IMO Problems, Theorems,

and Methods
Algebra
Mathematical Olympiad Series
ISSN: 1793-8570

Series Editors: Lee Peng Yee (Nanyang Technological University, Singapore)

Xiong Bin (East China Normal University, China)

Published

Vol. 25 IMO Problems, Theorems, and Methods: Algebra

by Jinhua Chen (East China Normal University, China) &
Bin Xiong (East China Normal University, China)

Vol. 24 Leningrad Mathematical Olympiads (1961–1991)

by Dmitri Fomin

Vol. 23 Solving Problems in Point Geometry:

Insights and Strategies for Mathematical Olympiad and Competitions
by Jingzhong Zhang (Guangzhou University, China &
Chinese Academy of Sciences, China) &
Xicheng Peng (Central China Normal University, China)

Vol. 22 Mathematical Olympiad in China (2021–2022):

Problems and Solutions
editor-in-chief Bin Xiong (East China Normal University, China)

Vol. 21 Problem Solving Methods and Strategies in High School

Mathematical Competitions
edited by Bin Xiong (East China Normal University, China) &
Yijie He (East China Normal University, China)

Vol. 20 Hungarian Mathematical Olympiad (1964–1997):

Problems and Solutions
by Fusheng Leng (Academia Sinica, China),
Xin Li (Babeltime Inc., USA) &
Huawei Zhu (Shenzhen Middle School, China)

Vol. 19 Mathematical Olympiad in China (2019–2020):

Problems and Solutions
edited by Bin Xiong (East China Normal University, China)

Vol. 18 Mathematical Olympiad in China (2017–2018):

Problems and Solutions
edited by Bin Xiong (East China Normal University, China)

The complete list of the published volumes in the series can be found at
http://www.worldscientific.com/series/mos
 Vol. 25 Mathematical
Olympiad
Series

IMO Problems, Theorems,

and Methods
Algebra

Authors
Jinhua Chen
Bin Xiong
East China Normal University, China

Proofreader
Jiu Ding
School of Mathematics and Natural Sciences,
University of Southern Mississippi, USA

Copy Editors
Lingzhi Kong, Liyu Zhang, and Ming Ni
East China Normal University Press, China

East China Normal

University Press World Scientific
Published by
East China Normal University Press
3663 North Zhongshan Road
Shanghai 200062
China

and

World Scientific Publishing Co. Pte. Ltd.

5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

Library of Congress Control Number: 2025004976

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.

Mathematical Olympiad Series — Vol. 25

IMO PROBLEMS, THEOREMS, AND METHODS
Algebra
Copyright © 2026 by East China Normal University Press and
World Scientific Publishing Co. Pte. Ltd.
All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance
Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy
is not required from the publisher.

ISBN 978-981-98-0327-9 (hardcover)

ISBN 978-981-98-0688-1 (paperback)
ISBN 978-981-98-0328-6 (ebook for institutions)
ISBN 978-981-98-0329-3 (ebook for individuals)

For any available supplementary material, please visit

https://www.worldscientific.com/worldscibooks/10.1142/14098#t=suppl

Desk Editors: Nambirajan Karuppiah/Angeline Husni

Typeset by Stallion Press

Email: [email protected]

Printed in Singapore
Preface

It is generally believed that formal mathematics competitions began with

a contest held in Hungary in 1894, an event that gradually garnered atten-
tion worldwide. People aptly liken mathematics competitions to “Mental
Gymnastics.” In 1934, the Soviet Union straightforwardly termed it the
“Mathematical Olympiad,” a designation that reflects the Olympic spirit
of pursuing excellence in intellect more vividly than the previous term,
“mathematics competition.”
By 1959, the internationalization of mathematics competitions had
matured, leading to the inception of the “International Mathematical
Olympiad” (IMO). The first IMO was held in Brasov, Romania in 1959.
As of 2023, the IMO has successfully been held 64 times, except for 1980
when it was not conducted.
The IMO is typically held in July each year, and the format has become
standardized: the official competition spans two days, with contestants
tackling three problems in 4.5 hours each day, each problem worth a max-
imum of 7 points, totaling 42 points. Each participating team consists of
six contestants, accompanied by a Leader and a Deputy Leader. Approxi-
mately half of the contestants receive medals, with about 1/12 of the contes-
tants earning gold medals, 2/12 receiving silver medals, and 3/12 obtaining
bronze medals.
The IMO is currently one of the most influential secondary school math-
ematics competitions worldwide. In recent years, over 100 countries and
regions have participated in this event, including all major nations globally.
Problems for the IMO are submitted by the participating teams and
then reviewed and selected by a problem selection committee organized by

v
vi IMO Problems, Theorems, and Methods: Algebra

the host country. This committee narrows down the submissions to approx-
imately 30 Shortlist problems, covering algebra, geometry, number theory,
and combinatorics, with about seven to eight problems on each topic. These
are then presented to the Jury Meeting, composed of team leaders, who dis-
cuss and vote to decide on the six problems that will constitute the official
competition paper. The host country does not provide any problems.
This event has played a significant role in promoting the exchange of
mathematical education among nations, enhancing the level of mathemati-
cal education, facilitating mutual learning and understanding among young
students worldwide, stimulating a broad interest in mathematics among
secondary school students, and identifying and nurturing mathematically
gifted students.
The development over more than 60 years is the result of the collective
efforts of mathematicians, organizers, and contestants, and is worthy of
reflection and study. Particularly deserving of study are the evolution of
the competition problems, the mathematical ideas, and methods involved.
Indeed, several colleagues from the International Mathematical Olympiad
Research Center at East China Normal University had envisioned research
and publication before the 60th IMO. For this purpose, we initiated sev-
eral seminars involving over ten people. For special reasons, this work was
delayed. Based on the mathematical domains covered by the IMO prob-
lems — algebra, geometry, number theory, and combinatorics — we planned
to compile the work into four volumes, with the general title IMO Problems,
Theorems, and Methods, to be included in the “IMO Study Series.”
Each volume begins with an introduction that provides an overview of
the IMO. Subsequent chapters introduce relevant foundational knowledge
and methods, followed by a reclassification and organization of past IMO
problems. For some problems, multiple solutions are provided, along with
a difficulty analysis. It is worth noting that some problems do not fit neatly
into a single topic, as they may involve both algebra and number theory,
or algebra and combinatorics. We primarily categorize them based on the
topic under which they were placed on the Shortlist.
The four volumes titled IMO Problems, Theorems, and Methods were
conceived with an overall writing plan proposed by myself, with the authors
collectively discussing and refining the plan. The majority of the initial
drafts were completed by Jinhua Chen (Algebra), Tianqi Lin (Geome-
try), Gengyu Zhang (Number Theory), and Guangyu Xu (Combinatorics).
The first three volumes were supplemented, consolidated, and finalized by
 Preface vii

myself, while the combinatorics volume was supplemented, consolidated,

and finalized by Zhenhua Qu.
We extend our gratitude to the leaders and contestants of the Chinese
IMO teams over the years, as some elegant solutions included in the book
were contributed by them. During the compilation of this book, we con-
sulted various domestic and international sources, which are too numerous
to acknowledge individually here.
While the authors have diligently studied the IMO problems and pro-
vided thoughtful strategies and solutions, errors and inaccuracies may occur
due to our limitations. We sincerely invite readers to offer corrections and
feedback.
The translation of the algebra volume in this series was done by Jinhua
Chen and Bin Xiong; the geometry volume was translated by Xinyuan
Yang; the number theory volume was translated by Gengyu Zhang; and
the combinatorics volume was translated by Zhenhua Qu and Jinhua Chen.
Jiu Ding revised the translations of all four books.

Bin Xiong
June 2024
This page intentionally left blank
About the Authors

Jinhua Chen is a doctoral candidate in Mathematical Education at East

China Normal University, with research interests in mathematics compe-
titions and gifted education. He serves as an external mentor for math-
ematics competition courses at several high schools. Chen attended the
Affiliated High School of South China Normal University and Guangdong
Olympic School during his secondary education. He has excelled in various
competitions, including the Chinese High School Mathematics League, the
American Mathematics Competition 12, the Chinese College Mathematics
Competition, and the COMAP Mathematical/Interdisciplinary Contest in
Modeling.
Bin Xiong is a professor and doctoral supervisor at the School of Math-
ematical Sciences, East China Normal University. He also serves as the
director of the Shanghai Key Laboratory of Core Mathematics and Prac-
tice, and the International Mathematical Olympiad Research Center. Pro-
fessor Xiong is an expert with the State Council Special Allowance, and has
been honored with the Shanghai May 1 Labor Medal as well as the Shang-
hai Model of Teaching and Educating. He has published over 100 scholarly
papers in renowned national and international journals and has authored or
co-authored more than 150 books. Additionally, Professor Xiong has served
as the leader and head coach of the Chinese IMO team more than 10 times,
and received the prestigious Paul Erdös Award in 2018 for his contribu-
tion to the development of mathematics competitions at the national and
international level.

ix
This page intentionally left blank
Contents

Preface v

About the Authors ix

Introduction to the IMO 1
1 Evolution of the IMO . . . . . . . . . . . . . . . . . . . . . . 3
2 Problems in the IMO . . . . . . . . . . . . . . . . . . . . . . 4
2.1 The number of problems . . . . . . . . . . . . . . . . 5
2.2 The difficulty level of problems . . . . . . . . . . . . 6
2.3 The classification of problems . . . . . . . . . . . . . 10
2.4 The proposal for problems . . . . . . . . . . . . . . . 14
3 Awards in the IMO . . . . . . . . . . . . . . . . . . . . . . . 15
3.1 Participation . . . . . . . . . . . . . . . . . . . . . . 16
3.2 Gold medals . . . . . . . . . . . . . . . . . . . . . . . 17
3.3 Special prizes . . . . . . . . . . . . . . . . . . . . . . 18
4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Chapter 1 Equation Problems 23

1.1 Common Theorems, Formulas, and Methods . . . . . . . . . 26
1.1.1 Common equations . . . . . . . . . . . . . . . . . . . 26
1.1.2 Common methods for solving equations . . . . . . . 28
1.1.3 Other important theorems . . . . . . . . . . . . . . . 30
1.2 Problems and Solutions . . . . . . . . . . . . . . . . . . . . 31
1.2.1 Finding solutions (without parameters) . . . . . . . . 31
1.2.2 Finding solutions (with parameters) . . . . . . . . . 38

xi
xii IMO Problems, Theorems, and Methods: Algebra

1.2.3 Proving relationships . . . . . . . . . . . . . . . . . . 43

1.2.4 Investigating conditions . . . . . . . . . . . . . . . . 53
1.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Chapter 2 Function Problems 67

2.1 Common Theorems, Formulas, and Methods . . . . . . . . . 70
2.1.1 Common elementary functions . . . . . . . . . . . . . 70
2.1.2 Basic properties of functions . . . . . . . . . . . . . . 73
2.1.3 Common function iterations . . . . . . . . . . . . . . 77
2.1.4 Common methods for determining function
iterations . . . . . . . . . . . . . . . . . . . . . . . . 78
2.1.5 Common functional equations . . . . . . . . . . . . . 82
2.1.6 Common methods for solving functional
equations . . . . . . . . . . . . . . . . . . . . . . . . 84
2.1.7 Other important theorems . . . . . . . . . . . . . . . 91
2.2 Problems and Solutions . . . . . . . . . . . . . . . . . . . . 92
2.2.1 Proving properties . . . . . . . . . . . . . . . . . . . 92
2.2.2 Determining values . . . . . . . . . . . . . . . . . . . 111
2.2.3 Deriving expressions . . . . . . . . . . . . . . . . . . 120
2.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

Chapter 3 Sequence Problems 155

3.1 Common Theorems, Formulas, and Methods . . . . . . . . . 158
3.1.1 Common sequences . . . . . . . . . . . . . . . . . . . 158
3.1.2 Common methods for finding general term formulas . 165
3.1.3 Common methods for summations of sequences . . . 168
3.1.4 Mathematical induction . . . . . . . . . . . . . . . . 169
3.1.5 Other important theorems . . . . . . . . . . . . . . . 170
3.2 Problems and Solutions . . . . . . . . . . . . . . . . . . . . 171
3.2.1 Determining values . . . . . . . . . . . . . . . . . . . 171
3.2.2 Existence problems . . . . . . . . . . . . . . . . . . . 180
3.2.3 Proving quantitative relationships . . . . . . . . . . . 194
3.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

Chapter 4 Inequality Problems 209

4.1 Common Theorems, Formulas, and Methods . . . . . . . . . 212
4.1.1 Solving inequalities . . . . . . . . . . . . . . . . . . . 212
 Contents xiii

4.1.2Mean inequalities and the Cauchy–Schwarz

inequality . . . . . . . . . . . . . . . . . . . . . . . . 213
4.1.3 Other famous inequalities . . . . . . . . . . . . . . . 219
4.1.4 Common methods for proving inequalities . . . . . . 225
4.2 Problems and Solutions . . . . . . . . . . . . . . . . . . . . 237
4.2.1 Solving inequalities . . . . . . . . . . . . . . . . . . . 237
4.2.2 Proving inequalities . . . . . . . . . . . . . . . . . . . 244
4.2.3 Determining value ranges . . . . . . . . . . . . . . . 304
4.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312

Chapter 5 Other Algebra Problems 317

5.1 Common Theorems, Formulas, and Methods . . . . . . . . . 319
5.1.1 Common trigonometric identities . . . . . . . . . . . 319
5.1.2 Common methods for proving trigonometric
identities . . . . . . . . . . . . . . . . . . . . . . . . . 320
5.1.3 Common polynomials . . . . . . . . . . . . . . . . . . 324
5.1.4 Common techniques for polynomials . . . . . . . . . 331
5.1.5 Other important theorems . . . . . . . . . . . . . . . 335
5.2 Problems and Solutions . . . . . . . . . . . . . . . . . . . . 336
5.2.1 Proving trigonometric identities . . . . . . . . . . . . 336
5.2.2 Finding polynomials . . . . . . . . . . . . . . . . . . 341
5.2.3 Proving properties . . . . . . . . . . . . . . . . . . . 344
5.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357

Appendix A IMO General Information 361

Appendix B IMO Algebra Problem Index 365

This page intentionally left blank
Introduction to the IMO

The International Mathematical Olympiad (IMO), established in the year

1959, represents one of the foremost intellectual endeavors at the highest
tier for youth on a global scale. Prior to 1959, numerous countries around
the world had already initiated the organization of mathematics competi-
tions, thereby laying the groundwork for the inception of the IMO.
In 1891, the renowned physicist and President of the Hungarian
Academy of Sciences, Loránd Eötvös (also known as Roland Eötvös),
founded the Hungarian Mathematical and Physical Society. In 1894, he
assumed the position of Minister of Education, and under his enthusiastic
support, the Hungarian Mathematical and Physical Society initiated sec-
ondary school mathematics competitions. This competition, also known as
the Eötvös Competition, offered winners the Eötvös Prize and the oppor-
tunity to pursue higher education. Subsequently, the Eötvös Competition
was not held during 1919–1921 and 1944–1946 due to the world political
events. In 1947, under the leadership of János Surányi, the Eötvös Com-
petition was reinstated and renamed the Kürschák Competition (named
after József Kürschák). This competition has played a significant role in
Hungary in nurturing numerous mathematicians and scientists, including
Győző Zemplén, Lipót Fejér, Theodore von Kármán, Alfréd Haar, Dénes
Kőnig, Marcel Riesz, Gábor Szegő, Tibor Radó, Edward Teller, and Tibor
Szele. Interestingly, George Pólya also participated in the competition, but
did not hand in his paper.
With the aim of identifying and nurturing mathematical talents, promi-
nent mathematicians such as Boris Delaunay (also known as Delone), Grig-
orii Fikhtengol’ts, Dmitry Faddeev, and others organized the inaugural
Leningrad Mathematical Olympiad (LMO) in 1934, under the initiative of

1
2 IMO Problems, Theorems, and Methods: Algebra

Boris Delaunay. Winners of this competition were granted the privilege

of direct admission to the Mathematics Department of Leningrad State
University without the need for entrance examinations. Following the
example set by the LMO, in 1935, renowned mathematicians Pavel Alek-
sandrov and Andrey Kolmogorov, alongside the entire faculty of the Mathe-
matics Department at Moscow State University, organized the first Moscow
Mathematical Olympiad (MMO).
Subsequently, various regions throughout the Soviet Union started host-
ing their own Mathematical Olympiads, ultimately laying the foundation
for the All-Russian Mathematical Olympiad, which was first conducted in
1961. In 1967, the responsibility for organizing All-Russian Mathematical
Olympiad was assumed by the Ministry of Education of the Soviet Union,
leading to a renaming of the All-Russian Mathematical Olympiad as the
All-Soviet-Union Mathematical Olympiad.
In fact, almost all of the best mathematicians born in the Soviet Union
after 1930 had participated in Mathematical Olympiads, usually achieving
first prizes. This distinguished group includes Fields Medal awardees such
as Sergei Novikov, Grigory Margulis, Vladimir Drinfeld, Maxim Kontsevich,
Grigori Perelman, and Stanislav Smirnov. Although having claimed that
he was never particularly interested in Mathematical Olympiads, Sergei
Novikov did secure a second prize in the MMO when he was in eighth
grade.
While the United States of America Mathematical Olympiad (USAMO)
was first held in 1972, the United States had a longstanding tradition
of organizing mathematics competitions prior to that. In 1921, William
Lowell Putnam published an article in the Harvard Graduates’ Maga-
zine, proposing the idea of conducting a university-level mathematics team
competition. Following his passing, the Putnam family established the
William Lowell Putnam Intercollegiate Memorial Fund to support the
organization of the William Lowell Putnam Mathematical Competition
(Putnam Competition), administered by the Mathematical Association of
America.
With the assistance of George David Birkhoff, the first Putnam Com-
petition took place in 1938, and it has been held annually since then; the
top five ranking participants are designated as Putnam Fellows. Due to
wartime conditions, the competition was not held from 1943 to 1945. In
1946, George Pólya, Tibor Radó, and Irving Kaplansky (Putnam Fellow in
1938) formed the Putnam Competition Committee, thus reestablishing the
competition, but the responsibility for administration was undertaken by
 Introduction to the IMO 3

Garrett Birkhoff, the son of George David Birkhoff, and his colleagues in
the Harvard University Department of Mathematics.
Many participants in the Putnam Competition have gone on to become
prominent mathematicians and scientists. John Milnor, David Mumford,
Daniel Quillen, Paul Cohen, John G. Thompson, and Manjul Bhargava
have been recipients of Fields Medal. Richard Feynman, Kenneth Ged-
des Wilson, Steven Weinberg, and Murray Gell-Mann have received Nobel
Prize in Physics, while John Nash was awarded the Nobel Prize in Eco-
nomic Sciences. Additionally, numerous Putnam Fellows have been elected
as members of the National Academy of Sciences in the United States.
Building upon the foundation of existing mathematics competitions in
many countries, particularly in Eastern European nations, Romania pro-
posed in 1956 the organization of an international mathematics competition
involving seven Eastern European countries. This proposal led to the inau-
gural IMO held in 1959.
The first IMO was held in Braşov, Romania, in 1959. As of 2023, the
IMO has been successfully held 64 times, with the exception of the year 1980
when it was not held for specific reasons. Apart from the 61st IMO, which
was postponed to September in 2020 due to the impact of the COVID-19
pandemic, the IMO typically takes place in July each year.
The IMO has emerged as the most influential secondary school math-
ematics competition at present. In recent years, the number of countries
and regions participating in this event has exceeded 100.

1 Evolution of the IMO

The first IMO, held in 1959, saw the participation of only 52 contestants
from seven countries, including Romania, Hungary, Czechoslovakia, Bul-
garia, Poland, the Soviet Union, and the German Democratic Republic.
Subsequently, new countries and regions gradually joined this prestigious
event. By the 20th IMO, also hosted by Romania in 1978, approximately
20 countries and regions participated (with 21 in the 19th, 17 in the 20th,
and 23 in the 21st). The number of participating contestants also reached
132. The historical participation trends are depicted in Figure 1.
As the influence of the IMO continued to expand, the number of par-
ticipating countries and regions, as well as the number of contestants, grew
rapidly. The most significant increase in the number of participating coun-
tries and regions occurred in the 34th IMO, which was held in Turkey in
1993. In comparison to the 33rd IMO held in Russia in 1992, the number
4 IMO Problems, Theorems, and Methods: Algebra

700
Number of the Participating Countries and Regions
600
Number of Contestants
500

400

300

200

100

0
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64

Figure 1 Numbers of Participating Countries and Regions, as Well as the

Contestants, in the First 64 IMOs

increased by 17 countries and regions, reaching a total of 73, with 413 con-
testants. By the 40th IMO, which was still hosted by Romania in 1999,
the number of participating countries and regions had reached 81, with 450
contestants.
The first instance of the number of participating countries and regions
surpassing one hundred occurred during the 50th IMO, held in Germany in
2009, with a total of 565 contestants. Among the first 64 IMOs, the biggest
number of participating countries and regions, as well as the largest number
of contestants, was observed in the 60th IMO, hosted by the United King-
dom in 2019, where 621 contestants from 112 countries and regions took
part. In the 64th IMO held in Japan in 2023, there were 618 contestants
from 112 countries and regions.
As evident from Appendix A, the IMO is primarily hosted by European
countries. Moreover, as the number of participating countries and regions
in the IMO has increased, it is no longer confined to the seven founding
member countries, and many new participating countries and regions have
also begun to organize the IMO.

2 Problems in the IMO

The IMO is scheduled to take place annually in July. Each participating
country or region officially sends a delegation consisting of six contestants,
along with one team leader and one deputy leader. The official competition
 Introduction to the IMO 5

spans two days, with each day featuring three problems to be solved within
a four-and-a-half-hour timeframe. Each problem carries a maximum score
of 7 points, resulting in a total maximum score of 42 points, while the total
maximum score of the team is 252 points.
In early IMOs, the number of problems and their individual point values
varied from one session to another. For instance, the 2nd and 4th IMOs
featured seven problems, while all other IMOs had six problems each. Addi-
tionally, in the 13th IMO, although the total score remained at 42 points,
the six problems were allocated point values of 5, 7, 9, 6, 7, and 8, respec-
tively. It was only from the 22nd IMO, held in the United States in 1981,
that the IMO problems have become standardized, with each problem car-
rying 7 points and a total of six problems.
The number of contestants in each delegation has also become stable at
six individuals starting from the 24th IMO held in France in 1983.

2.1 The number of problems

The mathematical domains covered by IMO problems encompass four
major topics: algebra, combinatorics, geometry, and number theory. These
are also the primary focus in various national mathematics competitions.
Across the 1st–64th IMOs, a total of 386 problems have been fea-
tured. Among them, geometry problems are the most numerous, with 123
problems, while number theory problems are the least, with 75 problems.
Algebra problems account for approximately one-quarter of the total, com-
prising 101 problems. Furthermore, as indicated in Table 1, the quantity of
algebra problems has remained relatively stable over each span of 10 IMOs.

Table 1 Numbers of Problems with Different Topics in the First 64 IMOs

Topic

Session Algebra Combinatorics Geometry Number Theory

1–10 20 6 29 7
11–20 20 12 18 10
21–30 14 16 18 12
31–40 13 16 15 16
41–50 15 11 20 14
51–60 13 19 18 11
61–64 6 7 6 5
Total 101 87 123 75
6 IMO Problems, Theorems, and Methods: Algebra

Remarkably, in the first 64 IMOs, there were five sessions when three
algebra problems were presented, specifically in the 3rd, 5th, 7th, 15th,
and 18th IMOs. In 29 IMOs, two algebra problems were featured, while
in 28 IMOs, only one algebra problem was included. In two IMOs, algebra
problem was absent, namely the 30th and 39th IMOs.

2.2 The difficulty level of problems

Typically, the difficulty of a problem is correlated with its problem number
in the IMO.
Starting from the 24th IMO, the point value of each problem and the
number of contestants per team have become standardized. Therefore, an
analysis of the average scores of the 246 problems in the 24th–64th IMOs is
presented in Table 2. It can be observed that the first and fourth problems
in each IMO are relatively easy, with average scores generally exceeding
3 points. The second and fifth problems are relatively challenging, with
average scores mainly ranging from 1 to 4 points. The third and sixth
problems are exceptionally difficult, with average scores generally falling
below 2 points.

Table 2 Numbers of Problems with Different Average Scores in

the 24th–64th IMOs

Problem Mean

Problem Number 0–1 1–2 2–3 3–4 4–7

Problem 1 0 0 5 13 23
Problem 2 2 6 18 9 6
Problem 3 22 12 4 3 0
Problem 4 0 0 7 17 17
Problem 5 4 10 14 10 3
Problem 6 24 11 6 0 0
Total 52 39 54 52 49

The 246 problems are categorized into four topics: algebra, combina-
torics, geometry, and number theory, as shown in Table 3. Notably, there
is a relatively large representation of combinatorics and geometry problems.
Combining this information with Table 1, it is evident that in the first 23
IMOs, algebra and geometry problems were predominant.
Furthermore, in early IMOs, geometry problems predominantly
appeared in the 1st/4th and 2nd/5th positions. However, starting from
 Introduction to the IMO 7

Table 3 Numbers of Problems with Different Topics in the 24th–64th IMOs

Topic

Algebra Combinatorics Geometry Number Theory

Session 1, 4 2, 5 3, 6 1, 4 2, 5 3, 6 1, 4 2, 5 3, 6 1, 4 2, 5 3, 6

24–30 4 1 5 4 3 5 5 7 0 1 3 4
31–40 4 5 4 5 4 7 6 9 0 5 2 9
41–50 3 8 4 3 2 6 10 6 5 5 4 5
51–60 5 6 2 4 7 8 7 3 6 3 4 4
61–64 1 3 2 2 2 3 3 1 2 2 2 1
Total 17 23 17 18 18 29 31 26 13 16 15 23

57 65 70 54

the 41st to 50th IMOs, geometry problems were more commonly found in
the 1st/4th and 3rd/6th positions. Similarly, algebra problems were more
frequent in the 1st/4th and 2nd/5th positions, combinatorics problems were
more prevalent in the 2nd/5th and 3rd/6th positions, while the quantity of
number theory problems across different problem numbers does not differ
significantly.
From Table 4, it can be observed that among the four topics, the num-
bers of problems with an average score ranging from 2 to 4 points are quite
similar. However, in the combinatorics topic, there is a higher quantity of
challenging problems, with 31 problems having an average score between
0 and 2 points. Conversely, the geometry topic has the largest number
of relatively easy problems, with 23 problems scoring above 4 points.
This discrepancy is largely due to the fact that there are 29 combinatorics
problems in the 3rd/6th positions, and 31 geometry problems in the 1st/4th
positions.

Table 4 Numbers of Problems with Different Average Scores by Topics in

the 24th–64th IMOs

Problem Mean

Topic 0–1 1–2 2–3 3–4 4–7 Total

Algebra 7 14 19 6 11 57
Combinatorics 20 11 12 14 8 65
Geometry 13 5 14 15 23 70
Number theory 12 9 9 17 7 54
Total 52 39 54 52 49 246
8 IMO Problems, Theorems, and Methods: Algebra

Furthermore, when considering Table 3 and Table 4, it becomes appar-

ent that among the four topics, the number of problems with an average
score ranging from 0 to 2 points closely aligns with the number of prob-
lems in the 3rd/6th positions. There are slightly more problems with an
average score between 2 to 4 points compared to those in the 2nd/5th posi-
tions, and slightly fewer problems with an average score exceeding 4 points
compared to those in the 1st/4th positions. This indicates that even the
seemingly easier problems in the IMO are not as straightforward as they
might appear.
Notably, among these 246 problems, the lowest average score is
attributed to IMO 58-3 (Combinatorics, proposed by Austria):
A hunter and an invisible rabbit play a game in a Euclidean plane.
The rabbit’s starting point, A0 , and the hunter’s starting point, B0 , are
the same. After n − 1 rounds of the game, the rabbit is at point An−1 and
the hunter is at point Bn−1 . In the nth round of the game, three things
occur in order.
(i) The rabbit moves invisibly to a point An such that the distance
between An−1 and An is exactly 1.
(ii) A tracking device reports a point Pn to the hunter. The only guar-
antee provided by the tracking device to the hunter is that the distance
between Pn and An is at most 1.
(iii) The hunter moves visibly to a point Bn such that the distance
between Bn−1 and Bn is exactly 1.
Is it always possible, no matter how the rabbit moves, and no matter
what points are reported by the tracking device, for the hunter to choose
her moves so that after 109 rounds she can ensure that the distance between
her and the rabbit is at most 100?
This unconventional problem received an average score of only 0.042
points. Only two contestants, Mikhail Ivanov from Russia and Linus
Cooper from Australia, achieved a perfect score of 7 points. Joe Benton
from the United Kingdom scored 5 points, Pavel Hudec from Czech Repub-
lic earned 4 points, Hadyn Ka Ming Tang from Australia, Yahor Dubovik
from Belarus, and Jeonghyun Ahn from South Korea each scored 1 point.
Furthermore, among the 20 lowest-scoring problems in the 24th–64th
IMOs, nearly all of them appeared in the 3rd/6th positions, as indicated in
Table 5. There were three algebra problems, eight combinatorics problems,
six geometry problems, and three number theory problems among them.
 Introduction to the IMO 9

Table 5 The 20 Problems with the Lowest Average Scores in the

24th–64th IMOs

Problem Mean Topic Problem Mean Topic

IMO 58-3 0.042 Combinatorics IMO 54-6 0.296 Combinatorics

IMO 48-6 0.152 Algebra IMO 48-3 0.304 Combinatorics
IMO 50-6 0.168 Combinatorics IMO 52-6 0.318 Geometry
IMO 47-6 0.187 Geometry IMO 53-6 0.336 Number Theory
IMO 57-3 0.251 Number Theory IMO 55-6 0.339 Combinatorics
IMO 49-6 0.260 Geometry IMO 56-6 0.355 Combinatorics
IMO 64-6 0.275 Geometry IMO 51-6 0.368 Algebra
IMO 59-3 0.278 Combinatorics IMO 62-3 0.372 Geometry
IMO 61-6 0.282 Combinatorics IMO 62-2 0.375 Algebra
IMO 58-6 0.294 Number Theory IMO 60-6 0.403 Geometry

These three algebra problems were:

• (IMO 48-6, proposed by the Netherland). Let n be a positive

integer. Consider

S = {(x, y, z) : x, y, z ∈ {0, 1, . . . , n}, x + y + z > 0}

as a set of (n + 1)3 − 1 points in a three-dimensional space. Determine the

smallest possible number of planes, the union of which contains S but does
not include (0, 0, 0).

•(IMO 51-6, proposed by Iran). Let a1 , a2 , a3 , . . . be a sequence of

positive real numbers. Suppose that for some positive integer s,

an = max{ak + an−k |1 ≤ k ≤ n − 1}

for all n > s. Prove that there exist positive integers l and N with l ≤ s,
such that an = al + an−l for all n ≥ N .

• (IMO 62-2, proposed by Canada). Show that the inequality

 n 
n  n 
n 

|xi − xj | ≤ |xi + xj |
i=1 j=1 i=1 j=1

holds for all real numbers x1 , x2 , . . . , xn .

10 IMO Problems, Theorems, and Methods: Algebra

2.3 The classification of problems

In the 1st–64th IMOs, there were 101 algebra problems, which can
be categorized into five specialized subjects: equation problems, func-
tion problems, sequence problems, inequality problems, and other algebra
problems.
There were 87 combinatorics problems, which can be categorized into six
specialized subjects: enumerative combinatorics problems, existence prob-
lems, extremal combinatorial problems, operation and logical reasoning
problems, combinatorial geometry problems, and graph theory problems.
There were 123 geometry problems, which can be categorized into seven
specialized subjects: similarity and congruence problems, circle problems,
power-of-point problems, special point and line problems, trigonometry
problems, solid geometry problems, and geometric inequality problems.
There were 75 number theory problems, which can be categorized into
three specialized subjects: divisibility-of-integer problems, modular arith-
metic problems, and indeterminate equation problems.
For the 101 algebra problems categorized and analyzed, as shown in
Table 6, it can be observed that the subjects of functions and inequalities
had the largest number of problems, with 28 and 33 problems, respectively.
In the first 10 IMOs, algebra problems primarily focused on equa-
tions and inequalities. Subsequently, the number of equation problems
decreased rapidly, while problems related to functions and sequences grad-
ually increased and maintained a certain frequency.
In the 41st–50th IMOs, algebra problems were primarily centered
around inequality problems, whereas in the 51st–60th IMOs, algebra prob-
lems primarily focused on function problems.

Table 6 Numbers of Algebra Problems in the First 64 IMOs

Subject

Session Equation Function Sequence Inequality Others

1–10 11 1 1 5 2
11–20 3 5 3 7 2
21–30 1 6 2 4 1
31–40 0 5 3 4 1
41–50 0 3 1 8 3
51–60 1 7 3 2 0
61–64 0 1 1 3 1
Total 16 28 14 33 10
 Introduction to the IMO 11

In the first 64 IMOs, there had been a total of 16 equation problems,

accounting for approximately 15.8% of all algebra problems. These prob-
lems can be primarily categorized into three types: (1) finding solutions of
equations and systems of equations, totaling seven problems; (2) proving
relationships satisfied by equations and systems of equations, totaling four
problems; (3) investigating conditions under which equations and systems
of equations have solutions, totaling five problems.
As shown in Table 7, in the 24th–64th IMOs, there was only one equa-
tion problem, which is IMO 57-5 (proposed by Russia). This indicates that
equation problems mainly appeared in the first 23 IMOs.

Table 7 Numbers of Equation Problems in the 24th–64th IMOs

Problem Number
Number of Problems
Equation Problem 1, 4 2, 5 3, 6 in the First 64 IMOs

Finding solutions 0 0 0 7
Proving relationships 0 0 0 4
Investigating conditions 0 1 0 5
Total 0 1 0 16

In the first 64 IMOs, there had been a total of 28 function problems,

approximately accounting for 27.7% of all algebra problems. These prob-
lems can be primarily categorized into three types: (1) proving properties
of functions, totaling nine problems; (2) determining numerical values of
function variables or outputs, totaling four problems; (3) deriving expres-
sions for functions that meet specific conditions, totaling 15 problems.
As shown in Table 8, in the 24th–64th IMOs, there were a total of
20 function problems. These problems were predominantly present in the
1st/4th positions as well as in the 2nd/5th positions, with the main focus

Table 8 Numbers of Function Problems in the 24th–64th IMOs

Problem Number
Number of Problems
Function Problem 1, 4 2, 5 3, 6 in the First 64 IMOs

Proving properties 1 2 1 9
Determining values 0 0 1 4
Deriving expressions 6 8 1 15
Total 7 10 3 28
12 IMO Problems, Theorems, and Methods: Algebra

being on deriving expressions for functions that meet specific conditions.

The other two types of function problems appeared less frequently in the
last 40 IMOs.
In the first 64 IMOs, there had been a total of 14 sequence problems,
approximately accounting for 13.8% of all algebra problems. These prob-
lems can be primarily categorized into three types: (1) determining the
value of a specific term or the number of terms, totaling four problems;
(2) addressing problems related to the existence of sequences, totaling
five problems; (3) proving quantitative relationships satisfied by sequences,
totaling five problems.
As shown in Table 9, in the 24th–64th IMOs, there were a total of
nine sequence problems. These problems were predominantly found in the
3rd/6th positions, with a relatively balanced distribution among different
problem types.

Table 9 Numbers of Sequence Problems in the 24th–64th IMOs

Problem Number
Number of Problems
Sequence Problem 1, 4 2, 5 3, 6 in the First 64 IMOs

Determining values 2 1 0 4
Existence problems 0 0 2 5
Proving quantitative relationships 0 0 4 5
Total 2 1 6 14

In the first 64 IMOs, there had been a total of 33 inequality prob-

lems, approximately accounting for 32.7% of all algebra problems. These
problems can be primarily categorized into three types: (1) solving inequal-
ities, totaling five problems; (2) proving inequalities, totaling 25 problems;
(3) determining value ranges, totaling three problems.
As shown in Table 10, in the 24th–64th IMOs, there were a total of
21 inequality problems. These problems were predominantly found in the
1st/4th positions as well as in the 2nd/5th positions. The primary focus of
these problems is to prove inequalities, with the other two types of inequal-
ity problems appearing less frequently in the last 40 IMOs.
In the first 64 IMOs, there had been a total of 10 other algebra prob-
lems, approximately accounting for 9.9% of all algebra problems. These
problems can be primarily categorized into three types: (1) proving trigono-
metric identities, totaling two problems; (2) finding polynomials, totaling
 Introduction to the IMO 13

Table 10 Numbers of Inequality Problems in the 24th–64th IMOs

Problem Number
Number of Problems
Inequality Problem 1, 4 2, 5 3, 6 in the First 64 IMOs

Solving inequalities 1 0 0 5
Proving inequalities 6 8 4 25
Determining value ranges 0 1 1 3
Total 7 9 5 33

two problems; (3) proving properties of polynomials and sets, totaling six
problems.
As shown in Table 11, in the 24th–64th IMOs, there were a total of
six other algebra problems, primarily focusing on proving properties of
polynomials and sets.

Table 11 Numbers of Other Algebra Problems in the 24th–64th IMOs

Problem Number
Number of Problems
Others 1, 4 2, 5 3, 6 in the First 64 IMOs

Proving trigonometric identities 0 0 0 2

Finding polynomials 0 1 0 2
Proving properties 1 1 3 6
Total 1 2 3 10

As shown in Table 12, among the algebra problems in the 24th–

64th IMOs, the largest proportion of problems is attributed to function
and inequality. However, the average scores of function problems are pri-
marily concentrated between 2 and 7 points, while for inequality prob-
lems, the average scores are primarily concentrated between 0 and 3
points. This suggests that inequality problems tend to be more challeng-
ing, while function problems include a greater number of relatively simple
problems.
Overall, the numbers of algebra problems in the 1st/4th, 2nd/5th, and
3rd/6th positions are roughly equal to the numbers of algebra problems
with average scores of 4–7, 2–4, and 0–2, respectively. However, upon a
further comparison of the distribution of problem numbers among the five
subjects, it is evident that inequality problems tend to be more challenging.
There are five inequality problems in the 2nd/5th positions, with average
14 IMO Problems, Theorems, and Methods: Algebra

Table 12 Numbers of Algebra Problems with Different

Average Scores in the 24th–64th IMOs

Problem Mean

Algebra Problem 0–1 1–2 2–3 3–4 4–7 Total

Equation Problems 0 1 0 0 0 1
Function problems 0 4 6 5 5 20
Sequence problems 1 3 3 0 2 9
Inequality problems 3 5 8 1 4 21
Others 3 1 2 0 0 6
Total 7 14 19 6 11 57
21 25 11

scores ranging from 0 to 2, and three inequality problems in the 1st/4th

positions, with average scores ranging from 2 to 4.
Similarly, there are four function problems in the 1st/4th positions, with
average scores ranging from 2 to 4, and one function problem in the 2nd/5th
positions, with an average score of 1–2. This indicates that even the easier
function problems are not necessarily easy.

2.4 The proposal for problems

Problems in the IMO are proposed by the participating countries and
regions, except the host. Usually the team leaders are in charge of sub-
mitting problems with a limit of six, and these problems are then subjected
to the selection by a selection committee composed of experts organized
by the host. Approximately 30 problems are chosen as shortlist problems,
with around eight problems in each of the four topics: algebra, geome-
try, combinatorics, and number theory. Subsequently, these problems are
submitted to the Jury, which is comprised of team leaders from each partic-
ipating country or region. The problems are discussed and voted upon to
select the official examination problems. Once the problems are finalized,
they are translated into five working languages: English, French, German,
Russian, and Spanish. Each leader then translates the problems into their
respective national languages, and contestants can choose from two lan-
guages in which to answer the problems.
Among the first 64 IMOs, algebra problems were contributed by 35 dif-
ferent countries and regions. The United Kingdom had the highest number
of problems proposed, with a total of 10. The Netherlands proposed eight
 Introduction to the IMO 15

problems, followed by Hungary with six problems. Poland, Bulgaria, the

Soviet Union, France, Sweden, and South Korea each proposed five prob-
lems, while Ireland contributed four problems. These 10 countries collec-
tively provided 58 problems. Remarkably, only Sweden (in the 15th IMO
in 1973) and South Korea (in the 46th IMO in 2005) are the two countries
that have proposed two algebra problems in the same IMO session.
Furthermore, as indicated in Appendix B, algebra problems in the first
15 IMOs were primarily proposed by the seven founding countries of the
IMO. From the 16th–35th IMOs, algebra problems were predominantly
contributed by the United Kingdom, with supplementary contributions
from other countries and regions. However, in the 36th–64th IMOs, alge-
bra problems demonstrated a more diverse range of proposing countries
and regions. This, to some extent, correlates with the expansion of the
IMO’s influence and the growth in the number of participating countries
and regions.

3 Awards in the IMO

In addition to selecting problems, the Jury has several other responsibilities,
including: establishing grading criteria, resolving discrepancies in grading
between leaders and coordinators, and determining the number of gold,
silver, and bronze medals, as well as the score thresholds. In each IMO,
approximately 1/12 of the contestants receive a gold medal, 2/12 receive
silver, and 3/12 receive bronze.
Apart from the gold, silver, and bronze medals, contestants who do not
receive medals but attain a score of 7 on at least one problem in the IMO
will receive an Honorable Mention. Contestants who deliver exceptionally
elegant solutions to specific problems in the IMO will receive a Special
Prize.
As depicted in Figure 2, starting from the 24th IMO, the cutoff scores for
gold, silver, and bronze medals have gradually stabilized. The gold medal
cutoff is approximately 29 points, the silver medal cutoff is around 22 points,
and the bronze medal cutoff is roughly 15 points. Furthermore, the average
score of all contestants closely aligns with the bronze medal cutoff. This
indicates that the problem difficulty is well-balanced.
Interestingly, in the first 64 IMOs, there were three occasions where the
gold medal cutoff was a perfect score, meaning only those who scored full
marks could earn a gold medal. These three IMOs were: the 11th IMO
(1969, Romania) with a perfect score of 40 points and three gold medalists;
16 IMO Problems, Theorems, and Methods: Algebra

50
45
40
35
30
25
20
15
10
5
0
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64

Mean Gold Cutoff Silver Cutoff

Bronze Cutoff Perfect Score

Figure 2 Medal Cutoff Scores in the First 64 IMOs

the 14th IMO (1972, Poland) with a perfect score of 40 points and eight
gold medalists; and the 28th IMO (1988, Cuba) with a perfect score of 42
points and 22 gold medalists.

3.1 Participation
In the first 64 IMOs, a total of 269 contestants took part in four or more
IMOs. Among them, two contestants attended seven IMOs, four contes-
tants attended six IMOs, 42 contestants attended five IMOs, and 221 con-
testants attended four IMOs, as shown in Table 13.

Table 13 Contestants with Six or More Participations in the First 64 IMOs

Participation Honorable Perfect

Contestant Country Year Gold Silver Bronze Mention Score

David Norway 1997–2003 1 3 1 1

Kunszenti-
Kovács
Yeoh Zi Song Malaysia 2014–2020 1 1 4 1
Zhuo Qun Canada 2010–2015 5 0 1 0 1
(Alex)
Song
Teodor von Serbia 2007–2012 4 1 1 0
Burg
Alexey Entin Israel 2000–2005 1 3 1 0
Tan Li Xuan Malaysia 2016–2021 0 2 2 1
 Introduction to the IMO 17

Coincidentally, in the 43rd IMO held in 2002, the gold medal cutoff
was set at 29 points, and David Kunszenti-Kovács achieved a total score of
exactly 29 points.
In the 44th IMO held in 2003, the gold medal cutoff was set at 29 points,
whereas Alexey Entin attained a total score of exactly 28 points.
In the 51st IMO held in 2010, the bronze medal cutoff was set at 15
points, and Zhuo Qun (Alex) Song achieved a total score of exactly 15
points.
In the 58th IMO held in 2017, the silver medal cutoff was set at 19
points, and Yeoh Zi Song achieved a total score of exactly 19 points. In the
59th IMO held in 2018, the silver medal cutoff was set at 25 points, and
Yeoh Zi Song’s total score was exactly 24 points. In the 61st IMO held in
2020, the gold medal cutoff was set at 31 points, and Yeoh Zi Song’s total
score was exactly 31 points.
Additionally, from 2002 to 2005 and in 2007, Sherry Gong participated
in the IMO, earning one bronze, two silver, and one gold medal. In the
48th IMO held in 2007, she ranked 7th individually. Notably, from 2002 to
2004, she was a member of the Puerto Rico IMO team, while in 2005 and
2007, she was a member of the United States IMO team.
From 2020 to 2023, Alex Chui participated in the IMO, securing two
gold and two silver medals. However, in 2020 and 2021, he was a member
of the Chinese Hong Kong IMO team, while in 2022 and 2023, he was a
member of the United Kingdom IMO team.
Other than Sherry Gong and Alex Chui, the remaining 267 contestants
hailed from 75 different countries and regions. Among them, there were 12
contestants from Cyprus and Moldova each, 11 from Malaysia, eight from
Trinidad and Tobago, seven from each of Estonia, Germany, Sri Lanka, and
North Macedonia, and six from Japan and Philippines each.

3.2 Gold medals

In the first 64 IMOs, a total of 49 contestants achieved three or more
gold medals. Among them, one contestant earned five gold medals, six
contestants earned four gold medals, and 42 contestants earned three gold
medals.
As shown in Table 14, Zhuo Qun (Alex) Song, Reid Barton, and Lisa
Sauermann have all achieved perfect scores.
Coincidentally, in the 54th IMO held in 2013, the gold medal cutoff
was 31 points, and Nipun Pitimanaaree achieved a total score of exactly
31 points.
18 IMO Problems, Theorems, and Methods: Algebra

Table 14 Contestants with Four or More Gold Medals in the First 64 IMOs

Participation Gold Perfect

Contestant Country Year Year Score Year

Zhuo Qun (Alex) Song Canada 2010–2015 2011–2015 2015

Reid Barton The United 1998–2001 1998–2001 2001
States of
America
Christian Reiher Germany 1999–2003 2000–2003
Lisa Sauermann Germany 2007–2011 2008–2011 2011
Teodor von Burg Serbia 2007–2012 2009–2012
Nipun Pitimanaaree Thailand 2009–2013 2010–2013
Luke Robitaille The United 2019–2022 2019–2022
States of
America

In the 41st IMO held in 2000, the top four contestants all achieved
perfect scores, and Reid Barton ranked fifth with a total score of 39. In the
43rd IMO held in 2002, the top three contestants all achieved perfect scores,
and Christian Reiher ranked fourth with a total score of 36. Furthermore,
in the 50th IMO held in 2009, Lisa Sauermann achieved a total score of 41,
securing the third position.
Moreover, Oleg Golberg participated in the IMO from 2002 to 2004,
achieving three gold medals. He consistently ranked within the top 10 in
terms of total scores. Notably, in 2002 and 2003, he was a member of the
Russia IMO team, and in 2004, he was a member of the United States IMO
team.
Apart from Oleg Golberg, the remaining 48 contestants hailed from
22 different countries and regions. Among them, there were four contes-
tants from each of Russia, Bulgaria, Germany, Hungary, Romania, and
the United States. Both South Korea and the United Kingdom had three
contestants, while Canada, Japan, Singapore, and the Soviet Union were
represented by two contestants each.

3.3 Special prizes

In the first 64 IMOs, only 44 contestants have received special prizes.
Among them, one contestant has received the special prize three times,
seven contestants have earned twice, and 36 contestants have achieved once.
It indicates that achieving a special prize is even more challenging than
securing a gold medal.
 Introduction to the IMO 19

As shown in Table 15, John Rickard, Imre Ruzsa, and Marc van Leeuwen
all achieved two special prizes in one IMO for their elegant solutions. Fur-
thermore, John Rickard, Imre Ruzsa, and László Lovász have all earned a
perfect score twice.

Table 15 Contestants with Multiple Special Prizes in the First 64 IMOs

Perfect
Participation Special Gold Score
Contestant Country Year Prize Year Year Year

John Rickard The United 1975–1977 1976, 1977 (2) 1975–1977 1975, 1977
Kingdom
József Hungary 1963–1966 1965, 1966 1964–1966 1966
Pelikán
László Lovász Hungary 1963–1966 1965, 1966 1964–1966 1965, 1966
László Babai Hungary 1966–1968 1966, 1968 1968 1968
Simon The United 1967–1969 1967, 1969 1967–1969 1969
Phillips Kingdom
Norton
Wolfgang The German 1967–1971 1970, 1971 1968, 1970, 1971 1970
Burmeister Democratic
Republic
Imre Ruzsa Hungary 1969–1971 1971 (2) 1970, 1971 1970, 1971
Marc van The Nether- 1977–1978 1978 (2)
Leeuwen lands

Coincidentally, in the 11th IMO held in 1969, only three gold medals
were awarded, with Imre Ruzsa ranking 4th and receiving a silver medal.
Similarly, in the 19th IMO held in 1977, which also resulted in only 13 gold
medals, Marc van Leeuwen ranked 14th and earned a silver medal.
Additionally, these 44 contestants hailed from 16 different countries and
regions. Among them, there were seven contestants from each of Hungary
and the United Kingdom, five from the German Democratic Republic, four
from Bulgaria and Poland each, three from each of Czechoslovakia and the
Soviet Union, and two from Finland and the United States each.
Special prizes were more frequently granted in the first 20 IMOs, with
a total of 27 special prizes earned by 22 contestants from the 11th–20th
IMOs. Subsequently, the frequency of special prize presentations declined.
Since Moldovan contestant Iurie Boreico received a special prize for his
brilliant solution to IMO 46-3 (Algebra, proposed by South Korea) in 2005,
no contestant has achieved this accolade to date.
From 2003 to 2007, Iurie Boreico consistently participated in the IMO,
earning three gold and two silver medals. He achieved a perfect score in
2005 and 2006. It’s noteworthy that the 44th IMO held in 2003 only yielded
20 IMO Problems, Theorems, and Methods: Algebra

37 gold medals, with Iurie Boreico placing 38th individually and receiving
a silver medal.

• (IMO 46-3, proposed by South Korea). Let x, y, z be three positive

reals such that xyz ≥ 1. Prove that

x5 − x2 y5 − y2 z5 − z2
+ + ≥ 0.
x5 + y 2 + z 2 y 5 + z 2 + x2 z 5 + x2 + y 2

4 Summary
The IMO stands as a distinguished intellectual competition for young
minds. According to a study by Agarwal R. and Gaule P., statistical anal-
ysis reveals that among contestants in the IMO (including those who did
not secure medals), 22% choose to pursue further studies in mathemat-
ics, ultimately obtaining doctoral degrees in the field. Additionally, 1% of
these contestants become presenters at the International Congress of Math-
ematicians, and 0.2% attain the Fields Medal. These statistics underscore
the vital role of the IMO in identifying and nurturing mathematical talent.
It’s essential not to perceive the IMO as a mere selection exam. Rather
than focusing solely on the brief two-day competition, the crucial aspect
lies in the learning and preparation undertaken before participating. As
the mathematician Paul Halmos aptly put it, what mathematics really con-
sists of is problems and solutions. Contestants, through their exploration
of Olympiad problems, not only enhance their mathematical abilities but
also experience the joy and satisfaction of problem-solving. This experience
plants the seeds of a future career in mathematics.
However, it’s important to acknowledge that Olympiad problems and
research problems in mathematics differ. Research problems often lack
readily available answers and may require the investment of countless days
and nights. Hence, the IMO is just one pathway in the growth of math-
ematical talents, and success in the IMO is not the sole qualification for
becoming an outstanding mathematician.
Although every contestant aims for a gold medal, their aspirations go
far beyond accolades. On this stage, they have the opportunity to showcase
their intellectual capabilities, revel in the mathematical exploration, and
relish competing with talented young minds from around the world, all
without the narrow goal of proving their superiority over others. While the
 Introduction to the IMO 21

competition results may vary, each contestant stands as a victor in their own
right and becomes a companion and witness to one another’s life journeys.
In contrast to the Olympics, where athletes’ careers are closely inter-
twined with the Games, the IMO is merely a chapter in the growth of these
gifted young individuals. Following the IMO, the door to a new mathemat-
ical world has already swung wide open for them.
This page intentionally left blank
Chapter 1

Equation Problems

Equations, in layman’s terms, are mathematical statements contain-

ing unknowns that demonstrate the equality relationship between two
mathematical expressions.
Ancient scholars had early investigations into equations, including
mathematicians from Babylon, Egypt, Greece, and China. By the time the
Indian mathematician Brahmagupta found a general solution, most ancient
civilizations already knew and understood solutions to quadratic equa-
tions. Moreover, the 5th-century Chinese mathematician Chongzhi Zu,
7th-century Chinese mathematician Xiaotong Wang, and 13th-century
Italian mathematician Fibonacci all approximated the positive roots of
cubic equations in the form x3 + px2 + qx = r, where p, q, r > 0.
However, before the 16th century, mathematicians had not discovered
a general formula for the roots of cubic equations. In fact, the history
of cubic equation roots is closely linked to the mathematics competition
among Italian mathematicians in the 16th century.
During that time, Italian mathematicians often openly challenged each
other to gain honors and benefits. The discoverer of the cubic equation
root formula, Nicolo Tartaglia, triumphed over his opponents with a general
method for solving the three equations:
x3 + px2 = q, x3 = px + q, x3 + q = px2 .
However, he kept his discovery undisclosed for over two decades, leading
to unfortunate consequences. Shortly afterward, Ludovico Ferrari found a
solution to quartic equations.
Equations of the fifth degree and higher do not have radical solutions,
a fact proven by the 19th-century French mathematician Evariste Galois,

23
24 IMO Problems, Theorems, and Methods: Algebra

who introduced the concept of groups and thereby initiated a new branch of
mathematics called “Abstract Algebra.” This not only elevated the study
of algebra to new heights but also had a significant impact on modern
mathematics, physics, chemistry, and other fields.
Today, the knowledge and applications of equations are integral to all
stages of mathematical education, ranging from elementary puzzles like the
classic chicken-rabbit cage problem to complex university-level differential
equations. Equations have evolved to become essential tools in mathemat-
ics, enabling the exploration and understanding of relationships between
various mathematical elements and uncovering the fundamental principles
that govern these relationships.
In the first 64 IMOs, there had been a total of 16 equation problems,
accounting for approximately 15.8% of all algebra problems. These prob-
lems can be primarily categorized into three types: (1) finding solutions of
equations and systems of equations, totaling seven problems; (2) proving
relationships satisfied by equations and systems of equations, totaling four
problems; (3) investigating conditions under which equations and systems
of equations have solutions, totaling five problems. The statistical distri-
bution of these three types of problems in the previous IMOs is presented
in Table 1.1.

