinates
of the vertices of the outer hexagon are
√ √ √ √
1 3 1 3 1 3 1 3
A(1, 0), B( , ), C(− , ), D(−1, 0), E(− , − ), F ( , − ).
2 2 2 2 2 2 2 2
Notice that
OA = OB = OC = OD = OE = OF = 1.
If we can determine the coordinates of the point Y , say Y (u, v), then we can determine OY ,
and we will have the ratio of the two hexagons as
2
OY
= OY 2 .
OA
To find u and v, we can find the equations of the lines containing the line segments
√
that intersect
3
at Y . First, the line containing segment AH contains the points H(0, 2 ) and A(1, 0). It is
routine to find the equation of the line containing these points:
√
3
y=− (x − 1). (11.116)
2
Similarly, the line containing the segment BI has equation (written in point-slope form)
√ √
3 3 1
y= + (x − ). (11.117)
2 5 2
The coordinates (u, v) must be a simultaneous solution to Equations (11.116) and (11.117).
We leave to the reader the straightforward verification that
1 3√
u= and v= 3.
7 7
Thus, r √
p 1 27 28 2
OY = u2 + v 2 = + = =√ .
49 49 7 7
Thus, the ratio in question is
4
OY 2 = ,
7
from which the answer follows as in Solution #1. 2
10. (1986 AIME, Problem #9) In ∆ABC shown below, AB = 425, BC = 450 and
CA = 510. Moreover, P is an interior point chosen so that the segments DE, F G
and HI are each of length d, contain P , and are parallel to the sides AB, BC and
CA, respectively. Find d.
Solution: The situation is pictured in Figure 42.
FIGURE 42 GOES HERE
The challenge of this problem is an overabundance of information arising from the many similar
triangles showing in ∆ABC. In order to determine the length of d, it suffices to determine,
without loss of generality, DP and P E, since
d = DP + EP. (11.118)
(It would be equally valid to acquire the distances GP and P F , or the distances HP and P I.)
The strategy to find DP and EP is to use similar triangles that relate these lengths to the
known lengths of the sides of the large triangle ABC. For instance, using the similarity of11
∆DGP and ∆ACB, note that
DP 425 5
= = .
DG 510 6
Therefore,
5
DP = DG. (11.119)
6
It is possible to express the length DG in terms of d by using the fact that
d = IP + P H
as follows:
AC = AD + DG + GC
= IP + DG + P H
= d + DG.
11 The triangles ∆DGP and ∆ACB are similar because the given sides of the two triangles are parallel, so that
corresponding angles are congruent.

Since AC = 510, we have

DG = 510 − d.
Therefore, together with Equation (11.119), we conclude that
5
DP = (510 − d). (11.120)
6
Likewise, using the similarity of ∆EHP and ∆BCA, we have
EP 425 17
= = .
EH 450 18
Therefore,
17
EP = EH.
18
Using a similar argument to what we used above to write DG in terms of d, the reader can
check that
17
EP = (450 − d). (11.121)
18
Adding the expressions we obtained for DP and EP in Equations (11.120) and (11.121) and
using Equation (11.118), we find that
5 17 16
d = DP + EP = (510 − d) + (450 − d) = 850 − d.
6 18 9
That is,
9
d= · 850 = 306.
25
2
11. (2007 AIME-2, Problem #9) Rectangle ABCD is given with AB = 63 and BC = 448.
Points E and F lie on AD and BC respectively, such that AE = CF = 84. The
inscribed circle of triangle BEF is tangent to EF at point P , and the inscribed
circle of triangle DEF is tangent to EF at point Q. Find P Q.
Solution: Figure 43 illustrates the problem.
FIGURE 43 GOES HERE
We have labelled the point where BE meets the inscribed circle of ∆BEF as G and the point
where BC meets the inscribed circle of ∆BEF as H. Note that this implies that
EG = EP and BG = BH and F P = F H.
(These equalities come from Theorem 8.3.6, the Two Tangent Theorem.) For notational con-
venience, let x = EP , y = BG, and z = F P . Applying the Pythagorean Theorem to the right
triangle ABE, note that
p p
x + y = 632 + 842 = 21 32 + 42 = 21 · 5 = 105
and
y + z = BG + F P = BH + F H = BC − F C = 448 − 84 = 364.
By symmetry, observe that EP = F Q. Thus,
P Q = F P − F Q = F H − EP = z − x = (y + z) − (x + y) = 364 − 105 = 259.
2
12. (2003 AIME-2, Problem #11) Triangle ABC is a right triangle with AC = 7,
BC = 24, and right angle at C. Point M is the midpoint of AB, and D is on the
same side of line AB as√C so that AD = BD = 15. Given that the area of ∆CDM
m n
can be expressed as , where m, n, and p are positive integers, m and p are
p
relatively prime, and n is not divisible by the square of any prime, find m + n + p.
Solution #1: This solution will make use of Equation (8.16), so we need to position the given
points on the coordinate plane. The computations with this approach will involve somewhat
large numbers, so readers preferring a more elegant computation-free approach are invited to
study Solution #2 below instead.
Proceeding with the present approach, we will assume that C is placed at the origin (0, 0) of
the xy-plane, A rests at (0, 7) and B rests at (24, 0) and proceed to determine the coordinates
of the other two vertices, M (x1 , y1 ) and D(x2 , y2 ), of the triangle CDM in question. From
there, we will be able to obtain the area of ∆CDM via the expression in Equation (8.16):
1
[CDM ] = |x1 y2 − x2 y1 |.
2
(See Figure 44.)
FIGURE 44 GOES HERE
First observe that the midpoint M of AB sits at (12, 72 ). Thus, x1 = 12 and y1 = 72 . Next we
derive the equation of the line containing the segment M D. Its slope is the negative reciprocal
of the slope of the line passing through A and B, since AB ⊥ M D. Since the line containing A
7 24
and B has slope − , the line containing segment M D has slope . (Recall that the slopes
24 7
of perpendicular lines are the negative reciprocals of one another.) Hence, its equation (in
point-slope form) is
7 24
y− = (x − 12).
2 7
That is,
7 24
y= + (x − 12).
2 7
We need to determine the coordinates of the point D(x2 , y2 ) on this line. We have
7 24
y2 = + (x2 − 12).
2 7
In addition, A has coordinates A(0, 7) and AD = 15. Applying the distance formula (8.3) to
the points A(0, 7) and D(x2 , y2 ), we obtain
2
(48x2 − 625)2

2 7 24
AD = 225 = x22 + + (x2 − 12) − 7 = x22 + .
2 7 196
Multiplying through by 196, this yields
44100 = 196x22 + (48x2 − 625)2 .
Expanding this expression and rearranging, we obtain the quadratic equation
2500x22 − 60000x2 + 346525 = 0.
We divide through by 25 to obtain
100x22 − 2400x2 + 13861 = 0.
Applying the quadratic formula, we obtain
7√
x2 = 12 ± 11.
10
7
√
From the figure, it is clear that x2 < 12, so that x2 = 12 − 10 11. Hence,
7 24 7 24 7 √ 7 12 √
y2 = + (x2 − 12) = − · 11 = − 11.
2 7 2 7 10 2 5
Therefore, we can now compute the area of triangle CDM via Equation (8.16):
1
= |x1 y2 − x2 y1 |
2
7 12 √ 7√

1 7
= (12)( − 11) − (12 − 11)( )
2 2 5 10 2
√ √

1 144
49
= − 11 + 11
2 5 20
527 √
= 11.
40
Therefore, m = 527, n = 11, and p = 40, so that m + n + p = 578. 2
Solution #2: We will find the area of ∆CDM by applying Equation (8.14), where we will
regard the base of the triangle as the side b := DM . This side length is easy to find, since it
is one leg of the right triangle AM D. We know that
1 1p 2 25
AM = AB = 7 + 242 = .
2 2 2
Since AD = 15, we obtain
s 2
5√

25
DM = 152 − = 11.
2 2
Now the height h of the triangle can be expressed as
h = CM sin θ, (11.122)
where θ := ∠CM D. Let us begin by determining CM . If we draw a line segment from M to

its orthogonal projection along BC, say at point P , we create a right triangle CM P whose
hypotenuse is CM .
INSERT FIGURE !!!!!!!!!!!!!!!!!!!
In fact, we have s 2
7 25
CM = 122 + = . (11.123)
2 2
Now we actually know all of the side lengths in ∆BCM , and the angle at M measure θ +90 (in
degrees). Thus, we can apply the Law of Cosines at this angle, along with Equations (11.123)
and (6.34):
BC 2 = CM 2 + BM 2 − 2(CM )(BM ) cos(θ + 90)
2 2 2
25 25 25
= + +2 sin θ
2 2 2
2
25
=2 (1 + sin θ) .
2
Substituting BC = 24 and rearranging the equation, we can routinely compute that
527
sin θ = .
625
Now we can find h in Equation (11.122):
25 527 527
h= · = .
2 625 50
Thus, we conclude that

1 1 5√ 527 527 √
[CDM ] =bh = · 11 · = 11.
2 2 2 50 40
Thus, as in Solution #1, we arrive at the answer 578. 2
13. (1990 AIME, Problem #12) A regular 12-gon is inscribed in a circle of radius 12.
The sum of the lengths
√ √ of all√ sides and diagonals of the 12-gon can be written
in the form a + b 2 + c 3 + d 6, where a, b, c, and d are positive integers. Find
a + b + c + d.
Solution: The 12-gon is shown in Figure 45.
FIGURE 45 GOES HERE
Observe that every side and every diagonal of the 12-gon has a length equal to the length of
one of these: AB, AC, AD, AE, AF , or AG. Therefore, we can break this problem into two
parts. First, we will determine each of these six lengths, and then we will count how many
sides and diagonals have each of those lengths.
To address the first part, we can use the Law of Cosines. For example,
√ !
2 2 2 2 2 3 ◦
AB = OA +OB −2(OA)(OB) cos(∠AOB) = 12 +12 −2·12·12·cos 30 = 288 1 − .
2
The reader can apply the Law of Cosines with exactly the same reasoning to determine that
√ !
2 2 2 2 3
AC = 144, AD = 288, AE = 432, AF = 288 1 + , AG = 24.
2
(Of course, the value of AG is simply the diameter of the circle containing the 12-gon.) Hence,
√ √ √ √
q q
AB = 12 2 − 3, AC = 12, AD = 12 2, AE = 12 3, AF = 12 2 + 3, AG = 24.
There are 12 segments each of lengths AB, AC, AD, AE, and AF , and six segments of length
AG. The sum, S, is given by
√ √ √ √
q q
S = 12 12 2 − 3 + 12 + 12 2 + 12 3 + 12 2 + 3 + 6 · 24
√ √ √ √
q q
= 144 2− 3+1+ 2+ 3+ 2+ 3+1 .
The key observation at this point is the fact that12

√ √ √ √
√ √
q q
6− 2 6+ 2
2− 3= and 2+ 3= . (11.124)
2 2
12 There is an alternate way to determine the expressions for AB and AF that avoids the need for this observation.
We will explain this in the Remark following this solution.