Table 1.1 Numbers of Equation Problems in the First 64 IMOs

Session

Content 1–10 11–20 21–30 31–40 41–50 51–60 61–64 Total

Finding solutions 7 0 0 0 0 0 0 7
Proving 2 2 0 0 0 0 0 4
relationships
Investigating 2 1 1 0 0 1 0 5
conditions
Algebra problems 20 20 14 13 15 13 6 101
The percentage 55.0% 15.0% 7.1% 0.0% 0.0% 7.7% 0.0% 15.8%
of equation
problems among
the algebra
problems

It is evident that equation problems predominantly appeared in early

IMOs, with a majority focusing on finding solutions of equations and sys-
tems of equations. Particularly, in the 1st–10th IMOs, the proportion of
 Equation Problems 25

equation problems among all algebra problems was very high, exceeding
50%, but rapidly decreased thereafter.
This can partly be attributed to the fact that the study of equations
spans various stages of primary and secondary school mathematics educa-
tion. The ability to formulate and solve appropriate equations has become
an essential skill and method in dealing with mathematical problems.
Therefore, in recent years, algebra problems rarely specifically address equa-
tion problems, with the focus shifting towards more flexible topics such as
solving functional equations and proving inequalities.
Problems involving finding solutions are generally not very difficult, and
commonly used methods include factoring, the method of undetermined
coefficients, trial and error, etc. It is important to note the verification of
extraneous roots during the solving process. As for problems about proving
relationships, the relationship between the unknowns and coefficients of the
equations or systems is somehow clear, and the direction from conditions
to conclusions is often more apparent. In contrast, problems investigating
conditions tend to be more open-ended.
This chapter will be divided into three parts. The first part introduces
common equations and methods, followed by presentations of important
theorems related to equations, mainly focusing on roots and coefficients.
These theorems are highly applicable when solving equations or proving
relationships satisfied by equations. Polynomials and equations are closely
related, with additional contents related to polynomials planned for the
fifth chapter, “Other Algebra Problems.”
The second part revolves around three types of problems: “finding solu-
tions of equations and systems of equations,” “proving relationships satis-
fied by equations and systems of equations,” and “investigating conditions
under which equations and systems of equations have solutions.” These
problems are presented in chronological order, and some problems include
various solutions, generalizations, and similar problems.
It is important to note that for each problem, the solutions are fol-
lowed by information on the scores, including the number of contestants in
each score range, the average score, and the scores of the top five teams.
However, early IMOs often lacked information on contestant scores, so the
number of contestants in each score range only represents the counted num-
ber of contestants, and some problems lack scores of the top five teams.
The third part provides a brief summary of this chapter.
26 IMO Problems, Theorems, and Methods: Algebra

Some of the equation problems are quite innovative, and readers are
encouraged to contemplate the solutions before referring to the answers.

1.1 Common Theorems, Formulas, and Methods

1.1.1 Common equations
(1) Linear equations of one variable
The standard form of a linear equation of one variable is ax = b, where
a = 0, and the equation has a unique solution x = ab .

(2) Quadratic equations of one variable

The standard form of a quadratic equation of one variable is ax2 + bx + c =
0, where a = 0, and a, b, and c are real numbers, with the discriminant
Δ = b2 − 4ac.

(i) When Δ > 0, the equation has two distinct real roots.
(ii) When Δ = 0, the equation has two identical real roots.
(iii) When Δ < 0, the equation has no real roots but two complex conjugate
roots.

The quadratic formula is given by

√
−b ± b2 − 4ac
x= .
2a
(3) Higher-degree equations
In general, polynomial equations with the highest power of the variable
greater than 2 are referred to as higher-degree equations. Their general
form is

an xn + · · · + ai xi + · · · + a1 x + a0 = 0,

where an = 0, and ai (i = 0, 1, 2, . . . , n) are real numbers.

In particular, for a cubic equation of one variable x3 + px + q = 0, the
discriminant is
 q 2  p 3
Δ= + .
2 3
(i) When Δ > 0, there is one real root and two complex roots.
(ii) When Δ = 0, there are three real roots. If p = q = 0, then there is a
triple root; if pq = 0, then two of the three real roots are equal.
 Equation Problems 27

(iii) When Δ < 0, there are three distinct real roots given by
⎧ √3
⎪
⎨x1 = 2 √r cos θ,
x2 = 2 3 r cos(θ + 120◦ ),
⎪
⎩ √
x3 = 2 3 r cos(θ + 240◦ ),

3
where r = − p3 and θ = 13 arccos − 2r q
.
(4) Fractional equations
Equations with unknowns in the numerators and denominators are termed
fractional (or rational) equations. Typically, methods such as clear-
ing denominators, substitution, and basic identity transformations are
employed to transform fractional equations into polynomial equations for
solving. It is important to note that this process may introduce extraneous
solutions.
(5) Irrational equations
An irrational equation, also known as a radical equation, is an equation
containing an unknown under a radical root symbol. The fundamental
approach to solving irrational equations involves transforming the “irra-
tional” into “rational.” Typically, techniques such as factoring, substitu-
tion, and completing the square are employed to convert the equation.
When dealing with irrational equations, caution must be exercised regard-
ing the possibility of extraneous solutions during the process.
(6) Systems of equations
A system of equations refers to a combination of two or more equations with
multiple unknowns. In secondary education, the most commonly encoun-
tered type is the n-variable linear system:
⎧
⎪a11 x1 + a12 x2 + · · · + a1n xn = b1 ,
⎪
⎪
⎪
⎨a x + a x + · · · + a x = b ,
21 1 22 2 2n n 2
⎪
⎪ ·································
⎪
⎪
⎩
as1 x1 + as2 x2 + · · · + asn xn = bs ,
where a11 , a12 , . . . , asn are coefficients, and b1 , b2 , . . . , bs are constants. The
number of equations may be equal to, less than, or greater than the number
of unknowns. There are three possible scenarios for the solutions of an n-
variable linear system: (i) no solution; (ii) a unique solution; (iii) an infinite
number of solutions.
28 IMO Problems, Theorems, and Methods: Algebra

1.1.2 Common methods for solving equations

(1) Completing the square
This method involves transforming an expression or part of it through iden-
tity transformations into a perfect square or the sum of several perfect
squares. It is frequently used in manipulating identities to uncover implicit
conditions in the problem. In particular, if there are nested square roots,
then one might consider completing the square to eliminate one of the
square roots.
For example, for solving the equation

√ √ 1
x+ y−1+ z − 2 = (x + y + z)
2

within the real number domain, completing the square can be employed to
transform the equation into

√  2 √ 2
2
x−1 + y−1−1 + z−2−1 = 0.

(2) Elimination method

The elimination method involves eliminating certain elements from multiple
relations through finite transformations to simplify the problem.
For example, for solving the system of linear equations in two variables:

3x + 7y = 2,
2x + y = 7,

using the elimination method, the system can be transformed into 3x +

7 (7 − 2x) = 2, significantly simplifying the problem.

(3) Method of undetermined coefficients

This method involves expressing a polynomial in another form containing
undetermined coefficients, resulting in an identity. Subsequently, based on
the properties of the identity, equations or systems of equations can be
derived to determine the values of the coefficients. This method is com-
monly applied in factorization.
For example, given the polynomial 2x4 − 3x3 + ax2 + 7x + b divisi-
ble by x2 + x − 2, the method of undetermined coefficients allows us to
 Equation Problems 29

assume

2x4 − 3x3 + ax2 + 7x + b = (x2 + x − 2)(mx2 + nx + k),

where solving for m, n, k allows us to determine the values of a, b.

(4) Formula method
For specific types of equations, applying corresponding formulas or conclu-
sions directly facilitates the solution of the equation.
For example, for solving the equation 2x2 + 7x + 5 = 0, using the
quadratic formula reveals the real roots as
√ √
−7 + 72 − 4 × 2 × 5 −7 − 72 − 4 × 2 × 5
x1 = and x2 = .
2×2 2×2
(5) Trial-and-error method
Also known as the trial root method, this involves making speculative
guesses for roots. For instance, in solving a polynomial equation with inte-
ger coefficients

an xn + · · · + a1 x + a0 = 0(an = 0),

by listing the divisors of the leading coefficient an and the constant term
a0 , denoted as q and p, respectively, and dividing them separately, poten-
tial rational roots pq for the equation can be obtained. Subsequently, these
values are substituted back into the equation for verification.
For example, for solving the equation 2x3 − 6x2 + 3x + 2 = 0, using the
trial-and-error method yields potential rational roots as ±1, ±2, ± 12 . Upon
testing, x = 2 is found to be a rational root, leading to the factorization of
2x3 − 6x2 + 3x + 2 = (x − 2)(2x2 − 2x − 1).
(6) Substitution method
When faced with a complex equation structure, considering certain parts
as a whole and then replacing them with a new variable is a com-
mon approach. This method is often employed to reduce the number of
unknowns or decrease the degree of the equation, simplifying complex
problems.
For example, for solving the equation (x2 − 2x)2 − 3(x2 − 2x) − 4 = 0,
using substitution, let y = x2 − 2x. The original equation is then simplified
to y 2 − 3y − 4 = 0, and solving this quadratic equation yields y1 = 4
and y2 = −1. Substituting the value of y into y = x2 − 2x provides the
corresponding values for x.
30 IMO Problems, Theorems, and Methods: Algebra

1.1.3 Other important theorems

Theorem 1.1 (Vieta’s Formulas). For a real-coefficient quadratic equa-
tion ax2 + bx + c = 0 with the roots x1 and x2 , where a = 0, the following
relations hold:
b c
x1 + x2 = − , x1 x2 = .
a a
For a real-coefficient polynomial equation of degree n given by an xn +
· · · + a1 x + a0 = 0 with the roots x1 , x2 , . . . , xn , where an = 0, the following
relations hold:
an−1
x1 + x2 + · · · + xn = − ,
an
an−2
xi xj = ,
an
1≤i<j≤n
an−3
xi xj xk = − ,
an
1≤i<j<k≤n

···························
(−1)n a0
x1 x2 · · · xn = an .

Theorem 1.2 (Inverse Vieta’s Formulas). If real numbers α and β

satisfy the following relations:
b c
α + β = − , αβ = ,
a a
then α and β are the roots of the quadratic equation ax2 + bx + c = 0,
where a, b, c are real numbers and a = 0.
Theorem 1.3 (Rational Root Theorem). For a polynomial equation
with integer coefficients
an xn + · · · + a1 x + a0 = 0(an = 0),
let pq be a rational root, where p and q are relatively prime. Then p is
an integer factor of the constant term a0 and q is an integer factor of the
leading coefficient an .
In particular, for a quadratic equation ax2 + bx + c = 0, where a, b, c
are integers,
(i) when a = 1, the rational roots of the equation must be integers;
(ii) integer roots must be divisors of c;
(iii) the discriminant Δ of the equation is a perfect square.
 Equation Problems 31

1.2 Problems and Solutions

1.2.1 Finding solutions (without parameters)
Problem 1.1 (IMO 1-2, proposed by Romania). For what real values
of x is
 
√ √
x + 2x − 1 + x − 2x − 1 = A,
√
given (a) A = 2, (b) A = 1, (c) A = 2, where only non-negative real
numbers are admitted for square roots?
Solution 1. It is easy to see that x ≥ 12 , and we also have

√ 1 √
x + 2x − 1 = √ | 2x − 1 + 1|
2
1 √
= √ ( 2x − 1 + 1),
2

√ 1 √
x − 2x − 1 = √ | 2x − 1 − 1|.
2
It is evident that
√ √ 1
| 2x − 1 − 1| = 1 − 2x − 1, ≤ x ≤ 1,
2
√ √
| 2x − 1 − 1| = 2x − 1 − 1, x ≥ 1.
Let
 
√ √
y= x + 2x − 1 + x − 2x − 1
1 √ √
= √ ( 2x − 1 + 1 + | 2x − 1 − 1|).
2
√ √ √ √
For 12 ≤ x ≤ 1, we have y = 2; for x ≥ 1, we have y = 2 2x − 1 ≥ 2.
Therefore, the result for this problem is (a) 12 ≤ x ≤ 1; (b) no solution;
(c) x = 32 .
√ √
Solution 2. Let y = x + 2x − 1+ x − 2x − 1. Squaring both sides
and simplifying, we get
y2 1
= x + |x − 1|, x≥ .
2 2
2 2
So, when 12 ≤ x ≤√1, we have 1 = y2 ; when 1 ≤ x, we have 2x − 1 = y2 .
By substituting y = 2, 1, 2, we can obtain the same result as Solution 1.
√  
Note. a + b = a+c 2 + a−c 2 2
2 , where c = a − b.
32 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 1 contestant scored 8 points, 2 contestants scored 7 points, 4 contestants scored
6 points, 1 contestant scored 5 points, 1 contestant scored 4 points, no contestant scored 3
points, no contestant scored 2 points, no contestant scored 1 point, and no contestant scored
0 point. The average score of this problem is 6.111, indicating that it was simple.
Among the top five teams in the team scores, the Romania team achieved a total score
of 249 points, the Hungary team achieved a total score of 233 points, the Czechoslovakia
team achieved a total score of 192 points, the Bulgaria team achieved a total score of 131
points, and the Poland team achieved a total score of 122 points.
The gold medal cutoff for this IMO was set at 37 points (with 3 contestants earning gold
medals), the silver medal cutoff was 36 points (with 3 contestants earning silver medals), and
the bronze medal cutoff was 33 points (with 5 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Bohuslav
Diviš from Czechoslovakia.

Problem 1.2 (IMO 4-4, proposed by Romania). Solve the equation

cos2 x + cos2 2x + cos2 3x = 1.

Solution 1. Substituting cos2 x = 12 (1 + cos 2x) and cos2 3x =

1
2 (1 + cos 6x) into the given equation, we get

0 = cos 2x + 2 cos2 2x + cos 6x = 2 cos 2x(cos 4x + cos 2x)

= 4 cos x cos 2x cos 3x.

For this, we obtain

π kπ π kπ π
x1 = kπ + , x2 = + , x3 = + .
2 2 4 3 6
Here, k takes all integers, and it is easy to see that the first solution x1
is included in the third solution x3 , so it can be omitted.

Solution 2. Let the trigonometric√ form of the complex number a + bi be

z = r(cos θ + i sin θ), where r = a2 + b2 and tan θ = ab .
From De Moivre’s formula, if r = 1, then z n = cos nθ + i sin nθ
and z −n = cos nθ − i sin nθ. So cos nθ = 12 (z n + z −n ) and sin nθ =
1 n −n
2i (z − z ). Replacing the unknown angle x with θ in the original
equation, we have

(z + z −1 )2 + (z 2 + z −2 )2 + (z 3 + z −3 )2 = 4.
 Equation Problems 33

Simplifying the above gives z −6 + z −4 + z −2 + 1 + z 2 + z 4 + z 6 = −1.

It’s worth noting that z 2 = 1, otherwise, θ would be a multiple of π,
and substituting it into the original equation wouldn’t make it valid. So,
z −6 −z 8 7 −7
1−z 2 = −1, and this leads to z − z = −(z − z −1 ), or
sin 7θ = − sin θ = sin(−θ),
which implies 7θ = nπ + (−1)n (−θ).
If n = 2k, then 8θ = 2kπ, so θ = kπ4 , where k is an integer that cannot
be divisible by 4.
If n = 2k +1, then 6θ = (2k + 1)π, so θ = (2k+1)π
6 , where k is an integer.
Note. The solutions θ = kπ 4 in Solution 2, where k cannot be divisible by
4, are equivalent to the solutions in Solution 1, x = kπ π π
2 + 4 and x = kπ + 2 ,
where k is an integer.
Score Situation This particular problem saw the following distribution of scores among
contestants: 8 contestants scored 5 points, 5 contestants scored 4 points, no contestant
scored 3 points, 3 contestants scored 2 points, 1 contestant scored 1 point, and no contestant
scored 0 point. The average score of this problem is 3.941, indicating that it was relatively
straightforward.
Among the top five teams in the team scores, the Hungary team achieved a total score
of 289 points, the Soviet Union team achieved a total score of 263 points, the Romania team
achieved a total score of 257 points, the Czechoslovakia team achieved a total score of 212
points, and the Poland team achieved a total score of 212 points.
The gold medal cutoff for this IMO was set at 41 points (with 4 contestants earning gold
medals), the silver medal cutoff was 34 points (with 12 contestants earning silver medals),
and the bronze medal cutoff was 29 points (with 15 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 46 points, namely Iosif
Bernstein from the Soviet Union.

Problem 1.3 (IMO 7-4, proposed by the Soviet Union). Find all
sets of four real numbers x1 , x2 , x3 , x4 such that the sum of any one and
the product of the other three is equal to 2.
Solution 1. For the system of equations
⎧
⎪
⎪ x1 + x2 x3 x4 = 2, (1)
⎪
⎪
⎪
⎨ x2 + x3 x4 x1 = 2, (2)
⎪
⎪ x3 + x4 x1 x2 = 2, (3)
⎪
⎪
⎪
⎩
x4 + x1 x2 x3 = 2, (4)
34 IMO Problems, Theorems, and Methods: Algebra

it is evident that xi = 0(i = 1, 2, 3, 4). Otherwise, for example, if x1 =

0, then from (2), (3), and (4), we have x2 = x3 = x4 = 2. However,
x2 x3 x4 = 2 from (1), which leads to a contradiction.
Now, from (1) and (2),
2 − x1 2 − x2
x3 x4 = = ,
x2 x1
so 2x1 − x21 = 2x2 − x22 , which implies (x1 − 1)2 = (x2 − 1)2 .
Therefore,
|x1 − 1| = |x2 − 1| = |x3 − 1| = |x4 − 1|. (5)
Next, we’ll discuss five cases:
Case 1: All xi ≥ 1. Then from (5),
x1 = x2 = x3 = x4 .
From (1), we have x1 + x31 = 2, so (x1 − 1)(x21 + x1 + 2) = 0. Since
x21 + x1 + 2 is always greater than 0, we see that x1 = x2 = x3 = x4 = 1.
Case 2: Exactly three xi ≥ 1. Without loss of generality, assume x1 < 1
and x2 = x3 = x4 ≥ 1. Then
1 − x1 = x2 − 1 = x3 − 1 = x4 − 1.
Therefore, x2 = x3 = x4 = 2 − x1 . From (1),
2 − x2 + x32 = 2,
which simplifies to x2 (x22 − 1) = 0. Solving this gives x2 = 1 = x1 , leading
to a contradiction.
Case 3: Exactly two xi ≥ 1. Without loss of generality, assume x1 , x2 <
1 ≤ x3 , x4 . From (5), we can deduce x1 = x2 and x3 = x4 = 2 − x1 .
From (3),
2 − x1 + x21 (2 − x1 ) = 2,
which can be further simplified to x1 (x1 − 1)2 = 0. Solving this equation
yields x1 = 1, leading to a contradiction.
Case 4: Exactly one xi ≥ 1. Without loss of generality, assume
x1 , x2 , x3 < 1 ≤ x4 . Then from (5),
x1 = x2 = x3 = 2 − x4 .
Now from (4),
2 − x1 + x31 = 2,
 Equation Problems 35

which simplifies to x1 (x21 − 1) = 0. Solving this equation gives x1 =

x2 = x3 = −1 and x4 = 3. Similarly, there are four sets of solutions:
(−1, −1, −1, 3), (−1, −1, 3, −1), (−1, 3, −1, −1), and (3, −1, −1, −1).

Case 5: All xi < 1. Then from (5),

x1 = x2 = x3 = x4 .

It is easy to see that x1 = 1, which contradicts the condition that x1 < 1.

In summary, there are totally five solutions: (1, 1, 1, 1), (−1, −1, −1, 3),
(−1, −1, 3, −1), (−1, 3, −1, −1), and (3, −1, −1, −1).

Solution 2. Using P to represent x1 x2 x3 x4 , we see that each equation has

the same form (as shown in Solution 1, P = 0):

P
xi + = 2 (i = 1, 2, 3, 4),
xi
√
which implies xi = 1 ± 1 − P (i = 1, 2, 3, 4). It is easy to see that P ≤ 1.

Case 1: If P = 1, then x1 = x2 = x3 = x4 = 1.

Case 2: If P < 1 and x1 , x2 , x3 , x4 are not all equal, then we discuss the
following three cases.
√ √
(i) xi (i = 1, 2, 3, 4) include two 1 + 1 − P and two 1 − 1 − P . Then

P = x1 x2 x3 x4
√ √
= (1 + 1 − P )2 (1 − 1 − P )2
= P 2.

This leads to P = 1, which is a contradiction.

√ √
(ii) xi (i = 1, 2, 3, 4) include three 1 + 1 − P and one 1 − 1 − P . Then

P = x1 x2 x3 x4
√ √
= (1 + 1 − P )3 (1 − 1 − P )
√
= P (1 + 1 − P )2 .

We still have P = 1, which is a contradiction.

36 IMO Problems, Theorems, and Methods: Algebra

√ √
(iii) xi (i = 1, 2, 3, 4) include one 1 + 1 − P and three 1 − 1 − P . Then

P = x1 x2 x3 x4
√ √
= (1 + 1 − P )(1 − 1 − P )3
√
= P (1 − 1 − P )2 .

We can solve out P = −3, and consequently, one of the xi (i = 1, 2, 3, 4)

is 3, while the other three are −1.
Note. This problem is the same as the seventh problem in 1967 British
Mathematical Olympiad Round 1. Furthermore, there are several similar
problems:
• (German Mathematical Olympiad 2023, Final Round, Prob-
lem 4). Determine all triples (a, b, c) of real numbers with
4 4 4
a+ =b+ =c+ .
b c a
• (Junior Balkan Mathematical Olympiad 2018, Problem 3). Let
k > 1 be a positive integer and n > 2018 be an odd positive integer.
Suppose that nonzero rational numbers x1 , x2 , . . . , xn are not all equal
and satisfy
k k k k k
x1 + = x2 + = x3 + = · · · = xn−1 + = xn + .
x2 x3 x4 xn x1
Find:
(a) the product x1 x2 · · · xn as a function of k and n;
(b) the least value of k, such that there exist n, x1 , x2 , . . . , xn satisfying
the given conditions.
• (Vojtěch Jarnı́k International Mathematical Competition 2017,
Category II, Problem 3). Let n ≥ 2 be an integer. Consider the
system of equations
2 2 2 2
x1 + = x2 + = · · · = xn−1 + = xn + . (1)
x2 x3 xn x1
(a) Prove that (1) has infinitely many real solutions (x1 , x2 , . . . , xn ) such
that the numbers x1 , x2 , . . . , xn are distinct.
(b) Prove that every solution (x1 , x2 , . . . , xn ) of (1), such that the num-
n
bers x1 , x2 , . . . , xn are not all equal, satisfies |x1 x2 · · · xn | = 2 2 .
 Equation Problems 37

• (Irish Mathematical Olympiad 2014, Problem 4). Three different

nonzero real numbers a, b, c satisfy the equations
2 2 2
a+ = b + = c + = p,
b c a
where p is a real number. Prove that abc + 2p = 0.
• (British Mathematical Olympiad 2008, 1st Round, Problem 2).
Find all real values of x, y, and z such that

(x + 1)yz = 12, (y + 1)zx = 4, and (z + 1)xy = 4.

• (All-Russian Mathematical Olympiad 1993, 4th Round, Grade

10, Problem 3). Solve for positive numbers the system
1 1 1
x1 + = 4, x2 + = 1, x3 + = 4,
x2 x3 x4
1 1 1
x4 + = 1, . . . , x99 + = 4, x100 + = 1.
x5 x100 x1
• (Canadian Mathematical Olympiad 1970, Problem 1). Find all
number triples (x, y, z) such that when any one of these numbers is added
to the product of the other two, the result is 2.

Score Situation This particular problem saw the following distribution of scores among
contestants: 37 contestants scored 6 points, 6 contestants scored 5 points, 7 contestants
scored 4 points, 4 contestants scored 3 points, 19 contestants scored 2 points, 5 contestants
scored 1 point, and 2 contestants scored 0 point. The average score for this problem is 4.188,
indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 43 points (with a total team score of 281 points), the Hungary
team scored 38 points (with a total team score of 244 points), the Romania team scored 39
points (with a total team score of 222 points), the Poland team scored 44 points (with a
total team score of 178 points), and the German Democratic Republic team scored 34 points
(with a total team score of 175 points).
The gold medal cutoff for this IMO was set at 38 points (with 8 contestants earning gold
medals), the silver medal cutoff was 30 points (with 12 contestants earning silver medals),
and the bronze medal cutoff was 20 points (with 17 contestants earning bronze medals).
In this IMO, only two contestants achieved a perfect score of 40 points, namely László
Lovász from Hungary and Pavel Bleher from the Soviet Union.
38 IMO Problems, Theorems, and Methods: Algebra

1.2.2 Finding solutions (with parameters)

Problem 1.4 (IMO 3-3, proposed by Bulgaria). Solve the equation
cosn x − sinn x = 1, where n is a positive integer.
Solution. When n is even, denote n = 2m. We have

cos2m x = 1 + sin2m x.

Since cos2m x ≤ 1 ≤ 1 + sin2m x, it can only be the case that

sin x = 0, cos x = ±1.

In this situation, x is of the form kπ, where k is an integer.

When n is odd and n ≥ 3, we can observe from

1 = cosn x − sinn x ≤ cos2 x + sin2 x = 1

that the equality holds if and only if cos x = 1 or sin x = −1. Consequently,
x = 2kπ or x = 2kπ − π2 , where k is an integer.
When n = 1, the original equation
√ becomes cos x − sin x = 1, which can
be rewritten as cos(x + 4 ) = 2 . This also leads to x = 2kπ or x = 2kπ− π2 ,
π 2

where k is an integer.
Note. There is a similar problem:

• (All-Russian Mathematical Olympiad 2002, Final Round,

Grade 11, Problem 3). Prove that if 0 < x < π2 and n > m, where
n, m are positive integers, then

2|sinn x − cosn x| ≤ 3|sinm x − cosm x|.

Score Situation This particular problem saw the following distribution of scores among
contestants: 1 contestant scored 7 points, 1 contestant scored 6 points, 1 contestant scored
5 points, 1 contestant scored 4 points, no contestant scored 3 points, no contestant scored
2 points, 1 contestant scored 1 point, and 4 contestants scored 0 point. The average score
of this problem is 2.556, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the Hungary team achieved a total score of
270 points, the Poland team achieved a total score of 203 points, the Romania team achieved
a total score of 197 points, the Czechoslovakia team achieved a total score of 159 points,
and the German Democratic Republic team achieved a total score of 146 points.
The gold medal cutoff for this IMO was set at 37 points (with 3 contestants earning gold
medals), the silver medal cutoff was 34 points (with 4 contestants earning silver medals), and
the bronze medal cutoff was 30 points (with 4 contestants earning bronze medals).
 Equation Problems 39

In this IMO, only one contestant achieved a perfect score of 40 points, namely Béla
Bollobás from Hungary.

Problem 1.5 (IMO 5-1, proposed by Czechoslovakia). Find all real

solutions of the equation

x2 − p + 2 x2 − 1 = x,

where p is a real parameter.

Solution. If p < 0, then

x2 − p + 2 x2 − 1 ≥ x2 − p > x.

In this case, the equation has no solution. Therefore, we can assume p ≥ 0,

x ≥ 1, and x2 ≥ p.
The original equation can be rewritten as

2 x2 − 1 = x − x2 − p.

Squaring and rearranging, we get

2x2 + p − 4 = −2x x2 − p,

and squaring both sides again, we obtain

8(2 − p)x2 = (p − 4)2 .

It’s easy to see that p must satisfy 0 ≤ p < 2, and in this case, there is
a solution
4−p
x= .
8(2 − p)
Substituting this into the original equation and simplifying, we get
|3p − 4| = 4 − 3p, which implies p ≤ 43 .
In conclusion, the equation has a√ unique solution if and only if
0 ≤ p ≤ 43 , and the solution is x = (4−p) 4−2p
8−4p .
Note. There is a similar problem:

• (Canadian Mathematical Olympiad 1998, Problem 2). Find all

real numbers x such that
 
1 1
x= x− + 1− .
x x
40 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 2 contestants scored 6 points, 3 contestants scored 5 points, 4 contestants
scored 4 points, 1 contestant scored 3 points, 5 contestants scored 2 points, no contestant
scored 1 point, and 1 contestant scored 0 point. The average score of this problem is 3.500,
indicating that it was relatively straightforward.
Among the top five teams in the team scores, the Soviet Union team achieved a total
score of 271 points, the Hungary team achieved a total score of 234 points, the Romania
team achieved a total score of 191 points, the Yugoslavia team achieved a total score of 162
points, and the Czechoslovakia team achieved a total score of 151 points.
The gold medal cutoff for this IMO was set at 35 points (with 7 contestants earning gold
medals), the silver medal cutoff was 28 points (with 11 contestants earning silver medals),
and the bronze medal cutoff was 21 points (with 17 contestants earning bronze medals).
In this IMO, no contestant achieved a perfect score of 40 points.

Problem 1.6 (IMO 5-4, proposed by the Soviet Union). Find all
solutions x1 , x2 , x3 , x4 , x5 of the system
⎧
⎪ x + x2 = yx1 ,
⎪ 5
⎪
⎪
⎪
⎪
⎨x1 + x3 = yx2 ,
x2 + x4 = yx3 ,
⎪
⎪
⎪
⎪
⎪
⎪
x3 + x5 = yx4 ,
⎩
x4 + x1 = yx5 ,
where y is a parameter.
Solution. First, we sum up all five equations to obtain
(x1 + x2 + x3 + x4 + x5 )(y − 2) = 0,
which implies that either x1 + x2 + x3 + x4 + x5 = 0 or y = 2.
If y = 2, then we can deduce from the original equation:
(x1 + x3 ) + (x2 + x4 ) = 2x2 + 2x3 ⇒ x1 + x4 = x2 + x3 ,
and x1 + x4 = 2x5 . Consequently,
x1 + x2 + x3 + x4 + x5 = 5x5 .
Furthermore, x1 + x2 + x3 + x4 + x5 = 5xi (i = 1, 2, 3, 4, 5), which implies
that
x1 = x2 = x3 = x4 = x5 = k,
where k can be any real number.
 Equation Problems 41

If y = 2, then

(x5 + x2 ) − (x3 + x5 ) = yx1 − yx4 ⇒ x2 − x3 = y(x1 − x4 )

and x1 + x4 = yx5 = y(yx1 − x2 ). These two equations, along with x1 +

x3 = yx2 and x2 + x4 = yx3 , can be combined to eliminate x4 , resulting in
the following system:

yx1 − x2 + x3 − y 2 x3 + yx2 = 0,
(1 − y 2 )x1 + yx2 + yx3 − x2 = 0,

and substituting x3 = yx2 − x1 and simplifying, we have

(y 2 + y − 1)x1 + (−y 3 + 2y − 1)x2 = 0,

(1 − y − y 2 )x1 + (y 2 + y − 1)x2 = 0.

Since −y 3 + 2y − 1 = (y 2 + y − 1)(1 − y), we can rewrite the above

equations as

(y 2 + y − 1)(x1 + (1 − y)x2 ) = 0,
(y 2 + y − 1)(x2 − x1 ) = 0.

If y 2 + y − 1 = 0, then x1 = x2 and x1 + (1 − y)x2 = (2 − y)x1 = 0.

Assuming y = 2, we find x1 = x2 = 0, and it can be further shown that
x3 = x4 = x5 = 0, which is the trivial solution.
If y 2 + y − 1 = 0, then x1 and x2 can take any values. And x3 , x4 , x5
can be expressed in terms of x1 , x2 , y as

x3 = yx2 − x1 , x4 = yx3 − x2 = y(yx2 − x1 ) − x2 = −yx2 − yx1 ,

x5 = yx1 − x2 ,
√
where y = 12 (−1 ± 5).
Score Situation This particular problem saw the following distribution of scores among
contestants: no contestant scored 6 points, 1 contestant scored 5 points, no contestant
scored 4 points, 1 contestant scored 3 points, 4 contestants scored 2 points, 6 contestants
scored 1 point, and 4 contestants scored 0 point. The average score of this problem is 1.375,
indicating that it was relatively challenging.
Among the top five teams in the team scores, the Soviet Union team achieved a total
score of 271 points, the Hungary team achieved a total score of 234 points, the Romania
team achieved a total score of 191 points, the Yugoslavia team achieved a total score of 162
points, and the Czechoslovakia team achieved a total score of 151 points.
42 IMO Problems, Theorems, and Methods: Algebra

The gold medal cutoff for this IMO was set at 35 points (with 7 contestants earning gold
medals), the silver medal cutoff was 28 points (with 11 contestants earning silver medals),
and the bronze medal cutoff was 21 points (with 17 contestants earning bronze medals).
In this IMO, no contestant achieved a perfect score of 40 points.
Problem 1.7 (IMO 8-5, proposed by Czechoslovakia). Solve the
system of equations
⎧
⎪
⎪ |a1 − a2 |x2 + |a1 − a3 |x3 + |a1 − a4 |x4 = 1,
⎪
⎪
⎨|a − a |x + |a − a |x + |a − a |x = 1,
2 1 1 2 3 3 2 4 4
⎪
⎪ |a3 − a1 |x1 + |a3 − a2 |x2 + |a3 − a4 |x4 = 1,
⎪
⎪
⎩
|a4 − a1 |x1 + |a4 − a2 |x2 + |a4 − a3 |x3 = 1,
where a1 , a2 , a3 , a4 are four different real numbers.
Solution. If we swap the subscripts of ai and aj , then the original system
remains unchanged. Therefore, we can assume a1 > a2 > a3 > a4 , and the
original system becomes
⎧
⎪ (a1 − a2 )x2 + (a1 − a3 )x3 + (a1 − a4 )x4 = 1, (1)
⎪
⎪
⎪
⎪
⎨ (a1 − a2 )x1 + (a2 − a3 )x3 + (a2 − a4 )x4 = 1, (2)
⎪
⎪ (a1 − a3 )x1 + (a2 − a3 )x2 + (a3 − a4 )x4 = 1, (3)
⎪
⎪
⎪
⎩
(a1 − a4 )x1 + (a2 − a4 )x2 + (a3 − a4 )x3 = 1. (4)
(1)–(2) yields (a1 − a2 )(x3 + x2 + x4 − x1 ) = 0.
(2)–(3) yields (a2 − a3 )(x3 + x4 − x1 − x2 ) = 0.
(3)–(4) yields (a3 − a4 )(x4 − x1 − x2 − x3 ) = 0.
Since ai are all distinct, we have
⎧
⎨x2 + x3 + x4 − x1 = 0,
⎪
x3 + x4 − x1 − x2 = 0,
⎪
⎩
x4 − x1 − x2 − x3 = 0.
1
Solving this system results in x1 = x4 = a1 −a4 and x2 = x3 = 0.
Note. Using the same method, we can solve a system of n equations:
⎧
⎪
⎪ |a1 − a2 |x2 + |a1 − a3 |x3 + · · · + |a1 − an |xn = 1,
⎪
⎪
⎨|a2 − a1 |x1 + |a2 − a3 |x3 + · · · + |a2 − an |xn = 1,
⎪· · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ·
⎪
⎪
⎪
⎩
|an − a1 |x1 + |an − a2 |x2 + · · · + |an − an−1 |xn−1 = 1.
1
If a1 > a2 > · · · > an , then the solution is x1 = xn = a1 −an and x2 = x3 =
· · · = xn−1 = 0.
 Equation Problems 43

Score Situation This particular problem saw the following distribution of scores among
contestants: 15 contestants scored 7 points, no contestant scored 6 points, 1 contestant
scored 5 points, 3 contestants scored 4 points, 1 contestant scored 3 points, 4 contestants
scored 2 points, 2 contestants scored 1 point, and 1 contestant scored 0 point. The average
score of this problem is 5.000, indicating that it was simple.
Among the top five teams in the team scores, the Soviet Union team achieved a total
score of 293 points, the Hungary team achieved a total score of 281 points, the German
Democratic Republic team achieved a total score of 280 points, the Poland team achieved a
total score of 269 points, and the Romania team achieved a total score of 257 points.
The gold medal cutoff for this IMO was set at 39 points (with 13 contestants earning gold
medals), the silver medal cutoff was 34 points (with 15 contestants earning silver medals),
and the bronze medal cutoff was 31 points (with 11 contestants earning bronze medals).
In this IMO, a total of 11 contestants achieved a perfect score of 40 points.

1.2.3 Proving relationships

Problem 1.8 (IMO 7-2, proposed by Poland). Consider the system
of equations
⎧
⎪
⎨a11 x1 + a12 x2 + a13 x3 = 0,
a21 x1 + a22 x2 + a23 x3 = 0,
⎪
⎩
a31 x1 + a32 x2 + a33 x3 = 0,
with unknowns x1 , x2 , x3 . The coefficients satisfy the conditions:
(a) a11 , a22 , a33 are positive numbers;
(b) the remaining coefficients are negative numbers;
(c) in each equation, the sum of the coefficients is positive.
Prove that the given system has only the solution x1 = x2 = x3 = 0.
Proof 1. Assume that the given system of equations has a non-trivial
solution, with some xi (i = 1, 2, 3) having the largest absolute value, say
i = 2, so that x2 = 0, |x2 | ≥ |x1 |, and |x2 | ≥ |x3 |.
According to the given conditions,
a22 |x2 | = |−a21 x1 − a23 x3 |
≤ −a21 |x1 | − a23 |x3 |
≤ −a21 |x2 | − a23 |x2 |.
Since x2 = 0, we have a22 ≤ −a21 − a23 , which contradicts the condition
a22 + a21 + a23 > 0. Thus, the original system of equations has the unique
solution x1 = x2 = x3 = 0.
44 IMO Problems, Theorems, and Methods: Algebra

Proof 2. A linear homogeneous system of equations has the unique zero

solution only when the determinant of the coefficient matrix is non-zero.
Let’s prove that
 
a11 a12 a13 
 
 
D = a21 a22 a23  = 0.
 
a a a 
31 32 33

The last column can be replaced with the sum of all three columns,
resulting in
 
a11 a12 S1 

 
D = a21 a22 S2  ,
 
a a32 S3 
31

where Si = ai1 + ai2 + ai3 > 0 (i = 1, 2, 3).

Using a cofactor expansion, we have
     
a  a  a a12 
 21 a22   11 a12   11
D = S1   − S2   + S3  
a31 a32  a31 a32  a21 a22 
= S1 (a21 a32 − a22 a31 ) + S2 (a12 a31 − a11 a32 ) + S3 (a11 a22 − a12 a21 ).

Note that the first two terms in this expression are positive, from con-
ditions (a) and (b). It is also apparent that from conditions (b) and (c),

a11 + a12 > a11 + a12 + a13 > 0 ⇒ a11 > −a12 = |a12 |,
a22 + a21 > a22 + a21 + a23 > 0 ⇒ a22 > −a21 = |a21 |.

Therefore a11 a22 > a12 a21 , making the third term in the above expres-
sion positive. Hence, D > 0, and as a result, the original system of equa-
tions has only the trivial solution x1 = x2 = x3 = 0.

Note. This problem is a special case of Hadamard’s theorem.

◦ Hadamard’s Theorem. For an n × n matrix A = (aij )n×n , if the

following conditions are satisfied:

(i) aij are complex numbers, where i, j ∈ {1, 2, . . . , n};

n
(ii) |aii | > j=1 |aij | for i = 1, 2, . . . , n,
j=i

then the determinant of A satisfies |A| = 0.

 Equation Problems 45

Furthermore, there is a similar problem:

• (Germany Team Selection Test 2004, Problem 7). Let aij , where
i, j ∈ {1, 2, 3}, be real numbers such that aij is positive for i = j and
negative for i = j. Prove the existence of positive real numbers c1 , c2 , c3
such that the numbers

a11 c1 + a12 c2 + a13 c3 , a21 c1 + a22 c2 + a23 c3 , a31 c1 + a32 c2 + a33 c3

are either all negative, all positive, or all zero.

Score Situation This particular problem saw the following distribution of scores among
contestants: 19 contestants scored 6 points, 6 contestants scored 5 points, 4 contestants
scored 4 points, 6 contestants scored 3 points, 10 contestants scored 2 points, 18 contestants
scored 1 point, and 17 contestants scored 0 point. The average score for this problem is
2.700, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 46 points (with a total team score of 281 points), the Hungary
team scored 36 points (with a total team score of 244 points), the Romania team scored 35
points (with a total team score of 222 points), the Poland team scored 25 points (with a
total team score of 178 points), and the German Democratic Republic team scored 17 points
(with a total team score of 175 points).
The gold medal cutoff for this IMO was set at 38 points (with 8 contestants earning gold
medals), the silver medal cutoff was 30 points (with 12 contestants earning silver medals),
and the bronze medal cutoff was 20 points (with 17 contestants earning bronze medals).
In this IMO, only two contestants achieved a perfect score of 40 points, namely László
Lovász from Hungary and Pavel Bleher from the Soviet Union.

Problem 1.9 (IMO 10-3, proposed by Bulgaria). Consider the sys-

tem of equations
⎧ 2
⎪
⎪
ax1 + bx1 + c = x2 ,
⎪
⎪
⎪
⎪
2
⎨ax2 + bx2 + c = x3 ,
·····················
⎪
⎪
⎪
⎪ ax2n−1 + bxn−1 + c = xn ,
⎪
⎪
⎩ 2
axn + bxn + c = x1 ,

with unknowns x1 , x2 , . . . , xn , where a, b, c are real and a = 0. Let Δ =

(b − 1)2 − 4ac.
46 IMO Problems, Theorems, and Methods: Algebra

Prove that for this system,

(a) if Δ < 0, then there is no solution;
(b) if Δ = 0, then there is exactly one solution;
(c) if Δ > 0, then there is more than one solution.

Proof. Consider the quadratic function Q(x) = ax2 + (b − 1)x + c. It is

clear that the discriminant of Q(x) is Δ = (b − 1)2 − 4ac, and its graph is
a parabola.
Summing up the equations of the original system, we get
n
Q(xi ) = 0.
i=1
(i) If Δ < 0, then Q(x) > 0(a > 0) or Q(x) < 0(a < 0), implying Q(xi ) >

0(a > 0) or Q(xi ) < 0(a < 0). It is impossible to have ni=1 Q(xi ) = 0.
Therefore, there is a contradiction, and the original system has no
solution.
(ii) If Δ = 0, then Q(x) ≥ 0(a > 0) or Q(x) ≤ 0(a < 0), implying Q(xi ) ≥
n
0(a > 0) or Q(xi ) ≤ 0(a < 0). To have i=1 Q(xi ) = 0, we must have
Q(xi ) = 0, meaning x1 = x2 = · · · = xn = r. Here, r is the unique
root of Q(x) = 0.
(iii) If Δ > 0, then Q(x) = 0 has two roots r1 and r2 . It is evident that
x1 = x2 = · · · = xn = r1 and x1 = x2 = · · · = xn = r2 are both
solutions of the original system.
Note. There are several similar problems:
• (All-Russian Mathematical Olympiad 2011, Grade 10, Problem
2). Nine quadratics are written on the board:
x2 + a1 x + b1 , x2 + a2 x + b2 , . . . , x2 + a9 x + b9 .
It is known that a1 , a2 , . . . , a9 and b1 , b2 , . . . , b9 are both arithmetic
sequences, and the sum of these nine quadratics has at least one real
root. What is the maximum number of these original nine quadratics
that can have no real roots?
• (Mexican Mathematical Olympiad 2011, Problem 3). Let n be a
positive integer. Find all real solutions (a1 , a2 , . . . , an ) to the system:
⎧ 2
⎪
⎪ a1 + a1 − 1 = a2 ,
⎪
⎪
⎨a2 + a − 1 = a ,
2 2 3
⎪
⎪ ··················
⎪
⎪
⎩ 2
an + an − 1 = a1 .
 Equation Problems 47

• (All-Russian Mathematical Olympiad 2009, Regional Round,

Grade 10, Problem 7). Let x1 , x2 , . . . , x2009 be positive real numbers
satisfying

x21 − x1 x2 + x22 = x22 − x2 x3 + x23 = · · · = x22008 − x2008 x2009 + x22009

= x22009 − x2009 x1 + x21 .

Prove that x1 = x2 = · · · = x2009 .

• (Austrian–Polish Mathematical Competition 2001, Problem 2).
Given an integer n > 2, solve in non-negative real numbers the system

xk + xk+1 = x2k+2 , k = 1, 2, . . . , n,

where xn+i = xi .
• (Polish Mathematical Olympiad 2000, Problem 1). For a given
integer n ≥ 2, find the number of non-negative real solutions of the
system of equations:
⎧
⎪ x1 + x2n = 4xn ,
⎪
⎪
⎪
⎪ 2
⎪
⎨x2 + x1 = 4x1 ,
x3 + x22 = 4x2 ,
⎪
⎪
⎪
⎪ ··················
⎪
⎪
⎩
xn + x2n−1 = 4xn−1 .

• (Canadian Mathematical Olympiad 1996, Problem 2). Find all

real solutions to the following system of equations:
⎧ 2
⎪ 4x
⎪
⎪ 1+4x2 = y,
⎨ 2
4y
⎪ 1+4y 2 = z,
⎪
⎪
⎩ 4z2
1+4z 2 = x.

• (Turkey Team Selection Test 1995, Problem 1). Given real num-
bers b ≥ a > 0, find all solutions of the system
⎧ 2
⎪ x1 + 2ax1 + b2 = x2 ,
⎪
⎪
⎪
⎪ 2 2
⎪
⎨x2 + 2ax2 + b = x3 ,
························
⎪
⎪
⎪
⎪ x2n−1 + 2axn−1 + b2 = xn ,
⎪
⎪
⎩ 2
xn + 2axn + b2 = x1 .
48 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 38 contestants scored 7 points, 9 contestants scored 6 points, 1 contestant
scored 5 points, 2 contestants scored 4 points, no contestant scored 3 points, 7 contestants
scored 2 points, 21 contestants scored 1 point, and 13 contestants scored 0 point. The
average score of this problem is 4.044, indicating that it was simple.
Among the top five teams in the team scores, the German Democratic Republic team
achieved a total score of 304 points, the Soviet Union team achieved a total score of 298
points, the Hungary team achieved a total score of 291 points, the United Kingdom team
achieved a total score of 263 points, and the Poland team achieved a total score of 262
points.
The gold medal cutoff for this IMO was set at 39 points (with 22 contestants earning gold
medals), the silver medal cutoff was 33 points (with 22 contestants earning silver medals),
and the bronze medal cutoff was 26 points (with 20 contestants earning bronze medals).
In this IMO, a total of 16 contestants achieved a perfect score of 40 points.

Problem 1.10 (IMO 18-2, proposed by Finland). Let p1 (x) = x2 − 2

and Pj (x) = P1 (Pj−1 (x)) for j = 2, 3, . . . . Show that, for any positive
integer n, the roots of the equation Pn (x) = x are real and distinct.
Proof 1. It is evident that pn (x) = x is a polynomial of degree 2n , implying
it has at most 2n distinct roots. Moreover, p1 (x) > x when x > 2, and
p2 (x) = p1 (p1 (x)) > p1 (x) > x.
Similarly, pn (x) > x > 2, and pn (x) > 2 > x when x < −2. Thus
the real roots of pn (x) = x lie within the closed interval [−2, 2]. Therefore,
we can assume that the roots of the equation pn (x) = x take the form
x = 2 cos t. Consequently,
p1 (2 cos t) = 4 cos2 t − 2 = 2 cos 2t,
p2 (2 cos t) = p1 (2 cos 2t) = 2 cos 4t.
In general, pn (2 cos t) = 2 cos 2n t. For the equation 2 cos 2n t = 2 cos t,
the following solutions exist:
2kπ
tk = , k = 0, 1, . . . , 2n−1 − 1,
2n − 1
2lπ
sl = n , l = 1, 2, . . . , 2n−1 ,
2 +1
leading to 0 = t0 < t1 < · · · < t2n−1 −1 < π, and 0 < s1 < s2 < · · · <
s2n−1 < π.
 Equation Problems 49

Since the cosine function is strictly monotonic on [0, π], the values

2 cos tk (k = 0, 1, . . . , 2n−1 − 1)

are 2n−1 distinct real numbers. Similarly, 2 cos sl (l = 1, 2, . . . , 2n−1 ) also

constitute 2n−1 distinct real numbers.
Moreover, for any k and l within the aforementioned range, it is impos-
sible to have tk = sl . Otherwise, there exists the equation

2n − 1 k
= .
2n + 1 l

Since k < 2n − 1 and l < 2n + 1, it indicates that the fraction on the left
side of the equation has not been reduced to its simplest form. However,
2n − 1 and 2n + 1 are consecutive odd numbers and coprime, meaning they
cannot have a common factor greater than 1. This implies that 2 cos tk and
2 cos sl are all 2n distinct real roots of the equation pn (x) = x.

Proof 2. It is easy to observe that the polynomial

pn (x) = (· · · ((x2 − 2)2 − 2)2 − · · ·)2 − 2

is of degree 2n and an even function, i.e., pn (−x) = pn (x).

As shown in Figure 1.1, p1 maps the interval [−2, 2] onto the interval
[−2, 2]. When x increases from −2 to 0, we see that p1 (x) decreases from 2
to −2. Furthermore, since p1 (x) is an even function, p1 (x) increases from
−2 to 2 when x increases from 0 to 2.

y y=x

2
p2
1
p1
–2 –1 O 1 2 x
–1

Figure 1.1 Graph of the Function P n (x)

50 IMO Problems, Theorems, and Methods: Algebra

Next, consider p2 (x) = p1 (p1 (x)). When x increases from −2 to 0,

clearly p1 (x) decreases from 2 to −2. Subsequently, p1 (p1 (x)) maps these
values onto the interval [−2, 2], and the graph for the interval −2 ≤ x ≤ 0
is symmetric about the y-axis with the graph for the interval 0 ≤ x ≤ 2.
Similarly, in the graph of p3 (x), the number of times its ordinate
decreases from 2 to −2 or increases from −2 to 2 is twice that of p2 (x),
and so forth. Therefore, in the graph of pn (x), the curve of pn (x) repeats
between −2 and 2 2n times.
Furthermore, the solutions of pn (x) = x correspond to the abscissas of
the intersection points of pn (x) and the line y = x. When −2 < x < 2,
this line precisely corresponds to the diagonal of the 4 × 4 square. As pn (x)
oscillates 2n times, there are 2n distinct intersection points, i.e., 2n distinct
real roots.
Moreover, since the degree of pn (x) is 2n , the number of roots of
pn (x) = x is no more than 2n . Therefore, all its roots are distinct real
numbers.

Note. The nth Chebyshev polynomial Tn (x), to be presented in Chap-

ter 5, is defined by Tn (cos θ) = cos nθ. It is easy to verify that pn (x) =
2T2n ( x2 ).