Therefore,
√ √ √ √
q q
S = 144 2− 3+1+ 2+ 3+ 2+ 3+1
"√ √ √ √ #
6− 2 √ √ 6+ 2
= 144 +1+ 2+ 3+ +1
2 2
h √ √ √ i
= 144 2 + 6 + 2 + 3 .
Therefore, a = 288, b = c = d = 144. Hence, the answer is
a + b + c + d = 288 + 144 + 144 + 144 = 720.
Remark: We can avoid the observation in Equation (11.124) by computing AB and AF

differently. Namely, we can use the right triangle formed by O, A, and the midpoint of AB to
get
AB = 2 · OA · sin 15◦
= 24 sin 15◦
= 24 sin(45 − 30)
√ √ √ !
2 3 2 1
= 24 · · − ·
2 2 2 2
√ √
= 6( 6 − 2),
and similarly, √ √
AF = 6( 6 + 2).
Substituting these values into the final calculation for S will yield the same answer that we
obtained above.
14. (2000 AIME, Problem #14) In triangle ABC, it is given that angles B and C are
congruent. Points P and Q lie on AC and AB, respectively, so that AP = P Q =
QB = BC. Angle ACB is r times as large as angle AP Q, where r is a positive real
number. Find the greatest integer that does not exceed 1000r.
Solution: Throughout this problem, we will measure all angles in degrees. An excellent
starting point is to draw a figure and label the important features. Denote
θ := m∠(AP Q).
Therefore,
rθ = m∠(ACB).
Next, let ` denote the common length
` = AP = P Q = QB = BC.
These notations are shown in Figure 46.

FIGURE 46 GOES HERE
For a sound solution strategy to follow, observe that because there are several segments of
equal length ` in the figure, it is possible to create isosceles triangles rather easily. Such
triangles have base angles of equal measure, which will help us determine all of the important
angle measures we need to solve this problem. For instance, ∆QAP and ∆QBP are both
isosceles, which implies that their base angles have equal measure. For instance, since ∆QBP
is isosceles, we can let
α := m∠(QBP ) = m∠(QP B).
Thus, from Theorem 8.2.1,
m∠BQP = 180 − 2α.
Since ∠AQP is supplementary to ∠BQP and ∆QAP is isosceles,
m∠(AQP ) = m∠(QAP ) = 2α.
Applying Theorem 8.2.1 to ∆QAP , we have
θ = 180 − 4α. (11.125)
Now consider ∠BP C. We observe that the three angles that meet at P must sum to 180.
That is
θ + α + m∠(BP C) = 180.
Combining this with Equation (11.125), we find that
m∠(BP C) = 3α.
Focusing our attention now on ∆BP C, we now know that BC = ` and the angle opposite this
side is 3α. This suggests trying to apply the Law of Sines, which relates side lengths to the
measures of the angles oppsoite those sides. However, before we can use the Law of Sines, we
should determine another side length. In isosceles triangle QBP , note that
P C = AQ = 2` cos(2α).
Thus, the Law of Sines gives us
2` cos(2α) `
= . (11.126)
sin(∠P BC) sin(3α)
We need to determine more information about the angle β := m∠(P BC). Observe from
Theorem 8.2.1 that
β + 3α + rθ = 180
and from the fact that ∆ABC is isosceles that
β + α = rθ.
Putting these two equations together, we have
4α + 2β = 180.
Therefore, we can write