Score Situation This particular problem saw the following distribution of scores among
contestants: 25 contestants scored 7 points, 5 contestants scored 6 points, 4 contestants
scored 5 points, no contestant scored 4 points, 4 contestants scored 3 points, 5 contestants
scored 2 points, 22 contestants scored 1 point, and 74 contestants scored 0 point. The
average score for this problem is 1.935, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 46 points (with a total team score of 250 points), the United
Kingdom team scored 35 points (with a total team score of 214 points), the United States
team scored 18 points (with a total team score of 188 points), the Bulgaria team scored 18
points (with a total team score of 174 points), and the Austria team scored 16 points (with
a total team score of 167 points).
The gold medal cutoff for this IMO was set at 34 points (with 9 contestants earning gold
medals), the silver medal cutoff was 23 points (with 28 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 45 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Laurent
Pierre from France.
 Equation Problems 51

Problem 1.11 (IMO 18-5, proposed by the Netherlands). Consider

the system of p equations in q = 2p unknowns x1 , x2 , . . . , xq :
⎧
⎪
⎪ a11 x1 + a12 x2 + · · · + a1q xq = 0,
⎪
⎪
⎨a x + a x + · · · + a x = 0,
21 1 22 2 2q q
⎪
⎪ · · · · · · · · · · · · · · · · · · · · · · · ··········
⎪
⎪
⎩
ap1 x1 + ap2 x2 + · · · + apq xq = 0,

with every coefficient aij a member of the set {−1, 0, 1}. Prove that the
system has a solution (x1 , x2 , . . . , xq ) satisfying:

(a) all xj (j = 1, 2, . . . , q) are integers;

(b) there is at least one value of j for which xj = 0;
(c) |xj | ≤ q(j = 1, 2, . . . , q).

Proof. Consider the arrays (x1 , x2 , . . . , xq ) substituted into the system,

satisfying |xj | ≤ p. According to the given conditions,

|bi | = |ai1 x1 + ai2 x2 + · · · + aiq xq | ≤ pq.

Thus, ai1 x1 +ai2 x2 +· · ·+aiq xq can take at most 2pq +1 possible values.
Consequently, the values corresponding to the left side of the system can
take at most (2pq + 1)p possible sets of integers (b1 , b2 , . . . , bp ).
For each xj , let it take 2p + 1 values ranging from −p to p, resulting in
(2p + 1)q possible sets of integers (b1 , b2 , . . . , bp ). Since

(2p + 1)q = (2p + 1)2p

= (4p2 + 4p + 1)p
> (4p2 + 1)p
= (2pq + 1)p ,

by the Pigeonhole Principle, there must exist two distinct arrays

(x1 , x2 , . . . , xq ) and (x1 , x2 , . . . , xq ) such that they correspond to the same
set of integers (b1 , b2 , . . . , bp ) on the left side of the system.
Consider the array (x1 − x1 , x2 − x2 , . . . , xq − xq ), which evidently sat-
isfies the conditions (a) and (b). Moreover, since |xj | ≤ p and |xj | ≤ p,
it follows that |xj − xj | ≤ 2p = q, and thus condition (c) is also satisfied.
Therefore, the proposition holds.
52 IMO Problems, Theorems, and Methods: Algebra

Note. This problem can also be solved from a matrix perspective. Let A
represent the coefficient matrix of the system of equations, 0 represent the
constant term vector, and X the unknown vector, that is, let
⎛ ⎞ ⎛ ⎞ ⎛ ⎞
a11 a12 · · · a1q 0 x1
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜a21 a22 · · · a2q ⎟ ⎜0 ⎟ ⎜x2 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
A=⎜ . . . ⎟, 0 = ⎜ . ⎟, X = ⎜ . ⎟.
⎜ .. .. .. ⎟ ⎜ .. ⎟ ⎜ .. ⎟
⎝ ⎠ ⎝ ⎠ ⎝ ⎠
ap1 ap2 · · · apq 0 xq

Then the original system of equations can be succinctly written as

AX = 0.
It is easily known that there are (2p + 1)q different integer vectors X
that satisfy |xj | ≤ p(j = 1, 2, . . . , q). For such X, it is evident that

|ai1 x1 + ai2 x2 + · · · + aiq xq | ≤ |ai1 x1 | + |ai2 x2 | + · · · + |aiq xq |

≤ |x1 | + |x2 | + · · · + |xq |
≤ pq.

Thus, the vector AX has at most (2pq + 1)p different values.

Since (2p + 1)q > (2pq + 1)p , there must exist two distinct vectors X 1
and X 2 such that AX 1 = AX 2 . Hence A(X 1 − X 2 ) = 0. Consequently,
the components of X 1 − X 2 are solutions.
Score Situation This particular problem saw the following distribution of scores among
contestants: 15 contestants scored 7 points, no contestant scored 6 points, no contestant
scored 5 points, 3 contestants scored 4 points, 2 contestants scored 3 points, 3 contestants
scored 2 points, 23 contestants scored 1 point, and 93 contestants scored 0 point. The
average score for this problem is 1.094, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 31 points (with a total team score of 250 points), the United
Kingdom team scored 11 points (with a total team score of 214 points), the United States
team scored 26 points (with a total team score of 188 points), the Bulgaria team scored 6
points (with a total team score of 174 points), and the Austria team scored 8 points (with a
total team score of 167 points).
The gold medal cutoff for this IMO was set at 34 points (with 9 contestants earning gold
medals), the silver medal cutoff was 23 points (with 28 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 45 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Laurent
Pierre from France.
 Equation Problems 53

1.2.4 Investigating conditions

Problem 1.12 (IMO 1-3, proposed by Hungary). Let a, b, c be real
numbers. Consider the quadratic equation in cos x:

a cos2 x + b cos x + c = 0.

Using the numbers a, b, c, form a quadratic equation in cos 2x, whose

roots are the same as those of the original equation. Compare the equations
in cos x and cos 2x for a = 4, b = 2, and c = −1.

Solution 1. Expressing the original equation as

a cos2 x + c = −b cos x,

squaring both sides, and multiplying by 4, we obtain

(a(1 + cos 2x) + 2c)2 = 2b2 (1 + cos 2x),

simplifying to

a2 cos2 2x + (2a2 + 4ac − 2b2 ) cos 2x + (a2 + 4ac − 2b2 + 4c2 ) = 0.

For the specific values a = 4, b = 2, and c = −1, the equation concerning

cos 2x becomes

4 cos2 2x + 2 cos 2x − 1 = 0.

This quadratic equation shares the same coefficients as the given equa-
tion involving cos x.

Solution 2. According to Vieta’s formulas, for the roots (which may be

identical) of the given equation, cos x1 +cos x2 = − ab and cos x1 ·cos x2 = ac .
Thus,

cos 2x1 + cos 2x2 = 2(cos2 x1 + cos2 x2 − 1)

= 2((cos x1 + cos x2 )2 − 2 cos x1 · cos x2 − 1)
 2 
b c
=2 − −2· −1
a a

2a2 + 4ac − 2b2

=− ,
a2
54 IMO Problems, Theorems, and Methods: Algebra

cos 2x1 · cos 2x2 = (2 cos2 x1 − 1)(2 cos2 x2 − 1)

= 4(cos x1 · cos x2 )2 − 2(cos2 x1 + cos2 x2 ) + 1
 2 
 c 2 b c
=4 −2 − −2· +1
a a a

a2 + 4ac − 2b2 + 4c2

= .
a2
Therefore, the required quadratic equation satisfied by cos 2x is

a2 cos2 2x + (2a2 + 4ac − 2b2 ) cos 2x + (a2 + 4ac − 2b2 + 4c2 ) = 0.

When a = 4, b = 2, and c = −1, this quadratic equation shares the

same coefficients as the given equation a cos2 x + b cos x + c = 0.

Score Situation This particular problem saw the following distribution of scores among
contestants: 4 contestants scored 7 points, no contestant scored 6 points, 2 contestants
scored 5 points, no contestant scored 4 points, no contestant scored 3 points, no contestant
scored 2 points, no contestant scored 1 point, and 3 contestants scored 0 point. The average
score of this problem is 4.222, indicating that it was simple.
Among the top five teams in the team scores, the Romania team achieved a total score
of 249 points, the Hungary team achieved a total score of 233 points, the Czechoslovakia
team achieved a total score of 192 points, the Bulgaria team achieved a total score of 131
points, and the Poland team achieved a total score of 122 points.
The gold medal cutoff for this IMO was set at 37 points (with 3 contestants earning gold
medals), the silver medal cutoff was 36 points (with 3 contestants earning silver medals), and
the bronze medal cutoff was 33 points (with 5 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Bohuslav
Diviš from Czechoslovakia.

Problem 1.13 (IMO 3-1, proposed by Hungary). Solve the system

of equations:
⎧
⎪
⎨x + y + z = a,
x2 + y 2 + z 2 = b2 ,
⎪
⎩
xy = z 2 ,

where a and b are constants. Give conditions that a and b must satisfy, so
that x, y, z (the solutions of the system) are distinct positive numbers.
 Equation Problems 55

Solution. Since the solutions of the system of equations are distinct pos-
itive numbers, it is evident that a > 0. Additionally, it is easy to observe
that
b2 = x2 + y 2 + z 2 = x2 + y 2 + 2xy − z 2 = (x + y + z)(x + y − z),
which leads to
b2
x+y−z = ,
a
(x + y + z) − (x + y − z) a2 − b 2
z= = ,
2 2a
a2 + b 2
x+y = ,
2a
(a2 − b2 )2
xy = .
4a2
a2 +b2 (a2 −b2 )2
Consider the quadratic equation f (x) = x2 − 2a x + 4a2 . Then,
1
Δ= ((a2 + b2 )2 − 4(a2 − b2 )2 )
4a2
1
= 2 (3a2 − b2 )(3b2 − a2 ) ≥ 0,
4a
and in this case, the system of equations has solutions
⎧ 1
⎪
⎪ x = 4a (a2 + b2 ± (3a2 − b2 )(3b2 − a2 )),
⎨
1
y = 4a (a2 + b2 ∓ (3a2 − b2 )(3b2 − a2 )),
⎪
⎪
⎩ 1
z = 2a (a2 − b2 ).
To ensure that the solutions are positive, z > 0, implying a2 > b2 .
Since x = y, we have Δ > 0. Additionally, as 3a2 − b2 > a2 − b2 > 0,
there holds 3b2 > a2 .
Therefore,
√
0 < |b| < a < 3|b|.
2 2
√
It is evident that a 4a
+b
> 2Δ . Thus, x > 0 and y > 0. Furthermore,
√
since x, y are distinct positive numbers, and z = xy, it is clear that x, y, z
are distinct positive numbers.
Note. From the given conditions, it is known that the distance from the
plane x + y + z = a > 0 to the origin is √a3 > 0. The radius of the sphere
56 IMO Problems, Theorems, and Methods: Algebra

centered at the origin with the equation x2 + y 2 + z 2 = b2 is |b|. The

plane and the sphere have a common point only when |b| ≥ √a3 . However,
√
when 3|b| = a, the plane and the sphere touch at a single
√ point, and the
coordinates of this point are equal. Therefore, 0 < a < 3|b| is a necessary
condition for x, y, z to be distinct positive numbers.
Furthermore, by using the AM-GM inequality
x+y √
> xy(x = y),
2
it is evident that
 
x+y 1 1 a2 − b 2 a2 − b 2 √
= (a − z) = a− > = z = xy.
2 2 2 2a 2a
√
This implies a2 < 3b2 . Given that a > 0, the condition 0 < a < 3|b| still
holds.
There are several similar problems:
• (British Mathematical Olympiad 2007, 1st Round, Problem 2).
Find all solutions in positive integers x, y, z to the simultaneous equations
x + y − z = 12 and x2 + y 2 − z 2 = 12.
• (British Mathematical Olympiad 1998, 2nd Round, Problem 4).
Find a solution of the simultaneous equations
xy + yz + zx = 12 and xyz = 2 + x + y + z,
in which all of x, y, z are positive, and prove that it is the only such
solution.
Show that a solution exists in which x, y, z are real and distinct.
• (British Mathematical Olympiad 1996, 2nd Round, Problem 4).
Let a, b, c, and d be positive real numbers such that
a + b + c + d = 12 and abcd = 27 + ab + ac + ad + bc + bd + cd.
Find all possible values of a, b, c, d satisfying these equations.
Score Situation This particular problem saw the following distribution of scores among
contestants: 1 contestant scored 6 points, 2 contestants scored 5 points, 2 contestants scored
4 points, 1 contestant scored 3 points, 1 contestant scored 2 points, no contestant scored 1
point, and 2 contestants scored 0 point. The average score of this problem is 3.222, indicating
that it was relatively straightforward.
Among the top five teams in the team scores, the Hungary team achieved a total score of
270 points, the Poland team achieved a total score of 203 points, the Romania team achieved
 Equation Problems 57

a total score of 197 points, the Czechoslovakia team achieved a total score of 159 points,
and the German Democratic Republic team achieved a total score of 146 points.
The gold medal cutoff for this IMO was set at 37 points (with 3 contestants earning gold
medals), the silver medal cutoff was 34 points (with 4 contestants earning silver medals), and
the bronze medal cutoff was 30 points (with 4 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Béla
Bollobás from Hungary.

Problem 1.14 (IMO 15-3, proposed by Sweden). Let a and b be real

numbers for which the equation

x4 + ax3 + bx2 + ax + 1 = 0

has at least one real solution. For all such pairs (a, b), find the minimum
value of a2 + b2 .

Solution 1. It is evident that x = 0 is not a root of the original equation.

Therefore, the original equation can be transformed into
 2  
1 1
x+ +a x+ + (b − 2) = 0.
x x

Let y = x + x1 . Then it is easy to observe that |y| ≥ 2. Solving the

equation y 2 + ay + (b − 2) = 0 yields

−a ± a2 − 4(b − 2)
y= ,
2
where at least one of the roots has an absolute value not less than 2. This
is equivalent to

|a| + a2 − 4(b − 2) ≥ 4,

or

a2 − 4(b − 2) ≥ 4 − |a|.

(i) When |a| ≥ 4, it follows that a2 + b2 ≥ 16.

(ii) When |a| < 4, squaring both sides and simplifying, we obtain

2|a| ≥ 2 + b.
58 IMO Problems, Theorems, and Methods: Algebra

If b ≤ −2, then a2 + b2 ≥ 4. Considering b > −2 and squaring both

sides again, we get
4a2 ≥ 4 + 4b + b2 ,
4a2 + 4b2 ≥ 5b2 + 4b + 4
 2
2 16
= 5 b+ +
5 5
16
≥ .
5
Therefore, when b = − 52 and a = ± 54 , the expression a2 + b2 attains
the minimum value of 45 . Upon verification, the original equation has real
roots x = ∓1.
Solution 2. It is evident that x = 0 is not a root of the original equation.
By the Cauchy−Schwarz inequality, we have
 
1 2 1 2
a + b + a (2x6 + x4 + 2x2 ) ≥ (ax3 + bx2 + ax)2 = (x4 + 1)2 .
2
2 2
Therefore,
2
2 2 (x4 + 1)2 x2 + x12
a +b ≥ 6 =
2x + x4 + 2x2 2 x2 + x12 + 1
 1 
1 2 1 1 4
= x + 2+ + 2 −1 .
2 x 2 x + x12 + 12
Since x2 + x12 + 12 ≥ 2 + 12 = 52 and a2 + b2 ≥ 12 ( 52 + 1
10 − 1) = 45 , the
equality holds when x = ∓1, b = − 52 , and a = ± 54 .
Note. In Solution 1, from |y| ≥ 2, it follows that
a2 − 4(b − 2) ≥ 4 − |a|.
Squaring both sides and rearranging yield b ≤ −2a − 2 or b ≤ 2a − 2.
Consequently, the pairs of numbers (a, b) that lead to solutions of the orig-
inal equation lie below the lines y = −2x − 2 or y = 2x − 2. The distance
from the origin to the lines y = −2x − 2 or y = 2x − 2 is exactly √25 . Thus,
√
a2 + b2 ≥ √25 .

Score Situation This particular problem saw the following distribution of scores among con-
testants: 42 contestants scored 8 points, 4 contestants scored 7 points, 8 contestants scored
6 points, 4 contestants scored 5 points, 4 contestants scored 4 points, 6 contestants scored
 Equation Problems 59

3 points, 9 contestants scored 2 points, 15 contestants scored 1 point, and 33 contestants

scored 0 point. The average score of this problem is 3.992, indicating that it was relatively
straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 53 points (with a total team score of 254 points), the Hungary
team scored 50 points (with a total team score of 215 points), the German Democratic
Republic team scored 31 points (with a total team score of 188 points), the Poland team
scored 16 points (with a total team score of 174 points), and the United Kingdom team
scored 38 points (with a total team score of 164 points).
The gold medal cutoff for this IMO was set at 35 points (with 5 contestants earning gold
medals), the silver medal cutoff was 27 points (with 15 contestants earning silver medals),
and the bronze medal cutoff was 17 points (with 48 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Sergei
Konyagin from the Soviet Union.

Problem 1.15 (IMO 21-5, proposed by Israel). Find all real numbers
a for which there exist non-negative real numbers x1 , x2 , x3 , x4 , x5 satisfying
the relations
5 5 5
kxk = a, k 3 xk = a2 , k 5 xk = a3 .
k=1 k=1 k=1

Solution 1. The relations can be rewritten as

5 5 5
a2 kxk + k 5 xk = 2a k 3 xk ,
k=1 k=1 k=1

and hence
5
k(a − k 2 )2 xk = 0.
k=1

This implies that each of the five non-negative terms is zero, i.e.,

k(a − k 2 )2 xk = 0, k = 1, 2, 3, 4, 5.

If x1 = x2 = x3 = x4 = x5 = 0, then a = 0; otherwise, xk = 0 and

a = k 2 for some k.
Thus, the possible values for a are 0, 1, 4, 9, 16, 25.
60 IMO Problems, Theorems, and Methods: Algebra

Solution 2. Using the Cauchy–Schwarz inequality, we have

 5
2
(a2 )2 = k 3 xk
k=1
 5
 5

5
≤ kxk k xk
k=1 k=1

= a4 .

√
The equality implies the existence of a real number λ such that k 5 xk =
√
λ kxk , or k 4 xk = λ2 xk . In other words,

(k 4 − λ2 )xk = 0, k = 1, 2, 3, 4, 5.

If xk = 0 for some k, then λ = k 2 . The remaining xi are all zero.

Replacing them into the original expression yields kxk = a and k 3 xk = a2 ,
implying a = k 2 = λ, while also satisfying k 5 xk = a3 .
The other case is xk = 0 for k = 1, 2, 3, 4, 5, leading to a = 0.
Thus, the possible values for a are 0, 1, 4, 9, 16, 25.

Score Situation This particular problem saw the following distribution of scores among
contestants: 46 contestants scored 7 points, 2 contestants scored 6 points, 7 contestants
scored 5 points, 4 contestants scored 4 points, 10 contestants scored 3 points, 31 contestants
scored 2 points, 42 contestants scored 1 point, and 24 contestants scored 0 point. The
average score for this problem is 3.127, indicating that it was relatively straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 41 points (with a total team score of 267 points), the Romania
team scored 40 points (with a total team score of 240 points), the Germany team scored 17
points (with a total team score of 235 points), the United Kingdom team scored 35 points
(with a total team score of 218 points), and the United States team scored 38 points (with
a total team score of 199 points).
The gold medal cutoff for this IMO was set at 37 points (with 8 contestants earning gold
medals), the silver medal cutoff was 29 points (with 32 contestants earning silver medals),
and the bronze medal cutoff was 20 points (with 42 contestants earning bronze medals).
In this IMO, a total of four contestants achieved a perfect score of 40 points.
 Equation Problems 61

Problem 1.16 (IMO 57-5, proposed by Russia). The equation

(x − 1)(x − 2) · · · (x − 2016) = (x − 1)(x − 2) · · · (x − 2016)
is written on the board, with 2016 linear factors on each side. What is the
least possible value of k for which it is possible to erase exactly k of these
4032 linear factors so that at least one factor remains on each side and the
resulting equation has no real solutions?
Solution. In order to ensure that the resulting equation has no real roots,
the same linear factor cannot appear on both sides of the equation. At least
one of them needs to be removed, requiring the removal of at least 2016
linear factors.
Next, it is explained that if all linear factors (x − k), where k ≡
2, 3(mod 4), are removed from the left side, and all linear factors (x − m),
where m ≡ 0, 1(mod 4), are removed from the right side, then the resulting
equation
503
 503

(x − 4j − 1)(x − 4j − 4) = (x − 4j − 2)(x − 4j − 3) (1)
j=0 j=0

will have no real roots.

Case 1: x ∈ {1, 2, . . . , 2016}.
In this case, one side of (1) is zero, and the other side is non-zero, making
(1) false.
Case 2: x ∈ (4k + 1, 4k + 2)∪(4k + 3, 4k + 4), where k ∈ {0, 1, 2, . . . , 503}.
For j ∈ {0, 1, 2, . . . , 503} and j = k, the inequalities
(x − 4j − 1)(x − 4j − 4) > 0 and (x − 4j − 2)(x − 4j − 3) > 0
hold.
If j = k, then
(x − 4j − 1)(x − 4j − 4) < 0 and (x − 4j − 2)(x − 4j − 3) > 0.
Multiplying these inequalities results in the left side of (1) being negative
and the right side being positive, making (1) false.
Case 3: x < 1 or x > 2016 or x ∈ (4k, 4k + 1), where k ∈ {1, 2, . . . , 503}.
For j ∈ {0, 1, 2, . . . , 503}, the inequalities
0 < (x − 4j − 1)(x − 4j − 4) < (x − 4j − 2)(x − 4j − 3)
are valid.
62 IMO Problems, Theorems, and Methods: Algebra

Multiplying these inequalities results in the left side of (1) being less
than the right side, making (1) false.
Case 4: x ∈ (4k + 2, 4k + 3), where k ∈ {0, 1, 2, . . . , 503}.
For j ∈ {1, 2, . . . , 503}, the inequalities

0 < (x − 4j + 1)(x − 4j − 2) < (x − 4j)(x − 4j − 1)

are true.
Additionally, x − 1 > x − 2 > 0 > x − 2015 > x − 2016. Multiplying
these inequalities results in
503
 503

(x − 4j − 1)(x − 4j − 4) < (x − 4j − 2)(x − 4j − 3) < 0.
j=0 j=0

Thus, (1) is false.

In conclusion, the minimum number of linear factors that need to be
removed is 2016.
Note. There are several similar problems:

• (Turkey Tean Selection Test 2023, Problem 8). Initially, the equa-
tion
1 1 1 1
   ··· =0
x−1 x−2 x−4 x − 22023
is written on the board. In each turn Asli and Zehra deletes one of the
stars in the equation and writes + or − instead. The first move is per-
formed by Asli and continues in order. What is the maximum number
of real solutions Asli can guarantee after all the stars have been replaced
by signs?
• (Chinese Team Selection Test 2014, Problem 15). Show that there
are no 2-tuples (x, y) of positive integers satisfying the equation

(x + 1)(x + 2) · · · (x + 2014) = (y + 1)(y + 2) · · · (y + 4028).

• (All-Russian Mathematical Olympiad 2010, Grade 11, Problem

1). Do there exist non-zero real numbers a1 , a2 , . . . , a10 for which
       
1 1 1 1 1
a1 + a2 + · · · a10 + = a1 − a2 − ···
a1 a2 a10 a1 a2
 
1
a10 − ?
a10
 Equation Problems 63

• (All-Russian Mathematical Olympiad 2005, Final Round,

Grade 11, Problem 1). Assume that the number of solutions N of
the equation

|x − a1 | + |x − a2 | + · · · + |x − a50 | = |x − b1 | + |x − b2 | + · · · + |x − b50 |

is finite, where ai , bj are distinct numbers. What is the greatest possible

value of N ?
• (All-Russian Mathematical Olympiad 2000, 4th Round, Grade
11, Problem 1). Prove that there exist distinct real numbers
a1 , a2 , . . . , a10 such that the equation

(x − a1 )(x − a2 ) · · · (x − a10 ) = (x + a1 )(x + a2 ) · · · (x + a10 )

has exactly five distinct real roots.

Score Situation This particular problem saw the following distribution of scores among
contestants: 81 contestants scored 7 points, 4 contestants scored 6 points, 2 contestants
scored 5 points, 50 contestants scored 4 points, 21 contestants scored 3 points, 55 contestants
scored 2 points, 36 contestants scored 1 point, and 353 contestants scored 0 point. The
average score for this problem is 1.678, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the United States team scored 41 points (with a total team score of 214 points), the South
Korea team scored 33 points (with a total team score of 207 points), the China team scored
42 points (with a total team score of 204 points), the Singapore team scored 42 points (with
a total team score of 196 points), and the Chinese Taiwan team scored 30 points (with a
total team score of 175 points).
The gold medal cutoff for this IMO was set at 29 points (with 44 contestants earning gold
medals), the silver medal cutoff was 22 points (with 101 contestants earning silver medals),
and the bronze medal cutoff was 16 points (with 135 contestants earning bronze medals).
In this IMO, a total of six contestants achieved a perfect score of 42 points.

1.3 Summary
In the first 64 IMOs, there were a total of 16 equation problems. These
problems can be broadly categorized into three types, as depicted in
Figure 1.2. The score details for these problems are presented in Table 1.2.
Due to the smaller number of participating teams and missing contestant
score information in early IMOs, there are several blanks in Table 1.2.
Problems 1.1–1.7 focus on “finding solutions of equations and systems
of equations;” among these seven problems, the one with the lowest average
64 IMO Problems, Theorems, and Methods: Algebra

8
7
Finding Solutions Proving Relationships
6
Investigating Conditions
5
4
3
2
1
0
1–10 11–20 21–30 31–40 41–50 51–60 61–64

Figure 1.2 Numbers of Equation Problems in the First 64 IMOs

score is Problem 1.6 (IMO 5-4), proposed by the Soviet Union. Problems
1.8–1.11 deal with “proving relationships satisfied by equations and systems
of equations;” among these four problems, the one with the lowest average
score is Problem 1.11 (IMO 18-5), proposed by the Netherlands. Problems
1.12–1.16 are about “investigating conditions under which equations and
systems of equations have solutions;” among these five problems, the one
with the lowest average score is Problem 1.16 (IMO 57-5), proposed by
Russia.
These 16 problems were proposed by 11 countries. The Soviet Union,
Romania, Bulgaria, Czechoslovakia, and Hungary each contributed two
problems.
From Table 1.2, it can be observed that in the first 64 IMOs, there were
four equation problems with an average score of 1–2 points; two problems
with an average score of 2–3 points; five problems with an average score
of 3–4 points; five problems with an average score above 4 points. Overall,
the equation problems were relatively simple, and the contestants scored
relatively high.
In the 24th–64th IMOs, there was only one equation problem, specifi-
cally Problem 1.16 (IMO 57-5), which had an average score of 1.678 points.
This indicates that equation problems were primarily featured in the first
23 IMOs, as shown in Table 1.3.
From Table 1.2, it can be observed that, excluding Problem 1.16 (IMO
57-5), the average score of the top five teams is typically about 1 point
higher than the average score of the problem. However, the average score
 Table 1.2 Score Details of Equation Problems in the First 64 IMOs

Problem 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8

Full points 8.000 5.000 6.000 7.000 6.000 6.000 7.000 6.000
Average score 6.111 3.941 4.188 2.556 3.500 1.375 5.000 2.700
Top five mean 4.950 3.975
6th–15th mean
16th–25th mean
Problem number 1-2 4-4 7-4 3-3 5-1 5-4 8-5 7-2
in IMO

Equation Problems
Proposing Romania Romania The Soviet Bulgaria Czecho- The Soviet Czecho- Poland
country Union slovakia Union slovakia

Problem 1.9 1.10 1.11 1.12 1.13 1.14 1.15 1.16

Full points 7.000 7.000 7.000 7.000 6.000 8.000 7.000 7.000
Average score 4.044 1.935 1.094 4.222 3.222 3.992 3.127 1.678
Top five mean 3.325 2.050 4.700 4.275 6.267
6th–15th mean 1.563 0.863 3.850 3.434 4.485
16th–25th mean 2.636
Problem number 10-3 18-2 18-5 1-3 3-1 15-3 21-5 57-5
in IMO
Proposing Bulgaria Finland The Netherlands Hungary Hungary Sweden Israel Russia
country

Note. Top five mean = Total score of the top five teams ÷ Total number of contestants from the top five teams,
6th–15th mean = Total score of the 6th–15th teams ÷ Total number of contestants from the 6th–15th teams,
16th–25th mean = Total score of the 16th–25th teams ÷ Total number of contestants from the 16th–25th teams.

65
66 IMO Problems, Theorems, and Methods: Algebra

Table 1.3 Numbers of Equation Problems in the 24th–64th IMOs

Problem Number
Number of Problems in
Equation Problem 1, 4 2, 5 3, 6 the First 64 IMOs

Finding solutions 0 0 0 7
Proving relationships 0 0 0 4
Investigating conditions 0 1 0 5
Total 0 1 0 16

of the 6th–15th teams tends to be close to or below the average score of the
problem, as seen in Problem 1.10 (IMO 18-2), Problem 1.11 (IMO 18-5),
Problem 1.14 (IMO 15-3), and Problem 1.15 (IMO 21-5).
This phenomenon is due to the smaller number of participating teams
in early IMOs. It was not until the 22nd IMO in 1981 that the number of
participating teams exceeded 25. Furthermore, the relatively low level of
difficulty associated with equation problems during this period suggests a
minimal variation in the average scores between the overall, top five, and
6th–15th teams.
Chapter 2

Function Problems

The concept of functions represents a type of correspondence between sets.

Since this concept was introduced, it has permeated various levels of math-
ematics, and the theory surrounding it has become quite comprehensive.
However, there are still many issues that remain worthy of exploration.
In mathematics competitions, problems often arise where the specific
form of a function is not given. Instead, certain properties, relations, or
functional equations are provided, and the task is to determine the func-
tion’s expression, calculate its values, or prove properties it possesses.
In the first 64 IMOs, there had been a total of 28 function problems,
accounting for approximately 27.7% of all algebra problems. These prob-
lems can be primarily categorized into three types: (1) proving properties of
functions, totaling nine problems; (2) determining numerical values of func-
tion variables or outputs, totaling four problems; (3) deriving expressions
for functions that meet specific conditions, totaling 15 problems. The sta-
tistical distribution of these three types of problems in the previous IMOs
is presented in Table 2.1.
From 21st–40th and 51st–60th IMOs, function problems accounted for
a high proportion, close to or exceeding 40%. Initially, most problems
involved proving properties of functions, requiring some mathematical anal-
ysis methods in advanced mathematics. Subsequently, problems on deter-
mining function values and solving functional equations appeared, with the
former being relatively straightforward when the properties of the function
were clear. Consequently, such problems gradually faded, while functional
equations became a focal point.

67
68 IMO Problems, Theorems, and Methods: Algebra

Table 2.1 Numbers of Function Problems in the First 64 IMOs

Session

Content 1–10 11–20 21–30 31–40 41–50 51–60 61–64 Total

Proving properties 1 4 1 1 0 2 0 9
Determining values 0 1 3 0 0 0 0 4
Deriving expressions 0 0 2 4 3 5 1 15
Algebra problems 20 20 14 13 15 13 6 101
The percentage of 5.0% 25.0% 42.9% 38.5% 20.0% 53.8% 16.7% 27.7%
function problems
among the algebra
problems

The study of functional equations dates back to the 14th-century math-

ematician Nicola Oresme. In 1352, in his paper “Tractatus de configura-
tionibus qualitatum et motuum,” Oresme employed the functional equation
y−x f (y) − f (x)
=
z−y f (z) − f (y)
to indirectly provide a definition of linear functions and use graphs to rep-
resent such functional relationships, predating René Descartes.
Although Oresme’s work on linear functions can be considered an early
example of functional equations, the field’s theoretical foundation is more
aptly marked by Cauchy’s work. In 1676, Newton extended the binomial
theorem to

(1 + x)z = 1 + C1z x + C2z x2 + C3z x3 + · · · ,

with z as any real number, but his derivation was not rigorous. Cauchy
further refined Newton’s work. On this basis, Cauchy defined the function

f (z) = 1 + C1z x + C2z x2 + C3z x3 + · · ·

and proved that f (z + w) = f (z)f (w) for any real numbers z and w, known
as Cauchy’s exponential functional equation.
Later, many renowned mathematicians, such as Jean Le Rond
d’Alembert, Léonard Euler, Carl Friedrich Gauss, Adrien-Marie Legendre,
Niels Henrik Abel, and David Hilbert, contributed to the study of func-
tional equations, leading to equations named after them. Today, func-
tional equations remain an actively developing branch of mathematics, with
widespread applications in various scientific and technical fields.
 Function Problems 69

Furthermore, functional equations frequently appeared in publications

like the Mathematical Association of America’s Mathematics Magazine and
the Canadian Mathematical Society’s Crux Mathematicorum, often solvable
with elementary mathematics.
Functional equations also appeared in various secondary and tertiary
mathematics competitions. For instance, the Putnam Mathematical Com-
petition has featured numerous such problems, the earliest being from the
4th Putnam Mathematical Competition in 1941:
• (B14). Show that any solution f (t) of the functional equation
f (x + y)f (x − y) = (f (x))2 + (f (y))2 − 1, (x, y are real),
is such that
f  (t) = ±m2 f (t), (m is a constant and m ≥ 0),
assuming the existence and continuity of the second derivative. Deduce
that f (t) is one of the functions ± cos mt and ± cosh mt.
These problems typically involve advanced mathematics and are thus
more common in university-level competitions.
The development of functional equations has also influenced Mathemat-
ical Olympiad problems, such as the 5th problem of the 14th IMO in 1972:
• (IMO 14-5). Let f and g be real-valued functions defined for all real
values of x and y, and satisfying the equation
f (x + y) + f (x − y) = 2f (x)g(y)
for all x, y ∈ R. Prove that if f (x) is not identically zero and |f (x)| ≤ 1
for all x ∈ R, then |g(y)| ≤ 1 for all y ∈ R.
This problem represents a special case of functional equations of interest
to applied mathematicians in the 18th century.
Over recent decades, problems involving deriving expressions have
appeared frequently in the IMO, almost every two to three years. This
is due to the fact that many significant problems in natural sciences can
ultimately be reduced to solving specific or classes of functional equations.
However, a complete and systematic theory of functional equations has
not yet been established, and general methods for solving them are scarce.
Thus, solving functional equations itself demands a good mathematical
foundation.
Function iterations, a special case of function compositions, often appear
in functional equations. Specifically, in cases where the recurrence relation
70 IMO Problems, Theorems, and Methods: Algebra

of a sequence {an } is known, an can be similarly regarded as the nth itera-

tion of the initial value a0 under the recurrence relation. Moreover, function
iterations can sometimes serve as a problem-solving approach, such as using
fixed-point properties to solve functional equations.
This chapter will be divided into three parts. The first part introduces
basic properties of functions and common functional equations, followed
by presentations of common methods for solving function iterations and
functional equations. These methods include the Cauchy method, method
of undetermined coefficients, fixed point method, and substitution method,
all of which find applications in the IMO.
The second part revolves around three types of problems: “proving
properties of functions,” “determining numerical values of function vari-
ables or outputs,” and “deriving expressions for functions that meet spe-
cific conditions.” These problems are presented in chronological order, and
some problems include various solutions, generalizations, and similar prob-
lems. Problems combining functional equations with number theory are
presented in IMO Problems, Theorems, and Methods: Number Theory.
It is important to note that for each problem, the solutions are fol-
lowed by information on the scores, including the number of contestants in
each score range, the average score, and the scores of the top five teams.
However, early IMOs often lacked information on contestant scores, so the
number of contestants in each score range only represents the counted num-
ber of contestants, and some problems lack scores of the top five teams.
The third part provides a brief summary of this chapter.
It should be noted that even functional equations with similar structures
may require entirely different approaches. In such cases, more intuitive
ideas can prove to be more effective. These include considering whether the
function is injective (one-to-one) or surjective (onto), determining whether
the function is odd or even, examining the set of points where the func-
tion equals the target function, and proving that it constitutes the entire
set. Although these ideas are straightforward and can be easily overlooked,
continuous exploration may outline the entire problem.

2.1 Common Theorems, Formulas, and Methods

2.1.1 Common elementary functions
Let A and B be non-empty sets. If for every element x in set A, there
exists a unique corresponding element y in set B based on a certain defined
correspondence f , then f : A → B is called a function from set A to set B,
 Function Problems 71

denoted as

y = f (x), x ∈ A.

Here, x is referred to as the independent variable, and its range A is called

the domain of the function. The corresponding value y of x is known as the
function value, and the set of function values {f (x)|x ∈ A} is termed the
range of the function. If for every y in the range, there exists a unique x
in the domain such that f (x) = y, then the function mapping y back to x
is called the inverse function of f , denoted as:

f −1 : {f (x)|x ∈ A} → A,

or

x = f −1 (y), y ∈ {f (x)|x ∈ A}.

It is important to note that the domain of function f is precisely the

range of its inverse function f −1 , and vice versa. Moreover, the inverse
function notation “f −1 ” should not be confused with the reciprocal f1 .
Common elementary functions include:
(1) Power functions
The function y = xa (a ∈ R) is called a power function, where the base x is
the independent variable and the exponent a is a constant, such as y = x
and y = x1 (x = 0).
(2) Exponential functions
The function y = ax (a > 0, a = 1) is called an exponential function, where
the exponent x is the independent variable and the base a is a constant.
For a > 1, the function y = ax is monotonically increasing; for 0 < a < 1,
the function y = ax is monotonically decreasing.
(3) Logarithmic functions
The function y = loga x(a > 0, a = 1) is called a logarithmic function,
where the positive real number x is the independent variable and the base
a is a constant.
A logarithmic function is the inverse function of an exponential func-
tion, and their graphs in the Cartesian coordinate system are symmetric
about the line y = x. For a > 1, the function y = loga x is monotoni-
cally increasing; for 0 < a < 1, the function y = loga x is monotonically
decreasing.
72 IMO Problems, Theorems, and Methods: Algebra

(4) Trigonometric functions

These include the sine function sin x, cosine function cos x, tangent function
tan x, cotangent function cot x, secant function sec x, and cosecant function
csc x.
In secondary mathematics, the sine, cosine, and tangent functions are
frequently encountered, and like other elementary functions, they exhibit
basic properties such as monotonicity, periodicity, and odd-even symmetry.
(5) Inverse trigonometric functions
Inverse trigonometric functions include the arcsine function arcsin x, arcco-
sine function arccos x, arctangent function arctan x, arccotangent function
arccot x, arcsecant function arcsec x, and arccosecant function arccsc x.
There is a difference between these functions and the inverse functions
of other common elementary functions, i.e., they are multivalued.
(6) Constant functions
Constant functions are those whose function values remain unchanged, i.e.,
the function values are a constant.
Consider two functions

z = f (y), where y ∈ E,
y = g(x), where x ∈ D.
If the set D ∗ = {x ∈ D|g(x) ∈ E} is non-empty, then a new function
can be defined with elements from D∗ as the independent variable, passing
through the successive actions of g and f , and elements from the range of
f as the function value, denoted as

z = f (g(x)), x ∈ D∗ .

This function is usually represented as f ◦ g and called the function

composition of f and g.
Function compositions follow the associative law, i.e., (f ◦ g) ◦ h =
f ◦ (g ◦ h), but not the commutative law, as f ◦ g and g ◦ f are generally
not the same. There is a related problem:

• (Romanian Master of Mathematics 2013, Problem 2). Does there

exist a pair (g, h) of functions g, h : R → R such that the only function
f : R → R satisfying f (g(x)) = g(f (x)) and f (h(x)) = h(f (x)) for all
x ∈ R is the identity function f (x) ≡ x?
 Function Problems 73

2.1.2 Basic properties of functions

In mathematics, an injective function, also known as injection or one-to-
one function, is a function f : A → B that maps distinct elements of A to
distinct elements, i.e., if x1 = x2 , then f (x1 ) = f (x2 ).
A surjective function, also known as surjection or onto function, is a
function f : A → B such that, for every element y ∈ B, there exists at least
one element x ∈ A satisfying y = f (x).
A bijective function, also known as bijection, is a function that is both
injective and surjective.
Actually, every bijective function f : Z → Z can be written in the way
f = u + v, where u, v : Z → Z are bijective functions (Romanian Master of
Mathematics 2008, Problem 2).

(1) Odd–even symmetry

Suppose the domain D of a function f (x) is symmetric about the origin.
If f (−x) = −f (x) for every x ∈ D, then f (x) is called an odd function. If
f (−x) = f (x) for every x ∈ D, then f (x) is called an even function.
Actually, every function f can be written in the way f = g + h, where g
is odd and h is even. Furthermore, the following propositions are relatively
straightforward to prove by using definitions.

Proposition 2.1. Only the zero function can be both odd and even.

Proposition 2.2. The sum of an odd function and an even function is

neither odd nor even, unless one of them is identically zero.

Proposition 2.3. The sum and difference of two odd (even) functions are
odd (even) functions, and an odd (even) function multiplied by a non-zero
constant remains odd (even).

Proposition 2.4. The product and quotient of two odd (even) functions
are even functions, while the product and quotient of an odd function and
an even function are odd functions.

Proposition 2.5. The composition of two odd functions is an odd function,

the composition of an even function and an odd function (the even function
acts second) is an even function, and the composition of any function and
an even function (the even function acts first) is an even function.
74 IMO Problems, Theorems, and Methods: Algebra

More generally, for a function f (x) with domain R, the following holds:

Proposition 2.6. If f (a + x) = f (b − x) is always true, then the graph of

f (x) is symmetric about the line x = a+b
2 .

Proposition 2.7. If f (a + x) = c − f(b − x) is always true, then the graph

of f (x) is symmetric about the point a+b c
2 , 2 .

Proposition 2.8. The function g(x) = f (a + x) + f (b − x) is symmetric

about the line x = b−a
2 .

 The function g(x) = f (a + x) − f (b − x) is symmetric

Proposition 2.9.
about the point b−a
2 ,0 .

(2) Monotonicity
Let the domain of a function f (x) be D. If for any x1 , x2 ∈ D1 ⊂ D,
it holds that f (x1 ) < f (x2 ) whenever x1 < x2 , then f (x) is said to be
(strictly) increasing on D1 ; if f (x1 ) > f (x2 ) whenever x1 < x2 , then f (x)
is said to be (strictly) decreasing on D1 .
If the function f (x) is increasing or decreasing on a subset of its domain
(usually an interval), then f (x) is said to be (strictly) monotonic on this
subset. This subset is referred to as the monotonic interval of f (x).
It should be noted that the monotonicity of a function pertains to a
subset of its domain and is a “local” property. Even if f (x) is increasing
(decreasing) on subintervals D1 and D2 of its domain, it cannot be directly
inferred that f (x) is increasing (decreasing) on D1 ∪ D2 .
Furthermore, in advanced mathematics, the definition of monotonicity
is slightly weaker, indicating that f (x1 ) ≤ f (x2 ) or f (x1 ) ≥ f (x2 ) for
x1 < x2 .
The following propositions are relatively straightforward to prove by
using definitions.

Proposition 2.10. If functions f (x) and g(x) are both increasing

(decreasing) on a common interval I, then the function f (x) + g(x) is also
increasing (decreasing) on I.

Proposition 2.11. If two positive-valued functions f (x) and g(x) are both
increasing (decreasing) on a common interval I, then the function f (x)g(x)
is also increasing (decreasing) on I.
 Function Problems 75

Proposition 2.12. If function f (x) is increasing (decreasing) on interval

I with the range f (I), then the inverse function f −1 (x) certainly exists and
is increasing (decreasing) on f (I).
Proposition 2.13. If functions f (x) and g(x) are both monotonic on
their respective intervals, then the function f (g(x)) is also monotonic on
the monotonic interval of g(x). If f (x) and g(x) have the same type of
monotonicity, then f (g(x)) is increasing. If f (x) and g(x) have opposite
types of monotonicity, then f (g(x)) is decreasing.
Moreover, there is a related problem:

• (Romanian Master of Mathematics 2011, Problem 1). Prove that

there exist two functions f, g : R → R such that f ◦g is strictly decreasing
and g ◦ f is strictly increasing.

(3) Periodicity
Let the domain of a function f (x) be D. If there exists a non-zero constant
T such that x + T ∈ D and f (x + T ) = f (x) for every x ∈ D, then f (x) is
termed a periodic function, and T is known as a period of the function f (x).
If among all positive periods of the function f (x), there exists the small-
est value T0 , then T0 is called the fundamental period of the periodic func-
tion f (x).
However, not every periodic function has a fundamental period. For
instance, a constant function can have any real number as a period, and
the function 
1, x ∈ Q,
d(x) =
0, x ∈/Q
can have any rational number as a period.
The following propositions are relatively straightforward to prove by
using definitions.
Proposition 2.14. If a function f (x) with domain R is symmetric about
the lines x = a and x = b, where a = b, then f (x) is periodic with a period
T = 2|a − b|.
In particular, f (x) being an even function is equivalent to f (x) being
symmetric about the line x = 0.
Proposition 2.15. If a function f (x) with domain R is symmetric about
the points (a, 0) and (b, 0), where a = b, then f (x) is periodic with a period
T = 2|a − b|.
76 IMO Problems, Theorems, and Methods: Algebra

In particular, f (x) being an odd function is equivalent to f (x) being

symmetric about the origin.

Proposition 2.16. If a function f (x) with domain R is symmetric about

the line x = a and the point (b, 0), where a = b, then f (x) is periodic with
a period T = 4|a − b|.

(4) Convexity and concavity

Let f (x) be a function defined on an interval I. If for any x, y ∈ I and any
t ∈ [0, 1],

f (tx + (1 − t)y) ≤ tf (x) + (1 − t)f (y),

then f (x) is called a convex function on I. Conversely, if for any x, y ∈ I

and any t ∈ [0, 1],

f (tx + (1 − t)y) ≥ tf (x) + (1 − t)f (y),

then f (x) is called a concave function on I.

It is important to note that the terminology for convex and concave
functions has varied considerably in different periods of the literature, what
is defined as a convex function in one text might be defined as a concave
function in another, necessitating a careful distinction.
The following propositions are relatively straightforward to prove by
using definitions.

Proposition 2.17. If f (x) is a convex function defined on I and c is a

positive real number, then cf (x) is also a convex function.

Proposition 2.18. If f (x) and g(x) are both convex functions defined on I,
then h1 (x) = f (x) + g(x) and h2 (x) = max{f (x), g(x)} are both convex
functions.

Proposition 2.19. If f (x) and g(x) are both convex functions and g(x) is
increasing, then g(f (x)) is a convex function.

Proposition 2.20. If f (x) is a convex function defined on I, then q(x) =

f (ax + b) is also a convex function, where a, b are real numbers, with a = 0.

Convex and concave functions are closely related to Jensen’s inequality,

which will be introduced in Chapter 4 on inequalities.
 Function Problems 77

(5) Continuity
A continuous function is one in which a sufficiently small change in the inde-
pendent variable results in a correspondingly small change in the function
value. However, this is not the rigorous definition of a continuous function,
which relies on the concept of limits in advanced mathematics.
Let x0 be a point in the domain D of the function f (x). If for any
sequence {xn } ⊂ D converging to x0 , the sequence {f (xn )} tends to f (x0 ),
then f (x) is said to be continuous at x0 .
Suppose the domain of the function f (x) is D, and the interval I is a
subset of D. If f (x) is continuous at every point x in I, then f (x) is said
to be continuous on the interval I.
Intuitively, functions whose graphs are unbroken curves are continuous.
For example, polynomial function, power function, exponential function,
logarithmic function, trigonometric function, and the absolute value func-
tion are all continuous on subintervals of their domains.
The following propositions are relatively straightforward to prove by
using definitions.

Proposition 2.21. If f and g are both continuous functions, then f ± g,

f g, and fg are all continuous functions; note that the domain of fg excludes
points x where g(x) = 0.

Proposition 2.22. The composition f ◦ g of two continuous functions f

and g is a continuous function.

It is not an exaggeration to say that the concept of continuous functions

has widespread applications in almost all disciplines related to mathematics.

2.1.3 Common function iterations

Let f : D → D be a function. For any x ∈ D, denote
f (0) (x) = x, f (n+1) (x) = f (f (n) (x)) for n ∈ N+ .
(n)
Then f (x) is called the nth iteration of the function f (x) over the
domain D, and n is called the iteration index.
It’s important to note that f (n) (x) is also commonly used to represent
the nth derivative of f (x). Therefore, it’s crucial to discern the context of
a problem for a proper understanding.
In the above defined function iteration, the iteration index is always a
non-negative integer. If f (x) is a bijection from D to D, negative iteration
indices can also be defined as follows.
78 IMO Problems, Theorems, and Methods: Algebra

Let f (x) be a bijection from D to D. Define f (−1) (x) = f −1 (x), i.e., the
inverse function of f (x). Then, for any positive integer n, define f (−n) (x)
as the nth iteration of f (−1) (x).
Common forms of function iterations include:

(i) The nth iteration of the function f (x) = x + c is f (n) (x) = x + nc.
n
(ii) The nth iteration of the function f (x) = xm is f (n) (x) = xm .
(iii) The nth iteration of the function f (x) = ax + b is
 
b b
f (n) (x) = an x − + .
1−a 1−a
x
(iv) The nth iteration of the function f (x) = a+bx is
x
f (n) (x) = 1−an .
an + bx · 1−a

(v) The nth iteration of the function f (x) = ax2 + bx + c (where a = 0

and 4ac = b2 − 2b) is
 2n
n b b
f (n) (x) = a2 −1 x + − .
2a 2a

Note that some functions’ iterations may exhibit a periodic behavior.

For example, for the function f (x) = x−1
x , it is easy to verify that f
(4)
(x) =
(1)
f (x) = f (x).

2.1.4 Common methods for determining function

iterations
The primary issue in function iterations is to determine the expression for
the nth iteration of a given function.
(1) Mathematical induction
This approach involves guessing and then proving. Initially, iterate the
function f (x) a few times, observe the pattern, conjecture the expression
for f (n) (x), and finally prove it using mathematical induction.

Example 2.1. Let f (x) = ax + b and find f (n) (x).

Solution. It is straightforward that

f (x) = ax + b,
 Function Problems 79

f (2) (x) = f (f (x)) = a(ax + b) + b = a2 x + ab + b,

f (3) (x) = f (f (2) (x)) = a(a2 x + ab + b) + b = a3 x + a2 b + ab + b,

leading to the conjecture f (n) (x) = an x + an−1 b + an−2 b + · · · + ab + b.

Next, prove it by mathematical induction. For n = 1, the proposition
holds.
Assume f (k) (x) = ak x + ak−1 b + ak−2 b + · · · + ab + b holds. Then

f (k+1) (x) = f (f (k) (x)) = a(ak x + ak−1 b + · · · + ab + b) + b

= ak x + ak−1 b + · · · + ab + b,

implying that the proposition also holds for n = k + 1, and thus for any
positive integer n.

(2) Recursive method

This involves solving by constructing a sequence. Let a0 = x and an =
f (n) (x) for all positive integers n. From an = f (n) (x) = f (an−1 ), find
an = g(a0 ). Then g(x) = f (n) (x).
√ √
Example 2.2. Let f (x) = 3 3 x( 3 x + 1) + x + 1 and find f (n) (x).
√
Solution. It is evident that f (x) = ( 3 x + 1)3 . Set a0 = x and an =
f (n) (x). Then
√
an = f (an−1 ) = ( 3 an−1 + 1)3 ,
√ √ √ √ √
which implies 3
an − 3 an−1 = 1. Hence, 3 an = 3 a0 +n = 3 x+n, leading
to
√
an = ( 3 x + n)3 ,
√
i.e., f (n) (x) = ( 3 x + n)3 .

(3) Similarity method

If there exist a function ϕ(x) and its inverse ϕ−1 (x) such that f (x) =
ϕ−1 (g(ϕ(x))), then the functions f (x) and g(x) are said to be conjugate or
similar. We call ϕ(x) the bridge function between f (x) and g(x), and it is
denoted as f ∼ g or f ∼ϕ g.
The significance of introducing a bridge function lies in its ability to
establish an equivalence relation, maintaining the similarity during the
80 IMO Problems, Theorems, and Methods: Algebra

process of function iterations. It is easy to observe the following properties:

(i) Reflexive Property: f ∼ f .

(ii) Symmetric Property: If f ∼ g, then g ∼ f .
(iii) Transitive Property: If f ∼ g and g ∼ h, then f ∼ h.
(iv) If f ∼ g, then f (n) ∼ g (n) , i.e., f (n) (x) = ϕ−1 (g (n) (ϕ(x))).

This implies that if functions f (x) and g(x) are similar, then finding the
nth iteration of f (x) is essentially equivalent to finding the nth iteration
of g(x). If the nth iteration of g(x) is easier to determine, then the problem
becomes more straightforward.
x
Example 2.3. Let f (x) = 1+ax and find f (n) (x).