2α = 90 − β. (11.127)
Referring back to Equation (11.126), we find that
` 2` cos(2α) 2` cos(90 − β)
= = = 2`,
sin(3α) sin β sin β
where we have used Equation (6.34) in the last step. Hence,
1
sin(3α) = .
2
Since β > 0, Equation (11.127) implies that 0 < α < 45. Thus, 0 < 3α < 135. The only angle
in this interval whose sine value is 1/2 is 3α = 30. Thus, α = 10. Therefore, β = 90 − 2α = 70.
Thus, by Equation (11.125), we have θ = 140. Finally, since ∆ABC is isosceles, we must have
rθ = α + β, so that 140r = 80. Hence,
80 4 4000
r= = , and thus, 1000r = ≈ 571.4.
140 7 7
The greatest integer that does not exceed this is 571. 2
15. (2005 AIME-2, Problem #14) In ∆ABC, AB = 13, BC = 15, and CA = 14. Point
D is on BC with CD = 6. Point E is on BC such that ∠BAE ∼ = ∠CAD. Given that
BE = p/q, where p and q are relatively prime positive integers, find q.
Solution #1: This solution is focused on comparing areas of different triangles in this prob-
lem. Let α denote both of the angles ∠DAC and ∠EAB, and let β = ∠EAD, as shown in
Figure 47.
FIGURE 47 GOES HERE
Notice that several different triangles can be considered, such as
∆ABC, ∆ABE, ∆ABD, ∆ADE, ∆ADC, ∆ACE,
all of which have the same height h if we measure the height as the length of the altitude
from A to BC. We can determine the area of the largest triangle, ∆ABC, by using Heron’s
Formula (8.15), since we know the semiperimeter
13 + 14 + 15
s= = 21.
2
We obtain p √
[ABC] = 21(21 − 13)(21 − 14)(21 − 15) = 21 · 8 · 7 · 6 = 84.
Therefore, the length h of the altitude from A to BC can be determined from Equation (8.14)
to be
2 · 84 56
= .
15 5
Hence, we can use Equation (8.14) to obtain
252 168
[ABD] = and [ACD] = . (11.128)
5 5
This question is asking us to determine BE, and in view of Equation (8.14) and the fact that
56
we know h = already, it suffices to find [ABE]. If we consider AB to be the base of ∆ABE,
5
then we have
13
[ABE] = (AE) sin α. (11.129)
2
To determine sin α, we can use ∆ACD, whose area is already known:
168 14
= [ACD] = (AD) sin α,
5 2
which implies that
168 1
sin α = · .
35 AD
Substituting this result into Equation (11.129), we find that
13 168 AE
[ABE] = · · . (11.130)
2 35 AD
AE
To determine , we will use triangles ∆ACE and ∆ABD as follows:
AD
[ACE] 7 · AE sin(α + β) 14 AE
= 13 = · .
[ABD] 2 · AD sin(α + β) 13 AD
Thus,
AE 13 [ACE]
= · .
AD 14 [ABD]
Substituting this into the formula (11.130) for [ABE] and using the value for [ABD] in (11.128)
and simplifying, we obtain
169 169
[ABE] = [ACE] = ([ABC] − [ABE]) .
294 294
We can solve this for
84 · 169
[ABE] = .
463
Thus,
2 · [ABE] 2 · 84 · 169 5 3 · 169 · 5
BE = = · = .
h 463 56 463
Thus, p = 3 · 169 · 5 and q = 463. Note that p and q are already relatively prime. The answer
is q = 463. 2
Solution #2: Refer again to Figure 47 above. This solution is focused on leveraging the
relationships between angles and side lengths in the various triangles presented by using the
Law of Sines. For instance, noting that m∠ADC = m∠ADE and using the Law of Sines, we
see that
6 14 14 14 13 14 DB
= = = · = · .
sin α sin(∠ADC) sin(∠ADE) 13 sin(∠ADE) 13 sin(α + β)
We can derive another ratio involving sin(α +β) by a similar calculation with the Law of Sines:
CE 14 14 14 13 14 BE
= = = · = · .
sin(α + β) sin(∠AEC) sin(∠AEB) 13 sin(∠AEB) 13 sin α
If we multiply the leftmost expressions of these two equations and the rightmost expressions
of these two equations, we obtain
2
6 CE 14 DB BE
· = · .
sin α sin(α + β) 13 sin(α + β) sin α
Simplifying, we have
2
14
6(CE) = (DB)(BE),
13
so that 2 2
6(CE) 13 6(15 − BE) 13
BE = = .
DB 14 9 14
From here, it is a routine matter to solve for BE as in Solution #1. 2
1. (2002 AIME-2, Problem #2) Three vertices of a cube are P = (7, 12, 10), Q = (8, 8, 1),
and R = (11, 3, 9). What is the surface area of the cube?
Solution: Since the surface area of a cube of side length a is given by formula (9.2) to be
A = 6a2 ,
we seek to determine a. To do this, it is useful to find the distances between the three given
points in the cube, in order to determine their relative positions:
p √
P Q = (8 − 7)2 + (8 − 12)2 + (1 − 10)2 = 98,
p √
PR = (11 − 7)2 + (3 − 12)2 + (9 − 10)2 = 98,
and p √
QR = (11 − 8)2 + (3 − 8)2 + (9 − 1)2 = 98.
Therefore, the three points are all equidistant. Thus, the three points lie on opposite corners
of two square faces that share a common edge. (See Figure 48.)
FIGURE 48 GOES HERE
Hence, applying the Pythagorean Theorem to one of the faces (of side length a), we have
p √
a2 + a2 = 98.
Therefore, a2 = 49 and a = 7. Thus, the surface area of the cube is 6a2 = 6 · 49 = 294. 2
2. (1985 AIME, Problem #2) When a right triangle is rotated about one leg, the
volume of the cone produced is 800π cm3 . When the triangle is rotated about the
other leg, the volume of the cone produced is 1920π cm3 . What is the length (in
cm) of the hypotenuse of the triangle?
Solution: Suppose the legs of the right triangle have lengths b and c. When rotated about
the leg of length b, we obtain a cone with height b and radius c. Therefore, from the formula
for the volume of a cone, we obtain
1 2
πc b = 800π. (11.131)
3
Similarly, from the rotation about the leg of length c, we obtain
1 2
πb c = 1920π.
3
Hence, we have
c2 b = 2400 and b2 c = 5760.
Solving each equation for bc and equating the results, we have
2400 5760
= bc = .
c b
Hence, we can solve for b in terms of c to obtain
5760c
b= = 2.4c.
2400
Substituting this expression for b into Equation (11.131) and simplifying, we obtain
0.8c3 = 800.
Thus, c3 = 1000, so that c = 10. Hence, the hypotenuse of the triangle is

p p √ √
b2 + c2 = 242 + 102 = 576 + 100 = 676 = 26 = 026.
3. (2004 AIME-2, Problem #3) A solid rectangular block is formed by gluing to-
gether N congruent 1-cm cubes face to face. When the block is viewed so that
three of its faces are visible, exactly 231 of the 1-cm cubes cannot be seen. Find
the smallest possible value of N .
Solution: Suppose the dimensions of the rectangular block are a, b, and c. Then N = abc.
Excluding the cubes showing on the three visible faces, there is a rectangular block of hidden
cubes measuring a − 1 by b − 1 by c − 1. Therefore, we are given that
(a − 1)(b − 1)(c − 1) = 231 = 11 · 7 · 3.
No special consideration is needed to distinguish the individual roles of a, b, and c, so that we

assume without loss of generality that a ≥ b ≥ c. We can then build the following table of
possible values for a − 1, b − 1, and c − 1:
a−1 b−1 c−1 N

231 1 1 928
77 3 1 624
33 7 1 544
21 11 1 528
11 7 3 384
From this exhaustive list of possibilities, we see that the smallest possible value of N is 384. 2
4. (2012 AIME-2, Problem #5) In the accompanying figure, the outer square S has
side length 40. A second square S 0 of side length 15 is constructed inside S with
the same center as S and with sides parallel to those of S. From each midpoint
of a side of S, segments are drawn to the two closest vertices of S 0 . The result is
a four-pointed starlike figure inscribed in S. The star figure is cut out and then
folded to form a pyramid with base S 0 . Find the volume of this pyramid.
Warning: Even though the four triangles that are each attached to one of the sides of S 0
appear in the figure to be equilateral, this cannot be assumed. In fact, as we will see in the
solution below, they are not equilateral!
Solution: Of course, we will want to use the formula for the volume of a pyramid given in
Equation (9.11). The base of the pyramid is S 0 , a square of side length 15. Thus, A = 152 =
225. The bulk of the work in this problem is to find the height of the pyramid.
INSERT FIGURE!!!!!!!!!!!!!! – and draw altitude length x on one of the trian-