Solution. Let g(x) = x + a and ϕ(x) = x1 . Then ϕ−1 (x) = 1

x. It is easy
to verify that f (x) = ϕ−1 (g(ϕ(x))), i.e., f ∼ g. Therefore,

f (n) (x) = ϕ−1 (g (n) (ϕ(x))) = ϕ−1 (ϕ(x) + na)

1 x
= 1 = .
x + na 1 + nax

(4) Fixed point method

In the process of deriving function iterations using the similarity method,
the key step is finding a bridge function ϕ(x). However, this is not easy in
most cases. Here, we introduce a method for identifying a class of bridge
functions, i.e., the fixed point method.
A fixed point of a function f (x) is a point that satisfies f (x) = x. It is
evident that fixed points have the following properties:

(i) If x0 is a fixed point of f (x), then x0 is also a fixed point of f (n) (x).
(ii) If f ∼ϕ g and x0 is a fixed point of f (x), then ϕ(x0 ) is a fixed point of
g(x).

Note that when using the similarity method, g(x) is often chosen as
x + 1, ax, ax2 , ax3 , etc., where the fixed points of g(x) are 0 or +∞.
Therefore, based on the properties of fixed points, the chosen bridge func-
tion ϕ(x) should map the fixed points of f (x) to 0 or +∞.
If f (x) has only one fixed point x0 , then consider ϕ(x) = x − x0 or
1
ϕ(x) = x−x 0
. If f (x) has two distinct fixed points α and β, then consider
x−α
ϕ(x) = x−β .
 Function Problems 81

x+6
Example 2.4. Let f (x) = x+2 and find f (n) (x).

Solution. Setting f (x) = x, we find that the fixed points of f (x) are
2 and −3. Let ϕ(x) = x−2 x+3 and g(x) = − 4 x, and it is easy to verify
1

ϕ(f (x)) = g(ϕ(x)).  n

Hence, f ∼ g and f (n) ∼ g (n) , and since g (n) (x) = − 41 x, we have

(2 · (−4)n + 3)x + 6 · ((−4)n − 1)

f (n) (x) = ϕ−1 (g (n) (ϕ(x))) = .
((−4)n − 1)x + (3 · (−4)n + 2)
In this way, we can deduce a bridge function’s expression based on the
function’s fixed points and then apply the similarity method to find the nth
iteration of the function. Similarly, we have:

(i) For functions like f (x) = a(x − h)k + h(a = 0, k ∈/ {0, 1}), a bridge
function can be ϕ(x) = x − h.
(ii) For functions like
√

k x k
axk + b
f1 (x) = ax + b, f2 (x) = √
k
k
, f 3 (x) = ,
k
ax + b x

where ab = 0, a bridge function can be ϕ(x) = xk .

(iii) For functions like
√
√ x (a k x + b)k
f1 (x) = (a k x + b)k , f2 (x) = √ , f3 (x) = ,
(a k x + b)k x
√
where ab = 0, a bridge function can be ϕ(x) = k x.
k
cx
(iv) For functions like f (x) = xk −c(x−c) k , where c = 0, a bridge function
x−c
can be ϕ(x) = cx .

Note that the expression of the nth iteration of most functions f (x) is
not easy to determine and can be quite complex. In such cases, it is more
common to study the properties of f (n) (x), such as given x0 , examining the
sequence

x0 , f (x0 ), f (2) (x0 ), . . . , f (n) (x0 ), . . .

and the quantitative relationships it satisfies. This sequence is called the

orbit of x0 , and the function sequence

x, f (x), f (2) (x), . . . , f (n) (x), . . .

82 IMO Problems, Theorems, and Methods: Algebra

is called the Picard sequence of the function f (x). There is a related

problem:

• (Romanian Master in Mathematics 2010, Problem 6). Given a

polynomial f (x) with rational coefficients, of degree d ≥ 2, we define the
sequence of sets f (0) (Q), f (1) (Q), . . . as
f (0) (Q) = Q, f (n+1) (Q) = f (f (n) (Q)) for n ≥ 0.
(Given a set S, we write f (S) for the set {f (x)|x ∈ S}.)

∞
Let f (ω) (Q) = f (n) (Q) be the set of numbers that are in all of the
n=0
sets f (n) (Q) with n ≥ 0. Prove that f (ω) (Q) is a finite set.

2.1.5 Common functional equations

The renowned Hungarian mathematician János Aczél defined functional
equations as follows: “Functional equations are equations in which both
sides contain a finite number of functions, some known and some unknown.”
In layman’s terms, functional equations are equations where the
unknowns are functions, but the number of solutions to a functional equa-
tion is not necessarily related to the number of unknown functions.
Furthermore, functional equations can be classified in various ways, such
as by the domain of the unknown functions, by the number of variables in
the functions, by the number of unknown functions, or by the number of
iterations the unknown functions undergo, among others.
Common functional equations include (some of which are named after
famous mathematicians):
(1) Cauchy’s functional equation
Let f : R → R be a continuous function. If for all real numbers x and y,
the Cauchy’s functional equation
f (x + y) = f (x) + f (y)
is satisfied, then for all x ∈ R, one has f (x) = f (1) · x.
Similarly, we can deduce the following functional equations:

(i) If f (x + y) = f (x) + f (y) + a, then f (x) = (f (1) + a)x − a.

(ii) If f (x + y) = f (x)f (y), then f (x) = (f (1))x .
(iii) If f (xy) = f (x) + f (y)(x, y > 0), then f (x) = f (a) · loga x(a > 0,
a = 1).
(iv) If f (xy) = f (x)f (y)(x, y > 0), then f (x) = (f (a))loga x (a > 0, a = 1).
 Function Problems 83

f (x)f (y) f (1)

(v) If f (x + y) = f (x)+f (y) (f (x) = 0), then f (x) = x (x = 0).

(vi) If f (x+y) = f (x)+f (y)+2 f (x)f (y)(x, y > 0), then f (x) = f (1)·x2 .
(vii) If f (x + y) = f (x) + f (y) + kxy, then f (x) = k2 x2 + (f (1) − k2 )x.
√
(viii) If (f (x + y))2 = (f (x))2 + (f (y))2 (x, y > 0), then f (x) = f (1) · x.
(ix) If (f (x+y))2 +(f (x−y))2 = 2(f (x))2 +2(f (y))2 , then f (x) = f (1)·|x|.

(2) Jensen’s functional equation

Jensen’s functional equation has the form
 
x+y f (x) + f (y)
f = ,
2 2
which can be seen as a mean-form of Cauchy’s functional equation. The
general solution is f (x) = ax + b.
(3) Linear functional equations
A linear functional equation has the form

f (ax + by + c) = pf (x) + qf (x) + r,

where a, b, c, p, q, r are constants.

(4) Cauchy’s exponential functional equation
Cauchy’s exponential functional equation has the form

f (x + y) = f (x) · f (y),

where f : R → R is a continuous function that is not identically zero.

(5) Pexider’s functional equation
Pexider’s functional equation has the form

f (x + y) = g(x) + h(y),

where f, g, h : R → R are continuous functions. Pexider’s functional equa-

tion is a natural generalization of Cauchy’s functional equation, and its
general solution is
⎧
⎨f (z) = cz + a + b,
g(x) = cx + a,
⎩
h(y) = cy + b,

where a, b, c are arbitrary constants.

84 IMO Problems, Theorems, and Methods: Algebra

(6) Aczel’s functional equation

Aczel’s functional equation has the form

f (x + y) = g(x) · k(y) + h(y),

which is a generalization of the Pexider’s functional equation.

(7) D’Alembert’s functional equation
D’Alembert’s functional equation has the form

g(x + y) + g(x − y) = 2g(x) · g(y).

Additionally, there are many other famous functional equations.

(i) Abel’s functional equation: f (g(x)) = f (x) + a.
(ii) Euler’s functional equation: f (tx, ty) = tk f (x, y).
(iii) Wilson’s functional equation: g(x + y) + g(x − y) = 2g(x)f (y).
(iv) Gauss’s functional equation: f ( x2 + y 2 ) = f (x)f (y).
(v) Davidson’s functional equation: f (xy) + f (x + y) = f (xy + x) + f (y).

2.1.6 Common methods for solving functional equations

Solving functional equations involves finding solutions under given con-
ditions or determining that no solutions exist. Typically, functional equa-
tions have more than one solution, and sometimes extraneous solutions may
arise (akin to extraneous roots). Therefore, it is necessary to substitute the
obtained solutions back into the original equations for verification.
Functional equations are closely related to function iterations, and some
methods for solving functional equations are similar to those for finding
expressions of the nth iteration of a function.
(1) Cauchy’s method
First, determine the solution f (q) when the independent variable is a ratio-
nal number q. Then, under certain assumptions or conditions about the
function f (x) (such as monotonicity, continuity, boundedness, etc.), find
the solution f (x) for any real number x.

Example 2.5. Suppose  a nonzero continuous function f (x) satisfies the

functional equation f ( x2 + y 2 ) = f (x)f (y), where x, y are real numbers.
2
Prove that f (x) = (f (1))x .

Proof. It is evident that f (x) is an even function, so it suffices to prove

the case for x ≥ 0.
 Function Problems 85

Setting x = y = 0 yields f (0) = (f (0))2 . Since f (x) = 0, it follows that

2
f (0) = 1, and thus f (0) = (f (1))0 .
Suppose n is a positive integer, first prove the following proposition:
√
f ( ny) = (f (y))n for n ∈ N+ , y ∈ R+ .

For n = 1, the proposition is trivially true. Assume it holds for n = k.

For n = k + 1,
√  √
f ( k + 1y) = f ( ky 2 + y 2 ) = f ( ky)f (y) = (f (y))k f (y) = (f (y))k+1 .

Thus, the proposition holds.

√ √ √
Setting y = 1 and y = n in f ( ny) = (f (y))n gives f ( n) = (f (1))n
√ n 2
and f (n) = (f ( n)) , so f (n) = (f (1))n .
For x = pq , where p and q are positive integers, the proposition yields
    q2
p p 2
f (p) = f q2 · = f and f (p) = (f (1))p .
q q
q2 2 p2
p p
Hence, f q = (f (1))p , and therefore f q = (f (1)) q2 .
2
From the continuity of f (x), it follows that f (x) = (f (1))x also holds
for irrational numbers x.
(2) Method of undetermined coefficients
This method involves assuming a form of the solution based on the struc-
ture of the functional equation, and then substituting this form into the
functional equation to determine the coefficients it contains.

Example 2.6. Suppose f (x) is a quadratic function satisfying

f (x + 1) − f (x) = 8x + 3 and f (0) = 5.

Find f (x).

Solution. Assume f (x) = ax2 + bx + c. From f (0) = 5, we have c = 5.

Also,

a(x + 1)2 + b(x + 1) + c − ax2 − bx − c = 8x + 3.

Comparing coefficients on both sides of the above equation yields a = 4

and b = −1.
Thus, f (x) = 4x2 − x + 5.
86 IMO Problems, Theorems, and Methods: Algebra

(3) Recursive method

The recursive method is a technique that involves studying functional equa-
tions through sequences. On the one hand, the function value at a point
and its iterated values can be treated as a sequence. By determining its
general term and combining the value range, the form of the function can
be deduced.

Example 2.7. Find all functions f : R+ ∪ {0} → R+ ∪ {0} such that for
any x ≥ 0,

f (f (x)) = 10x − 3f (x).

Solution. For any fixed x0 ≥ 0, let an = f (n) (x0 ), thus obtaining a non-
negative real number sequence {an }. This sequence satisfies the recurrence
relation an+2 = −3an+1 + 10an, with its characteristic equation being x2 +
3x − 10 = 0, and roots 2 and −5. Therefore, an = 2n A + (−5)n B, where
A, B are constants.   n 
Since an = 2n A + B − 52 , if B = 0, then an can be negative when
n is a sufficiently large odd integer. Hence, B must be zero, and an = 2n A.
Given a0 = x0 = A, we find f (n) (x0 ) = an = 2n x0 . In particular,
f (x0 ) = 2x0 , which holds true for any x0 ≥ 0. Upon verification, the only
function satisfying the conditions is f (x) = 2x.
On the other hand, the recursive relation can be used to determine a
series of function values. For instance, if the values of f (1), f (2), . . . , f (n)
are known and the value of f (n+1) is uniquely determined by the recurrence
relation, then one can also employ the characteristic equation.
(4) Fixed point method
This method is commonly used to solve functional equations involving func-
tion composition. If the function in the functional equation has a fixed
point, then the solutions can be derived by using this fixed point and the
properties of the function itself.

Example 2.8 (Putnam Mathematical Competition 1971, A2).

Find all polynomial functions P (x) satisfying

P (x2 + 1) = (P (x))2 + 1 and P (0) = 0

Solution. From the conditions, it is known that x0 = 0 is a fixed point of

P (x). If xk is a fixed point of P (x), then P (x2k + 1) = (P (xk ))2 + 1 = x2k + 1,
 Function Problems 87

making xk+1 = x2k + 1 also a fixed point of P (x). Thus, there exists an
infinite sequence

0, 1, 2, 5, 26, . . .

such that each term is a fixed point of P (x), i.e., roots of P (x) − x = 0.
However, since P (x) is a polynomial, P (x)−x = 0 can only have a finite
number of roots, implying P (x) − x is identically zero, i.e., P (x) = x.

Moreover, one may initially determine a fixed point x0 of f (x), express

it in terms of f (x) and x, and then find the specific value of x0 to derive the
expression of f (x). Examples include IMO 24-1 (Problem 2.14) and IMO
56-5 (Problem 2.25).
If there is more than one fixed point, then the problem becomes much
more complex, and the sought function could be a piecewise function.

(5) Method of equivalent transformations

Two functional equations (I) and (II) are equivalent if, by making appropri-
ate transformations of the independent variable, (I) can be converted into
(II) and vice versa. If one of them is already solved, then the solution for
the other can also be determined.
 
Example 2.9. Solve the functional equation 2f x+y 2 = f (x) + f (y).

Solution. Let f (0) = a. Then

 
x x+0 1 1
f =f = (f (x) + f (0)) = (f (x) + a).
2 2 2 2
    1
Also, since 12 (f (x) + f (y)) = f x+y
2 = f x+y2 + 0 = 2 (f (x + y) + a),
it follows that f (x + y) = f (x) + f (y) − a.
Let g(x) = f (x) − a. Then g(x + y) = g(x) + g(y), which is Cauchy’s
functional equation with the solution g(x) = x · g(1). Hence,

f (x) = g(x) + a = (f (1) − f (0))x + f (0).

Similarly, it can be shown that Lobachevsky’s functional equation

 
x+y 
f = f (x)f (y)
2

is equivalent to the functional equation f (x + y) = f (x)f (y), where

f (x) > 0.
88 IMO Problems, Theorems, and Methods: Algebra

(6) Estimation method

Sometimes, deeper relationships of magnitude are implicit in functional
equations or inequalities. These relationships are closely related to the solu-
tion to the problem. In particular, if it is known that f (x) ≥ g(x) and
f (x) ≤ g(x), then f (x) = g(x).

Example 2.10. Find all functions f : R+ → R+ such that for any positive
real numbers x and y,
  
x + f (y) 1 1
=f +f .
xf (y) y x
Solution. It is easy to see that f (x) = x is a solution of the functional
equation.
Suppose there exists x0 > 0 such that x10 > f x10 . Let x = x0 and
1 
y= 1
. Then
x0 −f x1
0

x + f (y) 1 1 1 1
= + = + ,
xf (y) x f (y) x0 f (y)
         
1 1 1 1 1 1
f +f =f −f +f =f .
y x x0 x0 x0 x0
1 1 1 1 1
Thus, x0 + f (y) =f x0 , i.e., x0 <f x0 , a contradiction.
Therefore, for any positive real number x0 , there is 1
x0 ≤f 1
x0 , i.e.,
x ≤ f (x).
1 1
Also, x1 + 1
f (y) =f 1
y +f x ≥ 1
y +f x ≥ 1
y + x1 , i.e., y ≥ f (y).
Consequently, the solution to the functional equation is f (x) = x.
(7) Calculus method
For some types of functional equations, it is sometimes possible to trans-
form the functional equation into a form like f  (x) = g(x) (where g(x) is
integrable). Then integrate to get

f (x) = g(x)dx + C,

and finally determine the expression of f (x) based on the given conditions.

Example 2.11. Find the real coefficient polynomial f (x) such that for any
real coefficient polynomial g(x), the equality f (g(x)) = g(f (x)) is true.
 Function Problems 89

Solution. Let g(x) = x + h. Then f (x + h) = f (x) + h, implying

f (x + h) − f (x)
=1 for h = 0.
h
Hence, f  (x) = 1, and f (x) = x + C, where C is a constant.
Let g(x) ≡ 0. Then f (0) = 0, so f (x) = x. Conversely, if f (x) = x,
then for any real coefficient polynomial g(x), one has f (g(x)) = g(f (x)).
Therefore, f (x) = x is the unique solution to the problem.
(8) Other elementary techniques
(i) Substitution
The basic idea of the substitution method is to appropriately replace the
independent variable in the functional equation with another variable. This
results in a new functional equation, from which the unknown function can
be derived.

Example 2.12 (Putnam Mathematical Competition 1971, B2).

Let f (x) be a real valued function defined for all real x except for x = 0
and x = 1 satisfying the functional equation
 
x−1
f (x) + f = 1 + x. (I)
x
Find all functions f (x) satisfying these conditions.

Solution. Replace x with x−1

x and substitute into (I) to get
   
x−1 1 x−1
f +f =1+ . (II)
x 1−x x
1
Replace x with 1−x and substitute into (I) to get
 
1 1
f + f (x) = 1 + . (III)
1−x 1−x
2 3
−x
From (I) + (III) − (II), f (x) = 1+x
2x(1−x) .
There is another similar problem:

• (Korean Mathematical Olympiad 1999, Final Round,

Problem 4). Find all functions f (x) such that
   
x−3 3+x
f +f = x for real x except for x = ±1.
x+1 1−x
90 IMO Problems, Theorems, and Methods: Algebra

(ii) Setting Values

Setting special values to the independent variable within the domain of the
function to uncover hidden conditions in the problem. These new condi-
tions are used to simplify the functional equation and subsequently deter-
mine the unknown function.

Example 2.13. Solve the functional equation: for any x, y ∈ R,

f (x + y) + f (x − y) = 2f (x) cos y.

Solution. Set x = 0 and y = t to get

f (t) + f (−t) = 2f (0) cos t. (I)

π π
Set x = 2 + t and y = 2 to get

f (π + t) + f (t) = 0. (II)
π π
Set x = 2 and y = 2 + t to get
π
f (π + t) + f (−t) = −2f sin t. (III)
2
π
From (I) + (II) − (III), f (t) = f (0) cos  πt+ f 2 sin t, i.e., f (x) =
a cos x + b sin x, where a = f (0) and b = f 2 are constants. Upon verifi-
cation, f (x) = a cos x + b sin x satisfies the conditions.
(iii) Mathematical Induction
This method is suitable for functional equations whose domain is the set of
positive integers. It begins with concrete calculations of f (1), f (2), f (3), . . .,
leading to a conjecture about the expression of f (n), and then use mathe-
matical induction to prove the conjecture.

Example 2.14. Solve the functional equation: f (n+1) = cos θ+f (n) sin θ,
where f (1) = cos θ with θ ∈ [0, 2π) and n ∈ N+ .

Solution. From the given conditions,

f (2) = cos θ + cos θ sin θ = cos θ(1 + sin θ),

f (3) = cos θ + cos θ(1 + sin θ) sin θ = cos θ(1 + sin θ + sin2 θ),
f (4) = cos θ + cos θ(1 + sin θ + sin2 θ) sin θ
= cos θ(1 + sin θ + sin2 θ + sin3 θ).
 Function Problems 91

From this, we conjecture

cos θ(1 − sinn θ)
f (n) = cos θ(1 + sin θ + sin2 θ + · · · + sinn−1 θ) = ,
1 − sin θ
which can be easily proven by mathematical induction.

Additionally, the properties of the function itself, such as monotonicity,

continuity, boundedness, periodicity, etc., are quite important in solving
functional equations. Whether explicitly stated as conditions in the prob-
lem or implicitly given from relationships the unknown function meets, the
properties of the function cannot be overlooked.

Example 2.15. Find all strictly monotonic functions f : R → R satisfying

f (f (x) + y) = f (x + y) + f (0)

for all real numbers x and y.

Solution. Setting y = −x, we have f (f (x)−x) = 2f (0). Since f is strictly

monotonic, it must be injective. For any real numbers x and y,

f (f (x) − x) = 2f (0) = f (f (y) − y),

so f (x) − x = f (y) − y, implying f (x) − x is a constant. Upon verification,

all solutions are f (x) = x + c, where c is any given real number.

If the solution of the functional equation is restricted to polynomial

functions, one can employ techniques specific to polynomials, such as ana-
lyzing degrees and coefficients, to solve polynomial functional equations.
Such techniques will be presented in Chapter 5.

2.1.7 Other important theorems

Theorem 2.1 (Bolzano’s Theorem). If a function y = f (x) is continu-
ous on a closed interval [a, b] and f (a) · f (b) < 0, then there exists a zero
of y = f (x) in the interval (a, b).

Theorem 2.2 (Boundedness Theorem). If a function f (x) is continu-

ous on a closed interval [a, b], then f (x) is bounded on [a, b].

Theorem 2.3 (Rolle’s Theorem). If a function f (x) is continuous on a

closed interval [a, b], differentiable on the open interval (a, b), and f (a) =
f (b), then there exists at least one point ξ in (a, b) such that f  (ξ) = 0.
92 IMO Problems, Theorems, and Methods: Algebra

Theorem 2.4 (Lagrange’s Mean Value Theorem). If a function f (x)

is continuous on a closed interval [a, b] and differentiable on the open interval
(a, b), then there exists at least one point ξ in (a, b) such that
f (b) − f (a)
f  (ξ) = .
b−a
Rigorous proofs of these theorems belong to the realm of advanced math-
ematics and are not required to be mastered here. However, understanding
them intuitively is relatively straightforward and helps deepen the under-
standing of functions. So they are provided for reference.

2.2 Problems and Solutions

2.2.1 Proving properties
Problem 2.1 (IMO 10-5, proposed by the German Democratic
Republic). Let f be a real-valued function defined for all real numbers x
such that, for some positive constant a, the equation
1 
f (x + a) = + f (x) − (f (x))2 (*)
2
holds for all x.

(a) Prove that the function f is periodic (i.e., there exists a positive number
b such that f (x + b) = f (x) for all x).
(b) For a = 1, give an example of a non-constant function with the required
properties.

Proof. (a) Since the equation (*) holds for any x ∈ R, we have
f (x + 2a) = f ((x + a) + a)
1 
= + f (x + a) − (f (x + a))2
2

1 

 + f (x) − (f (x))2
1  2
= +   
2  1 
− + f (x) − (f (x))2 + f (x) − (f (x))2
4

1 1
= + − f (x) + (f (x))2
2 4
 
1 1 
= +  − f (x).
2 2
 Function Problems 93

Since f (x + a) ≥ 1
2 for any x ∈ R, we see that f (x) ≥ 12 . Therefore
1 1
f (x + 2a) = + f (x) − = f (x).
2 2
Thus, f (x) is a periodic function with a period of 2a.
(b) When a = 1, the following is a periodic function with the prescribed
property:
1 1  πx 
f (x) = + sin .
2 2 2
Its period is 2.

Note. (i) If λ, μ are distinct constants and f (x) is a function on R satis-

fying
a 
f (x + λ) = ± b + af (x + μ) − (f (x + μ))2 ,
2
where a, b are real numbers and a = 0, then f (x) is a periodic function with
a period of 2|λ − μ|.
In particular, when a = 1, b = 0, λ > 0, and μ = 0, the function f (x)
above becomes the function in Problem 2.1.
(ii) Suppose {λn } is an arithmetic sequence with a non-zero common
difference and f (x) is a function on R satisfying

 n
a  
f (x + λ1 ) = ± b + (af (x + λi ) − (f (x + λi ))2 ),
2 i=2

where a, b are real numbers and a = 0. Then f (x) is a periodic function

with a period of n|d|, where d is the common difference of the arithmetic
sequence.
Score Situation This particular problem saw the following distribution of scores among
contestants: 38 contestants scored 7 points, 2 contestants scored 6 points, 20 contestants
scored 5 points, 3 contestants scored 4 points, 8 contestants scored 3 points, 1 contestant
scored 2 points, 2 contestants scored 1 point, and 17 contestants scored 0 point. The average
score of this problem is 4.593, indicating that it was simple.
Among the top five teams in the team scores, the German Democratic Republic team
achieved a total score of 304 points, the Soviet Union team achieved a total score of
298 points, the Hungary team achieved a total score of 291 points, the United Kingdom
team achieved a total score of 263 points, and the Poland team achieved a total score of
262 points.
94 IMO Problems, Theorems, and Methods: Algebra

The gold medal cutoff for this IMO was set at 39 points (with 22 contestants earning gold
medals), the silver medal cutoff was 33 points (with 22 contestants earning silver medals),
and the bronze medal cutoff was 26 points (with 20 contestants earning bronze medals).
In this IMO, a total of 16 contestants achieved a perfect score of 40 points.

Problem 2.2 (IMO 11-2, proposed by Hungary). Let a1 , a2 , . . . , an

be real constants, x a real variable, and

1 1
f (x) = cos(a1 + x) + cos(a2 + x) + cos(a3 + x)
2 4
1
+ ···+ cos(an + x).
2n−1

Given that f (x1 ) = f (x2 ) = 0, prove that x2 −x1 = mπ for some integer m.

Proof. For all real numbers x, obviously cos(ai + x) ≥ −1. Therefore,

1 1 1
f (−a1 ) = 1 + cos(a2 − a1 ) + cos(a3 − a1 ) + · · · + n−1 cos(an − a1 )
2 4 2
1 1 1
≥ 1 − − − · · · − n−1
2 4 2
1
= n−1 > 0.
2

Hence, f (x) is not identically zero. Furthermore,

n
 1
f (x) = (cos ak · cos x − sin ak · sin x)
2k−1
k=1
 n
  n

 1  1
= cos ak cos x − sin ak sin x
2k−1 2k−1
k=1 k=1

= A cos x − B sin x,
n 1 n 1
where A = k=1 2k−1 cos ak and B = k=1 2k−1 sin ak , and they cannot
both be zero. Otherwise, f (x) would be identically zero, leading to a con-
tradiction.
If A = 0, then

f (x1 ) = A cos x1 − B sin x1 = 0,

f (x2 ) = A cos x2 − B sin x2 = 0.
 Function Problems 95

In this case, if sin x1 = 0, then cos x1 = 0, which is a contradiction. Thus,

sin x1 = 0, and similarly, sin x2 = 0. Therefore, cot x1 = cot x2 = B
A.
If A = 0, then B = 0. From f (x1 ) = f (x2 ) = 0, we have sin x1 =
sin x2 = 0.
Both cases imply x2 − x1 = mπ, where m is an integer.
Note. We can also solve this problem from the perspective of vectors.
Suppose vectors v1 , v2 , . . . , vn begin at the origin (0, 0), with lengths
of 1, 12 , 14 , . . . , 2n−1
1
and angles a1 , a2 , . . . , an between themselves and the
x-axis, respectively. Then the given expression in the problem is the sum
of x-coordinates of these vectors rotated by angle x.
Let r be the length of V = v1 + v2 + · · · + vn . It is evident that r > 0,
and thus the angle between V rorated by angle x and the x-axis should be
2k+1
2 π.
Furthermore, there are several similar problems:
π
• (Estonia Team Selection Test 2002, Problem 5). Let 0 < α < 2
and x1 , x2 , . . . , xn be real numbers such that

sin x1 + sin x2 + · · · + sin xn ≥ n · sin α.

Prove that

sin(x1 − α) + sin(x2 − α) + · · · + sin(xn − α) ≥ 0.

• (William Lowell Putnam Mathematical Competition 1967, A1).

Let f (x) = a1 sin x + a2 sin 2x + · · · + an sin nx, where a1 , a2 , . . . , an are
real numbers. Given that |f (x)| ≤ | sin x| for all real x, prove that |a1 +
2a2 + · · · + nan | ≤ 1.

Score Situation This particular problem saw the following distribution of scores among
contestants: 34 contestants scored 7 points, 5 contestants scored 6 points, 12 contestants
scored 5 points, 32 contestants scored 4 points, 6 contestants scored 3 points, 5 contestants
scored 2 points, 5 contestants scored 1 point, and 13 contestants scored 0 point. The average
score of this problem is 4.366, indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Hungary team scored 53 points (with a total team score of 247 points), the German
Democratic Republic team scored 45 points (with a total team score of 240 points), the
Soviet Union team scored 52 points (with a total team score of 231 points), the Romania
team scored 36 points (with a total team score of 219 points), and the United Kingdom team
scored 43 points (with a total team score of 193 points).
96 IMO Problems, Theorems, and Methods: Algebra

The gold medal cutoff for this IMO was set at 40 points (with 3 contestants earning gold
medals), the silver medal cutoff was 30 points (with 20 contestants earning silver medals),
and the bronze medal cutoff was 24 points (with 21 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 40 points, namely Tibor
Fiala from Hungary, Vladimir Drinfeld from the Soviet Union, and Simon Phillips Norton from
the United Kingdom.

Problem 2.3 (IMO 14-5, proposed by Bulgaria). Let f and g be

real-valued functions defined for all real values of x and y and satisfying
the equation
f (x + y) + f (x − y) = 2f (x)g(y)
for all x, y ∈ R.
Prove that if f (x) is not identically zero and |f (x)| ≤ 1 for all x ∈ R,
then |g(y)| ≤ 1 for all y ∈ R.
Proof 1. Let M be the maximum value of |f (x)|. Suppose there exists a
y0 ∈ R such that |g(y0 )| = 1 + r, where r > 0. Then for any x ∈ R,
2|f (x)||g(y0 )| = |f (x + y0 ) + f (x − y0 )|
≤ |f (x + y0 )| + |f (x − y0 )|
≤ 2M.
Thus,
M M
|f (x)| ≤ = = M − δ < M, δ > 0,
|g(y0 )| 1+r
which contradicts the definition of M . Therefore, the conclusion holds.
Proof 2. Since f (x) is not identically zero, there exists a real number a
such that f (a) = 0. Substituting x = a + ny into the given functional
equation and rearranging, we get
f (a + (n + 1)y) − 2f (a + ny)g(y) + f (a + (n − 1)y) = 0. (1)
With y fixed, let f (a + ny) be denoted as fn and g(y) as g. The equa-
tion (1) can be written as a recurrence relation
fn+1 − 2gfn + fn−1 = 0. (2)
If the conclusion does not hold, then there exists a real number y0
such that |g(y0 )| > 1. For y = y0 , the equation r2 − 2g(y
 0 )r + 1 = 0 has
two distinct real
 roots r1 and r2 , where r1 = g(y 0 ) + (g(y0 ))2 − 1 and
r2 = g(y0 ) − (g(y0 )) − 1.
2
 Function Problems 97

Therefore, the solution to the recurrence relation (2) is fn = b1 r1n +b2 r2n .
Clearly, r1 > 1, and if b1 = 0, then limn→+∞ fn = ∞, which contradicts
the assumption. If b1 = 0, then b2 = 0, otherwise fn would be identically
zero. Furthermore, 0 < r2 < 1, so limn→−∞ fn = limn→−∞ b2 r2n = ∞,
which also contradicts the assumption.
In conclusion, |g(y)| ≤ 1 for all y ∈ R.
Note. It is evident that the recurrence relation fn+1 − 2gfn + fn−1 = 0 is
satisfied by the functions fn = cos nx and g = cos x. This is related to the
Chebyshev polynomials
Tn (cos θ) = cos nθ,
which will be presented in Chapter 5.
Furthermore, there are several similar problems:

• (All-Russian Mathematical Olympiad 2018, Regional Round,

Grade 11, Problem 7). A function f (x) is defined over the entire real
axis and for all real numbers x and y, it satisfies the condition
   
x+y x−y
f (x) + f (y) = 2f f .
2 2
Is f (x) necessarily an even function?
• (William Lowell Putnam Mathematical Competition 1997, B2).
Let f be a twice-differentiable real-valued function satisfying
f (x) + f  (x) = −xg(x)f  (x),
where g(x) ≥ 0 for all real x. Prove that |f (x)| is bounded.
• (Asian Pacific Mathematics Olympiad 1989, Problem 5). Let
g(x) be the composition inverse function to f (x), determine all functions
f from the reals to the reals such that:
(i) f (x) is strictly increasing;
(ii) f (x) + g(x) = 2x for all real x.
(Note: f and g are said to be composition inverse if f (g(x)) = x and
g(f (x)) = x for all real x).

Score Situation This particular problem saw the following distribution of scores among
contestants: 9 contestants scored 7 points, 2 contestants scored 6 points, 2 contestants
scored 5 points, no contestant scored 4 points, 1 contestant scored 3 points, 1 contestant
scored 2 points, 3 contestants scored 1 point, and 15 contestants scored 0 point. The average
score of this problem is 2.818, indicating that it had a certain level of difficulty.
98 IMO Problems, Theorems, and Methods: Algebra

Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 54 points (with a total team score of 270 points), the Hungary
team scored 40 points (with a total team score of 263 points), the German Democratic
Republic team scored 46 points (with a total team score of 239 points), the Romania team
scored 36 points (with a total team score of 208 points), and the United Kingdom team
scored 31 points (with a total team score of 179 points).
The gold medal cutoff for this IMO was set at 40 points (with 8 contestants earning gold
medals), the silver medal cutoff was 30 points (with 16 contestants earning silver medals),
and the bronze medal cutoff was 19 points (with 30 contestants earning bronze medals).
In this IMO, a total of eight contestants achieved a perfect score of 40 points.

Problem 2.4 (IMO 15-5, proposed by Poland). Suppose G is a set

of non-constant functions of a real variable x of the form

f (x) = ax + b, where a and b are real numbers,

and G has the following properties:

(a) if f and g are in G, then g ◦ f is in G; here (g ◦ f )(x) = g(f (x));

(b) if f is in G, then its inverse f −1 is in G; here the inverse of f (x) = ax+b
is f −1 (x) = x−b
a ;
(c) for every f in G, there exists a real number xf such that f (xf ) = xf .

Prove that there exists a real number k such that f (k) = k for all f in G.
Proof 1. First, we prove that

f (x) = x + b ∈ G ⇒ b = 0. (1)

From (c), there exists an xf ∈ R such that f (xf ) = xf + b = xf , so b = 0.

Next, we prove that if f (x) = ax + b, then b is uniquely determined
by a.
If g1 (x) = ax + b1 and g2 (x) = ax + b2 are two functions in G, then from
(a) and (b), g1−1 (g2 (x)) = (ax+ba2 )−b1 = x + b2 −b
a
1
is also a function in G.
From (1), it is known that b1 = b2 .
Finally, we prove that there exists a real number k such that f (k) = k
for all f (x) = ax + b ∈ G.
For a = 1, the function f (x) = x in G obviously satisfies f (k) = k for
all real numbers k. Consider m(x) = ax + b and n(x) = cx + d in G, where
a = 1 and c = 1. From (c), there exist xm , xn such that axm + b = xm and
b d
cxn + d = xn , i.e., xm = − a−1 and xn = − c−1 .
 Function Problems 99

From (a), m(n(x)) = acx + ad + b and n(m(x)) = acx + bc + d are both

in G, and ad + b = bc + d. Hence
b d
− =− .
a−1 c−1
Therefore, k = xm = xn .
Proof 2. From (a) and (b), it is known that G forms a group with respect
to the composition operation “◦.” The function i(x) = x serves as the
identity element of G, as for any real number x, it holds that
(f ◦ i)(x) = (i ◦ f )(x) = f (x).
For each f (x) = ax + b ∈ G, it can be associated with its coefficient a,
which is multiplicative, i.e., if f (x) = ax + b and g(x) = cx + d, then g ◦ f
is associated with ac.
It is evident that each linear function f (x) = ax + b(a = 1) has a fixed
b
point xf = 1−a , geometrically the intersection of the lines y = ax + b and
y = x.
For a = 1, the function f (x) = ax + b = x + b has a fixed point if and
only if b = 0. In this case, the function is i(x) = x, and every point is a
fixed point.
Now consider the composite functions m = f ◦ g and n = g ◦ f , where
f (x) = ax + b and g(x) = cx + d. It is easy to see that the slopes of m
and n are both ac, and the slopes of m−1 and n−1 are both ac 1
. Therefore,
−1 −1
the slope of m ◦ n is 1. Since m ◦ n ∈ G, it is the identity element of
G, so
m−1 ◦ n = i ⇒ m ◦ m−1 ◦ n = m ◦ i ⇒ n = m ⇒ f ◦ g = g ◦ f.
Thus, the functions in G are commutative with respect to the compo-
sition operation “◦.” If G contains only the identity element, then there
is no need to prove further. If G contains f (x) = ax + b(a = 1) and
b
g(x) = cx + d(c = 1), then the fixed points of these functions are 1−a
d
and 1−c , respectively. Since f ◦ g = g ◦ f and
(f ◦ g)(x) = f (cx + d) = acx + ad + b,
(g ◦ f )(x) = g(ax + b) = acx + bc + d,
b d
we have acx + ad + b = acx + bc + d. Thus 1−a = 1−c , indicating f (x) and
g(x) have a common fixed point.
100 IMO Problems, Theorems, and Methods: Algebra

Since f (x) and g(x) are arbitrary functions in G other than the identity
b
element, by setting k = 1−a , the number k is a common fixed point for all
functions in G.

Score Situation This particular problem saw the following distribution of scores among
contestants: 62 contestants scored 6 points, 3 contestants scored 5 points, 2 contestants
scored 4 points, 2 contestants scored 3 points, 2 contestants scored 2 points, 17 contestants
scored 1 point, and 37 contestants scored 0 point. The average score of this problem is 3.376,
indicating that it was relatively straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 48 points (with a total team score of 254 points), the Hungary
team scored 45 points (with a total team score of 215 points), the German Democratic
Republic team scored 40 points (with a total team score of 188 points), the Poland team
scored 42 points (with a total team score of 174 points), and the United Kingdom team
scored 24 points (with a total team score of 164 points).
The gold medal cutoff for this IMO was set at 35 points (with 5 contestants earning gold
medals), the silver medal cutoff was 27 points (with 15 contestants earning silver medals),
and the bronze medal cutoff was 17 points (with 48 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Sergei
Konyagin from the Soviet Union.

Problem 2.5 (IMO 19-6, proposed by Bulgaria). Let f (n) be a func-

tion defined on the set of all positive integers and having all its values in
the same set. Prove that if

f (n + 1) > f (f (n))

for each positive integer n, then

f (n) = n for each n.

Proof. We employ mathematical induction to prove that for any positive

integer n, if m ≥ n is a positive integer, then f (m) ≥ n.
For n = 1, the proposition is evidently true.
Suppose the proposition holds for n = k − 1, then for m ≥ k, by the
inductive hypothesis, f (m − 1) ≥ k − 1. Thus,

f (m) > f (f (m − 1)) ≥ k − 1,

which implies f (m) ≥ k. Therefore, the proposition holds for n = k.

 Function Problems 101

From this it follows that f (n) ≥ n and f (n + 1) > f (f (n)) ≥ f (n),

indicating that f (n) is monotonically increasing. If there exists a positive
integer n0 such that f (n0 ) > n0 , then f (n0 ) ≥ n0 + 1 and f (f (n0 )) ≥
f (n0 + 1), leading to a contradiction.
Hence, f (n) = n for any positive integer n.

Note. In this problem, a proof by contradiction can also be used to show

that if m ≥ n, then f (m) ≥ n.
Furthermore, there is a similar problem:

• (Belarus Team Selection Test 2000, Problem 4). Does there exist
a function f : N → N such that

f (f (n − 1)) = f (n + 1) − f (n)

for all n ≥ 2?

Score Situation This particular problem saw the following distribution of scores among con-
testants: 10 contestants scored 8 points, 1 contestant scored 7 points, no contestant scored
6 points, no contestant scored 5 points, 3 contestants scored 4 points, 3 contestants scored
3 points, no contestant scored 2 points, 5 contestants scored 1 point, and 15 contestants
scored 0 point. The average score for this problem is 3.054, indicating that it was relatively
straightforward.
Among the top five teams in the team scores, the United States team achieved a total
score of 202 points, the Soviet Union team achieved a total score of 192 points, the Hungary
team achieved a total score of 190 points, the United Kingdom team achieved a total score
of 190 points, and the Netherlands team achieved a total score of 185 points.
The gold medal cutoff for this IMO was set at 34 points (with 13 contestants earning gold
medals), the silver medal cutoff was 24 points (with 29 contestants earning silver medals),
and the bronze medal cutoff was 17 points (with 35 contestants earning bronze medals).
In this IMO, a total of five contestants achieved a perfect score of 40 points.

Problem 2.6 (IMO 28-4, proposed by Vietnam). Prove that there

is no function f from the set of non-negative integers into itself such that
f (f (n)) = n + 1987 for every n.

Proof 1. If such a function exists, then

f (n + 1987) = f (f (f (n))) = f (n) + 1987

102 IMO Problems, Theorems, and Methods: Algebra

for all n ∈ N. Following this pattern,

f (n + 1987k) = f ((n + 1987(k − 1)) + 1987)

= f (n + 1987(k − 1)) + 1987
= f (n) + 1987(k − 1) + 1987
= f (n) + 1987k.

Let r ∈ N with r ≤ 1986 and suppose

f (r) = 1987k + l, (1)

where k, l ∈ N and l ≤ 1986. By definition,

f (f (r)) = r + 1987, f (f (r)) = f (1987k + l) = f (l) + 1987k,

implying

r + 1987 = f (l) + 1987k. (2)

Since r ≤ 1986, it follows that k = 0 or k = 1.

If k = 1, then from (1) and (2),

f (r) = 1987 + l, f (l) = r < 1987 + l = f (r),

indicating r = l.
If k = 0, then from (1) and (2),

f (r) = l, f (l) = r + 1987 > l = f (r),

indicating r = l.
Hence, integers in the set {0, 1, 2, . . . , 1986} always pair up such that

f (l) = r, f (r) = 1987 + l, or f (r) = l, f (l) = 1987 + r,

with r = l. However, the set {0, 1, 2, . . . , 1986} has 1987 integers and 1987
is odd, so they cannot be paired up.
Therefore, such a function f does not exist.
Proof 2. By contradiction, suppose such a function f exists.
It is evident that f is injective: If f (n1 ) = f (n2 ), then f (f (n1 )) =
f (f (n2 )), implying n1 + 1987 = n2 + 1987, thus n1 = n2 . Therefore, f (0),
f (1), . . . , f (1986) are 1987 distinct integers. Let M = {f (0), f (1), . . . ,
f (1986)}.
 Function Problems 103

As shown in Proof 1, f (n + 1987k) = f (n) + 1987k. Hence, f (n) ≥

1987k for n ≥ 1987k. For any m ∈ {0, 1, . . . , 1986}, since f (f (m)) = m +
1987 < 2 × 1987, we get f (m) < 2 × 1987. Let I1 = {0, 1, . . . , 1986} and
I2 = {1987, 1988, . . . , 3973}. Then M ⊆ I1 ∪ I2 . If |M ∩ I1 | = m1 and
|M ∩ I2 | = m2 , then m1 + m2 = 1987.
For any f (a) ∈ M ∩I1 , there exists a b(0 ≤ b < 1987) such that f (a) = b.
Therefore,
f (b) = f (f (a)) = a + 1987,
implying f (b) ∈ M ∩ I2 . Since different f (a) correspond to different f (b),
it follows that m1 ≤ m2 .
Similarly, m1 ≥ m2 , thus m1 = m2 . Consequently, m1 + m2 = 2m1 =
1987, which is a contradiction.
Therefore, such a function f does not exist.
Note. There are several similar problems:

• (MathPath Summer Program Qualifying Test 2023, Problem

6). Let f (n) be a function that takes positive integers as inputs and also
gives positive integers as outputs. Is it possible to find a function such
that
f (f (n)) = n + 5
for every positive integer n?
Either prove that this task is impossible, or else describe how to con-
struct a function with the desired property.
• (Vietnam Mathematical Olympiad 1993, Problem 3). Find a
function f : N → N such that
f (f (n)) = 1993n1945 for all n ∈ N.
• (British Mathematical Olympiad 1992, 1st Round, Problem 5).
Let f be a function mapping positive integers into positive integers. Sup-
pose that
f (n + 1) > f (n) and f (f (n)) = 3n for all positive integers n.
Determine f (1992).
• (Japan Mathematical Olympiad 1991, Final Round, Prob-
lem 2). The mappings p, q from N+ to N+ are defined as follows:
p(1) = 2, p(2) = 3, p(3) = 4, p(4) = 1, p(n) = n(n ≥ 5),
q(1) = 3, q(2) = 4, q(3) = 2, q(4) = 1, q(n) = n(n ≥ 5).
104 IMO Problems, Theorems, and Methods: Algebra

(a) Suppose f is a mapping from N+ to N+ such that f (f (n)) = p(n)+2.

Give an example for f .
(b) Prove that f (f (n)) = q(n) + 2 does not hold for any mapping f from
N+ to N+ .

Score Situation This particular problem saw the following distribution of scores among
contestants: 91 contestants scored 7 points, 7 contestants scored 6 points, 12 contestants
scored 5 points, 1 contestant scored 4 points, 10 contestants scored 3 points, 21 contestants
scored 2 points, 20 contestants scored 1 point, and 75 contestants scored 0 point. The
average score for this problem is 3.523, indicating that it was relatively straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Romania team scored 42 points (with a total team score of 250 points), the Germany
team scored 42 points (with a total team score of 248 points), the Soviet Union team scored
42 points (with a total team score of 235 points), the German Democratic Republic team
scored 38 points (with a total team score of 231 points), and the United States team scored
36 points (with a total team score of 220 points).
The gold medal cutoff for this IMO was set at 42 points (with 22 contestants earning gold
medals), the silver medal cutoff was 32 points (with 42 contestants earning silver medals),
and the bronze medal cutoff was 18 points (with 56 contestants earning bronze medals).
In this IMO, a total of 22 contestants achieved a perfect score of 42 points.
Problem 2.7 (IMO 34-5, proposed by Germany). Does there exist a
function f : N+ → N+ such that:
(a) f (1) = 2;
(b) f (f (n)) = f (n) + n for all n ∈ N+ ;
(c) f (n) < f (n + 1) for all n ∈ N+ ?

Solution.
√ Such a function
√ exists. Let f (n) = αn + β for n ∈ N+ , where
α = 2 and β = 2 , and x denotes the greatest integer not exceed-
5+1 5−1

ing x. Next, we prove that f (n) satisfies the conditions:

√
f (1) = α + β =  5 = 2,
f (n + 1) = (n + 1)α + β = nα + α + β
≥ nα + 1 + β = f (n) + 1 > f (n).
Thus, f (n) satisfies conditions (a) and (c), and is a function from N+
to N+ .
f (f (n)) = ααn + β + β = (β + 1)αn + β + β
= αn + β + βαn + β + β
= f (n) + βαn + β + β.
 Function Problems 105

Since αn + β is not an integer,

βαn + β + β < β(αn + β) + β = n + β 2 + β = n + 1,

βαn + β + β > β(αn + β − 1) + β = n + β 2 > n.

Therefore, βαn + β + β = n, and thus f (f (n)) = f (n) + n.

In conclusion, f (n) = αn + β satisfies all conditions.
√ 
5+1 1
Note. The constructed f (n) is not unique; f (n) = 2 n + 2 also meets
the conditions. In fact, from f (n) > n and the given condition,
 −1
f (f (n)) f (n)
1< =1+ < 2,
f (n) n

suggesting that the ratio f (n)

n approaches a limit value α as n tends to
√
−1
infinity, satisfying α = 1 + α (α ≥ 1). Hence α = 5+1 2 .
This leads to the conjecture that f (n) is an integer close to αn, and
thus f (n) = αn + β is constructed, and upon verification, it is sufficient
that 1 − α−1 ≤ β ≤ α−1 .
Furthermore, there are several similar problems:

• (USA TST Selection Test 2017, Problem 6). A sequence of positive

integers (an )n≥1 is of Fibonacci type if it satisfies the recursive relation
an+2 = an+1 + an for all n ≥ 1. Is it possible to partition the set of
positive integers into an infinite number of Fibonacci type sequences?
• (From the “Problems in Mathematics” in the Hungarian jour-
nal KöMaL, January 2001, A.244). A sequence of numbers is called
of Fibonacci-type if each term, after the first two, is the sum of the
previous two.
Prove that the set of positive integers can be partitioned into the dis-
joint union of infinite Fibonacci-type sequences.
• (From the “Problems in Mathematics” in the Hungarian √ jour-
nal KöMaL, September 2000, B.3429). Let q = 1+2 5 and f : N →
N be a function satisfying

1
|f (n) − qn| <
q

for every n ∈ N. Prove that f (f (n)) = f (n) + n.

106 IMO Problems, Theorems, and Methods: Algebra

• (Estonia Team Selection Test 2000, Problem 4). Find all functions
f : N → N such that

f (f (f (n))) + f (f (n)) + f (n) = 3n

for all n ∈ N.
• (Estonia Team Selection Test 2000, Problem 6). We call an infi-
nite sequence of positive integers an F -sequence if every term of this
sequence (starting from the third term) equals the sum of the two pre-
ceding terms. Is it possible to decompose the set of all positive integers
into
(a) a finite;
(b) an infinite
number of F -sequences having no common members?
• (William Lowell Putnam Mathematical Competition 1988, A5).
Prove that there exists a unique function f from the set R+ of positive
real numbers to R+ such that

f (f (x)) = 6x − f (x) and f (x) > 0 for all x > 0.

• Problem 2.10 (IMO 20-3) in this chapter.

Score Situation This particular problem saw the following distribution of scores among
contestants: 91 contestants scored 7 points, 27 contestants scored 6 points, 19 contestants
scored 5 points, 32 contestants scored 4 points, 65 contestants scored 3 points, 44 contestants
scored 2 points, 92 contestants scored 1 point, and 43 contestants scored 0 point. The
average score for this problem is 3.383, indicating that it was relatively straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 39 points (with a total team score of 215 points), the Germany team
scored 40 points (with a total team score of 189 points), the Bulgaria team scored 42 points
(with a total team score of 178 points), the Russia team scored 38 points (with a total
team score of 177 points), and the Chinese Taiwan team scored 33 points (with a total team
score of 162 points).
The gold medal cutoff for this IMO was set at 30 points (with 35 contestants earning gold
medals), the silver medal cutoff was 20 points (with 66 contestants earning silver medals),
and the bronze medal cutoff was 11 points (with 97 contestants earning bronze medals).
In this IMO, only two contestants achieved a perfect score of 42 points, namely Hong
Zhou from China and Hung-Wu Wu from Chinese Taiwan.
 Function Problems 107

Problem 2.8 (IMO 52-3, proposed by Belarus). Let f : R → R be

a real-valued function defined on the set of real numbers that satisfies
f (x + y) ≤ yf (x) + f (f (x))
for all x, y ∈ R. Prove that f (x) = 0 for all x ≤ 0.
Proof 1. By setting y = f (x) − x in the given inequality, we obtain
f (f (x)) ≤ (f (x) − x)f (x) + f (f (x)),
from which it follows that (f (x)−x)f (x) ≥ 0. Consequently, for any x ∈ R,
(f (f (x)) − f (x))f (f (x)) ≥ 0.
By setting y = 0 in the given inequality, we have f (x) ≤ f (f (x)). Thus
f (f (x)) ≥ 0 or f (x) = f (f (x)) < 0. (1)

Case 1: There exists an x0 ∈ R such that f (x0 ) > 0.

For any y ∈ R,
f (x0 + y) ≤ yf (x0 ) + f (f (x0 )).
Hence, f (x0 + y) < 0 for any y < − f (f (x0 ))
f (x0 ) . Consequently, f (z) < 0 for
f (f (x0 ))
any real number z < x0 − f (x .
0 ) 
Therefore, when z < min 0, x0 − f (f (x0 ))
f (x0 ) , both z < 0 and f (z) < 0
are satisfied. Combining them with (f (z) − z)f (z) ≥ 0 leads to
 
f (f (x0 )) f (f (x0 ))
f (z) ≤ z < min 0, x0 − ≤ x0 − ,
f (x0 ) f (x0 )
implying f (f (z)) = f (x0 + (f (z) − x0 )) < 0.
 
f (f (x0 ))
Thus, f (f (z)) = f (z) < 0 for any z < min 0, x0 − f (x0 ) . As a
result, for any y ∈ R,
f (z + y) ≤ yf (z) + f (f (z)) = (y + 1)f (z).
Setting y = x0 − z yields f (x0 ) ≤ (1 + x0 − z)f (z). When z is sufficiently
small, 1 + x0 − z > 0 and f (z) < 0, leading to f (x0 ) < 0, a contradiction.
Case 2: f (x) ≤ 0 for any x ∈ R.
In this case,
f (f (x)) ≤ 0. (2)
If f (f (x)) ≥ 0 for any x ∈ R, then f (f (x)) = 0 for any x ∈ R.
Thus, f (0) = f (f (f (x))) = 0. From the conditions, f (x + y) ≤ yf (x) for
108 IMO Problems, Theorems, and Methods: Algebra

any x, y ∈ R, which implies f (0) ≤ −xf (x). For x < 0, it follows that
f (x) ≥ 0, leading to f (x) = 0 for x ≤ 0.
If there exists a z0 such that f (f (z0 )) < 0, then from (1) it is known
that f (z0 ) = f (f (z0 )) < 0. Therefore, there exists a t = f (z0 ) < 0 such
that f (t) = t. And if there exist t1 , t2 both satisfying f (t) = t, where
t1 < t2 < 0, then
t2 = f (t2 ) ≤ (t2 − t1 )f (t1 ) + f (f (t1 )) = (t2 − t1 )t1 + t1 ,
leading to (t2 − t1 )(1 − t1 ) ≤ 0, a contradiction. Thus, such t is unique, and
further, there exists an x1 ∈ R such that f (x1 ) = t.
(Otherwise, f (x) = t for all x ∈ R, implying t ≤ yt + t for any y ∈ R,
but it is not true for y > 0.)
For this x1 , we have f (f (x1 )) = f (x1 ), and from (1), f (f (x1 )) ≥ 0 > t.
Thus f (f (f (x1 ))) ≥ 0.
(Otherwise, from (1), 0 ≤ f (f (x1 )) = f (f (f (x1 ))) < 0, a contradiction.)
Combining with (2) implies
0 = f (f (x1 )) = f (f (f (x1 ))) = f (0).
Thus, it still holds that f (x + y) ≤ yf (x), leading to f (0) = 0 ≤ −xf (x).
Therefore, f (x) ≥ 0 for x < 0, which implies f (x) = 0 for x ≤ 0.
In conclusion, f (x) = 0 for any x ≤ 0.
Proof 2. Set y = t − x, and then the given inequality can be rewritten as
f (t) ≤ tf (x) − xf (x) + f (f (x)). (3)
In (3), by setting t = f (a), x = b and t = f (b), x = a, we have
f (f (a)) − f (f (b)) ≤ f (a)f (b) − bf (b),
f (f (b)) − f (f (a)) ≤ f (a)f (b) − af (a).
Adding the above two inequalities yields 2f (a)f (b) ≥ af (a) + bf (b).
Setting b = 2f (a) gives bf (b) ≥ af (a) + bf (b), so af (a) ≤ 0. Therefore,
f (a) ≤ 0 when a > 0, and f (a) ≥ 0 when a < 0.
Suppose there exists a real number x0 such  that f (x0 ) > 0. Then
 from
x0 f (x0 )−f (f (x0 ))
(3), it is known that f (t) < 0 for every t < min f (x0 ) ,0 , which
contradicts that f (t) ≥ 0 for t < 0.
Therefore, f (x) ≤ 0 for all real numbers x. Combined with f (x) ≥ 0
for x < 0, it follows that f (x) = 0 for any real number x < 0. Setting t = x
in (3), we have f (x) ≤ f (f (x)), and by taking x = −1, we obtain 0 ≤ f (0).
Since f (x) ≤ 0 for any real number x, it follows that f (0) = 0.
 Function Problems 109

In conclusion, f (x) = 0 for any x ≤ 0.

Note. There are several similar problems:

• (Asian Pacific Mathematics Olympiad 2002, Problem 5). Let R

denote the set of all real numbers. Find all functions f from R to R
satisfying:
(i) there are only finitely many s in R such that f (s) = 0;
(ii) f (x4 + y) = x3 f (x) + f (f (y)) for all x, y in R.
• (Romania Team Selection Test 2001, Problem 11). Prove that
there is no function f : (0, +∞) → (0, +∞) such that

f (x + y) ≥ f (x) + yf (f (x))

for all x, y ∈ (0, +∞).

Score Situation This particular problem saw the following distribution of scores among
contestants: 51 contestants scored 7 points, 5 contestants scored 6 points, 3 contestants
scored 5 points, 7 contestants scored 4 points, 13 contestants scored 3 points, 34 contestants
scored 2 points, 57 contestants scored 1 point, and 393 contestants scored 0 point. The
average score for this problem is 1.055, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 189 points), the United States
team scored 35 points (with a total team score of 184 points), the Singapore team scored
23 points (with a total team score of 179 points), the Russia team scored 23 points (with a
total team score of 161 points), and the Thailand team scored 23 points (with a total team
score of 160 points).
The gold medal cutoff for this IMO was set at 28 points (with 54 contestants earning gold
medals), the silver medal cutoff was 22 points (with 90 contestants earning silver medals),
and the bronze medal cutoff was 16 points (with 137 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 42 points, namely Lisa
Sauermann from Germany.

Problem 2.9 (IMO 54-5, proposed by Bulgaria). Let Q+ be the set

of positive rational numbers. Let f : Q+ → R be a function satisfying the
following three conditions:

(i) for all x, y ∈ Q+ , we have f (x)f (y) ≥ f (xy);

(ii) for all x, y ∈ Q+ , we have f (x + y) ≥ f (x) + f (y);
(iii) there exists a rational number a > 1 such that f (a) = a.
110 IMO Problems, Theorems, and Methods: Algebra

Prove that f (x) = x for all x ∈ Q+ .

Proof. Let N+ denote the set of all positive integers. In the inequality (i)
f (x)f (y) ≥ f (xy), (1)
setting x = 1 and y = a gives f (1) ≥ 1. Then, from
f (x + y) ≥ f (x) + f (y), (2)
applying mathematical induction on n, we can deduce for any n ∈ N+ and
x ∈ Q+ ,
f (nx) ≥ nf (x). (3)
In particular,
f (n) ≥ nf (1) ≥ n. (4)
p
For any x = q∈ Q+ and p, q ∈ N+ , f (q) · f ( pq )
≥ f (p) ≥ p, thus
f (x) > 0 for x ∈ Q+ .
From (2), f is strictly increasing, and for any rational number x ≥ 1,
f (x) ≥ f (x) ≥ x > x − 1,
where x denotes the greatest integer not exceeding x.
From (1), using mathematical induction gives (f (x))n ≥ f (xn ). There-
fore,
(f (x))n ≥ f (xn ) > xn − 1,
√
which implies f (x) > n xn − 1 for any rational number x > 1 and n ∈ N+ .
Thus, for any ratinoal number x > 1,
f (x) ≥ x. (5)
(If there exists a rational number x0 > 1 such that 1 < f (x0 ) = y < x0 ,
then for sufficiently large n,
xn0 − y n = (x0 − y)(xn−1
0 + xn−2
0 y + · · · + y n−1 ) > n(x0 − y),

from which xn0 − y n > 1. Therefore, y = f (x0 ) > n xn0 − 1 > y, a contra-
diction.)
From (1) and (5), an = (f (a))n ≥ f (an ) ≥ an , from which f (an ) = an .
Then, for any rational number x > 1, we can choose an n0 ∈ N+ such that
an0 − x > 1. In this case, from (2) and (5),
an0 = f (an0 ) ≥ f (x) + f (an0 − x) ≥ x + (an0 − x) = an0 .
Thus f (x) = x for any rational number x > 1. Combined with f (1) ≥ 1,
we conclude f (1) = 1.
 Function Problems 111

(Otherwise, if f (1) = t > 1, then

t = f (t) ≥ f (1) + f (t − 1) = t + f (t − 1) > t,

a contradiction.)
Finally, for any x ∈ Q+ and any n ∈ N+ , from (1) and (3) it follows
that

nf (x) = f (n)f (x) ≥ f (nx) ≥ nf (x),

  f (m)
so f (nx) = nf (x). Consequently, f m n =
m
n = n for any m, n ∈ N+ .

Note. The condition f (a) = a > 1 is essential. Indeed, for b ≥ 1, the

function f (x) = bx2 satisfies (1) and (2) for any x, y ∈ Q+ , with a unique
fixed point 1b ≤ 1.
Score Situation This particular problem saw the following distribution of scores among
contestants: 135 contestants scored 7 points, 19 contestants scored 6 points, 10 contestants
scored 5 points, no contestant scored 4 points, 11 contestants scored 3 points, 33 contestants
scored 2 points, 84 contestants scored 1 point, and 235 contestants scored 0 point. The
average score for this problem is 2.452, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 208 points), the South Korea
team scored 36 points (with a total team score of 204 points), the United States team scored
40 points (with a total team score of 190 points), the Russia team scored 40 points (with
a total team score of 187 points), and the North Korea team scored 41 points (with a total
team score of 184 points).
The gold medal cutoff for this IMO was set at 31 points (with 45 contestants earning gold
medals), the silver medal cutoff was 24 points (with 92 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 141 contestants earning bronze medals).
In this IMO, no contestant achieved a perfect score of 42 points.

2.2.2 Determining values

Problem 2.10 (IMO 20-3, proposed by the United Kingdom).
The set of all positive integers is the union of two disjoint subsets
{f (1), f (2), . . . , f (n), . . .} and {g(1), g(2), . . . , g(n), . . .}, where

f (1) < f (2) < · · · < f (n) < · · · ,

g(1) < g(2) < · · · < g(n) < · · · ,
and g(n) = f (f (n)) + 1 for all n ≥ 1.
Determine f (240).
112 IMO Problems, Theorems, and Methods: Algebra

Solution 1. Let A = {f (k)} and B = {g(k)}, where k ∈ N+ . Each pos-

itive integer is either in A or B, but not in both. For k = 1, we see that
f (1) is less than both f (k) and g(k), so f (1) = 1.
From the conditions, the nth positive integer not in A is g(n) =
f (f (n)) + 1, implying that in the positive integers 1, 2, . . . , g(n), there are
n integers in B and f (n) integers in A. Thus,

g(n) = f (n) + n, f (f (n)) = f (n) + n − 1.

Since g(n + 1) − g(n) = f (n + 1) + (n + 1) − f (n) − n ≥ 2, we have

g(n) + 1 ∈ A. Hence,

f (1) = 1,
g(1) = 2,
f (2) = 3,
f (f (2)) = f (3) = f (2) + 2 − 1 = 4,
f (f (3)) = f (4) = f (3) + 3 − 1 = 6,
f (f (4)) = f (6) = f (4) + 4 − 1 = 9,
f (f (6)) = f (9) = f (6) + 6 − 1 = 14,
f (f (9)) = f (14) = f (9) + 9 − 1 = 22,
f (f (14)) = f (22) = f (14) + 14 − 1 = 35,
f (f (22)) = f (35) = f (22) + 22 − 1 = 56,
f (f (35)) = f (56) = f (35) + 35 − 1 = 90,
g(35) = f (f (35)) + 1 = 91,
f (57) = 92,
f (f (57)) = f (92) = f (57) + 57 − 1 = 148,
f (f (92)) = f (148) = f (92) + 92 − 1 = 239,
f (f (148)) = f (239) = f (148) + 148 − 1 = 386,
g(148) = f (f (148)) + 1 = f (239) + 1 = 387,
f (240) = 388.

Therefore, f (240) = 388.

 Function Problems 113

Note. The expressions of these two functions are actually

√  √ 
5+1 5+3
f (n) = n and g(n) = n .
2 2

This is a special case of Beatty’s theorem.

◦ Beatty’s Theorem. If two irrational numbers α, β are both greater
than 1 and α1 + β1 = 1, then the functions f (n) = nα and g(n) = nβ
satisfy:

(i) f (n), g(n) are increasing functions from N+ to N+ ;

(ii) f (N+ ) ∪ g(N+ ) = N+ ;
(iii) f (N+ ) ∩ g(N+ ) = ∅.
√ √
5+1 5+3
In this problem, α = 2 and β = 2 .

Solution 2. By Beatty’s theorem, one can choose α, β such that {f (n)}

and {g(n)} satisfy

f (N+ ) ∪ g(N+ ) = N+ and f (N+ ) ∩ g(N+ ) = ∅.

The key is to derive g(n) = f (f (n)) + 1, leading to g(n) = f (n) + n, i.e.,

nβ = nα + n = nα + n

for all n ∈ N+ . If nβ = nα + n, i.e., β = α + 1, then 1
α
1
+ α+1 = 1, yielding
√ √
5+1 5+3
α= 2 , and thus β = 2 .
Therefore,
√ √
f (240) = 240α = 120 × (1 + 5) = 120 + 120 5
= 120 + 120 × 2.236 = 388.

Note. There is a similar problem:

• (From the “Problems in Mathematics” in the Hungarian √ jour-

nal KöMaL, January 2001, A.255). Let q = 1+2 5 and define a
function f on the set of positive integers as follows:
(a) start with f (1) = 2 and let n > 1;
(b) if the values f (1), f (2), . . . , f (n − 1) have already been defined and
n is not among these values, then set f (n) = f (n − 1) + 1;
(c) otherwise consider k, the smallest positive integer, for which f (k) =
n, and let f (n) = n + k.
Prove that qn − 1
q < f (n) ≤ qn + 1
q2 for any positive integer n.
114 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 8 contestants scored 8 points, 2 contestants scored 7 points, 1 contestant scored
6 points, 3 contestants scored 5 points, 7 contestants scored 4 points, 4 contestants scored
3 points, 6 contestants scored 2 points, 8 contestants scored 1 point, and 9 contestants
scored 0 point. The average score for this problem is 3.313, indicating that it was relatively
straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Romania team scored 40 points (with a total team score of 237 points), the United States
team scored 47 points (with a total team score of 225 points), the United Kingdom team
scored 45 points (with a total team score of 201 points), the Vietnam team scored 45 points
(with a total team score of 200 points), and the Czechoslovakia team scored 24 points (with
a total team score of 195 points).
The gold medal cutoff for this IMO was set at 35 points (with 5 contestants earning gold
medals), the silver medal cutoff was 27 points (with 20 contestants earning silver medals),
and the bronze medal cutoff was 22 points (with 38 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Mark
Kleiman from the United States.

Problem 2.11 (IMO 22-6, proposed by Finland). For all non-

negative integers x and y, let f (x, y) be a function satisfying:
(a) f (0, y) = y + 1;
(b) f (x + 1, 0) = f (x, 1);
(c) f (x + 1, y + 1) = f (x, f (x + 1, y)).
Determine f (4, 1981).
Solution. Set x = 0, and from (a) and (b), it is known that f (1, 0) =
f (0, 1) = 2. Substituting x = 0 and y = n − 1 into (c) and utilizing (a), we
have
f (1, n) = f (0, f (1, n − 1))
= f (1, n − 1) + 1
= n + f (1, 0)
= n + 2. (1)
From (c) and (1),
f (2, n) = f (1, f (2, n − 1))
= f (2, n − 1) + 2
= 2n + f (2, 0),
 Function Problems 115

where f (2, 0) = f (1, 1) = 1 + 2 = 3. Thus

f (2, n) = 2n + 3. (2)
From (c) and (2),
f (3, n) + 3 = f (2, f (3, n − 1)) + 3
= 2f (3, n − 1) + 6
= 2(f (3, n − 1) + 3)
= ··················
= 2n (f (3, 0) + 3),
where f (3, 0) = f (2, 1) = 5. Hence
f (3, n) = 2n+3 − 3. (3)
From (c) and (3),
f (4, n) + 3 = f (3, f (4, n − 1)) + 3
= 2f (4,n−1)+3
= ············
. f (4,0)+3
2
.. 2
=2 ,
with 2 repeating n times. Since f (4, 0) + 3 = f (3, 1) + 3 = 24 ,
.
.. 2
f (4, n) = −3 + 22 ,
with 2 repeating n + 3 times. .2
..
Therefore, f (4, 1981) = −3 + 22 , with 2 repeating 1984 times.
Note. This function, defined by a double recursion, is known as the Ack-
ermann function. Wilhelm Ackermann, a student of Hilbert, proved that
this function grows faster than any single-variable recursive function. Com-
puter scientists are also interested in it as it outpaces the growth of any
simple loop program.
Score Situation This particular problem saw the following distribution of scores among
contestants: 44 contestants scored 7 points, 3 contestants scored 6 points, no contestant
scored 5 points, 1 contestant scored 4 points, no contestant scored 3 points, 1 contestant
scored 2 points, no contestant scored 1 point, and 2 contestants scored 0 point. The average
score for this problem is 6.510, indicating that it was simple.
116 IMO Problems, Theorems, and Methods: Algebra

Among the top five teams in the team scores, the scores of this problem are as follows:
the United States team scored 55 points (with a total team score of 314 points), the Germany
team scored 56 points (with a total team score of 312 points), the United Kingdom team
scored 55 points (with a total team score of 301 points), the Austria team scored 55 points
(with a total team score of 290 points), and the Bulgaria team scored 55 points (with a total
team score of 287 points).
The gold medal cutoff for this IMO was set at 41 points (with 36 contestants earning gold
medals), the silver medal cutoff was 34 points (with 37 contestants earning silver medals),
and the bronze medal cutoff was 26 points (with 30 contestants earning bronze medals).
In this IMO, a total of 26 contestants achieved a perfect score of 42 points.
Problem 2.12 (IMO 23-1, proposed by the United Kingdom). A
function f (n) is defined for all positive integers n and takes on non-negative
integer values. Also, for all m and n,
f (m + n) − f (m) − f (n) = 0 or 1,
f (2) = 0, f (3) > 0, and f (9999) = 3333.
Determine f (1982).
Solution. From the given condition,
f (m + n) ≥ f (m) + f (n). (1)
Setting m = n = 1, we obtain f (2) ≥ 2f (1). However, since f (2) = 0
and f (1) is a non-negative integer, it follows that f (1) = 0. Setting m = 2
and n = 1, we get
f (3) − f (2) + f (1) = 0 or 1.
Given that f (3) > 0, it implies f (3) = 1. Now, we prove that f (3k) = k
for k ≤ 3333. From (1),
f (3k) ≥ f (3k − 3) + f (3)
≥ f (3k − 6) + 2f (3)
≥ ············
≥ kf (3)
= k.
If f (3k) > k, then f (3k) ≥ k + 1, leading to
f (9999) ≥ f (9999 − 3k) + f (3k)
≥ 3333 − k + k + 1
> 3333,
which contradicts the given conditions. Therefore, f (3k) = k.
 Function Problems 117

Hence, 1982 = f (3 × 1982) ≥ 3f (1982), i.e., f (1982) ≤ 1982 3 < 661.

However, f (1982) ≥ f (3 × 660) + f (2) = 660, and thus f (1982) = 660.

Note.
n
There exists a function that meets the conditions, such as f (n) =
3 .
Furthermore, there are several similar problems:

• (Korea Winter Program Test 2018, Problem 1). Find all functions
f : R → R satisfying the following conditions:

(i) f (x + y) − f (x) − f (y) ∈ {0, 1} for all x, y ∈ R;

(ii) f (x) = x for all x ∈ R.

• (Korean Mathematical Olympiad 2007, Final Round, Problem

6). Let f : N → N be a function satisfying kf (n) ≤ f (kn) ≤ kf (n)+k−1
for all k, n ∈ N.

(a) Prove that f (a) + f (b) ≤ f (a + b) ≤ f (a) + f (b) + 1 for all

a, b ∈ N.
(b) If f satisfies f (2007n) ≤ 2007f (n) + 200 for every n ∈ N, show that
there exists c ∈ N such that f (2007c) = 2007f (c).

Score Situation This particular problem saw the following distribution of scores among
contestants: 54 contestants scored 7 points, 4 contestants scored 6 points, no contestant
scored 5 points, 20 contestants scored 4 points, 11 contestants scored 3 points, 8 contestants
scored 2 points, 18 contestants scored 1 point, and 4 contestants scored 0 point. The average
score for this problem is 4.613, indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Germany team scored 28 points (with a total team score of 145 points), the Soviet Union
team scored 28 points (with a total team score of 137 points), the German Democratic
Republic team scored 28 points (with a total team score of 136 points), the United States
team scored 25 points (with a total team score of 136 points), and the Vietnam team scored
25 points (with a total team score of 133 points).
The gold medal cutoff for this IMO was set at 37 points (with 10 contestants earn-
ing gold medals), the silver medal cutoff was 30 points (with 20 contestants earning silver
medals), and the bronze medal cutoff was 21 points (with 31 contestants earning bronze
medals).
In this IMO, only three contestants achieved a perfect score of 42 points, namely Bruno
Haible from Germany, Grigori Perelman from the Soviet Union, and Le Tu QuocThang from
Vietnam.
118 IMO Problems, Theorems, and Methods: Algebra

Problem 2.13 (IMO 29-3, proposed by the United Kingdom). A

function f is defined on the positive integers by

f (1) = 1, f (3) = 3, f (2n) = f (n),

f (4n + 1) = 2f (2n + 1) − f (n), f (4n + 3) = 3f (2n + 1) − 2f (n),

for all positive integers n.

Determine the number of positive integers n, less than or equal to 1988,
for which f (n) = n.
Solution. Let the binary representation of n be (as as−1 · · · a1 )2 . We prove
that

f (n) = (a1 a2 · · · as )2 . (1)

For n ∈ {1, 2, 3}, it is evident that (1) holds. Since f (2n) = f (n), it
suffices to consider the case where n is odd. Assume (1) holds for n < k.
For n = k, we consider f (k).
(i) Suppose k = 4m + 1. Let 4m + 1 = (at at−1 · · · a0 )2 , where a0 = 1
and a1 = 0. Then

m = (at at−1 · · · a2 )2 , 2m + 1 = (at at−1 · · · a2 1)2 .

From the inductive hypothesis,

f (2m + 1) = (1a2 a3 · · · at )2 , f (m) = (a2 a3 · · · at )2 ,

leading to

f (4m + 1) = 2f (2m + 1) − f (m)

= (1a2 a3 · · · at )2 + (1a2 a3 · · · at )2 − (a2 a3 · · · at )2
= (10a2 a3 · · · at )2 + (a2 a3 · · · at )2 − (a2 a3 · · · at )2
= (10a2 a3 · · · at )2 = (a0 a1 a2 a3 · · · at )2 ,

confirming (1).
(ii) Let k = 4m + 3 with 4m + 3 = (at at−1 · · · a0 )2 , where a0 = a1 = 1.
Then

4m = (at at−1 · · · a2 00)2 ,

m = (at at−1 · · · a2 )2 ,
2m + 1 = (at at−1 · · · a2 1)2 .
 Function Problems 119

From the inductive hypothesis and given conditions,

f (4m + 3) = 3f (2m + 1) − 2f (m)
= 3(1a2 · · · at )2 − 2(a2 a3 · · · at )2
= (11a2 · · · at )2 + 2(a2 a3 · · · at )2 − 2(a2 a3 · · · at )2
= (11a2 a3 · · · at )2 = (a0 a1 a2 · · · at )2 ,
thereby confirming (1).
Since f (n) = n is equivalent to (a1 a2 · · · at )2 = (at at−1 · · · a1 )2 , where
n = (at at−1 · · · a1 )2 , the binary representation of n satisfying f (n) = n is
“symmetric.” Thus, if the binary representation of n has 2m digits, then
it is entirely determined by the first m digits.
Given the first digit 1, followed by m − 1 digits each chosen from 0 or 1,
there are 2m−1 symmetric binary numbers with 2m digits. Similarly, there
are 2m−1 symmetric binary numbers with 2m − 1 digits.
Since 210 < 1988 < 211 , there are 1 + 1 + 2 + 2 + 4 + 4 + 8 + 8 + 16 + 16 +
32 = 94 symmetric binary numbers less than 2048 that satisfy f (n) = n.
Since 1988 = (11111000100)2, there are only two symmetric binary numbers
exceeding 1988, namely (11111011111)2 and (11111111111)2.
Hence, there are 92 positive integers not exceeding 1988 such that
f (n) = n.
Note. There is a similar problem:

• (Asian Pacific Mathematics Olympiad 2008, Problem 4). Con-

sider a function f : N0 → N0 , where N0 is the set of all non-negative
integers, defined by the following conditions:
(i)f (0) = 0;
(ii)f (2n) = 2f (n);
(iii)f (2n + 1) = n + 2f (n) for all n ≥ 0.
(a) Determine the three sets L := {n|f (n) < f (n + 1)}, E :=
{n|f (n) = f (n + 1)}, and G := {n|f (n) > f (n + 1)}.
(b) For each k ≥ 0, find a formula for ak := max{f (n) : 0 ≤ n ≤ 2k } in
terms of k.
• (British Mathematical Olympiad 2007, 1st Round, Problem 6).
A function f is defined on the set of positive integers by
f (1) = 1, f (2n) = 2f (n),
and nf (2n + 1) = (2n + 1)(f (n) + n) for all n ≥ 1.
120 IMO Problems, Theorems, and Methods: Algebra

(i) Prove that f (n) is always an integer.

(ii) For how many positive integers less than 2007 is f (n) = 2n?

Score Situation This particular problem saw the following distribution of scores among
contestants: 31 contestants scored 7 points, 9 contestants scored 6 points, 4 contestants
scored 5 points, 3 contestants scored 4 points, 1 contestant scored 3 points, 13 contestants
scored 2 points, 132 contestants scored 1 point, and 75 contestants scored 0 point. The
average score for this problem is 1.731, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 36 points (with a total team score of 217 points), the Romania
team scored 35 points (with a total team score of 201 points), the China team scored 17 points
(with a total team score of 201 points), the Germany team scored 25 points (with a total team
score of 174 points), and the Vietnam team scored 12 points (with a total team score of
166 points).
The gold medal cutoff for this IMO was set at 32 points (with 17 contestants earning gold
medals), the silver medal cutoff was 23 points (with 48 contestants earning silver medals),
and the bronze medal cutoff was 14 points (with 65 contestants earning bronze medals).
In this IMO, a total of five contestants achieved a perfect score of 42 points.

2.2.3 Deriving expressions

Problem 2.14 (IMO 24-1, proposed by the United Kingdom). Find

all functions f defined on the set of positive real numbers which take positive
real values and satisfy the conditions:

(a) f (xf (y)) = yf (x) for all positive x and y;

(b) f (x) → 0 as x → +∞.

Solution. First, we prove that 1 is in the range of f . For any x0 > 1, set
y0 = f (x1 0 ) and substitute it into (a) to obtain

f (x0 f (y0 )) = y0 f (x0 ) = 1,

implying that 1 is within the range of f .

Next, suppose y0 ∈ R+ satisfies f (y0 ) = 1. Then, from (a),

f (x) = y0 f (x),

implying y0 = 1, i.e., f (1) = 1.

 Function Problems 121

For any positive real number x, let a = xf (x) and substitute y = x into
(a) to yield f (a) = a. If a > 1, then
f (a2 ) = f (af (a)) = af (a) = a2 .
n n
By mathematical induction, as n → +∞, f (a2 ) = a2 → +∞, contra-
dicting (b).
If a < 1, then
   
1 1
1 = f (1) = f f (a) = af ,
a a
 
implying f a1 = a1 . However, since a1 > 1, following the case for a > 1 by
setting a = a1 , we also come to a contradiction.
Therefore, a = 1, implying f (x) = x1 . It is easy to verify that f (x) = x1
meets the conditions.
Note. Another approach to deriving f (1) = 1 involves substituting values
into the condition. Setting x = y = 1 gives f (f (1)) = f (1), and setting x =
1 and y = f (1) yields f (f (f (1))) = f (1)f (1). Therefore, f (1) = f (1)f (1).
Since f (1) > 0, we have f (1) = 1.
Furthermore, there are several similar problems:

• (Japan Team Selection Test 2019, Problem 4) Let Q+ denote the

set of all positive rational numbers. Determine all functions f : Q+ →
Q+ satisfying
f (x2 (f (y))2 ) = f (y)(f (x))2 for all x, y ∈ Q+ .
• Problem 2.16 (IMO 31-4) in this chapter.

Score Situation This particular problem saw the following distribution of scores among
contestants: 12 contestants scored 7 points, 1 contestant scored 6 points, 1 contestant
scored 5 points, no contestant scored 4 points, 9 contestants scored 3 points, 4 contestants
scored 2 points, 13 contestants scored 1 point, and no contestant scored 0 point. The average
score of this problem is 3.575, indicating that it was relatively straightforward.
Among the top five teams in the team scores, the Germany team achieved a total score
of 212 points, the United States team achieved a total score of 171 points, the Hungary
team achieved a total score of 170 points, the Soviet Union team achieved a total score of
169 points, and the Romania team achieved a total score of 161 points.
The gold medal cutoff for this IMO was set at 38 points (with 9 contestants earning gold
medals), the silver medal cutoff was 26 points (with 27 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 57 contestants earning bronze medals).
In this IMO, a total of four contestants achieved a perfect score of 42 points.
122 IMO Problems, Theorems, and Methods: Algebra

Problem 2.15 (IMO 27-5, proposed by the United Kingdom). Find

all functions f , defined on the non-negative real numbers and taking non-
negative real values, such that:

(a) f (xf (y))f (y) = f (x + y) for all x, y ≥ 0;

(b) f (2) = 0;
 0 for 0 ≤ x < 2.
(c) f (x) =

Solution. Setting y = 2 and substituting it into (a), we obtain

f (x + 2) = 0. (1)
Since x ≥ 0, from (1), f (x) = 0 for x ≥ 2. Furthermore, from (c), x ≥ 2 if
f (x) = 0.
Hence, for x + y ≥ 2,
f (x + y) = 0 = f (xf (y))f (y),
implying either y ≥ 2 or xf (y) ≥ 2. For 0 ≤ y < 2, the inequality xf (y) ≥ 2
is equivalent to x + y ≥ 2, i.e.,
2
x≥ ⇔ x ≥ 2 − y,
f (y)
2
leading to f (y) = 2 − y, and therefore f (y) = 2−y
2
.
In conclusion, the function is
! 2
2−x , 0 ≤ x < 2,
f (x) =
0, x ≥ 2.

Note. This problem is the same as the 20th problem in 2004 Germany
Team Selection Test. Furthermore, there is a similar problem:

• (International Mathematics Competition for University Stu-

dents 2000, Problem 11). Let R+ be the set of positive real numbers.
Find all functions f : R+ → R+ such that for all x, y ∈ R+ ,
f (x)f (yf (x)) = f (x + y).

Score Situation This particular problem saw the following distribution of scores among
contestants: 57 contestants scored 7 points, 41 contestants scored 6 points, 8 contestants
scored 5 points, 13 contestants scored 4 points, 30 contestants scored 3 points, 14 contestants
scored 2 points, 17 contestants scored 1 point, and 30 contestants scored 0 point. The
average score for this problem is 4.152, indicating that it was simple.
 Function Problems 123

Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 41 points (with a total team score of 203 points), the United
States team scored 39 points (with a total team score of 203 points), the Germany team
scored 40 points (with a total team score of 196 points), the China team scored 30 points
(with a total team score of 177 points), and the German Democratic Republic team scored
42 points (with a total team score of 172 points).
The gold medal cutoff for this IMO was set at 34 points (with 18 contestants earning gold
medals), the silver medal cutoff was 26 points (with 41 contestants earning silver medals),
and the bronze medal cutoff was 17 points (with 48 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 42 points, namely Vladimir
Roganov and Stanislav Smirnov from the Soviet Union, and Géza Kós from Hungary.

Problem 2.16 (IMO 31-4, proposed by Turkey). Let Q+ be the set

of positive rational numbers. Construct a function f : Q+ → Q+ such that

f (x)
f (xf (y)) =
y

for all x and y in Q+ .

Solution. Suppose a function f satisfies

f (x)
f (xf (y)) = , where x, y ∈ Q+ , (1)
y

and f (x) ∈ Q+ for any x ∈ Q+ . If y1 , y2 ∈ Q+ such that f (y1 ) = f (y2 ),

then from (1),

f (x) f (x)
= f (xf (y1 )) = f (xf (y2 )) = ,
y1 y2

and since f (x) > 0, it follows that y1 = y2 . Hence f is injective from Q+

to Q+ .
Setting x = y = 1 in (1) gives f (f (1)) = f (1), and as f is injective,
f (1) = 1. Setting x = 1 in (1) gives

f (1) 1
f (f (y)) = = . (2)
y y

Setting y = f (t) (where t ∈ Q+ ) in (1) gives

f (x)
f (xf (f (t))) = ,
f (t)
124 IMO Problems, Theorems, and Methods: Algebra

x f (x)
and from (2), f (xf (f (t))) = f t = f (t) . Thus

f (xy) = f (x)f (y), where x, y ∈ Q+ . (3)

Conversely, if f : Q+ → Q+ satisfies (2) and (3), then f must satisfy (1).

Let P be the set of all prime numbers, with p1 = 2, p2 = 3, . . ., and
generally, pk being the kth prime number in ascending order. For a given
function g : P → Q+ , construct a function f : Q+ → Q+ as follows:
If x = pα 1 α2 αs
1 p2 · · · ps , where α1 , α2 , . . . , αs are integers, then define

f (x) = (g(p1 ))α1 (g(p2 ))α2 · · · (g(ps ))αs . (4)

Since α1 , α2 , . . . , αs can be 0 or negative integers, (4) defines f for any

positive rational number. Such a function obviously satisfies (3). To ensure
f satisfies (2), choose g : P → Q+ as follows:
!
pk+1 , k = 1, 3, 5, 7, . . . ,
g(pk ) = 1
(5)
pk−1 , k = 2, 4, 6, 8, . . . .

It is easy to verify that the function f defined by (4) and (5) satisfies
(2), and therefore such a defined f satisfies (1).

Note. The function f satisfying (1) is not unique.

Score Situation This particular problem saw the following distribution of scores among
contestants: 69 contestants scored 7 points, 3 contestants scored 6 points, 3 contestants
scored 5 points, 10 contestants scored 4 points, 46 contestants scored 3 points, 70 contestants
scored 2 points, 76 contestants scored 1 point, and 31 contestants scored 0 point. The
average score for this problem is 2.955, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 230 points), the Soviet Union
team scored 35 points (with a total team score of 193 points), the United States team scored
37 points (with a total team score of 174 points), the Romania team scored 30 points (with
a total team score of 171 points), and the France team scored 38 points (with a total team
score of 168 points).
The gold medal cutoff for this IMO was set at 34 points (with 23 contestants earning gold
medals), the silver medal cutoff was 23 points (with 56 contestants earning silver medals),
and the bronze medal cutoff was 16 points (with 76 contestants earning bronze medals).
In this IMO, a total of four contestants achieved a perfect score of 42 points.
 Function Problems 125

Problem 2.17 (IMO 33-2, proposed by India). Let R denote the set
of all real numbers. Find all functions f : R → R such that
f (x2 + f (y)) = y + (f (x))2 for all x, y ∈ R.

Solution 1. Suppose
f (x2 + f (y)) = y + (f (x))2 . (1)
First, we prove that f (0) = 0. Set x = 0 and t = (f (0))2 . Then
f (f (y)) = y + t. (2)
Replacing y with x2 + f (f (y)) in (2), we get
f (f (x2 + f (f (y)))) = x2 + f (f (y)) + t. (3)
From (1),
f (f (x2 + f (f (y)))) = f ((f (x))2 + f (y))
= y + (f (f (x)))2 . (4)
Combining (2), (3), and (4), we see that
x2 + y + 2t = y + (x + t)2 and 2t = t2 + 2tx
for any real number x. It follows that t = 0, i.e., f (0) = 0. Thus, (2)
becomes
f (f (y)) = y. (5)
For x ≥ 0, from (1) and (5),
√
f (x + y) = f (x + f (f (y))) = f (y) + (f ( x))2 ≥ f (y),
indicating that f (x) is an increasing function on R, i.e., f (x) ≥ f (y) for
x ≥ y.
If there exists an x0 such that f (x0 ) > x0 , then f (f (x0 )) ≥ f (x0 ) > x0 ,
which is a contradiction. Similarly, if there exists an x0 such that f (x0 ) <
x0 , then f (f (x0 )) ≤ f (x0 ) < x0 , which is also a contradiction.
Therefore, f (x) = x. Clearly, f (x) = x satisfies the given conditions.
Solution 2. From the conditions and the fact that y can take any real
number, f is surjective. Moreover, if f (y1 ) = f (y2 ), then
y1 + (f (x))2 = f (x2 + f (y1 )) = f (x2 + f (y2 )) = y2 + (f (x))2 ,
implying y1 = y2 , and thus f is injective. Therefore, f is bijective.
126 IMO Problems, Theorems, and Methods: Algebra

From the conditions,

y + (f (x))2 = f (x2 + f (y)) = f ((−x)2 + f (y)) = y + (f (−x))2 ,

For x = 0, we have x = −x and f (x) = f (−x). It follows that f (x) =

−f (−x) = 0. Since f is surjective, there must be f (0) = 0, and if we set
x = 0 in (1), then f (f (y)) = y.
The rest of the proof is similar to Solution 1.

Note. (i) The solution to the functional equation f (xf (y) + x2 ) = xy +

(f (x))2 for all x, y ∈ R is f (x) = x.
(ii) The solutions to the functional equation f (xl f (m) (y) + xn ) = xl y +
(f (x))n for all x, y ∈ R are as follows:

(a) except for n = 1 and l = 0, if and only if both m, n are odd, then there
are two solutions f (x) = x and f (x) = −x;
(b) except for n = 1 and l = 0, if m, n are not both odd, then there is only
one solution f (x) = x;
(c) for n = 1 and l = 0, the solutions are not yet fully determined.

Here, l ∈ {0, 1, 2, 3, . . .}, m, n ∈ {1, 2, 3, . . .}, and f (m) (x) represents the
mth iteration of f .
Furthermore, there are several similar problems:

• (Japan Team Selection Test 2023, Problem 5). Let R be the set
of real numbers and F be the set of all functions f : R → R such that

f (x + f (y)) = f (x) + f (y) for all x, y ∈ R.

Find all rational numbers q such that for every function f ∈ F , there
exists some z ∈ R satisfying f (z) = qz.
• (Chinese Team Selection Test 2021, Problem 17). Determine all
f : R → R such that

f (xf (y) + y 2021 ) = yf (x) + (f (y))2021 for all x, y ∈ R.

• (Korean Mathematical Olympiad 2021, Final Round, Problem

6). Find all functions f : R → R such that

f (x2 − g(y)) = (g(x))2 − y for all x, y ∈ R.

 Function Problems 127

• (European Girls’ Mathematical Olympiad 2021, Problem 2).

Find all functions f : Q → Q such that

f (xf (x) + y) = f (y) + x2

for all x, y ∈ Q. Here, Q denotes the set of rational numbers.

• (Korean Mathematical Olympiad 2015, Final Round, Problem
1). Find all functions f : R → R such that

f (x2015 + (f (y))2015 ) = y 2015 + (f (x))2015 for all x, y ∈ R.

• (Kyrgyzstan Mathematical Olympiad 2012, Problem 4). Find

all functions f : R → R such that

f ((f (x))2 + f (y)) = y + xf (x) for all x, y ∈ R.

• (USA Mathematical Olympiad Program 2006, Homework).

Find all functions f : R → R satisfying

f (x + f (y)) = x + f (f (y))

for all real numbers x and y, with the additional constraint f (2004) =
2005.
• (Japan Mathematical Olympiad 2004, Final Round,
Problem 2). Find all functions f : R → R such that
f (xf (x) + f (y)) = y + (f (x))2 for all x, y ∈ R.

• (Estonia Team Selection Test 2001, Problem 3). Let k be a fixed

real number. Find all functions f : R → R such that

kf (x2 + y) = f (y) + (f (x))2 for all x, y ∈ R.

Score Situation This particular problem saw the following distribution of scores among
contestants: 67 contestants scored 7 points, 3 contestants scored 6 points, 18 contestants
scored 5 points, 34 contestants scored 4 points, 43 contestants scored 3 points, 41 contestants
scored 2 points, 113 contestants scored 1 point, and 31 contestants scored 0 point. The
average score for this problem is 2.963, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 240 points), the United States
team scored 32 points (with a total team score of 181 points), the Romania team scored
42 points (with a total team score of 177 points), the Commonwealth of Independent States
team scored 30 points (with a total team score of 176 points), and the United Kingdom team
scored 24 points (with a total team score of 168 points).
128 IMO Problems, Theorems, and Methods: Algebra

The gold medal cutoff for this IMO was set at 32 points (with 26 contestants earning gold
medals), the silver medal cutoff was 24 points (with 55 contestants earning silver medals),
and the bronze medal cutoff was 14 points (with 74 contestants earning bronze medals).
In this IMO, a total of four contestants achieved a perfect score of 42 points.

Problem 2.18 (IMO 35-5, proposed by the United Kingdom). Let

S be the set of real numbers strictly greater than −1. Find all functions
f : S → S satisfying the two conditions:

(a) f (x + f (y) + xf (y)) = y + f (x) + yf (x) for all x and y in S;

(b) f (x)
x is strictly increasing on each of the intervals (−1, 0) and (0, +∞).

Solution. Since f (x)

x is strictly increasing on the interval (−1, 0), the equa-
tion f (x)
x = 1 has at most one solution x = u ∈ (−1, 0), i.e., there exists
at most one u ∈ (−1, 0) such that f (u) = u. If such u exists, then setting
x = y = u in (a) yields

f (u2 + 2u) = u2 + 2u.

Since u ∈ (−1, 0) and u2 + 2u = (u + 1)2 − 1, it follows that u2 + 2u ∈

(−1, 0). By the uniqueness,

u2 + 2u = u,

implying u = −1 or u = 0, which contradicts u ∈ (−1, 0).

Therefore, f (u) = u for any u ∈ (−1, 0). Similarly, f (u) = u for any
u ∈ (0, +∞). Thus, f (u) = u can only occur at u = 0.
Substituting x = y in (a), we obtain

f (x + (1 + x)f (x)) = x + (1 + x)f (x),

x
leading to x + (1 + x)f (x) = 0, i.e., f (x) = − 1+x .
x
It is easy to verify that f (x) = − 1+x satisfies all the given conditions.
x
Thus, the only function that satisfies the conditions is f (x) = − 1+x .

Score Situation This particular problem saw the following distribution of scores among
contestants: 84 contestants scored 7 points, 38 contestants scored 6 points, 23 contestants
scored 5 points, 22 contestants scored 4 points, 29 contestants scored 3 points, 40 contestants
scored 2 points, 54 contestants scored 1 point, and 95 contestants scored 0 point. The
average score for this problem is 3.221, indicating that it was relatively straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the United States team scored 42 points (with a total team score of 252 points), the China
team scored 40 points (with a total team score of 229 points), the Russia team scored
 Function Problems 129

40 points (with a total team score of 224 points), the Bulgaria team scored 41 points (with
a total team score of 223 points), and the Hungary team scored 31 points (with a total team
score of 221 points).
The gold medal cutoff for this IMO was set at 40 points (with 30 contestants earning gold
medals), the silver medal cutoff was 30 points (with 64 contestants earning silver medals),
and the bronze medal cutoff was 19 points (with 98 contestants earning bronze medals).
In this IMO, a total of 22 contestants achieved a perfect score of 42 points.

Problem 2.19 (IMO 40-6, proposed by Japan). Determine all func-

tions f : R → R such that

f (x − f (y)) = f (f (y)) + xf (y) + f (x) − 1 (1)

for all real numbers x and y.

Solution. Setting x = f (y) in (1), we get

f (0) = f (f (y)) + (f (y))2 + f (f (y)) − 1,

1+f (0)−(f (y))2
which leads to f (f (y)) = 2 . Furthermore, replacing x with
f (x) in (1), we get

f (f (x) − f (y)) = f (f (y)) + f (x)f (y) + f (f (x)) − 1

1 + f (0) − (f (y))2 1 + f (0) − (f (x))2
= + f (x)f (y) + −1
2 2
1
= f (0) − (f (x) − f (y))2 . (2)
2
Since the zero function does not satisfy (1), there exists a y0 ∈ R such
that f (y0 ) = 0. From (1),

f (x − f (y0 )) − f (x) = xf (y0 ) + f (f (y0 )) − 1.

Therefore, for any real number z, there exists an x0 ∈ R such that

f (x0 − f (y0 )) − f (x0 ) = z.

Setting x = x0 − f (y0 ) and y = x0 in (2) gives

1
f (z) = f (0) − z 2 . (3)
2
Substituting (3) into (1) yields f (0) = 1. Hence f (x) = 1 − 12 x2 .
Upon verification, f (x) = 1 − 12 x2 satisfies (1). Therefore, the only
function that meets the conditions is f (x) = 1 − 12 x2 .
130 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 11 contestants scored 7 points, 5 contestants scored 6 points, 7 contestants
scored 5 points, 8 contestants scored 4 points, 21 contestants scored 3 points, 28 contestants
scored 2 points, 225 contestants scored 1 point, and 145 contestants scored 0 point. The
average score for this problem is 1.151, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 15 points (with a total team score of 182 points), the Russia team
scored 11 points (with a total team score of 182 points), the Vietnam team scored 18 points
(with a total team score of 177 points), the Romania team scored 23 points (with a total
team score of 173 points), and the Bulgaria team scored 29 points (with a total team score
of 170 points).
The gold medal cutoff for this IMO was set at 28 points (with 38 contestants earning gold
medals), the silver medal cutoff was 19 points (with 70 contestants earning silver medals),
and the bronze medal cutoff was 12 points (with 118 contestants earning bronze medals).
In this IMO, no contestant achieved a perfect score of 42 points.

Problem 2.20 (IMO 43-5, proposed by India). Find all functions

f : R → R such that

(f (x) + f (z))(f (y) + f (t)) = f (xy − zt) + f (xt + yz) (1)

for all x, y, z, t ∈ R.
Solution 1. Setting x = y = z = 0 in (1) gives

2f (0)(f (0) + f (t)) = 2f (0). (2)

Substituting t = 0 into (2), we get 4(f (0))2 = 2f (0), which implies

1
f (0) = 0 or f (0) = .
2
If f (0) = 12 , then substituting it into (2) gives f (t) = 12 , i.e., f (x) = 1
2
for all x ∈ R. If f (0) = 0, then setting z = t = 0 in (1) gives

f (x)f (y) = f (xy). (3)

By setting x = y = 1 in (3), we get (f (1))2 = f (1), so f (1) = 0 or f (1) = 1.

If f (1) = 0, then (3) implies f (x) = f (x)f (1) = 0 for all x ∈ R.
If f (1) = 1, setting x = 0 and y = t = 1 in (1), we get

2f (z) = f (−z) + f (z),

f (z) = f (−z),
 Function Problems 131

indicating f (x) is an even function. Since setting x = y and z = t = 0 in

(1) gives

f (x2 ) = (f (x))2 ≥ 0,

and since f (x) is even, f (x) ≥ 0 for all x ∈ R.

Setting x = t and y = z in (1), we get

(f (x) + f (y))2 = f (x2 + y 2 ),

which implies

f (x2 + y 2 ) = (f (x))2 + 2f (x)f (y) + (f (y))2

≥ (f (x))2 = f (x2 ).

Hence, u ≥ v ≥ 0 implies f (u) ≥ f (v), i.e., f (x) is increasing for x > 0.

Setting y = z = t = 1 in (1), we find

2(f (x) + 1) = f (x − 1) + f (x + 1), (4)

and using (4) and mathematical induction, we see that f (n) = n2 for non-
negative integers n. From f (xy) = f (x)f (y), it follows that f (a) = a2 for
all positive rational numbers a.
Next, we prove that f (x) = x2 for x > 0. Otherwise, there exists x0 ∈ R
such that f (x0 ) = x20 . 
If f (x0 ) < x20 , by choosing a rational number a such that f (x0 ) <
a < x0 , then f (x0 ) ≥ f (a) = a2 > f (x0 ), which is a contradiction. If
f (x0 ) > x20 , then there is a similar contradiction. Therefore, f (x) = x2 for
x > 0. Since f (x) is even, f (x) = x2 for x ∈ R.
Upon verification, the solutions satisfying the conditions are f (x) = 0,
f (x) = 12 , and f (x) = x2 .
Solution 2. Setting y = t = 0 and z = x in (1), we obtain 4f (0)f (x) =
2f (0).

(i) If f (0) = 0, then f (x) = 12 for any x ∈ R, which is clearly a solution.

(ii) If f (0) = 0, then setting z = t = 0 gives f (xy) = f (x)f (y). If there
exists x0 = 0 such that f (x0 ) = 0, then for any x ∈ R,
   
x x
f (x) = f x0 · = f (x0 )f = 0,
x0 x0
which is also a solution.
132 IMO Problems, Theorems, and Methods: Algebra

(iii) If f (0) = 0 and f (x) = 0 for x = 0, then setting x = y = 0 gives

f (−zt) = f (z)f (t) = f (zt), (5)

indicating f (x) is even. Setting z = t = 1 in (5), we find f (1) =

(f (1))2 , and since f (1) = 0, it follows that f (1) = 1.
√ √ √ √
For x > 0, clearly f (x) = f ( x · x) = f ( x)f ( x) > 0. Setting
z = y and t = x in (1) gives

f (x2 + y 2 ) = (f (x) + f (y))2 .


Let g(x) = f (x). Then

g(0) = 0, g(1) = 1, g(−x) = g(x), g(xy) = g(x)g(y).

It is evident that g(x) > 0 for x = 0, and

g(x2 + y 2 ) = (g(x))2 + (g(y))2 = g(x2 ) + g(y 2 ),

implying g(x) satisfies Cauchy’s functional equation. Thus,


f (x) = g(x) = g(1) · x = x, and so f (x) = x2 .

Upon verification, the solutions satisfying the conditions are f (x) = 0,

f (x) = 12 , and f (x) = x2 .
Note. There are several similar problems:

• (Bolivia Team Selection Test 2021, Problem 1). Find all functions
f : Q → Q such that

f (x + y) + f (x − y) = 2f (x) + 2f (y)

for all x, y ∈ Q.
• (Korean Mathematical Olympiad 2015, 2nd Round, Prob-
lem 5). Find all functions f : R → R such that for all real numbers
x, y, z,

(f (x) + 1)(f (y) + f (z)) = f (xy + z) + f (xz − y).

• (United States of America Junior Mathematical Olympiad

2015, Problem 4). Find all functions f : Q → Q such that

f (x) + f (t) = f (y) + f (z)

for all rational numbers x < y < z < t that form an arithmetic progres-
sion.
 Function Problems 133

• (ELMO 2011, called “English Language Master’s Open” that

time, Problem 4). Find all functions f : R+ → R+ , where R+ denotes
the positive reals, such that whenever a > b > c > d > 0 are real numbers
with ad = bc,
f (a + d) + f (b − c) = f (a − d) + f (b + c).

Score Situation This particular problem saw the following distribution of scores among
contestants: 66 contestants scored 7 points, 25 contestants scored 6 points, 10 contestants
scored 5 points, no contestant scored 4 points, 21 contestants scored 3 points, 101 contestants
scored 2 points, 159 contestants scored 1 point, and 97 contestants scored 0 point. The
average score for this problem is 2.267, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 212 points), the Russia team
scored 42 points (with a total team score of 204 points), the United States team scored
34 points (with a total team score of 171 points), the Bulgaria team scored 36 points (with
a total team score of 167 points), and the Vietnam team scored 40 points (with a total team
score of 166 points).
The gold medal cutoff for this IMO was set at 29 points (with 39 contestants earning gold
medals), the silver medal cutoff was 23 points (with 73 contestants earning silver medals),
and the bronze medal cutoff was 14 points (with 120 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 42 points, namely Yunhao
Fu and Botong Wang from China, and Andrei Khaliavine from Russia.
Problem 2.21 (IMO 49-4, proposed by South Korea). Find all func-
tions f : (0, +∞) → (0, +∞) such that
(f (w))2 + (f (x))2 w2 + x2
=
f (y 2 ) + f (z 2 ) y2 + z 2
for all positive real numbers w, x, y, z satisfying wx = yz.
Solution. Setting w = x = y = z = 1 gives (f (1))2 = f (1). Hence
f (1) = 1. √
For any t > 0, setting w = t, x = 1, and y = z = t yields
(f (t))2 + 1 t2 + 1
= ,
2f (t) 2t
and after simplifying, we get (tf (t) − 1)(f (t) − t) = 0. Therefore, for each
t > 0,
1
f (t) = t or f (t) = . (1)
t
If there exist b, c ∈ (0, +∞) such that f (b) = b and f (c) = c , then from
1

(1) we know b and c are both not equal to 1, and f (b) = 1b and f (c) = c.
134 IMO Problems, Theorems, and Methods: Algebra

√ 1
+c2 b2 +c2
b2
Setting w = b, x = c, and y = z = bc, we get 2f (bc) = 2bc , i.e.,
c+b2 c3
f (bc) = b(b2 +c2 ) .
1
Since f (bc) = bc or f (bc) = bc , if f (bc) = bc, then
c + b 2 c3
bc = ,
b(b2 + c2 )
1
yielding b4 c = c and b = 1, which is a contradiction. If f (bc) = bc , then

1 c + b 2 c3
= ,
bc b(b2 + c2 )
yielding b2 c4 = b2 and c = 1, which is also a contradiction.
Thus, either f (x) = x for all x ∈ (0, +∞) or f (x) = x1 for all x ∈
(0, +∞). Upon verification, both f (x) = x and f (x) = x1 satisfy the given
conditions.
Score Situation This particular problem saw the following distribution of scores among
contestants: 227 contestants scored 7 points, 27 contestants scored 6 points, no contestant
scored 5 points, 128 contestants scored 4 points, 4 contestants scored 3 points, no contestant
scored 2 points, 80 contestants scored 1 point, and 69 contestants scored 0 point. The
average score for this problem is 4.402, indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 217 points), the Russia team
scored 42 points (with a total team score of 199 points), the United States team scored
41 points (with a total team score of 190 points), the South Korea team scored 42 points
(with a total team score of 188 points), and the Iran team scored 42 points (with a total
team score of 181 points).
The gold medal cutoff for this IMO was set at 31 points (with 47 contestants earning gold
medals), the silver medal cutoff was 22 points (with 100 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 120 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 42 points, namely
Xiaosheng Mu and Dongyi Wei from China, and Alex Zhai from the United States.