gles!!!!!!!!!!!!!!!!!!!!!!!!!!!
Observe that when the star figure is folded, the length x will become the hypotenuse of a right
triangle (see Figure !!!!!!) that can be used to find the height h of the pyramid. The base ` of
this triangle has half the length of the sides of S 0 , so that ` = 7.5. Thus, from the Pythagorean
Theorem, we have
7.52 + h2 = x2 .
Finally, we simply need to observe by looking at the length of S that
40 − 15
x= = 12.5.
2
Hence,
h2 = x2 − 7.52 = 12.52 − 7.52 = 156.25 − 56.25 = 100.
Therefore, h = 10, and the volume of the pyramid is
1
V = (225)(10) = 750.
3
2
5. (2003 AIME, Problem #5) Consider the set of points that are inside or within one
unit of a rectangular parallelepiped (box) that measures 3 by 4 by 5 units. Given
m + nπ
that the volume of this set is , where m, n, and p are positive integers, and
p
n and p are relatively prime, find m + n + p.
FIGURE 49 GOES HERE
Solution: We can decompose the set of points under consideration into some natural subsets:
(a) the parallelepiped itself
(b) rectangular boxes of height 1 that sit ”above” one face of the parallelepiped (there are
six of these)
(c) one-quarter cylinders of radius 1 that lie “above” each edge of the parallelepiped and
“between” two rectangular boxes (there are 12 of these)
(d) one-eighth spheres of radius 1 that lie “above” each corner of the parallelepiped and
“between” two of the one-quarter cylinders (there are eight of these)
Let us denote the contributions to the volume of the set coming from (a) – (d) by Vp , Vb , Ve ,
and Vc , respectively. Thus, we wish to compute
V = Vp + Vb + Ve + Vc .
For (a), we simply have that the volume of the rectangular parallelepiped itself is
Vp = 3 · 4 · 5 = 60.
For (b), consider the rectangular boxes of height 1 and base area equal to the area of the
face of the parallelepiped that the box sits above. The volume of such a box is therefore just
the area of the corresponding side of the parallelepiped. Hence, the contribution to the total
volume by all six of these boxes is simply the surface area of the parallelpiped:
Vb = 2(3 · 4 + 3 · 5 + 4 · 5) = 2(47) = 94.
For (c), we observe that along each of the 12 edges of the parallelepiped, there is a cylinder of
points (with base shaped like a quarter-circle with radius 1) lying between two of the one unit
extensions of the faces considered in (b). The volume of each of these cylinders is simply the
length of the corresponding edge multiplied by the area of the quarter-circle of radius 1, π/4.
There are four edges of each of the three lengths of the parallelepiped, so we have
π
Ve = (4 · 3 + 4 · 4 + 4 · 5) = 12π.
4
Finally, for (d), we account for one-eighth spheres protruding from each corner of the rectan-
gular parallelepiped. These spheres each have radius 1, and so one-eighth of such a sphere has
1 4 π
volume · π = . A total of eight of these results in an additional volume of
8 3 6
π 4π
Vc = 8 · = .
6 3
Summing the volumes found above, the total volume of the set of points in question is
4π 462 + 40π
V = Vp + Vb + Ve + Vc = 60 + 94 + 12π + = ,
3 3
so that m = 462, n = 40, and p = 3. Therefore, m + n + p = 505. 2
6. (2013 AIME, Problem #7) A rectangular box has width 12 inches, length 16
inches, and height m
n inches, where m and n are relatively prime positive integers.
Three faces of the box meet at a corner of the box. The center points of those
three faces are the vertices of a triangle with an area of 30 square inches. Find
m + n.
Solution #1: The idea behind this solution is to find the three side lengths of the triangle
m
in terms of the height h := , and then use Heron’s formula (8.16) to solve for h.
n
Figure ????? shows a picture of the box and the center points of three faces.
INSERT FIGURE !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
These points
√ form a triangle ABC, and from the Pythagorean Theorem, we can quickly find
that AB = 62 + 82 = 10,
s s
2 2
h 1 p h 1p 2
AC = + 62 = h2 + 144, BC = + 82 = h + 256.
2 2 2 2
To apply Heron’s formula, we must compute the semiperimeter of ∆ABC:
AB + AC + BC
s=
2
1 1p 2 1p 2
= 10 + h + 256 + h + 144
2 2 2
1p 2 1p 2
=5+ h + 256 + h + 144.
4 4
Then Heron’s formula (8.16) gives
2
[ABC] = s(s − AB)(s − AC)(s − BC)

1p 2 1p 2 1p 2 1p 2
= 5+ h + 256 + h + 144 −5 + h + 256 + h + 144 · · ·
4 4 4 4

1p 2 1p 2 1p 2 1p 2
··· 5 − h + 256 + h + 144 5+ h + 256 − h + 144 .
4 4 4 4
The best way to simplify the right-hand side is to multiply the first two factors together (and
simplify), and then do the same for the last two factors. The result of this work is

2 1 2 p 2 p 1 p p
[ABC] = (h + h + 256 h2 + 144) (−h2 + h2 + 256 h2 + 144)
8 8
1 4
−h + (h2 + 256)(h2 + 144)

=
64
1
400h2 + 256 · 144 .