Problem 2.22 (IMO 50-5, proposed by France). Determine all func-

tions f from the set of positive integers to the set of positive integers such
that, for all positive integers a and b, there exists a non-degenerate triangle
with the sides of lengths

a, f (b), and f (b + f (a) − 1).

Here, a triangle is non-degenerate if its vertices are not collinear.

 Function Problems 135

Solution 1. The only f that satisfies the conditions is f (n) = n for n ∈

N+ . From the discrete nature of integers, for any a, b ∈ N+ ,

f (b) + f (b + f (a) − 1) − 1 ≥ a, (1)

f (b) + a − 1 ≥ f (b + f (a) − 1), (2)
f (b + f (a) − 1) + a − 1 ≥ f (b). (3)

Setting a = 1 in (2) and (3), we find that f (b) = f (b + f (1) − 1) for any
b ∈ N+ . If f (1) = 1, then the above equation implies that f is periodic.
Considering the domain of f , we are sure that f is bounded. Let M be a
positive integer such that M ≥ f (n) for all n ∈ N+ (i.e., M is an upper
bound of f ). Setting a = 2M in (1) leads to a contradiction. Therefore
f (1) = 1.
Setting b = 1 in (1) and (2), we get f (f (n)) = n for n ∈ N+ . If there
exists t ∈ N+ such that f (t) < t, then clearly t ≥ 2 and f (t) ≤ t − 1.
Setting a = f (t) in (2) yields

f (b + t − 1) = f (b + f (a) − 1)
≤ f (b) + a − 1
≤ f (b) + t − 2.

Let M = (t − 1) · max f (i). For any integer n > M , let n0 be the

1≤i≤t−1
unique positive integer such that

1 ≤ n0 ≤ t − 1, n0 ≡ n(mod(t − 1)).

Then
t−2 M (t − 2)n
f (n) ≤ f (n0 ) + (n − n0 ) ≤ + < n.
t−1 t−1 t−1
Therefore, f (n) < n for integers n > M . Choose n1 ∈ N+ such
that n1 > M and n1 is not equal to any of f (1), f (2), . . . , f (M ). From
f (f (n1 )) = n1 , we know f (n1 ) > M , and from the above conclusion,
n1 > f (n1 ) > f (f (n1 )) = n1 , a contradiction.
This implies that f (t) ≥ t for any t ∈ N+ and further t = f (f (t)) ≥
f (t) ≥ t, implying the equality throughout, i.e., f (n) = n for n ∈ N+ .
Upon verification, f (n) = n for n ∈ N+ satisfies the conditions. There-
fore, the solution is f (n) = n for n ∈ N+ .
136 IMO Problems, Theorems, and Methods: Algebra

Solution 2. As in Solution 1, we know f (1) = 1 and f (f (n)) = n. It is

easy to deduce that f is surjective. Hence f is a bijection from N+ to N+ .
Otherwise, if f (n1 ) = f (n2 )(n1 = n2 ), then n1 = f (f (n1 )) = f (f (n2 )) =
n2 , a contradiction.
It is evident that f (2) > 1, and 2, f (b), f (b+f (2)−1) can form a triangle.
Suppose t = f (2) − 1 > 0, since f is a bijection, 0 < |f (b + t) − f (b)| < 2.
Thus for any b ∈ N+ ,

f (b + t) − f (b) = ±1.

By mathematical induction and since f is a bijection, f (b + nt) =

f (b) ± n for any n ∈ N+ . As f (b + nt) > 0, it must be that f (b + nt) =
f (b) + n, i.e., f (b + t) − f (b) = 1 for any b ∈ N+ .
Therefore, f (1), f (1 + t), f (1 + 2t), . . . , f (1 + nt), . . . cover all positive
integers. Since f is a bijection, 1, 1+t, 1+2t, . . . must also cover all positive
integers, so t = 1, implying f (b + 1) = f (b) + 1 for any b ∈ N+ .
Since f (1) = 1, we deduce that f (n) = n for any n ∈ N+ . In conclusion,
f (n) = n is the only solution.
Note. There are several similar problems:

• (Chinese Team Selection Test 2016, Problem 18). Find all func-
tions f : R+ → R+ satisfying the following condition: for any three dis-
tinct real numbers a, b, c, a triangle can be formed with side lengths a, b, c
if and only if a triangle can be formed with side lengths f (a), f (b), f (c).
• (British Mathematical Olympiad 2014, 1st Round, Problem 6).
Determine all functions f (n) from the positive integers to the positive
integers which satisfy the following condition: whenever a, b, and c are
positive integers such that a1 + 1b = 1c ,
1 1 1
+ = .
f (a) f (b) f (c)

Score Situation This particular problem saw the following distribution of scores among
contestants: 153 contestants scored 7 points, 7 contestants scored 6 points, 4 contestants
scored 5 points, 6 contestants scored 4 points, 33 contestants scored 3 points, 50 contestants
scored 2 points, 42 contestants scored 1 point, and 270 contestants scored 0 point. The
average score for this problem is 2.474, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 221 points), the Japan team
scored 42 points (with a total team score of 212 points), the Russia team scored 42 points
 Function Problems 137

(with a total team score of 203 points), the South Korea team scored 42 points (with a total
team score of 188 points), and the North Korea team scored 36 points (with a total team
score of 183 points).
The gold medal cutoff for this IMO was set at 32 points (with 49 contestants earning gold
medals), the silver medal cutoff was 24 points (with 98 contestants earning silver medals),
and the bronze medal cutoff was 14 points (with 135 contestants earning bronze medals).
In this IMO, only two contestants achieved a perfect score of 42 points, namely Dongyi
Wei from China and Makoto Soejima from Japan.

Problem 2.23 (IMO 51-1, proposed by France). Determine all func-

tions f : R → R such that
f (xy) = f (x)f (y) (1)
for all x, y ∈ R. Here, z denotes the greatest integer less than or equal
to z.
Solution. The solution is f (x) = C, where C is a constant and either
C = 0 or 1 ≤ C < 2.
Setting x = 0 in (1), we get
f (0) = f (0)f (y) (2)
for all y ∈ R. Thus, there are two cases:
Case 1: f (0) = 0.
From (2), f (y) = 1 for all y ∈ R. Therefore, (1) becomes f (xy) =
f (x). Setting y = 0 gives f (x) = f (0) = C = 0.
Furthermore, f (y) = 1 = C, and it follows that 1 ≤ C < 2.
Case 2: f (0) = 0.
If there exists 0 < α < 1 such that f (α) = 0, then setting x = α in (1)
gives
0 = f (0) = f (α)f (y)
for all y ∈ R, implying f (y) = 0 for all y ∈ R. Setting x = 1 in (1), we
find f (y) = f (1)f (y) = 0 for all y ∈ R, which contradicts f (α) = 0.
Therefore, f (α) = 0 for 0 ≤ α < 1. For any z ∈ R, there exists an
integer N such that α = Nz ∈ [0, 1). From (1),
f (z) = f (N α) = f (N )f (α) = 0
for all z ∈ R.
Upon verification, f (x) = C, where C is a constant and either C = 0 or
1 ≤ C < 2, satisfies the given conditions.
138 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 294 contestants scored 7 points, 54 contestants scored 6 points, 35 contestants
scored 5 points, 34 contestants scored 4 points, 16 contestants scored 3 points, 27 contestants
scored 2 points, 17 contestants scored 1 point, and 39 contestants scored 0 point. The
average score for this problem is 5.450, indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 41 points (with a total team score of 197 points), the Russia team
scored 41 points (with a total team score of 169 points), the United States team scored
40 points (with a total team score of 168 points), the South Korea team scored 42 points
(with a total team score of 156 points), the Kazakhstan team scored 42 points (with a total
team score of 148 points), and the Thailand team scored 42 points (with a total team score
of 148 points).
The gold medal cutoff for this IMO was set at 27 points (with 47 contestants earning gold
medals), the silver medal cutoff was 21 points (with 103 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 115 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 42 points, namely Zipei Nie
from China.

Problem 2.24 (IMO 53-4, proposed by South Africa). Find all func-
tions f : Z → Z such that, for all integers a, b, c that satisfy a + b + c = 0,
the following equality holds:

f (a)2 + f (b)2 + f (c)2 = 2f (a)f (b) + 2f (b)f (c) + 2f (c)f (a).

Here, Z denotes the set of integers.

Solution. Setting a = b = c = 0 gives 3f (0)2 = 6f (0)2 . Thus f (0) = 0.
Setting b = −a and c = 0 yields (f (a) − f (−a))2 = 0, implying f is an
even function, i.e., f (a) = f (−a) for all a ∈ Z.
Setting b = a and c = −2a, we obtain

2f (a)2 + f (2a)2 = 2f (a)2 + 4f (a)f (2a).

Therefore, for all a ∈ Z,

f (2a) = 0 or f (2a) = 4f (a). (1)

If f (r) = 0 for some r ≥ 1, then set b = r and c = −a − r to get

(f (a + r) − f (a))2 = 0,

implying f is a periodic function with period r, i.e., f (a + r) = f (a) for all

a ∈ Z. Particularly, if f (1) = 0, then f is a constant function, so f (a) = 0
 Function Problems 139

for all a ∈ Z. This function clearly satisfies the given conditions. Next,
assume f (1) = k = 0.
From (1), we have f (2) = 0 or f (2) = 4k.
If f (2) = 0, then f is a periodic function with period 2. Hence for all
n ∈ Z,
f (2n) = 0 and f (2n + 1) = k.
If f (2) = 4k = 0, then from (1), we know f (4) = 0 or f (4) = 16k.
Suppose f (4) = 0, the function f is periodic with period 4, and
f (3) = f (−1) = f (1) = k.
Hence f (4n) = 0, f (4n + 1) = f (4n + 3) = k, and f (4n + 2) = 4k for all
n ∈ Z.
Suppose f (4) = 16k = 0, setting a = 1, b = 2, and c = −3 gives
(f (3))2 − 10kf (3) + 9k 2 = 0,
implying f (3) ∈ {k, 9k}. Setting a = 1, b = 3, and c = −4 yields
(f (3))2 − 34kf (3) + 225k 2 = 0,
implying f (3) ∈ {9k, 25k}. Thus f (3) = 9k.
Next, we use mathematical induction to prove f (x) = kx2 for x ∈ Z.
For x ∈ {0, 1, 2, 3, 4}, the proposition already holds. Assume the proposi-
tion holds for x ∈ {0, 1, . . . , n}(n ≥ 4).
Setting a = n, b = 1, and c = −n − 1, we get
f (n + 1) ∈ {k(n + 1)2 , k(n − 1)2 }.
Setting a = n − 1, b = 2, and c = −n − 1, we obtain
f (n + 1) ∈ {k(n + 1)2 , k(n − 3)2 }.
Since k(n−1)2 = k(n−3)2 for n = 2, it follows that f (n+1) = k(n+1)2 ,
proving f (x) = kx2 for non-negative integers x. As f is even, f (x) = kx2
for x ∈ Z.
In conclusion, the possible functions are
f1 (x) = 0,
f2 (x) = kx2 ,

0, x ≡ 0 (mod2),
f3 (x) =
k, x ≡ 1 (mod2),
⎧
⎨0, x ≡ 0 (mod4),
f4 (x) = k, x ≡ 1 (mod2),
⎩
4k, x ≡ 2 (mod4),
where k is any non-zero integer.
140 IMO Problems, Theorems, and Methods: Algebra

It is easy to verify that f1 and f2 satisfy the given conditions. For f3 ,

if a, b, c are all even integers, then

f (a) = f (b) = f (c) = 0,

satisfying the conditions. If a, b, c include one even and two odd integers,
then the left side equals 2k 2 as does the right side, meeting the conditions.
For f4 , by considering the symmetry and a + b + c = 0, it suffices to
consider the cases where (f (a), f (b), f (c)) are (0, k, k), (4k, k, k), (0, 0, 0),
and (0, 4k, 4k), all of which evidently satisfy the given conditions.

Score Situation This particular problem saw the following distribution of scores among
contestants: 143 contestants scored 7 points, 44 contestants scored 6 points, 26 contestants
scored 5 points, 47 contestants scored 4 points, 74 contestants scored 3 points, 95 contestants
scored 2 points, 65 contestants scored 1 point, and 53 contestants scored 0 point. The
average score for this problem is 3.766, indicating that it was relatively straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the South Korea team scored 39 points (with a total team score of 209 points), the China
team scored 31 points (with a total team score of 195 points), the United States team scored
38 points (with a total team score of 194 points), the Russia team scored 41 points (with a
total team score of 177 points), the Canada team scored 39 points (with a total team score of
159 points), and the Thailand team scored 39 points (with a total team score of 159 points).
The gold medal cutoff for this IMO was set at 28 points (with 51 contestants earning gold
medals), the silver medal cutoff was 21 points (with 88 contestants earning silver medals),
and the bronze medal cutoff was 14 points (with 137 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 42 points, namely Jeck Lim
from Singapore.

Problem 2.25 (IMO 56-5, proposed by Albania). Let R be the set of

real numbers. Determine all functions f : R → R satisfying the equation

f (x + f (x + y)) + f (xy) = x + f (x + y) + yf (x)

for all real numbers x and y.

Solution. Denote the equation in the problem as P (x, y). Assume f is a

function that satisfies the condition. Considering P (x, 1), we have

f (x + f (x + 1)) = x + f (x + 1). (1)

Hence, x + f (x + 1) is a fixed point of f for any x ∈ R. Next, we study

the following two cases.
 Function Problems 141

Case 1: f (0) = 0.
Considering P (0, y), we have

f (f (y)) + f (0) = f (y) + yf (0).

If y0 is a fixed point of f , then setting y = y0 in the above equation gives

y0 = 1. Thus, x + f (x + 1) = 1, implying f (x) = 2 − x for all x ∈ R. It is
easy to verify that f (x) = 2 − x is a function that satisfies the conditions.

Case 2: f (0) = 0.
Considering P (x + 1, 0), we have

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

Considering P (1, y), we obtain

f (1 + f (y + 1)) + f (y) = 1 + f (y + 1) + yf (1). (3)

Setting x = −1 in (1) gives f (−1) = −1, and then setting y = −1 in

(3) yields f (1) = 1. Thus, the equation (3) can be rewritten as

f (1 + f (y + 1)) + f (y) = 1 + f (y + 1) + y. (4)

If both y0 and y0 + 1 are fixed points of f , then setting y = y0 in (4)

implies that y0 + 2 is also a fixed point of f . Therefore, from (1) and (2),
x + f (x + 1) + 2 is a fixed point of f for any x ∈ R, i.e.,

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

Replacing x with x − 2 in the above equality, we get

f (x + f (x − 1)) = x + f (x − 1).

Then considering P (x, −1) gives

f (x + f (x − 1)) = x + f (x − 1) − f (x) − f (−x).

The above two equalities imply f (−x) = −f (x), i.e., f is an odd

function.
Considering P (−1, −y) and using f (−1) = −1, we have

f (−1 + f (−y − 1)) + f (y) = −1 + f (−y − 1) + y.

142 IMO Problems, Theorems, and Methods: Algebra

Since f is an odd function, the above equality can be rewritten as

−f (1 + f (y + 1)) + f (y) = −1 − f (y + 1) + y,

and adding this equality to (4), we know f (y) = y for any y ∈ R. It is easy
to verify that f (x) = x is a function that satisfies the conditions.
In conclusion, there are two functions that satisfy the conditions:
f (x) = x and f (x) = 2 − x.
Note. There is a similar problem:

• (Japan Team Selection Test 2022, Problem 9). Find all functions
f : R → R such that f (−1) = −1 and

f (x + f (y)) + f (x)f (y) = f (x + y) + yf (x) for all x, y ∈ R.

Score Situation This particular problem saw the following distribution of scores among
contestants: 30 contestants scored 7 points, 3 contestants scored 6 points, 4 contestants
scored 5 points, 8 contestants scored 4 points, 90 contestants scored 3 points, 34 contestants
scored 2 points, 255 contestants scored 1 point, and 153 contestants scored 0 point. The
average score for this problem is 1.513, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the United States team scored 25 points (with a total team score of 185 points), the China
team scored 22 points (with a total team score of 181 points), the South Korea team scored
25 points (with a total team score of 161 points), the North Korea team scored 22 points
(with a total team score of 156 points), and the Vietnam team scored 28 points (with a total
team score of 151 points).
The gold medal cutoff for this IMO was set at 26 points (with 39 contestants earning gold
medals), the silver medal cutoff was 19 points (with 100 contestants earning silver medals),
and the bronze medal cutoff was 14 points (with 143 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 42 points, namely Zhuo Qun
Alex Song from Canada.

Problem 2.26 (IMO 58-2, proposed by Albania). Let R be the set

of real numbers. Determine all functions f : R → R such that for all real
numbers x and y,

f (f (x)f (y)) + f (x + y) = f (xy).

Solution 1. If f is a solution, then −f is also a solution. Therefore, we

can assume f (0) ≤ 0. Let the given equation be denoted as P (x, y). From
 Function Problems 143

P (0, 0),

f (f (0)2 ) = 0. (1)

For any real number x = 1, there exists a y ∈ R satisfying x + y = xy,

x x
i.e., y = x−1 . From P x, x−1 ,
  
x
f f (x)f =0 for x = 1. (2)
x−1
Obviously, f has at least one zero, namely (f (0))2 .
Case 1: f (0) = 0.
From P (x, 0) it follows that f (x) = 0 for all x ∈ R, so f is identically
zero.
Case 2: f (0) < 0.
Conclusion 1: f (a) = 0 if and only if a = 1.
Indeed, from (1) we know there exists an a ∈ R such that f (a) = 0.
If a = 1, then setting x = a in (2) yields f (0) = 0, contradicting the
assumption that f (0) = 0. From Conclusion 1 and (1), (f (0))2 = 1 and
f (0) = −1. From P (x, 1),

f (0) + f (x + 1) = f (x),

i.e., f (x + 1) = f (x) + 1. By mathematical induction, it is easy to show

that for any integer n and any x ∈ R,

f (x + n) = f (x) + n. (3)

Conclusion 2: f is injective.
By contradiction, suppose there exist a = b such that f (a) = f (b). From
(3), for any integer N , we have f (a + N + 1) = f (b + N ) + 1. Choosing
an integer N > −b, we see that there exist x0 , y0 ∈ R satisfying x0 + y0 =
a + N + 1 and x0 y0 = b + N . Since a = b, we have (x0 − 1)(y0 − 1) = 0.
From P (x0 , y0 ),

f (f (x0 )f (y0 )) + f (a + N + 1) = f (b + N ),

and thus f (f (x0 )f (y0 )) + 1 = f (f (x0 )f (y0 ) + 1) = 0.

From Conclusion 1, f (x0 )f (y0 ) + 1 = 1, i.e., f (x0 )f (y0 ) = 0. However,
x0 = 1 and y0 = 1, and from Conclusion 1, f (x0 ) = 0 and f (y0 ) = 0, a
contradiction.
144 IMO Problems, Theorems, and Methods: Algebra

For any t ∈ R, from P (t, −t),

f (f (t)f (−t)) + f (0) = f (−t2 ).
Thus, f (f (t)f (−t)) = f (−t2 ) + 1 = f (−t2 + 1). As f is injective,
f (t)f (−t) = −t2 + 1. (4)
From P (t, 1 − t),
f (f (t)f (1 − t)) + f (1) = f (t(1 − t)).
Thus, f (f (t)f (1 − t)) = f (t(1 − t)). As f is injective,
f (t)f (1 − t) = t(1 − t). (5)
Since f (1 − t) = 1 + f (−t), comparing (4) and (5) gives f (t) = t − 1. If
f (0) = 1, then f (t) = 1 − t.
It is easy to verify that f1 (x) = 0, f2 (x) = x − 1, and f3 (x) = 1 − x all
satisfy the conditions.
Solution 2. After Conclusion 1 in Solution 1, the following process can
replace Conclusion 2.
Define g(x) = f (x + 1) = f (x) + 1, and replace x and y with x + 1 and
y + 1 respectively in the original functional equation. For any x, y ∈ R,
f (f (x + 1)f (y + 1)) + f (x + y + 2) = f (xy + x + y + 1),
i.e.,
g(g(x)g(y)) + g(x + y) = g(xy + x + y). (6)
By Conclusion 1, “0 is the only zero of g,” we prove that g(x) = x.
Conclusion 3: Let n ∈ Z and x ∈ R. Then

(i) g(x + n) = g(x) + n, and g(x) = n if and only if x = n;

(ii) g(nx) = ng(x).

For (i), note that g(x + n) = g(x) + n is the same as (3) in Solution 1.
Thus
g(x) = n ⇔ g(x − n) = 0 ⇔ x − n = 0.
n
For (ii), when x = 0, it is obvious. For x = 0, substitute y = x into (6),
yielding
n n n
g g(x)g +g x+ =g n+x+
x x x
n n
⇔ g g(x)g = n ⇔ g(x)g = n,
x x
 Function Problems 145

n
1
i.e., g(x) = g( n
for x = 0. Taking n = 1, we get g x = 1
g(x) , and repla-
x)
cing x with nx in the above equation, we obtain g(nx) = g n1 = ng(x).
(x)
Thus proving the conclusion.
Conclusion 4: g is an additive function, i.e., for any a, b ∈ R,

g(a + b) = g(a) + g(b).

First, replace x with −x and y with −y in (6), and substitute n = −1

into Conclusion 3(ii), we know that g is an odd function. Thus

g(g(x)g(y)) − g(x + y) = −g(−xy + x + y).

Subtracting this from (6), we get

2g(x + y) = g(xy + x + y) + g(−xy + x + y).

Then, according to Conclusion 3(ii) (for n = 2), we rewrite the above

equation as: for α = xy + x + y and β = −xy + x + y,

g(α + β) = g(α) + g(β).

Note that for any α, β ∈ R satisfying

 2
α+β α−β
−4· ≥ 0,
2 2
we can always find x and y such that
α+β α−β
x+y = , xy = .
2 2
Thus for any a, b ∈ R we can always find an integer n (which can be 1
or −1) such that

g(na) + g(nb) = g(na + nb) ⇔ ng(a) + ng(b) = ng(a + b).

Hence, the conclusion holds.

Therefore, taking y = 1 in (6) and using Conclusion 3, we obtain

g(g(x)g(1)) + g(x + 1) = g(2x + 1)

⇔ g(g(x)) + g(x) + 1 = 2g(x) + 1
⇔ g(g(x)) = g(x).

Since g is additive, g(g(x) − x) = 0, and with the assumption “0 is the

only zero of g,” it follows that g(x) = x.
146 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 61 contestants scored 7 points, 8 contestants scored 6 points, 10 contestants
scored 5 points, 79 contestants scored 4 points, 138 contestants scored 3 points, 26 contes-
tants scored 2 points, 110 contestants scored 1 point, and 183 contestants scored 0 point.
The average score for this problem is 2.304, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the South Korea team scored 39 points (with a total team score of 170 points), the China
team scored 25 points (with a total team score of 159 points), the Vietnam team scored
36 points (with a total team score of 155 points), the United States team scored 29 points
(with a total team score of 148 points), and the Iran team scored 32 points (with a total
team score of 142 points).
The gold medal cutoff for this IMO was set at 25 points (with 48 contestants earning gold
medals), the silver medal cutoff was 19 points (with 90 contestants earning silver medals),
and the bronze medal cutoff was 16 points (with 153 contestants earning bronze medals).
In this IMO, no contestant achieved a perfect score of 42 points.
Problem 2.27 (IMO 60-1, proposed by South Africa). Let Z be the
set of integers. Determine all functions f : Z → Z such that, for all integers
a and b,
f (2a) + 2f (b) = f (f (a + b)).

Solution. Let’s denote the equation given in the problem as P (a, b). From
P (0, x),
f (0) + 2f (x) = f (f (x)). (1)
From P (x, 0),
f (2x) + 2f (0) = f (f (x)). (2)
Comparing (1) and (2), we get
f (2x) = 2f (x) − f (0). (3)
Substituting (1) and (3) into P (a, b), we have
2f (a) − f (0) + 2f (b) = f (0) + 2f (a + b). (4)
Let g(x) = f (x) − f (0). Then g(0) = 0, and (4) can be rewritten as
g(a + b) = g(a) + g(b). (5)
By Cauchy’s functional equation, g(n) = g(1)·n for any integer n. Thus
f can be expressed as f (x) = kx + c, where k, c ∈ Z. Substituting this back
into P (a, b), we see that for all integers a and b,
2k(a + b) + 3c = k 2 (a + b) + (k + 1)c.
 Function Problems 147

This holds if and only if 2k = k 2 , and 3c = (k + 1)c. Therefore, k = 2

or 0. If k = 2, then c can be any integer. If k = 0, then c = 0.
Thus, the functions that satisfy the conditions are f (x) = 0 or f (x) =
2x + c, where c is any integer.
Note. There are several similar problems:

• Find all continuous functions f : R → R such that

f (f (x + y)) = f (x) + f (y) for all x, y ∈ R.

• (Japan Team Selection Test 2021, Problem 12). Find all functions
f : Z → Z satisfying
2
+b2 )
f (a (a + b) = af (a) + bf (b) for all a, b ∈ Z.

Here, f (0) (n) = n and f (k) (n) = f (f (k−1) (n)).

Score Situation This particular problem saw the following distribution of scores among
contestants: 382 contestants scored 7 points, 52 contestants scored 6 points, 5 contestants
scored 5 points, 14 contestants scored 4 points, 24 contestants scored 3 points, 6 contestants
scored 2 points, 65 contestants scored 1 point, and 73 contestants scored 0 point. The
average score for this problem is 5.179, indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 40 points (with a total team score of 227 points), the United States
team scored 42 points (with a total team score of 227 points), the South Korea team scored
42 points (with a total team score of 226 points), the North Korea team scored 41 points
(with a total team score of 187 points), and the Thailand team scored 42 points (with a total
team score of 185 points).
The gold medal cutoff for this IMO was set at 31 points (with 52 contestants earning gold
medals), the silver medal cutoff was 24 points (with 94 contestants earning silver medals),
and the bronze medal cutoff was 17 points (with 156 contestants earning bronze medals).
In this IMO, a total of six contestants achieved a perfect score of 42 points.

Problem 2.28 (IMO 63-2, proposed by the Netherlands). Let R+

denote the set of positive real numbers. Find all functions f : R+ → R+
such that for each x ∈ R+ , there is exactly one y ∈ R+ satisfying

xf (y) + yf (x) ≤ 2.

1
Solution 1. The unique function that satisfies the conditions is f (x) = x
for all x ∈ R+ .
148 IMO Problems, Theorems, and Methods: Algebra

First, we verify that the function f (x) = x1 meets the conditions. For
any x, y ∈ R+ , using the AM-GM inequality, we know
x y
xf (y) + yf (x) = + ≥ 2,
y x
with equality if and only if x = y. This indicates that the only y ∈ R+
satisfying xf (y) + yf (x) ≤ 2 is y = x.
Next, we prove that if f satisfies the conditions, then f (x) = x1 .
If a pair of positive real numbers (x, y) satisfies xf (y) + yf (x) ≤ 2, then
we call it a “good pair.” If (x, y) is a good pair, then (y, x) is also a good
pair.
We assert that for any good pair (x, y), it must be that x = y. Oth-
erwise, suppose x = y. Then the conditions imply that neither (x, x) nor
(y, y) are good pairs, i.e., xf (x) > 1 and yf (y) > 1. Using the AM-GM
inequality, we have
 
xf (y) + yf (x) ≥ 2 xf (y) · yf (x) = 2 xf (x) · yf (y) > 2,

contradicting the assumption that (x, y) is a good pair.

From this assertion, it follows that for each x ∈ R+ , we see that (x, x)
is the only good pair that includes it. Thus xf (x) ≤ 1, i.e., f (x) ≤ x1 for
all x ∈ R+ . Let y = f (x)
1
. Then

1 1
xf (y) + yf (x) ≤ x · + · f (x) = xf (x) + 1 ≤ 2,
y f (x)

1
indicating that x, f (x) is also a good pair.
1
By the assertion, x = f (x) , and thus f (x) = x1 .

Solution 2. The verification for f (x) = x1 is the same as in Solution 1.

Next, we use another approach to proving that if f satisfies the conditions,
then f (x) = x1 .
For each x ∈ R+ , denote the unique y ∈ R+ satisfying xf (y) +
yf (x) ≤ 2 as σ(x). First, we prove two conclusions:

(i) f is strictly decreasing, i.e., f (y1 ) > f (y2 ) for any y1 < y2 .
Otherwise, suppose f (y1 ) ≤ f (y2 ). Let x = σ(y2 ). Then

xf (y1 ) + y1 f (x) < xf (y2 ) + y2 f (x) ≤ 2,

contradicting the uniqueness of σ(x).

 Function Problems 149

(ii) For any x ∈ R+ , let y = σ(x). Then xf (y) + yf (x) = 2.

Otherwise, suppose xf (y) + yf (x) < 2. Let y  = y + ε, where
2 − xf (y) − yf (x)
ε=
f (x)
is a positive number. Using conclusion (i), we have
xf (y  ) + y  f (x) < xf (y) + yf (x) + εf (x) = 2,
contradicting the uniqueness of σ(x).
From conclusion (ii), for any x, y ∈ R+ ,
xf (y) + yf (x) ≥ 2.
Taking y = x in the above inequality gives f (x) ≥ x1 , and therefore
σ(x) x
2 = f (x)σ(x) + f (σ(x))x ≥ + ,
x σ(x)
which is only possible if σ(x) = x.
Substituting σ(x) = x into conclusion (ii) yields f (x) = x1 .
Note. There is a similar problem:

• (Korean Mathematical Olympiad 2023, Final Round, Problem

2). Let R+ denote the set of positive real numbers. Suppose a function
f : R+ → R+ satisfies the following condition:
For each x ∈ R+ , there exists a y ∈ R+ such that
(x + f (y))(y + f (x)) ≤ 4,
and the number of y is finite. Prove that f (x) > f (y) for any positive
real numbers x < y.

Score Situation This particular problem saw the following distribution of scores among
contestants: 303 contestants scored 7 points, 19 contestants scored 6 points, 3 contestants
scored 5 points, 7 contestants scored 4 points, 41 contestants scored 3 points, 23 contestants
scored 2 points, 89 contestants scored 1 point, and 104 contestants scored 0 point. The
average score for this problem is 4.306, indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 252 points), the South Korea
team scored 42 points (with a total team score of 208 points), the United States team scored
42 points (with a total team score of 207 points), the Vietnam team scored 42 points (with
a total team score of 196 points), and the Romania team scored 41 points (with a total team
score of 194 points).
150 IMO Problems, Theorems, and Methods: Algebra

The gold medal cutoff for this IMO was set at 34 points (with 44 contestants earning gold
medals), the silver medal cutoff was 29 points (with 101 contestants earning silver medals),
and the bronze medal cutoff was 23 points (with 140 contestants earning bronze medals).
In this IMO, a total of 10 contestants achieved a perfect score of 42 points.

2.3 Summary
Functional equations constitute a rich and broadly applied branch of math-
ematics. In 2000, one of the seven Millennium Prize Problems released by
the Clay Mathematics Institute in the United States concerned the issue
of smoothness and existence of solutions for the three-dimensional Navier-
Stokes equations.
In the first 64 IMOs, there were a total of 28 function problems.
These problems can be broadly categorized into three types, as depicted in
Figure 2.1. The score details for these problems are presented in Table 2.2.
Due to the smaller number of participating teams and missing contestant
score information in early IMOs, there are several blanks in Table 2.2.

6
Proving Properties Determining Values Deriving Expressions
5

0
1–10 11–20 21–30 31–40 41–50 51–60 61–64

Figure 2.1 Numbers of Function Problems in the First 64 IMOs

Problems 2.1–2.9 focus on “proving properties of functions;” among

these nine problems, the one with the lowest average score is Problem 2.8
(IMO 52-3), proposed by Belarus. Problems 2.10–2.13 deal with “deter-
mining numerical values of function variables or outputs;” among these
four problems, the one with the lowest average score is Problem 2.13 (IMO
29-3), proposed by the United Kingdom. Problems 2.14–2.28 are about
 Function Problems 151

“deriving expressions for functions that meet specific conditions;” among

these 15 problems, the one with the lowest average score is Problem 2.19
(IMO 40-6), proposed by Japan.
These 28 problems were proposed by 17 countries, with the United King-
dom contributing the most, totaling six problems. Bulgaria proposed three
problems, while India, France, South Africa, and Albania each contributed
two problems.
From Table 2.2, it can be observed that in the first 64 IMOs, there were
four function problems with an average score of 1–2 points; seven problems
with an average score of 2–3 points; eight problems with an average score
of 3–4 points; nine problems with an average score above 4 points. Overall,
the function problems were relatively simple, and the contestants scored
relatively high.
In the 24th–64th IMOs, there were a total of 20 function problems.
Among these, four had an average score of 1–2 points; six had an average
score of 2–3 points; five had an average score of 3–4 points; five had an
average score above 4 points. Further analysis of the problem numbers of
these 20 function problems, as shown in Table 2.3, reveals that these prob-
lems frequently appeared as the 1st/4th or 2nd/5th problem. The majority
of these problems, totaling 15, were of the type deriving expressions for
functions that meet specific conditions. The other two types of function
problems were less frequent, with only five problems appearing.
Excluding Problem 2.15 (IMO 24-1), the function problems from the
25th–64th IMOs are arranged in order of their average scores, from left
to right, and a scatter plot of the score details is presented, as shown in
Figure 2.2.
From Table 2.2 and Figure 2.2, it is observable that in the function prob-
lems, the average score of the top five teams generally exceeds the average
score of the problem by 2.5 points, the average score of the 6th–15th teams
typically surpasses the average score by 2 points, and the average score of
the 16th–25th teams usually exceeds the average score by 1.5 points. This
characteristic is more pronounced in difficult problems. However, in simpler
problems, the disparity between the average scores of the top five teams,
the 6th–15th teams, and the 16th–25th teams is not significant, as seen
in problems such as Problem 2.28 (IMO 63-2), Problem 2.21 (IMO 49-4),
Problem 2.27 (IMO 60-1), and Problem 2.23 (IMO 51-1).
From Figure 2.2, it can also be observed that there are several func-
tion problems where the average score is higher than the average score
of the 16th–25th teams, such as Problem 2.13 (IMO 29-3), Problem 2.16
 152
Table 2.2 Score Details of Function Problems in the First 64 IMOs

Problem 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8

Full points 7.000 7.000 7.000 6.000 8.000 7.000 7.000 7.000
Average score 4.593 4.366 2.818 3.376 3.054 3.523 3.383 1.055

IMO Problems, Theorems, and Methods: Algebra

Top five mean 5.725 5.175 4.975 6.667 6.400 4.867
6th–15th mean 2.588 5.800 5.333 3.242
16th–25th mean 2.983 4.426 1.091
Problem number in IMO 10-5 11-2 14-5 15-5 19-6 28-4 34-5 52-3
Proposing country The Hungary Bulgaria Poland Bulgaria Vietnam Germany Belarus
German
Demo-
cratic
Republic

Problem 2.9 2.10 2.11 2.12 2.13 2.14 2.15 2.16

Full points 7.000 8.000 7.000 7.000 7.000 7.000 7.000 7.000
Average score 2.452 3.313 6.510 4.613 1.731 3.575 4.152 2.955
Top five mean 6.633 5.025 6.900 6.700 4.167 6.400 6.067
6th–15th mean 5.983 2.921 6.096 5.525 2.864 5.450 3.833
16th–25th mean 3.700 4.492 4.175 1.276 3.655 2.926
Problem number in IMO 54-5 20-3 22-6 23-1 29-3 24-1 27-5 31-4
Proposing country Bulgaria The United Finland The United The United The United The United Turkey
Kingdom Kingdom Kingdom Kingdom Kingdom
(Continued )
 Table 2.2 (Continued )

Problem 2.17 2.18 2.19 2.20 2.21 2.22 2.23 2.24

Full points 7.000 7.000 7.000 7.000 7.000 7.000 7.000 7.000
Average score 2.963 3.221 1.151 2.267 4.402 2.474 5.450 3.766
Top five mean 5.667 6.467 3.200 6.467 6.967 6.800 6.889 6.306
6th–15th mean 4.350 5.100 2.083 4.550 6.383 6.212 6.648 5.333
16th–25th mean 3.817 4.067 1.136 3.500 6.417 5.467 6.667 5.450
Problem number in IMO 33-2 35-5 40-6 43-5 49-4 50-5 51-1 53-4

Function Problems
Proposing country India The United Japan India South France France South
Kingdom Korea Africa

Problem 2.25 2.26 2.27 2.28

Full points 7.000 7.000 7.000 7.000

Average score 1.513 2.304 5.179 4.306
Top five mean 4.067 5.367 6.900 6.967
6th–15th mean 3.067 4.318 6.864 6.800
16th–25th mean 2.133 3.870 6.741 6.467
Problem number in IMO 56-5 58-2 60-1 63-2
Proposing country Albania Albania South The
Africa Netherlands

Note. Top five mean = Total score of the top five teams ÷ Total number of contestants from the top five teams,
6th–15th mean = Total score of the 6th–15th teams ÷ Total number of contestants from the 6th–15th teams,
16th–25th mean = Total score of the 16th–25th teams ÷ Total number of contestants from the 16th–25th teams.

153
154 IMO Problems, Theorems, and Methods: Algebra

Table 2.3 Numbers of Function Problems in the 24th–64th IMOs

Problem Number
Number of Problems in
Function Problem 1, 4 2, 5 3, 6 the First 64 IMOs

Proving properties 1 2 1 9
Determining values 0 0 1 4
Deriving expressions 6 8 1 15
Total 7 10 3 28

0
0 2 4 6 8 10 12 14 16 18 20

Average Score Average Score of Top Five Teams

Average Score of 6th–15th Teams Average Score of 16th–25th Teams

Figure 2.2 Score Details of Function Problems in the 25th–64th IMOs

(IMO 31-4), Problem 2.6 (IMO 28-4), and Problem 2.15 (IMO 27-5). This
phenomenon is due to the smaller number of participating teams in early
IMOs. It was not until the 30th IMO in 1989 that the number of partici-
pating teams exceeded 50. Therefore, it is common to see situations where
the average score is close to or even higher than the average score of the
16th–25th teams during this period.
Chapter 3

Sequence Problems

When it comes to sequences, one might immediately think of the story of

the renowned mathematician Gauss discovering the formula for the sum of
an arithmetic sequence. However, the appearance of formulas for the sum
of arithmetic and geometric sequences dates back much earlier.
The invention of numbers naturally led ancient scholars to study
sequences. The Nine Chapters on the Mathematical Art recorded such a
problem: To calculate the total distance traveled by a good horse over fif-
teen days, this distance can be determined by multiplying 15 with the sum
of the first day’s distance and seven times the daily incremental distance.
Here, the daily incremental distance is the constant additional distance
d the horse travels each day beyond the previous day’s distance, and let
the first day’s distance be a1 . Then,the total distance
 traveled by the
(n−1)d
good horse in n days is given by Sn = a1 + 2 n. This indicates that
Chinese mathematicians of that time were already aware of the formula for
the sum of an arithmetic sequence.
By the 19th century, people began to focus on more types of sequence
summation problems and derived sum formulas. These efforts are mainly
reflected in two lengthy entries, “Algebra” and “Series,” written by the
Scottish mathematician William Wallace for the Encyclopedia Britannica.
In “Algebra,” Wallace used the method of adding in reverse order to
derive the formula for the sum of an arithmetic sequence and used the

155
156 IMO Problems, Theorems, and Methods: Algebra

method of displacement subtraction to derive the formula for the sum of

a geometric sequence. In “Series,” he also derived sum formulas for other
sequences, such as sequences with terms

k × (k + 1) × · · · × (k + p − 1) p 1
, k , and .
1 × 2 × ···× p k × (k + 1) × · · · × (k + p − 1)

After the emergence of the concept of functions, sequences can be

regarded as a special type of functions whose domains are the set of natural
numbers. Viewing sequences from this perspective facilitates the study of
their monotonicity, boundedness, and relationships with more advanced
mathematical concepts. For instance, a problem from the 1980 British
Mathematical Olympiad demonstrates the application of the concept of
“limits” in this context: Find the set of real numbers a0 for which the infi-
nite sequence {an } of real numbers defined by an+1 = 2n − 3an (n ≥ 0) is
strictly increasing.
Additionally, one can approach the divisibility properties of sequences
from a number theory perspective. For instance, a problem from the 29th
IMO Shortlist states: An integer sequence is defined by a0 = 0, a1 = 1, and
an = 2an−1 + an−2 (n > 1). Prove that 2k |an if and only if 2k |n. Further-
more, sequences frequently appear in problems related to combinatorics,
inequalities, and other topics.
In the first 64th IMOs, there had been a total of 14 sequence prob-
lems, approximately accounting for 13.8% of all algebra problems. These
problems can be primarily categorized into three types: (1) determining
the value of a specific term or the number of terms, totaling four prob-
lems; (2) addressing problems related to the existence of sequences, totaling
five problems; (3) proving quantitative relationships satisfied by sequences,
totaling five problems. The statistical distribution of these three types of
problems in the previous IMOs is presented in Table 3.1.
It can be observed that there were not many sequence problems, with
only 1–3 appearing in every 10 IMOs. To some extent, this is because
sequence problems are typically intertwined with number theory and com-
binatorics, rarely focusing on the algebraic properties of sequences, and
hence not being included in Table 3.1.
Although the number of sequence problems is not particularly large,
this does not imply they are always straightforward. In fact, many are
 Sequence Problems 157

Table 3.1 Numbers of Sequence Problems in the First 64 IMOs

Session
Content 1–10 11–20 21–30 31–40 41–50 51–60 61–64 Total

Determining values 1 0 0 1 0 2 0 4
Existence problems 0 2 1 1 0 0 1 5
Proving quantitative 0 1 1 1 1 1 0 5
relationships
Algebra problems 20 20 14 13 15 13 6 101
The percentage of sequence 5.0% 15.0% 14.3 % 23.1% 6.7% 23.1% 16.7% 13.8%
problems among the
algebra problems

quite challenging. For example, the average score of Problem 3.14 (IMO
51-6) is merely 0.368 points, with as many as 470 contestants scoring
zero.
Addressing sequence problems typically involves two primary methods.
The first is the recursive method, such as the use of mathematical induction.
The second is the algebraic method, which demands proficiency in tech-
niques including substitution, identity transformations, and finding the
general term of a sequence. In mathematics competitions, many sequence
problems often start with algebraic transformations.
This chapter will be divided into three parts. The first part intro-
duces some common types and properties of sequences. Although peri-
odic sequences are usually associated with divisibility and congruence, their
properties are also reviewed here. Subsequently, commonly used methods
for finding the general term and sum of sequences are discussed, along with
several theorems that are easily overlooked.
The second part revolves around three types of problems: “determining
the value of a specific term or the number of terms,” “addressing problems
related to the existence of sequences,” and “proving quantitative relation-
ships satisfied by sequences.” These problems are presented in chronolog-
ical order, and some problems include various solutions, generalizations,
and similar problems.
It is important to note that for each problem, the solutions are fol-
lowed by information on the scores, including the number of contestants in
each score range, the average score, and the scores of the top five teams.
However, early IMOs often lacked information on contestant scores, so the
158 IMO Problems, Theorems, and Methods: Algebra

number of contestants in each score range only represents the counted num-
ber of contestants, and some problems lack scores of the top five teams.
The third part provides a brief summary of this chapter.

3.1 Common Theorems, Formulas, and Methods

3.1.1 Common sequences
(1) Arithmetic sequences
An arithmetic sequence {an } is defined such that, starting from the second
term, each term’s difference with its preceding term equals a constant. This
constant is referred to as the common difference of the arithmetic sequence
and is usually denoted by d. The general term is given by an = a1 +(n−1)d.

(2) Geometric sequences

A geometric sequence {an } is defined such that, starting from the second
term, each term’s ratio with its preceding term equals a constant. This
constant is referred to as the common ratio of the geometric sequence and
is usually denoted by q. The general term is given by an = a1 · q n−1 .
Let the sum of the first n terms of the sequence {an } be denoted as Sn .
The following propositions hold:

Proposition 3.1. If Sn = An2 + Bn + C, then {an } is an arithmetic

sequence if and only if C = 0.

Proposition 3.2. {an } is an arithmetic sequence with a common differ-

ence d if and only if {aan } is a geometric sequence with a common ratio
ad , where a > 0 and a = 1.

Proposition 3.3. {an } is an arithmetic sequence if and only if Sn =

n(a1 +an )
2 .

Proposition 3.4. {an } is an arithmetic sequence with a common differ-

ence d if and only if Snn is an arithmetic sequence with a common dif-
ference d2 .

Proposition 3.5. If {an } is an arithmetic sequence, then Sn , S2n −

Sn , S3n − S2n , . . . form an arithmetic sequence.
 Sequence Problems 159

Proposition 3.6. If {an } is an arithmetic sequence, then Sm+n =

(Sn −Sm )(n+m)
n−m .

Furthermore, if Sn = m and Sm = n, then Sm+n = −(m + n). If

Sn = Sm (m = n), then Sm+n = 0.

Proposition 3.7. If m + n = p + q, where m, n, p, q are positive integers,

then am + an = ap + aq .

Furthermore, if m + n = 2p, then am + an = 2ap .

Proposition 3.8. If {an } and {bn } are both arithmetic sequences, then
{kan + b} and {man + nbn } are also arithmetic sequences, where k, b, m, n
are constants.

Proposition 3.9. Let An and Bn denote the sums of the first n terms of
arithmetic sequences {an } and {bn } respectively. Then abnn = A 2n−1
B2n−1 .

Proposition 3.10. If sequences {an } and {bn } satisfy the condition that
for n ≥ 1,
a1 + 2a2 + · · · + nan
bn = ,
1 + 2 + ··· + n
then {bn } is an arithmetic sequence if and only if {an } is an arithmetic
sequence.

Proposition 3.11. If real numbers a and d satisfy d = 0 and ad ≥ 0, then

the arithmetic sequence

a + d, a + 2d, . . . , a + nd, . . .
a
has three terms in a geometric progression if and only if d is a rational
number.

Proposition 3.12. If Sn = Aq n + B(AB = 0, q = 0, 1), then {an } is a

geometric sequence if and only if A + B = 0.

Proposition 3.13. {an } is a geometric sequence with a positive common

ratio q if and only if {loga an } is an arithmetic sequence with a common
difference loga q, where a > 0 and a = 1.
160 IMO Problems, Theorems, and Methods: Algebra

Proposition 3.14. {an } is a geometric sequence with a common ratio q

if and only if Sn+1 = a1 + qSn .

Proposition 3.15. If {an } is a geometric sequence, then Sn , S2n −

Sn , S3n − S2n , . . . also form a geometric sequence.

Proposition 3.16. If m + n = p + q, where m, n, p, q are positive integers,

then am · an = ap · aq .

Furthermore, if m + n = 2p, then am · an = a2p .

Proposition 3.17. If {an } is a geometric sequence, then am , am+k ,

am+2k , . . . , am+nk , . . . is also a geometric sequence, where m, k are positive
integers.

(3) High-order arithmetic sequences

For a given sequence {an }, taking the differences between the adjacent
terms, a new sequence is obtained:

a2 − a1 , a3 − a2 , . . . , an+1 − an , . . . ;

this sequence is called the first order difference sequence of {an }.

If this sequence is denoted as {bn }, where bn = an+1 − an , and the
differences between the adjacent terms of {bn } are taken, then a sequence
is obtained:

b2 − b1 , b3 − b2 , . . . , bn+1 − bn , . . . ;

this sequence is termed the second order difference sequence of the original
sequence {an }.
Following this pattern, for any p ∈ N+ , the pth order difference sequence
of {an } can be defined. If the pth order difference sequence of {an } is a
non-zero constant sequence, then {an } is referred to as a pth order arith-
metic sequence. Particularly, a first order arithmetic sequence is commonly
known as an arithmetic sequence, while second and higher-order arithmetic
sequences are collectively termed high-order arithmetic sequences. The fol-
lowing propositions hold:

Proposition 3.18. If {an } is a pth order arithmetic sequence, then its first
order difference sequence is a (p − 1)th order arithmetic sequence.
 Sequence Problems 161

Proposition 3.19. A sequence {an } is pth order arithmetic if and only if

its general term is a polynomial of degree p with respect to n.

Proposition 3.20. If {an } is a pth order arithmetic sequence, then the

sum of its first n terms is a degree p + 1 polynomial with respect to n.

The most common problems in high-order arithmetic sequences involve

finding the general term formula and the sum of the first n terms.

(4) Recursive sequences

There are two prevalent methods for defining a sequence: one is by provid-
ing its general term formula, and the other is by specifying the initial terms
along with the recurrence relation.
For any n ∈ N+ , a sequence {an } determined by the recurrence relation
an+k = f (an+k−1 , an+k−2 , . . . , an )
is called a kth order recursive sequence. If f is linear, then the sequence
is called a kth order linear recursive sequence; otherwise, it is called a kth
order nonlinear recursive sequence.

Proposition 3.21. A kth order arithmetic sequence {an } is determined by

the recurrence relation
k k−1
an+k+1 = Ck+1 an+k − Ck+1 an+k−1 + · · · + (−1)k an .

Proposition 3.22. If a characteristic equation

xk = λ1 xk−1 + λ2 xk−2 + · · · + λk (λk = 0)
has k distinct roots x1 , x2 , . . . , xk , then for the corresponding recurrence
relation
an+k = λ1 an+k−1 + λ2 an+k−2 + · · · + λk an ,
the general term formula of the sequence {an } is given by
an = c1 xn1 + c2 xn2 + · · · + ck xnk ,
where c1 , c2 , . . . , ck are the solutions of the following system of equations:
⎧
⎪
⎪ c x + c2 x2 + · · · + ck xk = a1 ,
⎪ 1 1
⎪
⎨ c x2 + c x2 + · · · + c x2 = a ,
1 1 2 2 k k 2
⎪······························
⎪
⎪
⎪
⎩
c1 xk1 + c2 xk2 + · · · + ck xkk = ak .
162 IMO Problems, Theorems, and Methods: Algebra

Proposition 3.23. If a characteristic equation

xk = λ1 xk−1 + λ2 xk−2 + · · · + λk (λk = 0)

has a k-fold root λ, then for the corresponding recurrence relation

an+k = λ1 an+k−1 + λ2 an+k−2 + · · · + λk an ,

the general term formula of the sequence {an } is given by

an = (c1 + c2 n + · · · + ck nk−1 )λn ,

where c1 , c2 , . . . , ck are the solutions of the following system of equations:

⎧
⎪
⎪ (c1 + c2 + · · · + ck ) λ = a1 ,
⎪
⎪
⎪
⎨ (c1 + 2c2 + · · · + 2k−1 ck )λ2 = a2 ,
⎪
⎪ ·································
⎪
⎪
⎪
⎩
c1 + kc2 + · · · + k k−1 ck λk = ak .

Proposition 3.24. If a characteristic equation

xk = λ1 xk−1 + λ2 xk−2 + · · · + λk (λk = 0)

m
has m distinct roots x1 , x2 , . . . , xm , where xi is a ki -fold root, and i=1 ki =
k, then for the corresponding recurrence relation

an+k = λ1 an+k−1 + λ2 an+k−2 + · · · + λk an ,

the general term formula of the sequence {an } is given by

an = p1 (n)xn1 + p2 (n)xn2 + · · · + pm (n)xnm ,

where n ∈ N+ , pi (n) = ci1 + ci2 n + · · · + ciki nki −1 (1 ≤ i ≤ m), and

cij (1 ≤ i ≤ m, 1 ≤ j ≤ ki ) are determined by the following system of
equations:
⎧
⎪
⎪ p (1)x1 + p2 (1)x2 + · · · + pm (1)xm = a1 ,
⎪ 1
⎪
⎪
⎨ p1 (2)x2 + p2 (2)x2 + · · · + pm (2)x2 = a2 ,
1 2 m
⎪
⎪ ··········································
⎪
⎪
⎪
⎩
p1 (k)xk1 + p2 (k)xk2 + · · · + pm (k)xkm = ak .
 Sequence Problems 163

Proposition 3.25. If {an } is a kth order linear recursive sequence with a

characteristic equation

f (x) = xk − λ1 xk−1 − λ2 xk−2 − · · · − λk = 0(λk = 0),

and the sum of its first n terms is denoted as Sn , then the sequence {Sn }
is a (k + 1)th order linear recursive sequence with the recurrence relation

Sn+k+1 = (1 + λ1 )Sn+k + (λ2 − λ1 )Sn+k−1 + · · · + (λk − λk−1 )Sn+1 − λk Sn ,

and the characteristic equation is given by (x − 1)f (x) = 0.