=
64
Since [ABC]2 = 302 = 900, from this last expression we can quickly compute that h = 7.2 =
36
. Hence, m = 36 and n = 5, and m + n = 41 = 041. 2
5
Solution #2: We use the same notations for A, B, and C as in Figure ???? above. The
first solution above involved a somewhat messy calculation involving Heron’s formula. We can
avoid this if we simply use Equation (8.14) for the area of ∆ABC, treating the segment AB
as the base of the triangle. Then we have
1
30 = [ABC] = (10)h,
2
where h is the height of the triangle, measured as the distance from vertex C to its projection
along segment BC. Thus, we have h = 6.
Use the altitude from C to AB to decompose ∆ABC into two right triangles that meet at D,
the point along AB that the altitude from C meets its. Using the Pythagorean Theorem on
each of the two triangles ∆ACD and ∆BCD, we obtain
AB = AD + DB
p p
= AC 2 − 36 + BC 2 − 36
r r
h2 + 144 h2 + 256
= − 36 + − 36
4 4
r r
h2 h2
= + + 28.
4 4
Thus, r 2
h2 h2

h h
+ + 28 = 10, or 10 − = + 28.
2 4 2 4
36
A short calculation now provides the answer h = . Hence, we obtain the same answer that
5
we did in Solution #1 above. 2
7. (2000 AIME, Problem #8) A container in the shape of a right circular cone is
12 inches tall and its base has a 5-inch radius. The liquid that is sealed inside is
9 inches deep when the cone is held with its point down and its base horizontal.
When the cone is held with its point up and its base horizontal, the liquid is
√
m − n 3 p inches deep, where m, n, and p are positive integers and p is not divisible
by the cube of any prime number. Find m + n + p.
Solution: Using the formula for the volume of a right circular cone,
1 2
V = πr h,
3
where r denotes the radius of the cone and h denotes the height of the cone, we conclude that
the container has a total volume of
1
V = π · 52 · 12 = 100π.
3
When the cone is held with point down, the liquid inside the container forms another (similar)
cone with height 9, as depicted in Figure ???????????
INSERT FIGURE !!!!!!!!!!!!!!!!!!!!!!
9 3
Therefore, this cone of liquid has a radius that is = of the radius of the original cone,
12 4
3 15
which is r = · 5 = . Thus, the volume of the liquid is
4 4
2
1 15 675π
V = π· ·9= .
3 4 16
675π
Thus, the amount of space in the container that is not occupied by the liquid is 100π − =
16
925π
. When the container is held with the pointed end up, this amount of space forms a cone
16
inside the top portion of the container, again similar to the cone container itself. We now
determine the height of this cone:
2
925π 1 1 5 25π 3
= πr2 h = π h h= h .
16 3 3 12 432
Thus,
√
r
3 925 · 432 3
h= = 3 · 37.
16 · 25
√
Hence, the depth of liquid (in the occupied portion of the container) is 12 − 3 · 3 37. Hence, we
conclude that m = 12, n = 3, and p = 37. Therefore, m + n + p = 12 + 3 + 37 = 52 = 052. 2
8. (1984 AIME, Problem #9) In tetrahedron ABCD, edge AB has length 3 cm. The
area of face ABC is 15 cm2 and the area of face ABD is 12 cm2 . These two faces
meet each other at a 30◦ angle. Find the volume of the tetrahedron in cm3 .
Solution: As shown in Figure ?????, we can regard the tetrahedron as oriented to have base
ABC, so that its volume is
1
Volume = Ah = 5h,
3
where A is the area of the base and h is the height of the tetrahedron (see Equation (9.11)).
INSERT FIGURE !!!!!!!!!!!!!!!!!
Since the angle between face ABC and ABD is 30◦ , if a denotes the length of the altitude of
triangle ABD with base AB = 3, then
1
h = a sin 30◦ = a.
2
Therefore, since the area of ABD is
3a
[ABD] = 12 = ,
2
we find that a = 8. Hence, h = 4, and the volume of the tetrahedron is 5h = 20 = 020. 2
9. (2005 AIME, Problem #9) Twenty-seven unit cubes are each painted orange on
a set of four faces so that the two unpainted faces share an edge. The 27 cubes
are then randomly arranged to form a 3 × 3 × 3 cube. Given that the probability
pa
that the entire surface of the larger cube is orange is b c , where p, q, and r are
q r
distinct primes and a, b, and c are positive integers, find a + b + c + p + q + r.
Solution: We consider four different types of unit cubes: (1) unit cubes in the center of each
face of the larger cube, (2) unit cubes lying along the center of an edge between two faces of
the larger cube, (3) unit cubes lying in the corner of the larger cube, and (4) the one unit cube
that lies hidden in the center of the 3 × 3 × 3 cube.
There are six faces, and hence six unit cubes in category (1). For each of these unit cubes,
4 2
there is a = probability that an orange face is showing on the exterior of the 3 × 3 × 3
6 3
cube.
There are 12 edges on a cube, and hence 12 unit cubes in category (2). For each of these unit
cubes, consider the edge that is the exterior edge of the larger 3 × 3 × 3 cube. Of the 12 edges
in the unit cube, seven of them are adjacent to an unpainted face. Therefore, the probability
that both exposed faces of the unit cube are painted is 5/12.
Next, there are eight corners in the large cube, and therefore eight unit cubes in category
(3). In order for all visible sides of such a unit cube to be painted, the edge between the
unpainted sides must be placed in one of the three positions in the back of the cube where
neither adjoining face is exposed. Therefore, the probability that a corner unit cube is placed
3 1
so that its visible faces are all painted is = .
12 4
Finally, note that the one unit cube in category (4) can be placed in any arbitrary configuration
with no danger of exposing an unpainted face.
The configuration of any given unit cube is independent of the configuration chosen for all
other unit cubes, so by part 5 of Theorem 3.2.1, we simply multiply the probabilities obtained
above for each of unit cubes:
6 12 8
2 5 1 26 · 512 512
= 6 12 8
= 34 18 .
3 12 4 3 · 12 · 4 2 ·3
Thus, we have
p + q + r + a + b + c = 5 + 2 + 3 + 12 + 34 + 18 = 74 = 074.
2
10. (2010 AIME, Problem #11) Let R be the region consisting of the set of points in
the coordinate plane that satisfy both |8 − x| + y ≤ 10 and 3y − x ≥ 15. When R is
revolved around the line whose equation is 3y − x = 15, the volume of the resulting
mπ
solid is √ , where m, n, and p are positive integers, m and n are relatively prime,
n p
and p is not divisible by the square of any prime. Find m + n + p.
Solution: Let us examine the (unbounded region of the coordinate plane that satisfies |8 −
x| + y ≤ 10. The boundary of this region is determined from the equation |8 − x| + y = 10.
For x ≤ 8, this equation simplifies to 8 − x + y = 10, or y = x + 2. For x ≥ 8, the boundary
of the region is defined by y = 18 − x. The two lines
y =x+2 and y = 18 − x
intersect at the point (8, 10). Let us compute the intersection points of these boundary lines
with the line 3y − x = 15, or
1
y = x + 5. (11.132)
3
Plugging
y= x + 2 into Equation (11.132), we can quickly compute that the intersection point
9 13
is , . On the other hand, plugging y = 18 − x into Equation (11.132), we obtain the
2 2
39 33
intersection point , .
4 4
Once the region R, drawn in Figure 50, is revolved around the line in Equation (11.132), it
creates two right circular cones, say of heights h1 and h2 , that share a base whose radius r
is determined by the distance from the point (8, 10) to the line whose equation is given in
Equation (11.132).
FIGURE 50 GOES HERE
The volume of the solid in question is therefore

1 2 1 1
V = πr h1 + πr2 h2 = πr2 h,
3 3 3
where
h = h1 + h2

9 13 39 33
is the distance between the points , and , :
2 2 4 4
s 2 2 √ √
21 7 441 + 49 490 7√
h= + = = = 10.
4 4 4 4 4
To find the radius r for the cone, we can use the formula for the distance from a point to a
line, or we can derive it as follows: The line passing through (8, 10) and perpendicular to the
line in Equation (11.132) has equation
y − 10 = −3(x − 8).
That is,
y = −3x + 34.
Equating this to (11.132), we have
x
5+ = −3x + 34,
3
so that
10
x = 29.
3
We conclude that x = 8.7, and then y = 7.9. Thus, applying the distance formula (8.3) to the
points (8, 10) and (8.7, 7.9), we find that the radius is
p √ √ 7
r = (0.7)2 + (2.1)2 = .49 + 4.41 = 4.9 = √ .
10
Hence, the volume of the solid in question is

2 √ !
1 1 7 7 10 343π
Volume = πr2 h = π √ = √ .
3 3 10 4 12 10
Hence, m = 343, n = 12, and p = 10. Therefore, the answer is m + n + p = 343 + 12 + 10 = 365.
2
11. 1998 AIME, Problem #10) Eight spheres of radius 100 are placed on a flat surface
so that each sphere is tangent to two others and their centers are the vertices of a
regular octagon. A ninth sphere is placed on the flat surface so that it is √ tangent
to each of the other eight spheres. The radius of this last sphere is a + b c, where
a, b, and c are positive integers, and c is not divisible by the square of any prime.
Find a + b + c.
Solution: Since the first eight spheres are mutually tangent, the distance from the center of
one of those spheres to the center of an adjacent one is exactly twice the radius, or 200. Thus,
the centers of the first eight spheres form a regular octagon with side length 200. As Figure
51 shows, the triangle formed by the center of the octagon, O, and two of the adjacent centers
of the spheres is isosceles with base 200 and the angle 45◦ at O.
FIGURE 51 GOES HERE
The length of each of the other two sides of the isosceles triangle, which we denote by x, can
be computed from the Law of Cosines:
√
2002 = 2x2 − 2x2 cos 45◦ = (2 − 2)x2 .
Thus,
200
x= p √ .
2− 2
If we denote the radius of the ninth sphere by r, observe from Figure 51 that
(r + 100)2 = x2 + y 2 = x2 + (r − 100)2 .
Thus,
r2 + 200r + 10000 = x2 + r2 − 200r + 10000.
Simplifying this, we have
400r = x2 ,
so that
x2 1 40000 100 √ √
r= = · √ = √ = 50(2 + 2) = 100 + 50 2.
400 400 2 − 2 2− 2
Hence, a = 100, b = 50, and c = 2, so that the answer is a + b + c = 152. 2
12. 2004
√ AIME-2, Problem #11) A right circular cone has a radius 600 and height
200 7. A fly starts at a point on the surface of the cone whose distance from the
vertex of the cone is 125, and crawls along the surface of the cone to a √
point on
the exact opposite side of the cone whose distance from the vertex is 375 2. Find
the least distance that the fly could have crawled.
Solution: Let A denote the point where the fly starts, and let B be the point on the opposite
side of the cone where the fly reaches. The trick is to cut the cone along a straight line from
the vertex O of the cone to the base passing through the point A, and then unfold the surface
of the cone into a sector of a circle C, as shown in Figure 52.
FIGURE 52 GOES HERE
The radius of C is precisely the slant height of the cone, which can be computed using the
Pythagorean Theorem:
q √ √ √
s = 6002 + (200 7)2 = 360000 + 280000 = 640000 = 800.
The circumference of C is therefore 2π(800) = 1600π, whereas the portion of the sector created
by the cone has circumference equal to the circumference of the base of the cone, or 2π(600) =
◦
1200π. Thus, the sector occupies exactly 1200π 3
1600π = 4 of the circle, and therefore spans 270 .
◦
Since A and B are on opposite sides of the cone, and A is on one edge of the given 270 sector,
we see from Figure 52 that B is on the sector such that the triangle formed √ by A, O, and B
makes an angle of 135◦ at O. We are given that OA = 125 and OB = 375 2, so we can use
the Law of Cosines to find AB, which is the least distance the fly could have crawled:
p
AB = (OA)2 + (OB)2 − 2(OA)(OB) cos 135◦
q √ √ √
= 1252 + (375 2)2 − 2 · 125 · 375 2(− 2/2)
√
= 125 1 + 18 + 6 = 625.
2
13. (2005 AIME-2, Problem #10) Given that O is a regular octahedron, that C is
the cube whose vertices are the centers of the faces of O, and that the ratio of
the volume of O to that of C is m/n, where m and n are relatively prime positive
Solution: Let us assume that each side of the octahedron O has length s We begin by
computing the volume of O, which consists of two pyramids that meet in a square base in the
center. (See Figure ????)
INSERT FIGURE !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