Example 3.1. Given a sequence {an } such that a1 = 3, a2 = 8, and

an = 2an−1 + 2an−2 for n = 3, 4, . . ., find the general term formula for
{an }.

Solution. The characteristic equation corresponding

√ to the recurrence
√
relation is x2 = 2x + 2, with two roots x1 = 1 + 3 and x2 = 1 − 3.
Therefore,
√ n √
an = c1 (1 + 3) + c2 (1 − 3)n ,

which leads to the system of equations

√ √
c1 (1 + 3) + c2 (1 − 3) = 3,
√ √
c1 (1 + 3)2 + c2 (1 − 3)2 = 8.
√ √
3+2 3 3−2 3
Solving this system yields c1 = 6 and c2 = 6 . Thus,
√ √
3+2 3 √ n 3−2 3 √
an = (1 + 3) + (1 − 3)n .
6 6
In particular, for a recursive system given by

xn = axn−1 + byn−1 ,
yn = cxn−1 + dyn−1 ,

the characteristic equation is determined by

λ2 − (a + d)λ + (ad − bc) = 0.

164 IMO Problems, Theorems, and Methods: Algebra

If the characteristic equation has two distinct roots λ1 and λ2 , then the
general term formulas for {xn } and {yn } are
xn = Aλn1 + Bλn2 ,
yn = Cλn1 + Dλn2 ,
where the constants A, B, C, D are determined by the initial conditions:
x1 = Aλ1 + Bλ2 , x2 = Aλ21 + Bλ22 ,
and
y1 = Cλ1 + Dλ2 y2 = Cλ21 + Dλ22 .
If the characteristic equation has a double root λ, then the general term
formulas for {xn } and {yn } become:

xn = (An + B)λn ,
yn = (Cn + D)λn ,
where the constants A, B, C, D are determined by the initial conditions:
 
x1 = (A + B) λ, x2 = (2A + B) λ2 ,
and
y1 = (C + D) λ y2 = (2C + D) λ2 .
xn −axn−1
Actually, by substituting yn−1 = b into yn = cxn−1 + dyn−1 ,
we obtain
xn+1 = (a + d)xn + (bc − ad)xn−1 .

(5) Periodic sequences

For a sequence {an }, if there exist specific positive integers T and n0 , such
that an+T = an for all n ≥ n0 , then {an } is called a periodic sequence
with a period T starting from the n0 th term. If n0 = 1, then the sequence
{an } is called purely periodic. If n0 ≥ 2, then the sequence {an } is called
eventually periodic or uttimately periodic.
Proposition 3.26. The range of values for a periodic sequence is a finite
set.
Proposition 3.27. If T is a period of a periodic sequence {an }, then for
any positive integer k, the integer kT is also a period of {an }.
Proposition 3.28. Periodic sequences have a minimum period.
The minimum period of a periodic sequence is called its least period or
exact period.

Proposition 3.29. If T is the least period of a periodic sequence {an },

and T1 is another period of {an }, then T |T1 .
 Sequence Problems 165

Let {an } be an integer sequence and m be a given positive integer. If

bn is the remainder when an (n ≥ 1) is divided by m, i.e., bn ≡ an (modm)
and bn ∈ {0, 1, 2, . . . , m − 1}, then the sequence {bn } is called the modular
sequence of {an } with respect to m, denoted as {an (modm)}.
For an integer sequence {an }, if there exist positive integers T and N
such that an+T ≡ an (modm) for all n ≥ N, then {an (modm)} is called a
modular periodic sequence, and T is called a period of the modular periodic
sequence.
The smallest value of T is referred to as the least period, denoted as
T (m). If N = 1, then the sequence {an (modm)} is called purely modular
periodic. If N ≥ 2, then the sequence {an (modm)} is called eventually or
ultimately modular periodic.
Proposition 3.30. If T = T (m) is the least period of a modular periodic
sequence, then T1 is another period if and only if T |T1 .
Proposition 3.31. If {an (modm)} and {bn (modm)} are both modular
periodic sequences, then {an ± bn (modm)} and {an bn (modm)} are also
modular periodic sequences.
Proposition 3.32. Suppose {an (modm)} is a modular periodic sequence.
If m |m, then {an (modm )} is also a modular periodic sequence, and
T (m ) |T (m).
Proposition 3.33. If {an (modm1 )} and {an (modm2 )} are both mod-
ular periodic sequences, then {an (modm)} is also a modular periodic
sequence, and T (m) = [T (m1 ), T (m2 )], where m = [m1 , m2 ]. Here [a, b]
denotes the least common multiple of a and b.

3.1.2 Common methods for finding general term formulas

(1) Iteration method
(i) an+1 = an + f (n).
For n ≥ 2,
n−1

an = an−1 +f (n−1) = an−2 +f (n−2)+f (n−1) = · · · = a1 + f (k).
k=1
(ii) an+1 = an · f (n).
For n ≥ 2,
n−1

an = an−1 · f (n − 1) = an−2 · f (n − 2) · f (n − 1) = · · · = a1 f (k).
k=1
166 IMO Problems, Theorems, and Methods: Algebra

(iii) an+1 = pan + f (n), where p = 0, 1.

Let bn = apnn and g(n) = pfn+1
(n)
. Then bn+1 = bn + g(n), which can be
transformed into (i).
(iv) an+1 = pan + q, where p and q are constants, pq = 0, and p = 1.
Subtracting an = pan−1 +q from an+1 = pan +q, we obtain an+1 −an =
p(an − an−1 ). Let bn = an+1 − an . Then it reduces to (ii), where
f (n) = p is a constant function.

Similarly, the following recurrence relations can also be reduced to the

aforementioned types:

(v) an+1 = f (n)an + g(n).

(vi) an+1 = pama n
n +q
, where p = 0 and m = 0.
(vii) an+2 = pan+1 + qan , where pq = 0.
(viii) an+1 = paqn , where p = 0 and q = 1.

(2) Fixed-point method

(i) an+1 = pan + q, where p and q are constants, pq = 0, and p = 1.
Consider the fixed point of the function f (x) = px+q, i.e., the solution
q
of px + q = x is λ = 1−p . We have an+1 − λ = p(an − λ), and then

an − λ = pn−1 (a1 − λ).

(ii) an+1 = paann+q , where pq = 0 and a1 = 0.

Taking the reciprocal of both sides of the recurrence relation yields
an+1 = q · an + p. Let bn = an . Then it reduces to (i).
1 1 1

αan +β
(iii) an+1 = an +γ , where α, β, γ are constants.
αx+β
Consider the fixed points of the function f (x) = x+γ , i.e., the solu-
αx+β
tions of the equation x = x+γ are x1 , x2 .
 n−1
an −x1
• If x1 = x2 , then an −x2 = aa11 −x α−x1
−x2 · α−x2
1
.
• If x1 = x2 , then 1
an −x1 = 1
a1 −x1 + (n − 1) · 1
α−x1 .

Example 3.2. Consider a sequence {an } defined as follows: a1 = 2 and

an+1 = 2a n +6
an +1 for n = 1, 2, . . . . Determine the general term formula for
{an }.
 Sequence Problems 167

2x+6
Solution. Find the fixed points by setting x = x+1 , yielding x1 = 3 and
x2 = −2. Therefore,
2an + 6 −an + 3 2an + 6 4an + 8
an+1 − 3 = −3= , an+1 + 2 = +2= .
an + 1 an + 1 an + 1 an + 1
Dividing the two equations, we get
an+1 − 3 an − 3
=− ,
an+1 + 2 4(an + 2)
 
which implies that aan+1 −3
n+1 +2
is a geometric sequence with a common
a1 −3 an −3
ratio of − 14 and an initial term of a1 +2 = − 4 .
1
Consequently, an +2 =
n−1 n n
− 14 − 41 = − 14 . Thus, an = 3(−4) +2
(−4)n −1 .

(3) Characteristic equation method

For an+2 = pan+1 + qan , where pq = 0, let α, β be the roots of the charac-
teristic equation x2 = px + q.

• If α = β, then an = 1
α−β (a2 − βa1 )αn−1 − (a2 − αa1 )β n−1 .
a1 a2 −αa1
• If α = β, then an = α + (n − 1) · α2 αn .

(4) Vieta’s formulas method


For an+1 = pan + (p2 − 1)a2n + 1, where p is a positive integer, consider an
as a constant. Then, an−1 and an+1 are the roots of the quadratic equation

x2 − 2pan x + a2n − 1 = 0.

Using Vieta’s formulas, we have an+1 = 2pan − an−1 .

There is a related problem:

• (British Mathematical Olympiad 2002, 2nd Round, Problem 3).

Prove that a sequence defined by
1  
y0 = 1 and yn+1 = 3yn + 5yn2 − 4 for n ≥ 0
2
consists only of integers.
168 IMO Problems, Theorems, and Methods: Algebra

3.1.3 Common methods for summations of sequences

(1) Formula method
For some sequences, such as arithmetic and geometric sequences, there exist
corresponding summation formulas for the sum of the first n terms, denoted
as Sn .

(2) Reverse summation method

This method is generally applicable in cases where the sum of the kth term
and the kth term from the end in a sequence is equal to the sum of the first
and last terms. For instance, if α = a1 + an = ak + an+1−k with 1 ≤ k ≤ n,
then Sn = nα2 .

(3) Telescoping method

This method is generally suitable when the general term an can be decom-
posed into the difference of two terms, such as an = f (n) − f (n − k), where
k is an integer. Common telescoping forms include:
 
n − n+k .
1 1 1 1
(i) an = n(n+k) = k
√ √
1
√ n+k− n
(ii) an = √ = .
n+ n+k k
 
(iii) an = 1
n(n+1)(n+2) = 1
2
1
n(n+1) − 1
(n+1)(n+2) .
 
(2n)2
(iv) an = (2n−1)(2n+1) =1+ 1
2
1
2n−1 − 1
2n+1 .
 
2n−1
(v) an = (2n +1)(2n+1 +1) = 1
2
1
2n +1 − 2n+1
1
+1 .
n+2
(vi) an = 2n n(n+1) = 1
2n−1 n − 1
2n (n+1) .

Additionally, the telescoping method is closely related to the bounding

method.
 
(vii) 1
k3 < 1
k3 −k = 1
k(k2 −1) = 1
k(k−1)(k+1) = 1
2
1
k(k−1) − 1
k(k+1) .

(viii) 1
k2 < 4
4k2 −1 < 1
k2 −1 ≤ 1
k2 −k < 1
k2 −2k .
1 1 1 1
(ix) k2 > k2 +k > k2 +2k > k2 +3k+2 .
√ √
(x) √1
k
= 2
√
2 k
< √ 2
√
k+ k−1
= 2( k − k − 1).
 Sequence Problems 169

√ √
(xi) √1
k
= 2
√
2 k
> √ = 2( k + 1 − k).
2
√
k+ k+1
 
1
(xii) k√ k
< √1
k k−1
= √ √ 1√
k· k· k−1
= √1
k
√1
k−1
− √1
k
√ 1
√
k− k−1
 
< 2 √k−11
− √1k .
 
1
(xiii) k√ > √1 = √ √ 1√ = √1 √1 − √ 1 √ 1 √
k k k+1 k· k· k+1 k k k+1 k+1− k
 
> 2 √1k − √k+11
.

(4) Shifting and subtracting method

This method is typically employed when the general term is in the form
of “arithmetic times geometric.” If an = (an + b)q n−1 , then the sum of
a
the first n terms is given by Sn = (An + B)q n − B, where A = q−1 and
b−A
B = q−1 .

3.1.4 Mathematical induction

(1) First principle of mathematical induction
Let P (n) be a proposition concerning all positive integers n. If

• P (1) is true;
• assuming P (k) is true, one can deduce that P (k + 1) is also true;

then the proposition P (n) is true for any positive integer n.

(2) Second principle of mathematical induction

Let P (n) be a proposition concerning all positive integers n. If

• P (1) is true;
• assuming that P (k) holds for all positive integers k less than n, one can
deduce that P (n) is also true;

then the proposition P (n) is true for any positive integer n.

As the general term an of a sequence is related to the positive integer
n, sequence problems often employ mathematical induction. This includes
finding the general term an , proving that an is an integer, odd number, or a
170 IMO Problems, Theorems, and Methods: Algebra

perfect square, and establishing recurrence relations or inequalities between

adjacent terms of the sequence.

3.1.5 Other important theorems

Theorem 3.1 (Well-Ordering Principle). For any non-empty subset
T of the set of positive integers N+ , there exists a minimum element.
In other words, there exists a positive integer t0 ∈ T such that t0 ≤ t
for any t ∈ T .
Furthermore, the proof by infinite descent, also known as Fermat’s
method of descent, is closely related to the Well-Ordering Principle. If the
proposition P (n) concerning positive integers n holds true for n = n0 , and
it can be deduced that the proposition also holds for infinitely many integers
n = n1 , n2 , . . . (n0 > n1 > n2 · · · ), then the proposition is false.
Fermat used this method to prove that the equation x4 + y 4 = z 4 has
no positive integer solution. He assumed the existence of a set of positive
integer solution (x, y, z) and then derived another smaller set of positive
integer solution (x , y  , z  ), where z  < z (or x + y  + z  < x + y + z). By
repeating this process, a new smaller solution can be obtained each time,
and an infinitely decreasing sequence of positive integers

z > z  > z  > · · ·

will be generated. However, a strictly decreasing sequence of positive inte-

gers can only have a finite number of terms, leading to a contradic-
tion and demonstrating that the original equation has no positive integer
solution.

Corollary 3.1. Let M be a non-empty subset of the set of positive integers

N+ and suppose that M has an upper bound, i.e., there exists an a ∈ N+
such that x ≤ a for any x ∈ M , then M has a maximum element.

Corollary 3.2. Any finite set of real numbers necessarily contains both a
minimum element and a maximum element.

Theorem 3.2. A set M composed of n real numbers can be expressed as

M = {x1 , x2 , . . . , xn }, where x1 < x2 < · · · < xn .
Theorem 3.3 (Monotone Convergence Theorem). A monotonically
increasing (decreasing) sequence that is bounded above (below) has a limit,
and this limit is the least upper (greatest lower) bound, also known as
supremum (infimum).
 Sequence Problems 171

Theorem 3.4 (Nested Interval Theorem). Given a series of closed

intervals {[an , bn ]} satisfying the conditions:

• [an+1 , bn+1 ] ⊂ [an , bn ] for n = 1, 2, . . . and

• limn→+∞ (bn − an ) = 0,

there exists a real number ξ = limn→+∞ an = limn→+∞ bn , and ξ is the

unique common point of all the closed intervals.
Theorem 3.5 (Boundedness Theorem). If a sequence {an } has a limit,
then the sequence {an } is bounded.
Theorem 3.6. A sequence {an } has a limit if and only if every non-trivial
subsequence of {an } has the same limit.

3.2 Problems and Solutions

3.2.1 Determining values
Problem 3.1 (IMO 9-5, proposed by the Soviet Union). Consider
a sequence {Cn }, where

C1 = a1 + a2 + · · · + a8 ,
C2 = a21 + a22 + · · · + a28 ,
························
Cn = an1 + an2 + · · · + an8 ,
························

in which a1 , a2 , . . . , a8 are real numbers not all equal to zero. Suppose that
an infinite number of terms of the sequence {Cn } are equal to zero. Find
all positive integers n for which Cn = 0.

Solution. Assume that a1 has the largest absolute value. Then there must
 = 1) such that ai = −a1 . Otherwise, for sufficiently large n,
exist some ai (i
n
ai
it holds that a1 either are constantly 1 or tend to zero (i = 1). Thus,
  n  n 
n a2 a8
|Cn | = |a1 | 1 + + ···+ > 0,
a1 a1
which contradicts the given conditions.
Without loss of generality, let a2 = −a1 . Since n that makes Cn = 0
must be odd, an1 + an2 = 0. Removing ak1 and ak2 from Ck , the remaining
{Cn } still has infinitely many terms equal to zero. Following the above
172 IMO Problems, Theorems, and Methods: Algebra

reasoning, we can assume a3 = −a4 , a5 = −a6 , and a7 = −a8 . Therefore,

when n is any positive odd numbers, Cn = 0.
Note. If the given n expressions are of the form Ck = ak1 + ak2 + · · · + akt ,
where t is any positive integer, then the solution remains the set of all
positive odd numbers.
There are several similar problems:
• (Germany Team Selection Test 2004, Problem 16). Let n be a
positive integer. Find all complex numbers x1 , x2 , . . . , xn satisfying the
following system of equations:
⎧
⎪
⎪ x1 + 2x2 + · · · + nxn = 0,
⎪
⎨ x2 + 2x2 + · · · + nx2 = 0,
1 2 n
⎪
⎪ ···························
⎪
⎩ n
x1 + 2xn2 + · · · + nxnn = 0.
• (All-Russian Mathematical Olympiad 1996, Final Round,
Grade 10, Problem 4). Prove that if nonzero numbers a1 , a2 , . . . , am
satisfy for each k = 0, 1, . . . , n(n < m − 1),

a1 + 2k a2 + 3k a3 + · · · + mk am = 0,

then the sequence a1 , a2 , . . . , am contains at least n+1 pairs of consecutive

terms having opposite signs.
• (Japan Mathematical Olympiad 1995, Final Round, Problem
5). Let k, n be integers such that 1 ≤ k ≤ n, and let a1 , a2 , . . . , ak be
numbers satisfying the following equations:
⎧
⎪
⎪ a1 + a2 + · · · + ak = n,
⎨ 2
a1 + a22 + · · · + a2k = n,
⎪
⎪ ························
⎩ k
a1 + ak2 + · · · + akk = n.
Prove that

(x + a1 ) (x + a2 ) · · · (x + an ) = xk +C1n xk−1 +C2n xk−2 +· · ·+Ck−1 k

n x+Cn ,

where Cin is a binomial coefficient which means

n · (n − 1) · · · · · (n − i + 2) · (n − i + 1)
.
i · (i − 1) · · · · · 2 · 1

Score Situation This particular problem saw the following distribution of scores among
contestants: 13 contestants scored 7 points, no contestant scored 6 points, 2 contestants
 Sequence Problems 173

scored 5 points, 3 contestants scored 4 points, 3 contestants scored 3 points, 1 contestant

scored 2 points, 8 contestants scored 1 point, and 7 contestants scored 0 point. The average
score of this problem is 3.568, indicating that it was relatively straightforward.
Among the top five teams in the team scores, the Soviet Union team achieved a total score
of 275 points, the German Democratic Republic team achieved a total score of 257 points,
the Hungary team achieved a total score of 251 points, the United Kingdom team achieved
a total score of 231 points, and the Romania team achieved a total score of 214 points.
The gold medal cutoff for this IMO was set at 38 points (with 11 contestants earning gold
medals), the silver medal cutoff was 30 points (with 14 contestants earning silver medals),
and the bronze medal cutoff was 22 points (with 26 contestants earning bronze medals).
In this IMO, a total of five contestants achieved a perfect score of 42 points.

Problem 3.2 (IMO 36-4, proposed by Poland). Find the maximum

value of x0 for which there exists a sequence x0 , x1 , . . . , x1995 of positive
reals with x0 = x1995 , such that for i = 1, 2, . . . , 1995,
2 1
xi−1 + = 2xi + .
xi−1 xi

Solution. Rewrite the recurrence relation as

1
(xi−1 − 2xi ) + (2xi − xi−1 ) = 0.
xi xi−1

Therefore, xi = xi−12
1
or xi = xi−1 .
Suppose, starting from x0 , there are k − t transformations of the first
kind and t transformations of the second kind to reach xk . It is evident
(−1)t
that xk = 2s x0 , where s ≡ k − t(mod2).
Now consider the case k = 1995. If t is even, then x0 = x1995 = 2s x0 .
Since x0 > 0, it follows that s = 0, implying 0 ≡ 1995 − t(mod2), which
contradicts the assumption that t is even. Therefore, t must be odd, and
s
x0 = x1995 = 2s x−10 , yielding x0 = 2 .
2

At this point, s ≡ 1995 − t ≡ 0(mod2). Hence s is even. However,

since the two kinds of transformations occur alternatively, there may be a
cancelation effect, so s ≤ |s| ≤ 1995.
Thus, s ≤ 1994, and x0 ≤ 2997 . When x0 = 2997 , the recurrence can be
realized, i.e.,
1
xi = 2997−i , i = 0, 1, 2, . . . , 1994, x1995 = = 2997 .
x1994

In conclusion, the maximum value of x0 is 2997 .

174 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 168 contestants scored 7 points, 48 contestants scored 6 points, 37 contestants
scored 5 points, 25 contestants scored 4 points, 18 contestants scored 3 points, 16 contestants
scored 2 points, 57 contestants scored 1 point, and 43 contestants scored 0 point. The
average score for this problem is 4.592, indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 36 points (with a total team score of 236 points), the Romania team
scored 42 points (with a total team score of 230 points), the Russia team scored 42 points
(with a total team score of 227 points), the Vietnam team scored 42 points (with a total
team score of 220 points), and the Hungary team scored 42 points (with a total team score
of 210 points).
The gold medal cutoff for this IMO was set at 37 points (with 30 contestants earning gold
medals), the silver medal cutoff was 29 points (with 71 contestants earning silver medals),
and the bronze medal cutoff was 19 points (with 100 contestants earning bronze medals).
In this IMO, a total of 14 contestants achieved a perfect score of 42 points.

Problem 3.3 (IMO 58-1, proposed by South Africa). For each

integer a0 > 1, define a sequence a0 , a1 , a2 , . . . by
√ √
an , if an is an integer,
an+1 = for each n ≥ 0.
an + 3, otherwise,

Determine all values of a0 for which there is a number A such that

an = A for infinitely many values of n.
Solution. The values of a0 satisfying the conditions are all multiples of 3.
Since an+1 is determined solely by an , the sequence {an } has infinitely
many equal terms if and only if it is a periodic sequence, which is also
equivalent to the sequence being bounded.
Notably, perfect squares leave a residue of 0 or 1 when divided by 3.
If a term ak ≡ 2(mod3), then ak is not a perfect square, and ak+1 =
ak + 3, with ak+1 ≡ 2(mod3). Recursively, am+1 = am + 3 for m ≥ k, and
the sequence strictly increases from the kth term onward, resulting in an
unbounded sequence. In particular, if a0 ≡ 2( mod 3), then such an a0 does
not satisfy the conditions.
√
If 3|ak , then whether ak+1 = ak or ak+1 = ak + 3, it holds that 3|ak+1 .
Recursively, am is always a multiple of 3 for m ≥ k. In particular, if 3|a0 ,
then every term in the sequence is divisible by 3.
Assume 3|a0 , and choose a perfect square N 2 > a0 such that 3|N . We
claim that an ≤ N 2 for every n. By contradiction, assume there exists a
 Sequence Problems 175

term greater than N 2 , and take the smallest k such that ak > N 2 . Since
3|ak , we have ak ≥ N 2 + 3. However, ak−1 ≤ N 2 and ak − ak−1 ≤ 3, so
ak−1 = N 2 . According to the definition, ak = N , leading to a contradic-
tion. Therefore, there is no ak > N 2 , and the sequence {an } is bounded,
satisfying the conditions for such a0 .
Lastly, consider the case a0 ≡ 1(mod 3). It is easy to see that if ak is not
divisible by 3 and ak > 1, then 3 does not divide ak+1 and ak+1 > 1. Thus,
every term in the sequence {an } is not divisible by 3 and greater than 1.
Suppose {an } is bounded, i.e., it eventually becomes periodic. Then each
term in the sequence is congruent to 1 modulo 3. Taking the largest term
ak in the period, we see that ak is a perfect square; otherwise, ak+1 =
a k + 3 > ak .
Let ak = N 2 . Then ak+1 = N , where N ≡ 1(mod3) and N > 1, i.e.,
N ≥ 4. As there exists j > k such that aj = ak = N 2 > (N − 2)2
and ak+1 = N ≤ (N − 2)2 , take the smallest l > k + 1 such that al >
(N − 2)2 , and then al−1 ≤ (N − 2)2 < al , implying al = al−1 + 3. Since
al ≡ al−1 ≡ 1(mod3), it follows that al−1 = (N − 2)2 ≡ 1(mod3), and by
definition, al = N − 2 ≡ 2(mod3), resulting in a contradiction. Therefore,
a0 ≡ 1(mod3) does not satisfy the conditions.
In conclusion, the values of a0 satisfying the conditions are all multiples
of 3.
Note. There are several similar problems:
• (Indian National Mathematical Olympiad 2016, Problem 3).
Let N denote the set of all natural numbers. Define a function T : N →
N by T (2k) = k and T (2k + 1) = 2k + 2. We write T 2 (n) = T (T (n))
and in general T k (n) = T k−1 (T (n)) for any k > 1.
(a) Show that for each n ∈ N, there exists k such that T k (n) = 1.
(b) For k ∈ N, let ck denote the number of elements in the set
{n|T k (n) = 1}. Prove that ck+2 = ck+1 + ck , for k ≥ 1.
• (IMO Shortlist 2015, Number Theory, Problem 4). Suppose that
a0 , a1 , . . . and b0 , b1 , . . . are two sequences of positive integers satisfying
a0 , b0 ≥ 2 and
an+1 = gcd (an , bn ) + 1, bn+1 = lcm(an , bn ) − 1
for all n ≥ 0. Prove that the sequence {an } is eventually periodic; in
other words, there exist integers N ≥ 0 and t > 0 such that an+t = an
for all n ≥ N .
176 IMO Problems, Theorems, and Methods: Algebra

Here, gcd(x, y) means the greatest common divisor of x and y and

lcm(x, y) means the least common multiple of x and y.
• (Balkan Mathematical Olympiad 2006, Problem 4). Let m be a
positive integer. Find all positive integers a such that the sequence {an }
defined by a0 = a and
an
2 , if an is even,
an+1 = for n = 0, 1, 2, . . .
an + m, if an is odd,

is perodic (there exists an integer d > 0 such that an+d = an for all n).
• (All-Russian Mathematical Olympiad 2000, 4th Round, Grade
10, Problem 6). Given a positive integer a0 , construct a sequence {an }
as follows:
a2n − 5, if an is odd,
an+1 = an
2 , if an is even.
Prove that for any odd integer a0 > 5 and real number M > 0, there is
a positive integer k such that ak > M .
• (Baltic Way Mathematical Contests 1997, Problem 5). In a
sequence u0 , u1 , . . . of positive integers, u0 is arbitrary, and for any non-
negative integer n,
un
2 , for even un ,
un+1 =
a + un , for odd un ,

where a is a fixed odd positive integer. Prove that the sequence is periodic
from a certain step.

Score Situation This particular problem saw the following distribution of scores among
contestants: 446 contestants scored 7 points, 25 contestants scored 6 points, 54 contestants
scored 5 points, 12 contestants scored 4 points, 5 contestants scored 3 points, 17 contestants
scored 2 points, 16 contestants scored 1 point, and 40 contestants scored 0 point. The
average score for this problem is 5.943, indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the South Korea team scored 42 points (with a total team score of 170 points), the China
team scored 42 points (with a total team score of 159 points), the Vietnam team scored 42
points (with a total team score of 155 points), the United States team scored 42 points (with
a total team score of 148 points), and the Iran team scored 42 points (with a total team
score of 142 points).
 Sequence Problems 177

The gold medal cutoff for this IMO was set at 25 points (with 48 contestants earning gold
medals), the silver medal cutoff was 19 points (with 90 contestants earning silver medals),
and the bronze medal cutoff was 16 points (with 153 contestants earning bronze medals).
In this IMO, no contestant achieved a perfect score of 42 points.

Problem 3.4 (IMO 59-2, proposed by Slovakia). Find all integers

n ≥ 3 for which there exist real numbers a1 , a2 , . . . , an+2 , such that
an+1 = a1 and an+2 = a2 , and
ai ai+1 + 1 = ai+2
for i = 1, 2, . . . , n.

Solution 1. The values of n satisfying the conditions are all multiples

of 3.
On the one hand, if n is a multiple of 3, then let n = 3k. Choose
a3i−2 = a3i−1 = −1 and a3i = 2 for i = 1, 2, . . . , k, along with an+1 =
an+2 = −1. It can be easily verified that for i = 1, 2, . . . , n, the relation
ai ai+1 + 1 = ai+2 satisfies the requirements.
On the other hand, assume there exist a1 , a2 , . . . , an+2 satisfying the
conditions. Extend them into an infinite sequence with period n at both
ends. Then, ai ai+1 + 1 = ai+2 for any integer i. We successively prove the
following conclusions:

(i) There is no i such that ai > 0 and ai+1 > 0.

If there exists such an i with ai > 0 and ai+1 > 0, then ai+2 =
ai ai+1 + 1 > 1. By induction, am > 1 for all m ≥ i + 2, implying
am+2 = am am+1 + 1 > am am+1 > am+1 .
Thus, am+1 < am+2 < am+3 < · · · < am+n+1 . However, this contra-
dicts am+n+1 = am+1 .
(ii) There is no i such that ai = 0.
If there exists such an i with ai = 0, then ai+1 = ai−1 ai + 1 = 1 and
ai+2 = ai ai+1 + 1 = 1, which contradicts (i).
(iii) There is no i such that ai < 0, ai+1 < 0, and ai+2 < 0.
If there exists such an i with ai < 0, ai+1 < 0, and ai+2 < 0, then
ai+2 = ai ai+1 + 1 > 0, a contradiction.
(iv) There is no i such that ai > 0, ai+1 < 0, and ai+2 > 0.
If there exists such an i with ai > 0, ai+1 < 0, and ai+2 > 0, then
from (i), ai−1 < 0 and ai+3 < 0. Since 0 < ai+2 = 1 + ai ai+1 < 1 and
|ai+1 ai+2 | = |ai+3 − 1| > 1, we have |ai+1 | > 1, implying ai+1 < −1.
178 IMO Problems, Theorems, and Methods: Algebra

Moreover, |ai ai+1 | = |ai+2 − 1| < 1 implies |ai | < 1, yielding

0 < ai < 1. Then, from ai−2 ai−1 = ai − 1 < 0, we get ai−2 > 0.
Since ai+1 < 0 and 0 < ai < 1,
ai+1 − 1
ai−1 = < ai+1 − 1 < ai+1 .
ai
Thus, from ai > 0, ai+1 < 0, and ai+2 > 0, we deduce ai−1 < 0,
ai−2 > 0, and ai−1 < ai+1 . By induction, it is easy to see that for
k ≥ 0, ai−2k > 0, ai−2k+1 < 0, and

ai+1 > ai−1 > ai−3 > ai−5 > · · · .

Particularly, ai+1 > ai+1−2n , which contradicts the fact that the
defined sequence {am } (m ∈ Z) has n as its period.

Consequently, it is known from (i), (ii), (iii), and (iv) that in any con-
secutive three terms ai , ai+1 , ai+2 , exactly one term is positive, and the
other two are negative. Thus, the sign of this sequence follows a period of
3. Since there is at least one positive term, say ai > 0, then ai+n > 0,
leading to 3|n.
Solution 2. The example part follows Solution 1. Here, we present an
alternative proof for 3|n. Extend a1 , a2 , . . . , an into an infinite sequence
with period n at both ends. Then, for any i,

ai ai+1 ai+2 = (ai+2 − 1) ai+2 = a2i+2 − ai+2 ,

ai ai+1 ai+2 = ai (ai+3 − 1) = ai ai+3 − ai ,

leading to a2i+2 − ai+2 = ai ai+3 − ai .

Summing up the above equalities for i = 1, 2, . . . , n (and noting the
periodicity), we have
n
 n

a2i = ai ai+3 ,
i=1 i=1

n
which is equivalent to i=1 (ai − ai+3 )2 = 0.
Therefore, ai = ai+3 for any i. Hence, the sequence {ai } also has a
period of 3. If n is not a multiple of 3, then the sequence {ai } has a period
of 1, making it a constant sequence. However, the equation x2 + 1 = x
has no real solution, and there is no constant sequence satisfying the given
conditions. Thus, 3|n.
 Sequence Problems 179

Note. There are several similar problems:

• (MathPath Summer Program Qualifying Test 2024,
Problem 6). Given some positive real number x, let x denote the
integer part of x and x denote the decimal part. Next, starting with a
positive real number a0 , define a sequence via
ak+1 = ak ak + 1 for all k ≥ 0.
Show that regardless of the initial value a0 , the sequence eventually sta-
bilizes, meaning that an = an+1 = an+2 = · · · for some n onwards.
• (United States of America Mathematical Olympiad 2021, Prob-
lem 5). Let n ≥ 4 be an integer. Find all positive real solutions to the
following system of 2n equations:
1 1
a1 = + , a2 = a1 + a3 ,
a2n a2
1 1
a3 = + , a4 = a3 + a5 ,
a2 a4
1 1
a5 = + , a6 = a5 + a7 ,
a4 a6
········································

1 1
a2n−1 = + , a2n = a2n−1 + a1 .
a2n−2 a2n
• (British Mathematical Olympiad 2020, 2nd Round, Problem 4).
A sequence b1 , b2 , b3 . . . of nonzero real numbers has the property that
b2n+1 − 1
bn+2 =
bn
for all positive integers n.
Suppose that b1 = 1 and b2 = k, where 1 < k < 2. Show that there
is some constant B, depending on k, such that −B ≤ bn ≤ B for all
n. Also show that, for some 1 < k < 2, there is a value of n such that
bn > 2020.

Score Situation This particular problem saw the following distribution of scores among
contestants: 157 contestants scored 7 points, 7 contestants scored 6 points, 16 contestants
scored 5 points, 18 contestants scored 4 points, 66 contestants scored 3 points, 87 contestants
scored 2 points, 85 contestants scored 1 point, and 158 contestants scored 0 point. The
average score for this problem is 2.946, indicating that it had a certain level of difficulty.
180 IMO Problems, Theorems, and Methods: Algebra

Among the top five teams in the team scores, the scores of this problem are as follows:
the United States team scored 41 points (with a total team score of 212 points), the Russia
team scored 42 points (with a total team score of 201 points), the China team scored 37 points
(with a total team score of 199 points), the Ukraine team scored 34 points (with a total team
score of 186 points), and the Thailand team scored 38 points (with a total team score of 183
points).
The gold medal cutoff for this IMO was set at 31 points (with 48 contestants earning gold
medals), the silver medal cutoff was 25 points (with 98 contestants earning silver medals),
and the bronze medal cutoff was 16 points (with 143 contestants earning bronze medals).
In this IMO, only two contestants achieved a perfect score of 42 points, namely Agnijo
Banerjee from the United Kingdom and James Lin from the United States.

3.2.2 Existence problems

Problem 3.5 (IMO 12-3, proposed by Sweden). Real numbers
a0 , a1 , . . . , an , . . . satisfy the condition:

1 = a0 ≤ a1 ≤ a2 ≤ · · · ≤ an ≤ · · · .

Numbers b1 , b2 , . . . , bn , . . . are defined by

n 
 
ak−1 1
bn = 1− √ .
ak ak
k=1

(a) Prove that 0 ≤ bn < 2 for all n.

(b) Given c with 0 ≤ c < 2, prove that there exist numbers a0 , a1 , . . . with
the above properties such that bn > c for n large enough.

ak−1
Proof. (a) It is evident that ak ≤ 1, implying bn ≥ 0. Moreover,

   √  √ 
ak−1 1 ak−1 ak−1 1
1− √ = 1+ √ 1− √ √
ak ak ak ak ak
√  √  
ak−1 ak−1 1 1
= √ 1+ √ √ −√
ak ak ak−1 ak
 
1 1
≤2 √ −√ .
ak−1 ak
 Sequence Problems 181

Thus,
n   n  
ak−1 1 1 1
bn = 1− √ ≤2 √ −√
ak ak ak−1 ak
k=1 k=1
   
1 1 1
=2 √ −√ =2 1− √ < 2.
a0 an an
(b) Choose an appropriate positive number d ∈ (0, 1) such that
ak = d−2k , yielding √1ak = dk . It follows that.
n 
  n

ak−1 1
bn = 1− √ = 1 − d2 dk
ak ak
k=1 k=1
n
1−d
= d 1 − d2 = d (1 + d) (1 − dn ) .
1−d
c
For any positive number c < 2, let d = 2 , and it is evident that
d(1 + d) > c. Since 0 < d < 1, we have lim (1 − dn ) = 1, i.e., for n
n→+∞
sufficiently large,
√
n 2 c c
1−d > √ √ = .
2+ c d(1 + d)
Therefore, there exist infinitely many n such that bn > c.
Note. There are several similar problems:
• (Turkey Mathematical Olympiad 1999, 2nd Round, Problem 4).
Find all sequences a1 , a2 , . . . , a2000 of real numbers with 2000
n=1 an = 1999
such that
1
< an < 1 and an = an−1 (2 − an−1 ) for all n > 1.
2
• (Chinese Northern Mathematical Olympiad 2006, Problem 8).
Given a sequence {an } such that a0 = 12 and an+1 = an + 2006
1
a2n for
n ∈ N, prove that 1 − 1
2008 < a2006 < 1.

Score Situation This particular problem saw the following distribution of scores among
contestants: 5 contestants scored 8 points, no contestant scored 7 points, 1 contestant
scored 6 points, no contestant scored 5 points, 1 contestant scored 4 points, no contestant
scored 3 points, 1 contestant scored 2 points, 1 contestant scored 1 point, and 26 contestants
scored 0 point. The average score of this problem is 1.514, indicating that it was relatively
challenging.
182 IMO Problems, Theorems, and Methods: Algebra

Among the top five teams in the team scores, the Hungary team achieved a total score
of 233 points, the German Democratic Republic team achieved a total score of 221 points,
the Soviet Union team achieved a total score of 221 points, the Yugoslavia team achieved a
total score of 209 points, and the Romania team achieved a total score of 208 points.
The gold medal cutoff for this IMO was set at 37 points (with 7 contestants earning gold
medals), the silver medal cutoff was 30 points (with 11 contestants earning silver medals),
and the bronze medal cutoff was 19 points (with 40 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 40 points, namely Wolfgang
Burmeister from the German Democratic Republic, Imre Ruzsa from Hungary, and Andrei
Hodulev from the Soviet Union.

Problem 3.6 (IMO 15-6, proposed by Sweden). Let a1 , a2 , . . . , an

be n positive numbers and q be a given real number such that 0 < q < 1.
Find n numbers b1 , b2 , . . . , bn satisfying:

(a) ak < bk for k = 1, 2, . . . , n;

b
bk < q for k = 1, 2, . . . , n − 1;
1
(b) q < k+1
(c) b1 + b2 + · · · + bn < 1+q
1−q (a1 + a2 + · · · + an ).

Solution. For k ∈ {1, 2, . . . , n}, let

bk = a1 q k−1 + a2 q k−2 + · · · + ak−1 q + ak + ak+1 q + · · · + an q n−k .

(a) Clearly, bk > ak for k ∈ {1, 2, . . . , n}.

(b) For k ∈ {1, 2, . . . , n − 1},

qbk − bk+1 = (q 2 − 1)(ak+1 + · · · + an q n−k−1 ) < 0,

and

qbk+1 − bk = q 2 − 1 (a1 q k−1 + · · · + ak ) < 0,

bk+1
so q < bk < 1q .
(c) Note that

n

b1 + b2 + · · · + bn = (a1 q k−1 + a2 q k−2 + · · · + ak + ak+1 q
k=1

+ · · · + an q n−k )
 Sequence Problems 183

< (a1 + a2 + · · · + an )(1 + 2q + 2q 2 + · · · + 2q n−1 )

 
1 − q n−1
= (a1 + a2 + · · · + an ) 1 + 2q ·
1−q
 
2q
< (a1 + a2 + · · · + an ) 1 +
1−q
1+q
= (a1 + a2 + · · · + an ).
1−q

Thus, b1 , b2 , . . . , bn satisfy the given conditions.

Note. This problem has certain connections with advanced mathematics.

For n-dimensional vectors A = (a1 , a2 , . . . , an ) and A = (a1 , a2 , . . . , an ),
solutions satisfying the given conditions are B = (b1 , b2 , . . . , bn ) and
B  = (b1 , b2 , . . . , bn ) respectively. It is evident that for vectors A + A
and CA (where C is a positive real number), the solutions satisfying the
given conditions are B + B  and CB respectively. Therefore, if for unit
vectors

u 1 = (1, 0, 0, . . . , 0) ,
u 2 = (0, 1, 0, . . . , 0) ,
·····················
u n = (0, 0, 0, . . . , 1) ,

the corresponding solutions v 1 , v 2 , . . . , v n can be found, then for the vector

(a1 , a2 , . . . , an ), numbers b1 , b2 , . . . , bn satisfying the given conditions would
be components of the vector a1 v 1 + a2 v 2 + · · · + an v n .

Score Situation This particular problem saw the following distribution of scores among con-
testants: 10 contestants scored 8 points, 2 contestants scored 7 points, 2 contestants scored
6 points, 3 contestants scored 5 points, 3 contestants scored 4 points, 1 contestant scored
3 points, 2 contestants scored 2 points, 1 contestant scored 1 point, and 101 contestants
scored 0 point. The average score of this problem is 1.128, indicating that it was relatively
challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 30 points (with a total team score of 254 points), the Hungary
team scored 11 points (with a total team score of 215 points), the German Democratic
Republic team scored 14 points (with a total team score of 188 points), the Poland team
scored 8 points (with a total team score of 174 points), and the United Kingdom team scored
25 points (with a total team score of 164 points).
184 IMO Problems, Theorems, and Methods: Algebra

The gold medal cutoff for this IMO was set at 35 points (with 5 contestants earning gold
medals), the silver medal cutoff was 27 points (with 15 contestants earning silver medals),
and the bronze medal cutoff was 17 points (with 48 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Sergei
Konyagin from the Soviet Union.

Problem 3.7 (IMO 23-3, proposed by the Soviet Union).

Consider an infinite sequence {xn } of positive real numbers with the fol-
lowing properties:

x0 = 1 and xi+1 ≤ xi for all i ≥ 0.

(a) Prove that for every such sequence, there is an n ≥ 1 such that

x20 x2 x2
+ 1 + · · · + n−1 ≥ 3.999.
x1 x2 xn

(b) Find such a sequence for which

x20 x2 x2
+ 1 + · · · + n−1 < 4.
x1 x2 xn

Solution 1. (a) We consider two cases.

(i) If the sequence {xn } does not converge to 0, then there must
exist a positive number δ > 0 such that xi ≥ δ for all i. Therefore,
x2i δ2
xi+1 ≥ xi+1 ≥ δ .
2

Choosing n such that nδ 2 ≥ 3.999 is sufficient.

(ii) If {xn } converges to 0, then

(xi − 2xi+1 )2 = x2i − 4xi xi+1 + 4x2i+1 ≥ 0,

x2i
implying xi+1 ≥ 4(xi − xi+1 ). Thus,

x20 x2 x2
+ 1 + · · · + n−1 ≥ 4(x0 − x1 ) + · · · + 4(xn−1 − xn ) = 4 − 4xn .
x1 x2 xn

Since limn→+∞ xn = 0, there exists a positive integer N0 such that

4xn ≤ 0.001 for n ≥ N0 . Hence 4 − 4xn ≥ 3.999.
 Sequence Problems 185

1
(b) Choose xi = 2i , as
x20 x2 x2 1 1 1
+ 1 + · · · + n−1 = 2 + 1 + + · · · + n−2 = 4 − n−2 < 4.
x1 x2 xn 2 2 2
Solution 2. We construct a positive sequence {Cn } such that for each
sequence satisfying
x0 ≥ x1 ≥ · · · ≥ xn ≥ · · · > 0,
the following inequality holds:
x20 x2 x2
+ 1 + · · · + n−1 ≥ Cn x0 .
x1 x2 xn
Hence,
  
x20 x21 x2 x2 x2
+ + · · · + n−1 + n ≥ 0 + Cn x1 ≥ 2 Cn x0 ,
x1 x2 xn xn+1 x1
√
implying Cn+1 = 2 Cn .
x2
Given that x01 ≥ x0 , we can choose C1 = 1. Consequently, the general
1
term of {Cn } is given by Cn = 4 × 2− 2n−2 for n = 1, 2, . . ..
To show that Cn ≥ 3.999, we need to find an n ≥ 1 such that
2n−2
4
3.999 ≥ 2. Since
 2n−2  2n−2
4 1 2n−2
> 1+ >1+ ,
3.999 4000 4000
212 4096
when n = 14, we have 1 + 4000 =1+ 4000 > 2.
x20 x21 x2n−1
Therefore, + x1 + · · · + xn ≥ Cn x0 = Cn > 2 for n ≥ 14. The
x2
construction for (b) is the same as in Solution 1.
x2 x2 x2
Solution 3. Let Sn = x01 + x12 + · · · + xn−1
n
. Since the sequence {xn } is a
non-increasing sequence of positive real numbers,

x2 x2 x1
S2 = 0 + 1 ≥ 2x0 ≥ 2x0 = 2,
x1 x2 x2

x20 x2 x2 x2
S3 = + 1 + 2 ≥ 0
x1 x2 x3 x1

x2 x2 √ 1
+ 2x1 ≥ 0 + 2x1 ≥ 2 2x0 = 21+ 2 ,
x3 x1
186 IMO Problems, Theorems, and Methods: Algebra

x20 x2 x2 x2 x2 x2
S4 = + 1 + 2 + 3 ≥ 0 + 1 + 2x2
x1 x2 x3 x4 x1 x2

x2 √ √ 1 1
≥ 0 + 2 2x1 ≥ 2 2 2 = 21+ 2 + 4 ,
x1
1 1
and so on, leading to the general inequality Sn ≥ 21+ 2 +···+ 2n−2 .
As limn→∞ 1 + 12 + · · · + 2n−2
1
= 2, there exists a positive integer N
such that Sn > 3.999 for n > N .
The construction for (b) is the same as in Solution 1.
Note. (i) For any positive sequence {xn } with x0 = 1 and xi+1 ≤ xi for
∞ x2i
i ≥ 0, i=0 xi+1 ≥ 4 is valid.
(ii) For a positive sequence {xn } with x0 = 1 and xi+1 ≤ xi for i ≥ 0,
∞  t
xr+t
i r+t r r+t
≥ ·
xt
i=0 i+1
t r
  ri
t
is true, where r, t ∈ R+ . The equality is achieved when xi = r+t for
i ≥ 0.
Score Situation This particular problem saw the following distribution of scores among
contestants: 27 contestants scored 7 points, 5 contestants scored 6 points, 2 contes-
tants scored 5 points, 9 contestants scored 4 points, 8 contestants scored 3 points,
36 contestants scored 2 points, 1 contestant scored 1 point, and 31 contestants scored
0 point. The average score for this problem is 3.042, indicating that it was relatively
straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Germany team scored 27 points (with a total team score of 145 points), the Soviet Union
team scored 23 points (with a total team score of 137 points), the German Democratic
Republic team scored 28 points (with a total team score of 136 points), the United States
team scored 27 points (with a total team score of 136 points), and the Vietnam team scored
20 points (with a total team score of 133 points).
The gold medal cutoff for this IMO was set at 37 points (with 10 contestants earning gold
medals), the silver medal cutoff was 30 points (with 20 contestants earning silver medals),
and the bronze medal cutoff was 21 points (with 31 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 42 points, namely Bruno
Haible from Germany, Grigori Perelman from the Soviet Union, and Le Tu QuocThang from
Vietnam.
 Sequence Problems 187

Problem 3.8 (IMO 32-6, proposed by the Netherlands). An infi-

nite sequence x0 , x1 , x2 , . . . of real numbers is said to be bounded if there
is a constant C such that |xi | ≤ C for every i ≥ 0.
Given any real number a > 1, construct a bounded infinite sequence
x0 , x1 , x2 , . . . such that

|xi − xj | ≥ |i − j|−a

for every pair of distinct non-negative integers i and j.

Solution 1. From the problem requirement, the constructed sequence {xi }

should be related to the index i to ensure that the absolute value of the
difference between two numbers |xi − xj |(i = j) is not too small relative to
the difference in their indices |i − j|.
Here, using binary representation, let i be a non-negative integer repre-
sented as

i = b0 + b1 · 2 + b2 · 2 2 + · · · + br · 2 r ,

where b0 , b1 , . . . , br are either 0 or 1. Define

yi = b0 + b1 · 2−a + b2 · 2−2a + · · · + br · 2−ra for i = 0, 1, 2, . . . .

It is evident that |yi | < 1−21−a , making {yi } a bounded sequence. Next,
estimate |yi − yj | for i = j.
Assume j = c0 + c1 · 2 + c2 · 22 + · · · + cs · 2s . Since i = j, there exists
a non-negative integer t0 such that bt0 = ct0 and bt = ct for 0 ≤ t < t0 .
Consequently, |i − j| ≥ 2t0 , and

|yi − yj | > 2−t0 a − (2−(t0 +1)a + 2−(t0 +2)a + · · · )

2a − 2
= 2−t0 a ·
2a − 1
a
−a 2 − 2
≥ |i − j| · a .
2 −1

From the above expression, it is evident that choosing the sequence

a
xi = 22a −1
−2 yi for i ≥ 0 satisfies the specific conditions.
188 IMO Problems, Theorems, and Methods: Algebra

Solution 2. We first prove that if the following lemma holds, then the
problem is true for a = 1.

Lemma. For any given real quadratic irrational number α (meaning α

is not a rational number and is a root of some quadratic polynomial with
integer coefficients), there exists a positive constant c (possibly dependent
on α) such that for any integers p = 0 and q,
c
|pα − q| ≥ , (1)
|p|
 
 
or equivalently α − pq  ≥ pc2 .
This implies that such irrational numbers α cannot be approximated too
closely by rational numbers pq . If the lemma holds, then take the sequence

xi = c−1 (αi − αi ) for i = 0, 1, 2, . . . ,

where t denotes the greatest integer not exceeding the real number t.
Clearly, 0 ≤ xi < c−1 , making {xi } a bounded sequence. From (1), for
i = j,

|xi − xj | = c−1 |α(i − j) − ( αi − αj )| ≥ |i − j|−1 ,

which implies that the bounded infinite sequence {xi } satisfies the require-
ments of the given problem when a = 1.
Clearly, |xi − xj | ≥ |i − j|−1 > |i − j|−a for a > 1.
√ √ −1
Next, we prove that when α = 2 and c = 2 + 2 , inequality (1)
holds for any integers p = 0 and q.
(i) Assume 1 ≤ pq ≤ 2. From 2p2 − q 2 = 0,
√  √ 
   
1 ≤ |2p2 − q 2 | =  2p + q  ·  2p − q 
 
q  √
√ 

= |p| ·  2 +·  2p − q 
p
 √  √ 
 
≤ 2 + 2 |p| ·  2p − q  .

q
(ii) Assume p < 1. It is evident that
   √
√ q  √ √ −1 (2 + 2)−1
 2− > 2−1> 2+ 2 ≥ .
 p p2
 Sequence Problems 189

q
(iii) Assume p > 2. We have

  √
q   √  
−1
√ √ −1 2+ 2
 2− > 2 − 2 > 2 + 2 ≥ .
 p p2

Thus, (1) holds. Therefore, we can take

√ √ √ 
xi = (2 + 2)( 2i − 2i ) for i = 0, 1, 2, . . . .

Solution 3. For a real number a > 1, it is evident that there exists a finite
constant Ca such that
1 1 1
+ a + · · · + a < Ca for n = 1, 2, 3, . . . . (2)
1a 2 n
The definition of the sequence x0 , x1 , x2 , . . . will be inductive. Let x0 =
0, fix an integer k ≥ 1, and assume we have already defined real numbers
x0 , x1 , . . . , xk−1 satisfying the conditions
a
|xi | ≤ Ca , |xi − xj | · |i − j| ≥ 1. (3)

for i, j ∈ {0, 1, . . . , k − 1} with i = j. Next, we need to find an xk satisfying

the conditions (3); this leads to
a
|xk | ≤ Ca , |xk − xi | · |k − i| ≥ 1.

Thus, xk ∈ / (xi − δi , xi + δi ), where δi = (k−i) 1

a for i = 0, 1, . . . , k − 1.

From (2), consider the open intervals Ii = (xi − δi , xi + δi ). Their joint

length does not exceed 2δ0 + 2δ1 + · · · + 2δk−1 , which is less than 2Ca .
Indeed, we can find a number xk ∈ [−Ca , Ca ] that does not belong to any
of the intervals I0 , I1 , . . . , Ik−1 . This means that conditions (3) are satisfied
for i, j ∈ {0, 1, . . . , k} with i = j, completing the induction step.
Thus, we can obtain an infinite sequence x0 , x1 , x2 , . . ..