The area of the base of each pyramid is s2 , so we only need to determine the height of each
pyramid. To do this, form a right triangle whose vertices consist of the top of the pyramid,
the center of its base, and one corner of the square base. The hypotenuse of this triangle is the
s
length s, and the distance between the two vertices on the base of the pyramid is √ . Thus,
2
the height of the pyramid is the remaining leg of the right triangle, and it has length
r
s2 s
s2 − =√ .
2 2
Now we can use the formula for the volume of a pyramid:
s3

1 1 s
Ah = s2 √ = √ .
3 3 2 3 2
√
2 3
We double this to obtain the volume of O: s .
3
Next, we must compute the volume of the cube C. For this, it suffices to find the distance
between two adjacent vertices. From symmetry, observe that the corners of the cube must
touch the octahedron precisely in the center of the equilateral triangles forming the sides of
the octahedron. The key observation is that this touching point occurs along the central axis
of each equilateral triangle, one-third of the way up from the bottom. When this equilateral
triangle is folded towards the meeting point of the four triangles at the top of the pyramid, the
coordinates of this point relative to the square base move one-third of the way from the edge of
the square base to the center of the square base. Figure ???? shows the location (with respect
to the square base) of the four center points of the equilateral triangles above the square base.
We see that the distance between two adjacent points will be
s
2 2 √
1 1 2
s + s = s.
3 3 3
Hence, the volume of C is

√ !3 √
2 2 2 3
s = s .
3 27
Thus, the ratio of the volume of O to the volume of C is
√
2 3
3 s 9
√ = .
2 2 3 2
27 s

About the Author

Scott A. Annin received his B.S. in mathematics and physics from the University of Nebraska in
1995. He earned his Ph.D. in mathematics from the University of California at Berkeley in 2002,
specializing in noncommutative algebra. He has been a professor at California State University in
Fullerton ever since. Annin has received several teaching awards at both Berkeley and Fullerton,
including college and university-wide honors at Fullerton in 2007 and 2008, respectively. In 2009,
he received the MAA’s prestigious Henry L. Alder Award for Outstanding Teaching by a Beginning
Faculty Member in Mathematics. In addition to being a former AIME contestant himself (while
a student at Lincoln East High School in Lincoln, Nebraska), Annin has led workshops over the
years for countless many AIME contestants. He has also increased awareness of the AMC program
in general among high school teachers at both the state and national levels at math education
and MAA meetings. Annin has also co-authored the widely used undergraduate mathematics text
“Differential Equations and Linear Algebra” (3rd edition) with Stephen Goode.

A Gentle Introduction To The American Invitational Mathematics Examination

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Gentle Introduction To The American Invitational Mathematics Examination

Uploaded by

Copyright:

Available Formats

A Gentle Introduction to the American Invitational

March 16, 2014

3.3 Examples: Tournaments, Socks, and Dice . . . . . . . . . . . . . . . . . . . . . . . . 49

5 Sequences and Series 75

6 Logarithmic and Trigonometric Functions 97

7 Complex Numbers and Polynomials 121

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

8 Plane Geometry 149

9 Spatial Geometry 183

10 Hints for the Exercises 193

10.3.1 Hints to Get Started . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

11 Solutions to Exercise Sets 217

A Gentle Approach to the American Invitational Mathematics Examination is a celebration of math-

The American Mathematics Competitions

younger. That exam is not the subject of this volume.

An Integer Answer from 000 to 999

dxe denotes the smallest integer greater than or equal to x

The Structure of this Book

Advice for Success on the AIME

who love mathematics or want to grow to love mathematics.

A Word from the Author

130 − x = 164 − x − r = 188 − x − w.

liters of acid to jar A, and it adds

1.2 Distance-Rate-Time Problems

Distance = Rate × Time . (1.8)

2b + 3j + 4s = 74 and 4b + 2j + 3s = 91. (1.9)

Eliminating fractions, we obtain

Amount of Product Manufactured = Productivity Rate × Time . (1.14)

The system of equations

1.3 Systems of Equations

Expanding the left side, we obtain eight terms:

Find the value of

16x1 + 25x2 + 36x3 + 49x4 + 64x5 + 81x6 + 100x7 . (1.19)

equations in the system and adding the resulting equations together.

an2 + b(n + 1)2 + c(n + 2)2 = (n + 3)2 .

Expanding this equation, we obtain

(a + b + c)n2 + (2b + 4c)n + (b + 4c) = n2 + 6n + 9.

Equating the coefficients on both sides of the equation, we find that

Hints begin on Page 194. Solutions begin on Page 217.

In the newspaper story covering the event, it was reported that

• the winner caught 15 fish;

“Mathematics may be defined as the economy of counting. There is no problem in the

2.2.1 Permutations with Repetition

The number of ways to make an ordered arrangement of k objects

The visualization of this principle is seen below.

|n choices | |n choices | |n choices| |n choices|

Theorem 2.2.1. (The Multiplication Principle) If a task has independent subtasks T1 , T2 , . . . ,

2.2.2 Permutations without Repetition

|n choices | |n − 1 choices | |n − k + 2 choices| |n − k + 1 choices|

Solution: Label the five men M1 , M2 , M3 , M4 , and M5 , with spouses (respectively) W1 , W2 , W3 ,

2.3.1 Combinations without Repetition

2.3.2 Combinations with Repetition

A selection of six of each fruit, on the other hand, would appear as

while a selection of 18 apples would appear as

Example 2.3.5. How many quadruples (x, y, z, w) of nonnegative integers satisfy

2.4 More Challenging Combinatorics Problems

Subtask # Description # of Ways

Subtask # Description # of Ways

Therefore, the leftmost three nonzero digits of n are 376.

Case 2: We assume

Case 3: We assume

x01 + x02 + · · · + x0g+1 = b − g + 1,

with the condition that

Case 2: We assume

Case 3: We assume