Note. Solution 1’s construction necessarily requires a > 1, while Solution

2 refines the problem by setting a = 1, and Solution 2 was proposed by
Wei Luo from China.
In addition, the series

∞
1 1 1 1
p
= 1 + p + p + ···+ p + ···
n=1
n 2 3 n
190 IMO Problems, Theorems, and Methods: Algebra

in Solution 3 is called a p-series, and we can use

 
1 1 1 1
< − p−1 , where k ≥ 2 is an integer,
kp p − 1 (k − 1)p−1 k
to prove that its sum is finite for p > 1. For p ≤ 1, this sum is
infinity.
Furthermore, for p > 1, the sum of the p-series is also known as the
Riemann zeta function ζ(p), which is a monotone decreasing function of p.
n+1
B2n (2π)2n
For any positive even integer 2n, ζ(2n) = (−1) 2(2n)! , where B2n is
the 2nth Bernoulli number.
There are several similar problems:
• (From the “Problems in Mathematics” in the Hungarian jour-
nal KöMaL, October 2014, A.624). (a) Prove that for every infinite
sequence x1 , x2 , . . . ∈ [0, 1] there exists some C > 0 such that for every
positive integer r, there are positive integers n, m satisfying |n − m| ≥ r
C
and |xn − xm | < |n−m| .
(b) Show that for every C > 0 there exists an infinite sequence
C
x1 , x2 , . . . ∈ [0, 1] and a positive integer r such that |xn − xm | > |n−m|
for every pair n, m of positive integers with |n − m| ≥ r.

• (Turkey Team Selection Test 2014, Problem 1). Find the number
of (a1 , a2 , . . . , a2014 ) permutations of the (1, 2, . . . , 2014) such that i+ai ≤
j + aj for all 1 ≤ i < j ≤ 2014.
• (Turkey Team Selection Test 2005, Problem 4). Show that for any

integer n ≥ 2 and all integers a1 , a2 , . . . , an , the product i<j (aj − ai )

is divisible by i<j (j − i).
• (IMO Shortlist 2002, Algebra, Problem 2). Let a1 , a2 , . . . be an
infinite sequence of real numbers, for which there exists a real number c
with 0 ≤ ai ≤ c for all i, such that
1
|ai − aj | ≥ for all i, j with i = j.
i+j
Prove that c ≥ 1.
• (All-Russian Mathematical Olympiad 1998, 4th Round, Grade
11, Problem 8). Given a sequence {an } (n ∈ N) of natural numbers
such that every natural number occurs and for any distinct natural num-
bers m and n,
1 |an − am |
< < 1998,
1998 |n − m|
prove that |an − n| < 2000000 for all natural numbers n.
 Sequence Problems 191

• (Canadian Mathematical Olympiad 1985, Problem 5). Let 1 <

x1 < 2 and define xn+1 = 1 + xn − 12 x2n for n = 1, 2, . . .. Prove that

 √ 
 1
xn − 2 < n for n ≥ 3.
2

• (All Soviet Union Mathematical Olympiad 1978). Prove that

there exists an infinite sequence {xi } such that for all m and all k (k = m),
the inequality |xm − xk | > |m−k|
1
is valid.

Score Situation This particular problem saw the following distribution of scores among
contestants: 50 contestants scored 7 points, 3 contestants scored 6 points, 5 contes-
tants scored 5 points, 10 contestants scored 4 points, 16 contestants scored 3 points,
27 contestants scored 2 points, 29 contestants scored 1 point, and 172 contestants scored
0 point. The average score for this problem is 1.808, indicating that it was relatively
challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 42 points (with a total team score of 241 points), the China
team scored 32 points (with a total team score of 231 points), the Romania team scored 32
points (with a total team score of 225 points), the Germany team scored 36 points (with a
total team score of 222 points), and the United States team scored 31 points (with a total
team score of 212 points).
The gold medal cutoff for this IMO was set at 39 points (with 20 contestants earning gold
medals), the silver medal cutoff was 31 points (with 51 contestants earning silver medals),
and the bronze medal cutoff was 19 points (with 84 contestants earning bronze medals).
In this IMO, a total of nine contestants achieved a perfect score of 42 points.

Problem 3.9 (IMO 64-3, proposed by Malaysia). For each integer

k ≥ 2, determine all infinite sequences of positive integers a1 , a2 , . . . for
which there exists a polynomial P of the form

P (x) = xk + ck−1 xk−1 + · · · + c1 x + c0 ,

where c0 , c1 , . . . , ck−1 are non-negative integers, such that P (an ) =

an+1 an+2 . . . an+k for every integer n ≥ 1.
192 IMO Problems, Theorems, and Methods: Algebra

Solution. All arithmetic sequences with a non-negative common difference

meet the conditions.
On the one hand, if {an } is an arithmetic sequence with a common
difference d, then letting P (x) = (x + d)(x + 2d) · · · (x + kd) meets the
conditions.
On the other hand, for any positive integer sequence a1 , a2 , . . . , an and
a polynomial P (x) = xk + ck−1 xk−1 + · · · + c1 x + c0 (c0 , c1 , . . . , ck−1 are
non-negative integers) satisfying the equation P (an ) = an+1 an+2 · · · an+k ,
there exists the following claim:

Claim. The sequence {an }(n ∈ Z+ ) is non-decreasing.

Proof of the Claim. By contradiction, assume there exists an m ∈ Z+

such that am > am+1 .
From the conditions,

P (am+1 ) am+2 am+3 · · · am+k+1 am+k+1

= = .
P (am ) am+1 am+2 · · · am+k am+1

Since each term of P (x) has non-negative coefficients, P (am ) >

P (am+1 ). Thus am+1 > am+k+1 . Let m1 ≥ m + 1 be the smallest posi-
tive integer such that am+1 > am1 +1 . Then am1 ≥ am+1 > am1 +1 . Denote
m1 as f (m), and f (m) is defined only when am > am+1 . Let f (t) (m) denote
f (f (· · · f (m) · · · )) with f iterated t times. The above analysis implies that
f (t) (m) is well-defined for any t ∈ Z+ , and

am+1 > af (m)+1 > af (2) (m)+1 > · · · .

Thus, we have found an infinite strictly decreasing sequence of positive

integers {af (t) (m)+1 }(t ∈ Z+ ), where there is a contradiction. Hence the
claim holds.
If P (x) = xk , from the claim, {an } is a constant sequence.
Suppose P (x) = xk . For any n ∈ Z+ and 1 ≤ i ≤ k, we have
an+1 an+2 · · · an+k ≥ an+i · ak−1
n , and

an+1 an+2 · · · an+k = P (an ) = akn + ak−1

n · ck−1 + · · · + c0
⎛ ⎞
k−1

≤ ak−1
n
⎝an + cj ⎠ .
j=0
 Sequence Problems 193

Therefore, an+i ≤ an + c0 + c1 + · · · + ck−1 . Suppose

Sn = (an+1 − an , an+2 − an , . . . , an+k − an ).
Then Sn ∈ {0, 1, . . . , M }k , where M = k−1 j=0 cj .
By the Pigeonhole Principle and the claim, there exist 0 ≤ d1 ≤ d2 ≤
· · · ≤ dk such that there are infinitely many n ∈ Z+ for which an+i = an +di
for all 1 ≤ i ≤ k. For these n, we have P (an ) = (an + d1 )(an + d2 ) · · ·
(an + dk ).
Since P (x) = xk , it follows that among an+1 , an+2 , . . . , an+k , there must
be one greater than an , implying {an } is unbounded. Therefore, there are
infinitely many distinct values of an such that P (an ) = (an + d1 )(an +
d2 ) · · · (an + dk ). Hence,
P (x) = (x + d1 )(x + d2 ) · · · (x + dk ).
If Sm = (d1 , d2 , . . . , dk ) = (d1 , d2 , . . . , dk ), then there are only finitely
many m ∈ Z+ satisfying
k
 k
 k

P (am ) = am+i = (am + di ) = (am + di ).
i=1 i=1 i=1
Hence, there exists N ∈ Z+ such that Sn = {d1 , d2 , . . . , dk } for all
n ≥ N , i.e., for all n ≥ N and 1 ≤ i ≤ k, it holds that an+i = an + di .
Therefore,
i
di = an+i − an = (an+j − an+j−1 ) = id1 .
j=1

Thus, {an }(n ≥ N ) is an arithmetic sequence with common difference

d1 , implying there exists an integer a such that an = a + nd1 for all n ≥ N .
For n = N − 1,
P (aN −1 ) = aN aN +1 · · · aN +k−1 = P (aN − d1 ).
Since P (x) is monotonically increasing on (0, +∞), we see that aN −1 =
aN − d1 = a + (N − 1)d1 .
By mathematical induction, an = a + nd1 for all n ∈ Z+ .
Note. There are several similar problems:

• (Turkey Mathematical Olympiad 2020, 2nd Round,

Problem 5). Find all polynomials P (x) with real coefficients such that
one can find an integer valued sequence a0 , a1 , a2 , . . . satisfying P (x) =
ax2  for all real numbers x.
Here t denotes the greatest integer not exceeding the real number t.
194 IMO Problems, Theorems, and Methods: Algebra

• (Czech And Slovak Mathematical Olympiad, 3rd Round, Prob-

lem 6). Let n ≥ 1 be an integer and a1 , a2 , . . . , an be positive integers.
Let f : Z → R be a function such that f (x) = 1 for each integer x < 0
and

f (x) = 1 − f (x − a1 )f (x − a2 ) · · · f (x − an )

for each integer x ≥ 0. Show that there exist positive integers s and t
such that f (x + t) = f (x) for any integer x > s.

Score Situation This particular problem saw the following distribution of scores among
contestants: 73 contestants scored 7 points, 3 contestants scored 6 points, 6 contestants
scored 5 points, 8 contestants scored 4 points, 23 contestants scored 3 points, 7 contestants
scored 2 points, 102 contestants scored 1 point, and 396 contestants scored 0 point. The
average score of this problem is 1.256, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 240 points), the United States
team scored 42 points (with a total team score of 222 points), the South Korea team scored
42 points (with a total team score of 215 points), the Romania team scored 42 points (with
a total team score of 208 points), and the Canada team scored 25 points (with a total team
score of 183 points).
The gold medal cutoff for this IMO was set at 32 points (with 54 contestants earning gold
medals), the silver medal cutoff was 25 points (with 90 contestants earning silver medals),
and the bronze medal cutoff was 18 points (with 170 contestants earning bronze medals).
In this IMO, a total of five contestants achieved a perfect score of 42 points.

3.2.3 Proving quantitative relationships

Problem 3.10 (IMO 18-6, proposed by the United Kingdom). A
sequence {un } is defined by

5
u0 = 2, u1 = , un+1 = un (u2n−1 − 2) − u1 for n = 1, 2, . . . .
2

Prove that for positive integers n,

2n −(−1)n
un = 2 3 ,

where x denotes the greatest integer not exceeding x.

 Sequence Problems 195

Proof. We prove the following relation using mathematical induction:

2n −(−1)n 2n −(−1)n
un = 2 3 + 2− 3 . (1)
n n
Let an = 2 −(−1)
3 . For n = 0, we have u0 = 20 + 20 . For n = 1, there is
1
u1 = 2 + 2 . The formula (1) holds for n = 0, 1. Assume the formula (1)
holds for n ≤ k. Then for n = k + 1,
5
uk+1 = (2ak + 2−ak )((2ak−1 + 2−ak−1 )2 − 2) −
2
5
= 2ak +2ak−1 + 2−ak +2ak−1 + 2ak −2ak−1 + 2−ak −2ak−1 − ,
2
and note that
1 k
ak + 2ak−1 = (2 − (−1)k + 2k − 2(−1)k−1 )
3
1 k+1
= 2 − (−1)k+1
3
= ak+1 ,
1 k
2ak−1 − ak = (2 − 2(−1)k−1 − 2k + (−1)k )
3
= (−1)k ,
ak − 2ak−1 = − (2ak−1 − ak )
= (−1)k+1 .
k k+1
Since 2(−1) + 2(−1) − 52 = 0, there is uk+1 = 2ak+1 + 2−ak+1 . As 2n ≡
(−1)n (mod3), we see that 2an is always an integer. Clearly, 2−an < 1.
2n −(−1)n
Therefore, un = 2an = 2 3 for n = 1, 2, . . ..
Note. If the expression for un is not given in the problem statement and
is asked to find the expression for un instead, one can identify a pattern
by computing the first few terms of un and assume un = 2an + 2−an .
Substituting this into un+1 = un (u2n−1 −2)−u1, one can expand and obtain
an+1 = an + 2an−1 , with the associated characteristic equation x2 − x − 2 =
n n
0. This leads to an = 2 −(−1)
3 .
Score Situation This particular problem saw the following distribution of scores among
contestants: 67 contestants scored 7 points, 18 contestants scored 6 points, 5 contestants
scored 5 points, 3 contestants scored 4 points, 4 contestants scored 3 points, 1 contestant
scored 2 points, 10 contestants scored 1 point, and 31 contestants scored 0 point. The
average score for this problem is 4.590, indicating that it was simple.
196 IMO Problems, Theorems, and Methods: Algebra

Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 45 points (with a total team score of 250 points), the United
Kingdom team scored 50 points (with a total team score of 214 points), the United States
team scored 43 points (with a total team score of 188 points), the Bulgaria team scored 55
points (with a total team score of 174 points), and the Austria team scored 55 points (with
a total team score of 167 points).
The gold medal cutoff for this IMO was set at 34 points (with 9 contestants earning gold
medals), the silver medal cutoff was 23 points (with 28 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 45 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Laurent
Pierre from France.

Problem 3.11 (IMO 26-6, proposed by Sweden). For every real

number x1 , construct a sequence x1 , x2 , . . . by setting
 
1
xn+1 = xn xn + for each n ≥ 1.
n

Prove that there exists exactly one value of x1 for which

0 < xn < xn+1 < 1

for every n.

Proof 1. To prove 0 < xn < xn+1 < 1, it suffices to demonstrate 1 − 1

n <
xn < 1. Rewrite the recurrence relation as

1 1
xn = xn+1 + 2
− . (1)
4n 2n

If 1 − 1
n+1 < xn+1 < 1, then evidently

 
1 1 1 1 1
xn = xn+1 + 2 − < 1+ + 2 − = 1,
4n 2n n 4n 2n
 
1 1 1 1 1 1 1
xn > 1 − + 2− > 1− + 2 − =1− .
n + 1 4n 2n n 4n 2n n
 Sequence Problems 197

Therefore, for any positive integer n, if xn ∈ 1 − n1 , 1 , then it necessarily

follows that
 
1
xn−1 ∈ 1 − ,1 ,
n−1
 
1
xn−2 ∈ 1 − ,1 ,
n−2
·····················
x1 ∈ (0, 1).
It is known that when xk is given, xn (n > k) is determined. Consider xn
as a non-negative polynomial function Xn,k (x) with the leading coefficient
equal to 1 in terms of xk . Clearly, Xn,k (x) is a monotonically increasing
function, and xn ∈ 1 − n1 , 1 , Xn,k (0) = 0, and Xn,k (1) ≥ 1. It follows
that
1
1 − < an < xk < bn < 1 for n = k + 1, k + 2, . . . , (2)
k
where an and bn are values of xk when xn = 1 − n1 and xn = 1 respectively.
From Xn,k (bn ) = 1,
 
1
Xn+1,k (bn ) = Xn,k (bn ) Xn,k (bn ) + > 1.
n
As Xn+1,k (x) is an increasing function, Xn+1,k (x) = 1 has a solution
bn+1 < bn , indicating that bk+1 , bk+2 , . . . form a decreasing sequence.
Similarly, ak+1 , ak+2 , . . . form an increasing sequence, but always
an < bn . Hence {an } has a limit a and {bn } has a limit b with a ≤ b.
To ensure (2) holds, i.e., xk exists and is unique, it suffices to prove a = b.
Furthermore,
 
1 1
b − a = lim bn − lim an < · · · < bk − ak < 1 − 1 − = ,
n→+∞ n→+∞ k k
so b − a = 0 from k → +∞.
Thus, we obtain a series of open intervals whose lengths tend to zero.
For xk within all open intervals (an , bn ), it is unique, which implies that x1
is also uniquely determined.
Note. This problem is essentially a variation of the Nested Interval Theo-
rem in advanced mathematics.
It is important to note that if [an , bn ] is replaced with (an , bn ), then the
existence of ξ = lim an = lim bn still holds, but it may not belong to
n→+∞ n→+∞
198 IMO Problems, Theorems, and Methods: Algebra

any open interval. For instance, consider (an , bn ) = 0, n1 . Nevertheless,

if an < an+1 and bn+1 < bn hold for any n, then there still exists a unique
ξ within all these open intervals.

Proof 2. Construct a sequence of polynomials {Pn (x)} such that

 
1
P1 (x) = x, Pn+1 (x) = Pn (x) Pn (x) + for n ≥ 1.
n

It is evident that Pn (x) is a positive coefficient polynomial of degree

2n−1 . From Proof 1, the problem can be reduced to proving the existence
of a unique positive real number t such that

1
1− < Pn (t) < 1, n = 1, 2, . . . .
n

Since Pn (x) is an increasing function for x ≥ 0 and Pn (0) = 0, we can find

a unique pair an , bn (an < bn ) such that Pn (an ) = 1 − n1 and Pn (bn ) = 1.
Similar to Proof 1, it is known that an+1 > an and bn+1 < bn . Also,
Pn (x) is a positive coefficient polynomial of degree 2n−1 , and Pn (x) is either
zero or a positive coefficient polynomial of degree 2n−1 − 2, i.e., Pn (x) ≥ 0
for x ≥ 0. Thus, Pn (x) is a convex function, and the graph of Pn (x) for
0 ≤ x ≤ bn lies below the line y = b1n x. Consequently, Pn (x) ≤ bxn for
0 ≤ x ≤ bn and

1 an
Pn (an ) = 1 − ≤ .
n bn

Since 0 < bn ≤ 1, we have bn − bnn ≤ an , implying that bn −an ≤ bn

n ≤ 1
n
for any n.
The subsequent steps are identical to those in Proof 1.

Score Situation This particular problem saw the following distribution of scores among
contestants: 23 contestants scored 7 points, 4 contestants scored 6 points, 10 contestants
scored 5 points, 9 contestants scored 4 points, 13 contestants scored 3 points, 19 contestants
scored 2 points, 74 contestants scored 1 point, and 57 contestants scored 0 point. The
average score for this problem is 2.019, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Romania team scored 34 points (with a total team score of 201 points), the United
States team scored 40 points (with a total team score of 180 points), the Hungary team
scored 25 points (with a total team score of 168 points), the Bulgaria team scored 23 points
 Sequence Problems 199

(with a total team score of 165 points), and the Vietnam team scored 12 points (with a total
team score of 144 points).
The gold medal cutoff for this IMO was set at 34 points (with 14 contestants earning gold
medals), the silver medal cutoff was 22 points (with 35 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 52 contestants earning bronze medals).
In this IMO, only two contestants achieved a perfect score of 42 points, namely Géza
Kós from Hungary and Daniel Tătaru from Romania.

Problem 3.12 (IMO 37-6, proposed by France). Let p, q, and n be

three positive integers with p + q < n. Let (x0 , x1 , . . . , xn ) be an (n + 1)-
tuple of integers satisfying the following conditions:
(a) x0 = xn = 0.
(b) For each i with 1 ≤ i ≤ n, either xi − xi−1 = p or xi − xi−1 = −q.
Show that there exist indices i < j with (i, j) = (0, n), such that xi = xj .

Proof. First, assume without loss of generality that (p, q) = 1. This is

because if (p, q) = d with p = dp1 and q = dq1 , then (p1 , q1 ) = 1. We only
need to consider xi = xdi for i = 0, 1, . . . , n with xi − xi−1 = p1 or −q1 .
Clearly, n > p1 + q1 and all conditions are satisfied.
Since xi − xi−1 = p or −q for i = 1, 2, . . . , n, if among these n differences
xi − xi−1 , a of them are p and b of them are −q, then xn = ap − bq, where
a + b = n. Thus, ap = bq.
From (p, q) = 1, we can assume a = kq and b = kp. Moreover, since
n = a+b = k(p+q) and n > p+q, it follows that k ≥ 2. Let yi = xi+p+q −xi
for i = 0, 1, . . . , (k − 1)(p + q).
As xi − xi−1 = p or −q ≡ p(mod(p + q)), setting i = 1, 2, . . . and
summing up, we obtain
xi ≡ ip(mod(p + q)).
Thus, xj − xi ≡ (j − i) p(mod (p + q)). Particularly, yi ≡ 0(mod (p + q)),
meaning each yi is a multiple of p + q. Also
yi+1 − yi = (xi+1+p+q − xi+p+q ) − (xi+1 − xi ),
where the numbers in brackets are either p or −q, so yi+1 − yi can only be
p + q, 0, or −(p + q).
From y0 + yp+q + y2(p+q) + · · · + y(k−1)(p+q) = xn = 0, consider the sign
of y0 . If y0 > 0, then there must be a yi(p+q) < 0. If y0 < 0, then there
must be a yi(p+q) > 0.
y0 y1 y
Thus, p+q , p+q , . . . , n−p−q
p+q are all integers, and the difference between
the adjacent terms can only be 1, 0, or −1. Moreover, it cannot be all
200 IMO Problems, Theorems, and Methods: Algebra

positve or all negative. By the discrete version of the Intermediate Value

Theorem, there must be a yt = 0. Taking i = t and j = t + p + q meets the
requirement.
Score Situation This particular problem saw the following distribution of scores among
contestants: 99 contestants scored 7 points, 10 contestants scored 6 points, 6 contes-
tants scored 5 points, 3 contestants scored 4 points, 15 contestants scored 3 points, 16
contestants scored 2 points, 79 contestants scored 1 point, and 196 contestants scored 0
point. The average score for this problem is 2.243, indicating that it had a certain level of
difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Romania team scored 30 points (with a total team score of 187 points), the United States
team scored 36 points (with a total team score of 185 points), the Hungary team scored 36
points (with a total team score of 167 points), the Russia team scored 24 points (with a
total team score of 162 points), and the United Kingdom team scored 30 points (with a total
team score of 161 points).
The gold medal cutoff for this IMO was set at 28 points (with 35 contestants earning gold
medals), the silver medal cutoff was 20 points (with 66 contestants earning silver medals),
and the bronze medal cutoff was 12 points (with 99 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 42 points, namely Ciprian
Manolescu from Romania.

Problem 3.13 (IMO 50-3, proposed by the United States).

Suppose that s1 , s2 , s3 , . . . is a strictly increasing sequence of positive inte-
gers such that the subsequences

ss1 , ss2 , ss3 , . . . and ss1 +1 , ss2 +1 , ss3 +1 , . . .

are both arithmetic progressions. Prove that the sequence s1 , s2 , s3 , . . . is

itself an arithmetic progression.

Proof. From the conditions, it is evident that both ss1 , ss2 , ss3 , . . . and
ss1 +1 , ss2 +1 , ss3 +1 , . . . are strictly increasing sequences of positive integers.
Let ssk = a + (k − 1)d1 and ssk +1 = b + (k − 1)d2 for k = 1, 2, . . .,
where a, b, d1 , d2 are positive integers. From sk < sk + 1 ≤ sk+1 and the
monotonicity of {sn }, for any positive integer k,

ssk < ssk +1 ≤ ssk+1 ,

i.e.,

a + (k − 1)d1 < b + (k − 1)d2 ≤ a + kd1 ,

 Sequence Problems 201

which can be further simplified to

a − b < (k − 1)(d2 − d1 ) ≤ a + d1 − b.

From the arbitrariness of k, it follows that d2 −d1 = 0. Let d = d2 = d1 ,

and b − a = c ∈ N+ .
If d = 1, then ssk+1 = ssk + 1 ≤ ssk +1 by the monotonicity of {sn }.
Hence sk+1 ≤ sk + 1. Since sk+1 > sk , it follows that sk+1 = sk + 1, i.e.,
{sn } is an arithmetic sequence, and the conclusion holds.
Now, assume d > 1, and we prove that sk+1 − sk = c for any positive
integer k. If not, then there are two cases:

Case 1: There exists a positive integer k such that sk+1 − sk < c.

Since sk+1 − sk can only take integer values, there must be a minimum
among them; assume si+1 − si = c0 is the minimum. Then,

sa+id − sa+(i−1)d+1 = sssi+1 − sssi +1

= (a + (si+1 − 1)d) − (b + (si − 1)d)
= c0 d − c. (1)

On the other hand, from (a + id) − (a + (i − 1)d + 1) = d − 1,

sa+id − sa+(i−1)d+1 ≥ c0 (d − 1)

(using the minimality of c0 ). Comparing with (1), we get c0 ≥ c, a contra-

diction.

Case 2: There exists a positive integer k such that sk+1 − sk > c.

Since sk+1 − sk can only take integer values, and for a given k,

sk+1 − sk ≤ ssk+1 − ssk = d,

there must be a maximum among them; assume sj+1 − sj = c1 is the

maximum. Then,

sa+jd − sa+(j−1)d+1 = sssj+1 − sssj +1

= (a + (sj+1 − 1)d) − (b + (sj − 1) d)
= c1 d − c. (2)
202 IMO Problems, Theorems, and Methods: Algebra

On the other hand, from (a + jd) − (a + (j − 1)d + 1) = d − 1,

sa+jd − sa+(j−1)d+1 ≤ c1 (d − 1)
(using the maximality of c1 ). Comparing with (2), we get c1 ≤ c, a contra-
diction.
Hence, sk+1 −sk = c for any positive integer k, i.e., {sn } is an arithmetic
sequence.
Note. This problem can also be approached from the perspective of func-
tions. The problem is equivalent to: Given that f : Z+ → Z+ is a
strictly increasing function, bn = f (f (n)) is an arithmetic sequence, and
cn = f (f (n) + 1) is also an arithmetic sequence. Prove that an = f (n) is
an arithmetic sequence.
Score Situation This particular problem saw the following distribution of scores among
contestants: 51 contestants scored 7 points, 2 contestants scored 6 points, 5 contestants
scored 5 points, 2 contestants scored 4 points, 5 contestants scored 3 points, 16 contestants
scored 2 points, 127 contestants scored 1 point, and 357 contestants scored 0 point. The
average score for this problem is 1.019, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 221 points), the Japan team
scored 34 points (with a total team score of 212 points), the Russia team scored 31 points
(with a total team score of 203 points), the South Korea team scored 22 points (with a total
team score of 188 points), and the North Korea team scored 24 points (with a total team
score of 183 points).
The gold medal cutoff for this IMO was set at 32 points (with 49 contestants earning gold
medals), the silver medal cutoff was 24 points (with 98 contestants earning silver medals),
and the bronze medal cutoff was 14 points (with 135 contestants earning bronze medals).
In this IMO, only two contestants achieved a perfect score of 42 points, namely Dongyi
Wei from China and Makoto Soejima from Japan.

Problem 3.14 (IMO 51-6, proposed by Iran). Let a1 , a2 , a3 , . . . be a

sequence of positive real numbers. Suppose that for some positive integer
s,
an = max{ak + an−k |1 ≤ k ≤ n − 1} (1)
for all n > s. Prove that there exist positive integers l and N with l ≤ s,
such that an = al + an−l for all n ≥ N .

Proof. By the given conditions, an can be expressed as an = aj1 + aj2

for each n > s, where j1 , j2 < n and j1 + j2 = n. If j1 > s, then aj1 can
 Sequence Problems 203

be further expressed as the sum of two terms in the sequence, and so on.
Eventually, an can be represented as

an = ai1 + · · · + aik , (2)

1 ≤ ij ≤ s, i1 + · · · + ik = n. (3)

Let ai1 , ai2 be the two terms at the last step in expressing an in the form
(2). Then i1 + i2 > s, and (3) becomes

1 ≤ ij ≤ s, i1 + i2 + · · · + ik = n, i1 + i2 > s. (4)

On the other hand, when indices i1 , i2 , . . . , ik satisfy (4), let sj = i1 +

i2 + · · · + ij . From (1),

an = ask ≥ ask−1 + aik ≥ ask−2 + aik−1 + aik ≥ · · · ≥ ai1 + ai2 + · · · + aik .

Therefore, for each n > s,

an = max { ai1 + ai2 + · · · + aik | (i1 , . . . , ik ) satisfies (4)} .

 
Define m = max aii | 1 ≤ i ≤ s , and let l ≤ s be a fixed positive
integer such that m = all . Construct the sequence {bn }:

bn = an − mn for n = 1, 2, . . . ,

so that bl = 0. When n ≤ s, by the definition of m, we have bn ≤ 0. When

n > s,

bn = an − mn
= max {ak + an−k | 1 ≤ k ≤ n − 1 } − mn
= max {bk + bn−k + mn | 1 ≤ k ≤ n − 1 } − mn
= max {bk + bn−k | 1 ≤ k ≤ n − 1 } ≤ 0,

which implies bn ≤ 0 for n = 1, 2, . . ., and for n > s,

bn = max{bk + bn−k |1 ≤ k ≤ n − 1}.

If bk = 0 for k = 1, 2, . . . , s, then bn = 0 for any positive integer n.

Hence an = mn for n = 1, 2, . . ., and so the proposition holds.
204 IMO Problems, Theorems, and Methods: Algebra

If not, let M = max |bi | and ε = min{|bi | : 1 ≤ i ≤ s, bi < 0}. Thus,

1≤i≤s
for n > s,

bn = max{bk + bn−k |1 ≤ k ≤ n − 1} ≥ bl + bn−l = bn−l ,

leading to 0 ≥ bn ≥ bn−l ≥ · · · ≥ −M .
For the sequence {bn }, by (2) and (3), each bn belongs to the set
"
T = {bi1 + bi2 + · · · + bik : 1 ≤ i1 , i2 , . . . , ik ≤ s} [−M, 0],

and T is a finite set. In fact, for any x ∈ T , let x = bi1 + · · · + bik (1 ≤

i1 , . . . , ik ≤ s). Then there are at most M ε non-zero terms among bij . Oth-
erwise, x < M ε (−ε) = −M , a contradiction. Hence x has a finite number
of representations of such sum.
Therefore, for each t ∈ {1, 2, . . . , l}, the sequence

bs+t , bs+t+l , bs+t+2l , . . .

is increasing and takes a finite number of values, so it becomes constant

from some index onwards. Hence, the sequence {bn } becomes periodic with
period l from some index N0 onwards, i.e.,

bn = bn−l = bl + bn−l for n > N0 + l,

so an = bn + mn = (bl + ml) + (bn−l + m(n − l)) = al + an−l for n > N0 + l.

Score Situation This particular problem saw the following distribution of scores among
contestants: 15 contestants scored 7 points, 4 contestants scored 6 points, 6 contestants
scored 5 points, no contestant scored 4 points, 3 contestants scored 3 points, 4 contestants
scored 2 points, 14 contestants scored 1 point, and 470 contestants scored 0 point. The
average score for this problem is 0.368, indicating that it was extremely difficult.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 25 points (with a total team score of 197 points), the Russia team
scored 20 points (with a total team score of 169 points), the United States team scored
12 points (with a total team score of 168 points), the South Korea team scored 6 points
(with a total team score of 156 points), the Kazakhstan team scored 12 points (with a total
team score of 148 points), and the Thailand team scored 1 point (with a total team score of
148 points).
The gold medal cutoff for this IMO was set at 27 points (with 47 contestants earning gold
medals), the silver medal cutoff was 21 points (with 103 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 115 contestants earning bronze medals).
 Sequence Problems 205

In this IMO, only one contestant achieved a perfect score of 42 points, namely Zipei Nie
from China.

3.3 Summary
Through various definitions, we can derive many interesting sequences that
are not only simple in form but also widely applicable. To a certain extent,
sequences also bridge secondary school mathematics with university-level
mathematics. For instance, the concepts of series and limits in caculus are
closely related to sequences.
In the first 64 IMOs, there were a total of 14 sequence problems.
These problems can be broadly categorized into three types, as depicted in
Figure 3.1. The score details for these problems are presented in Table 3.2.
Due to the smaller number of participating teams and missing contestant
score information in early IMOs, there are several blanks in Table 3.2.

2.5
Determining Values Existence Problems
Proving Quantitative Relationships
2

1.5

0.5

0
1–10 11–20 21–30 31–40 41–50 51–60 61–64

Figure 3.1 Numbers of Sequence Problems in the First 64 IMOs

Problems 3.1–3.4 focus on “determining the value of a specific term

or the number of terms;” among these four problems, the one with the
lowest average score is Problem 3.4 (IMO 59-2), proposed by Slovakia.
Problems 3.5–3.9 deal with “addressing problems related to the existence
of sequences;” among these five problems, the one with the lowest average
score is Problem 3.6 (IMO 15-6), proposed by Sweden. Problems 3.10–
3.14 are about “proving quantitative relationships satisfied by sequences;”
among these five problems, the one with the lowest average score is Problem
3.14 (IMO 51-6), proposed by Iran.
 206
Table 3.2 Score Details of Sequence Problems in the First 64 IMOs

Problem 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8

Full points 7.000 7.000 7.000 7.000 8.000 8.000 7.000 7.000

IMO Problems, Theorems, and Methods: Algebra

Average score 3.568 4.592 5.943 2.946 1.514 1.128 3.042 1.808
Top five mean 6.800 7.000 6.400 2.200 6.250 5.767
6th–15th mean 6.433 6.909 6.150 0.663 3.925 3.233
16th–25th mean 5.967 6.778 5.133 1.750 1.833
Problem number in IMO 9-5 36-4 58-1 59-2 12-3 15-6 23-3 32-6
Proposing The Soviet Poland South Slovakia Sweden Sweden The Soviet The Nether-
country Union Africa Union lands

Problem 3.9 3.10 3.11 3.12 3.13 3.14

Full points 7.000 7.000 7.000 7.000 7.000 7.000

Average score 1.256 4.590 2.019 2.243 1.019 0.368
Top five mean 6.433 6.200 4.467 5.200 5.100 2.111
6th–15th mean 3.883 4.413 2.800 3.967 2.470 0.907
16th–25th mean 2.267 1.233 3.697 2.100 0.567
Problem number in IMO 64-3 18-6 26-6 37-6 50-3 51-6
Proposing country Malaysia The United Kingdom Sweden France The United States Iran

Note. Top five mean = Total score of the top five teams ÷ Total number of contestants from the top five teams,
6th–15th mean = Total score of the 6th–15th teams ÷ Total number of contestants from the 6th–15th teams,
16th–25th mean = Total score of the 16th–25th teams ÷ Total number of contestants from the 16th–25th teams.
 Sequence Problems 207

These 14 problems were proposed by 11 countries. Sweden proposed

three problems, while the Soviet Union contributed two problems.
From Table 3.2, it can be observed that in the first 64 IMOs, there was
one sequence problem with an average score of 0–1 point; five problems
with an average score of 1–2 points; three problems with an average score
of 2–3 points; two problems with an average score of 3–4 points; three prob-
lems with an average score above 4 points. Overall, the sequence problems
tended to be relatively difficult.
In the 24th–64th IMOs, there were a total of nine sequence problems.
Among these, one had an average score of 0–1 point; three had an average
score of 1–2 points; three had an average score of 2–3 points; no problem
had an average score of 3–4 points; two had an average score above 4 points.
Further analysis of the problem numbers of these nine sequence problems,
as shown in Table 3.3, reveals that these problems frequently appeared as
the 3rd/6th problem, with a relatively even distribution across different
types of problems.

Table 3.3 Numbers of Sequence Problems in the 24th–64th IMOs

Problem Number
Number of Problems
Sequence Problem 1, 4 2, 5 3, 6 in the First 64 IMOs

Determining values 2 1 0 4
Existence problems 0 0 2 5
Proving quantitative relationships 0 0 4 5

Total 2 1 6 14

From Table 3.2, it can be observed that in the sequence problems, the
average score of the top five teams is generally about 3 points higher than
the average score of the problem. However, in simpler problems, the dif-
ferences in average scores among the top five teams, 6th–15th teams, and
16th–25th teams are not significant, as seen in problems such as Problem
3.2 (IMO 36-4), Problem 3.3 (IMO 58-1), and Problem 3.4 (IMO 59-2).
From Table 3.2, it can also be observed that in the sequence problems
of the first 32 IMOs, the average score of the 16th–25th teams is close to or
below the average score of the problem, such as Problem 3.7 (IMO 23-3),
Problem 3.8 (IMO 32-6), and Problem 3.11 (IMO 26-6). However, after the
32nd IMO, the average score of the 16th–25th teams is about 1.5 points
208 IMO Problems, Theorems, and Methods: Algebra

higher than the average score of the problem. This phenomenon is due to
the smaller number of participating teams in early IMOs. It was not until
the 30th IMO in 1989 that the number of participating teams exceeded
50. Therefore, during this period, it was common to see situations where
the average score was close to or even higher than the average score of the
16th–25th teams.
Chapter 4

Inequality Problems

American mathematician Loren C. Larson once stated, “Inequalities are

useful in virtually all areas of mathematics, and inequality problems are
among the most beautiful.” Indeed, mathematics, as a discipline studying
the quantitative relationships and spatial forms of the real world, often
manifests these relationships through inequalities, and they hold true in
most mathematical domains.
The study of inequalities has a long history, primarily originating from
geometry, with examples such as the triangle inequality, AM-GM inequality
for n = 2, isoperimetric inequality, and others. Notably, the isoperimetric
inequality was known to Archimedes and even earlier Greek mathematicians
before the common era, but its first rigorous proof did not appear until
the 19th century, over two millennia later. In particular, estimating the
value of π was one of the significant contributions of inequalities to ancient
mathematics.
However, before Newton and Cauchy, the development of inequalities
progressed slowly, and only  a few were known. Newton discovered

a1 a2 ···ar
pr−1 pr+1 ≤ p2r , where pr = r 1≤a1 <a2 <···<ar ≤n
Crn , and Cauchy used
induction to prove the AM-GM inequality. Subsequently, many mathemati-
cians have studied inequalities, leading to numerous famous inequalities
named after them.
Two significant events were noteworthy in the development of inequali-
ties. One was the introduction of the Chebyshev inequality by mathemati-
cian P. L. Chebyshev in his 1882 paper. It states
 b  b  b  b
f (x)g(x)p(x)dx p(x)dx ≥ f (x)p(x)dx g(x)p(x)dx,
a a a a

209
210 IMO Problems, Theorems, and Methods: Algebra

where f (x), g(x), p(x) are integrable functions on [a, b], with p(x) ≥ 0,
and f (x), g(x) are both increasing or decreasing on [a, b]. Chebyshev also
applied this inequality to probability theory.
The other pivotal moment was during the 1928 resignation speech of
renowned mathematician Godfrey Harold Hardy as the president of the
London Mathematical Society. He emphasized the elegance of inequalities
and his fondness for them, leading to a high proportion of inequality-related
papers in the society’s journal. He proposed that mastering elementary
inequalities is one of the necessary skills for the study of function theory.
Along with John Edensor Littlewood and George Pólya, Hardy co-authored
“Inequalities,” which cataloged some of the famous historical inequalities
and their origins, significantly impacting the field’s development.
Subsequently, the study of inequalities gradually formed into a
systematic scientific theory, significantly influencing other mathemati-
cal domains, and led to numerous publications on inequalities. Notably,
the Journal of Inequalities and Applications was founded in 1997, and
Mathematical Inequalities and Applications in 1998. These academic jour-
nals marked a new phase in the development of inequalities.
Due to the significant value and impact of inequalities, they frequently
appear in various mathematics competitions. For example, inequality prob-
lems in the IMO are often special cases of some famous inequalities, with their
difficulty significantly reduced compared to the original famous inequalities.
Moreover, these problems typically have special, simpler solutions.
In the first 64 IMOs, there had been a total of 33 inequality problems,
approximately accounting for 32.7% of all algebra problems. These prob-
lems can be primarily categorized into three types: (1) solving inequalities,
totaling five problems; (2) proving inequalities, totaling 25 problems; (3)
determining value ranges, totaling three problems. The statistical distribu-
tion of these three types of problems in the previous IMOs is presented in
Table 4.1.
In the 11th–20th, 31st–40th, and 41st–50th IMOs, the proportion of
inequality problems among all algebra problems exceeded 30%, especially in
the 41st–50th IMOs, where the proportion of inequality problems surpassed
50%, making it a focal topic of the IMO. It can also be observed from the
table that in early IMOs, there were problems involving solving inequalities,
but this type of problems gradually disappeared over the time. However,
the type of proving inequalities continued to thrive.
Remarkably, in recent IMOs, inequality problems with parameters
have emerged, and this trend is evident in some National Mathematical
 Inequality Problems 211

Table 4.1 Numbers of Inequality Problems in the First 64 IMOs.

Session

Content 1–10 11–20 21–30 31–40 41–50 51–60 61–64 Total

Solving inequalities 3 1 1 0 0 0 0 5
Proving inequalities 2 5 3 3 7 2 3 25
Determining value 0 1 0 1 1 0 0 3
ranges
Algebra problems 20 20 14 13 15 13 6 101
The percentage of 25.0% 35.0% 28.6% 30.8% 53.3% 15.4% 50.0% 32.7%
inequality problems
among the algebra
problems

Olympiads. Also, the difficulty and flexibility of proving inequalities have

increased.
Among the inequality-proving problems, nine were related to three-
variable inequalities, with four appearing in the 41st–50th IMOs. After
the 51st IMO, such problems no longer appeared. The primary reason for
this is the relatively mature methods available for proving three-variable
inequalities, while n-variable inequalities pose more challenges.
Additionally, cyclic symmetric inequalities frequently appear in the
IMO, where the assumption of symmetry plays a crucial role in these prob-
lems. When encountering problems with uncertain equality conditions, it
is advisable to first seek simplification methods, such as adjusting to make
the smallest number zero, or using partial derivatives to deduce certain
properties.
This chapter will be divided into three parts. The first part introduces
some famous inequalities, including the AM-GM inequality, Cauchy–
Schwarz inequality, Hölder’s inequality, rearrangement inequality, Cheby-
shev’s inequality, etc. Simultaneously, there is a certain explanation of
the applications of these inequalities. Following that, common methods
for proving inequalities are presented, such as the comparison method,
adjustment method, SOS method, local inequality method, etc.
The second part revolves around three types of problems: “solving
inequalities,” “proving inequalities,” and “determining value ranges.”
These problems are presented in chronological order, and some problems
include various solutions, generalizations, and similar problems.
212 IMO Problems, Theorems, and Methods: Algebra

It is important to note that for each problem, the solutions are fol-
lowed by information on the scores, including the number of contestants in
each score range, the average score, and the scores of the top five teams.
However, early IMOs often lacked information on contestant scores, so
the number of contestants in each score range only represents the counted
number of contestants, and some problems lack scores of the top five teams.
The third part provides a brief summary of this chapter.
Inequalities come in a wide variety, and while this chapter introduces
some methods for proving inequalities, ranging from straightforward basic
approaches to techniques that require manipulation of the inequalities,
this book is not a compendium of problem-solving methods, nor can it
enumerate all possible strategies. Often, the most practical and effective
methods are those discovered and understood by readers through their own
exploration. Therefore, when appreciating problems from past IMOs, read-
ers are encouraged to ponder the problems on their own before consulting
the solutions. It’s possible that the readers’ own solutions may even be
more elegant.

4.1 Common Theorems, Formulas, and Methods

4.1.1 Solving inequalities
Solving inequalities is similar to solving equations, where there might be
extraneous solutions or potential loss of solutions. Therefore, while solving
inequalities, it is essential to apply equivalent transformations. When deal-
ing with inequalities involving radicals, exponentials, and logarithms, it is
crucial to leverage the properties of functions.
For solving polynomial inequalities of one variable of higher degree, the
“root plotting method” can be employed. Generally, consider a polynomial

f (x) = (x − a1 )(x − a2 ) · · · (x − an ),

with its n real roots ordered as a1 < a2 < · · · < an . The number line is
divided into n + 1 intervals:

(−∞, a1 ), (a1 , a2 ), . . . , (an−1 , an ), (an , +∞).

Counting these intervals from right to left, we obtain the following:

(i) f (x) > 0 in odd-numbered intervals;

(ii) f (x) < 0 in even-numbered intervals.
 Inequality Problems 213

4.1.2 Mean inequalities and the Cauchy–Schwarz

inequality
(1) Mean inequalities
Let a1 , a2 , . . . , an be n positive real numbers. Define
n √
Hn = 1 1 1 , Gn = n a1 a2 · · · an ,
a1 + a2 + · · · + an

a 1 + a2 + · · · + an a21 + a22 + · · · + a2n
An = , Qn = ,
n n
where Hn , Gn , An , Qn respectively represent the harmonic mean, geometric
mean, arithmetic mean, and quadratic mean of these n numbers. These four
means are related as follows:
Hn ≤ Gn ≤ An ≤ Qn ,
with equality holding if and only if a1 = a2 = · · · = an .
√
The inequality a1 + a2 + · · · + an ≥ n n a1 a2 · · · an , also known as the
AM-GM inequality, is sometimes utilized in proving inequalities through a
skillful manipulation of “terms” and “coefficients.” Generally, the structure
of the inequality determines the structure of the terms, while the condition
for the equality determines the coefficients of the terms.

Example 4.1. Given x, y, z > 0 with x + y + z = 3, prove that

x2 y2 z2 3
+ + ≥ .
(y + z)3 (z + x)3 (x + y)3 8
Proof. By the AM-GM inequality,
1 1 x2 3
x(y + z) + (y + z)2 + ≥ x,
16 32 (y + z)3 8
and similarly for the other two inequalities. Therefore,
x2 y2 z2
3
+ 3
+
(y + z) (z + x) (x + y)3
3 2(x2 + y 2 + z 2 ) + 6(xy + yz + zx)
≥ (x + y + z) − .
8 32
Since 3(xy + yz + zx) ≤ (x + y + z)2 ,
2(x2 + y 2 + z 2 ) + 6(xy + yz + zx)
2
= 2(x + y + z)2 + 2(xy + yz + zx) ≤ 2(x + y + z)2 + (x + y + z)2 .
3
Substituting x + y + z = 3 into the above, the proposition holds.
214 IMO Problems, Theorems, and Methods: Algebra

When dealing with inequality problems, elimination and degree reduc-

tion are commonly used techniques. In this case, two terms x(y + z) and
(y + z)2 are added to the original inequality, allowing the application of the
AM-GM inequality. This approach eliminates the denominator while also
utilizing the condition x + y + z = 3. However, if only one term (y + z)3 is
added, leading to
1 x2 1
(y + z)3 + 3
≥ x,
64 (y + z) 4
then it becomes challenging to handle after rearrangement.
The coefficients for the two terms x(y + z) and (y + z)2 are determined
by the equality condition x = y = z = 1.
Furthermore, there are several common corollaries to the AM-GM
inequality:

Corollary 4.1. Let a, b, c, d be non-negative real numbers. Then

(a + b + c + d)2
ab + bc + cd + da ≤ .
4
Corollary 4.2. Let a, b, c, d be non-negative real numbers. Then
(a + b + c + d)3
abc + bcd + cda + dab ≤ .
16
Corollary 4.3. Given an integer n ≥ 4, let x1 , x2 , . . . , xn be non-negative
real numbers. Then
(x1 + x2 + · · · + xn )2
x1 x2 + x2 x3 + · · · + xn x1 ≤ .
4
Proof. Without loss of generality, assume x1 = max{x1 , x2 , . . . xn }. Then

x1 x2 + x2 x3 + · · · + xn x1 ≤ x1 x2 + x2 x3 + x3 x4 + x1 (x4 + x5 + · · · + xn )
= x1 (x2 + x4 + x5 + · · · + xn ) + x3 (x2 + x4 )
≤ (x1 + x3 )(x2 + x4 + x5 + · · · + xn )
 2
x1 + x2 + · · · + xn
≤ .
2
Example 4.2. Given an integer n ≥ 3, let a1 , a2 , . . . , a2n and b1 , b2 , . . . , b2n
be 4n non-negative real numbers satisfying

a1 + a2 + · · · + a2n = b1 + b2 + · · · + b2n > 0,

 Inequality Problems 215

and ai ai+2 ≥ bi + bi+1 for any i = 1, 2, . . . , 2n, where a2n+1 = a1 ,

a2n+2 = a2 , and b2n+1 = b1 .

Find the minimum value of a1 + a2 + · · · + a2n .

Solution. For n = 3,
6
 6

(a1 + a3 + a5 )2 ≥ 3(a1 a3 + a3 a5 + a5 a1 ) ≥ 3 bi = 3 ai ,
i=1 i=1
6
 6

(a2 + a4 + a6 )2 ≥ 3(a2 a4 + a4 a6 + a6 a2 ) ≥ 3 bi = 3 ai .
i=1 i=1

Multiplying these two inequalities and taking the square root, we obtain
6  2
a 1 + a2 + · · · + a6
3 ai ≤ (a1 + a3 + a5 )(a2 + a4 + a6 ) ≤ ,
i=1
2
6
and hence i=1 ai ≥ 12. The equality holds when ai = bi = 2 for
i = 1, 2, 3, 4, 5, 6.
For n ≥ 4, from Corollary 4.1.3,
  2n 2n
(a1 + a3 + · · · + a2n−1 )2
≥ a1 a3 + a3 a5 + · · · + a2n−1 a1 ≥ bi = ai ,
4 i=1 i=1

 2n
 2n
(a2 + a4 + · · · + a2n )2
≥ a2 a4 + a4 a6 + · · · + a2n a2 ≥ bi = ai .
4 i=1 i=1

Multiplying these two inequalities and taking the square root, we get
2n
 (a1 + a3 + · · · + a2n−1 )(a2 + a4 + · · · + a2n )
ai ≤
i=1
4

(a1 + a2 + · · · + a2n )2
≤ .
16
2n
Therefore i=1 ai ≥ 16. The equality holds when

a1 = a2 = a3 = a4 = 4, ai = 0(i = 5, 6, . . . , 2n), b2 = 16,

bi = 0(i = 2).

This was the second problem in the 2020 Chinese High School Mathe-
matics League, illustrating a flexible application of the AM-GM inequality
216 IMO Problems, Theorems, and Methods: Algebra

and related corollaries can greatly simplify problems. However, in some

cases, it is necessary to transform the inequality first before using the AM-
GM inequality.

Example 4.3. Given a, b, c > 0 with a + b + c = 3, prove that

a b c 3
+ + ≥ .
1 + b2 1 + c2 1 + a2 2

In fact, directly applying the AM-GM inequality to the denominators

is not feasible, because it changes the direction of the inequality, i.e.,

a b c a b c 3
+ + ≤ + + ≥ .
1 + b2 1 + c2 1 + a2 2b 2c 2a 2

If we transform the inequality first, then

a ab2 ab2 ab
= a − ≥ a − =a− .
1 + b2 1 + b2 2b 2

Since 3(ab + bc + ca) ≤ (a + b + c)2 = 9,

a b c ab + bc + ca 3
+ + ≥a+b+c− ≥ .
1 + b2 1 + c2 1 + a2 2 2

This method of initially altering the sign of the expression and then
applying the AM-GM inequality to the denominator or similar structures
without changing the direction of the inequality is also known as the
Cauchy Reverse Technique. Similarly, this technique can be used to solve
the following problem:
Given a, b, c, d > 0 with a + b + c + d = 4, prove that

a b c d
+ + + ≥ 2.
1 + b2 c 1 + c2 d 1 + d2 a 1 + a2 b

(2) Weighted AM-GM inequality

Let a1 , a2 , . . . , an and λ1 , λ2 , . . . , λn be positive real numbers, where
n
i=1 λi = λ. Then

λ1 a1 + λ2 a2 + · · · + λn an
aλ1 1 aλ2 2 · · · aλnn ≤
λ
,
λ
with equality holding if and only if a1 = a2 = · · · = an .
 Inequality Problems 217

(3) Power mean inequality

Let a1 , a2 , . . . , an be n positive real numbers, and α > β > 0. Then
1
  α1
aβ1 + aβ2 + · · · + aβn
β
aα α α
1 + a2 + · · · + an
≤ ,
n n

with equality holding if and only if a1 = a2 = · · · = an .

(4) Cauchy–Schwarz inequality

Let a1 , a2 , . . . , an , b1 , b2 , . . . , bn be real numbers. Then
(a1 b1 + a2 b2 + · · · + an bn )2 ≤ (a21 + a22 + · · · + a2n )(b21 + b22 + · · · + b2n ),
and the equality holds if and only if bi = λai for i = 1, 2, . . . , n.
The Cauchy–Schwarz inequality is a well-known inequality, which can
be extended to complex numbers. In fact, the Cauchy–Schwarz inequality
has a clear geometric interpretation, deriving from the cosine of the angle
between two vectors in an n-dimensional space or the area of a triangle in
an n-dimensional space formed by two vectors with a common initial point.
There are various methods to prove the Cauchy–Schwarz inequality, and
one approach is using a quadratic function.
Let f (x) = (a1 x − b1 )2 + (a2 x − b2 )2 + · · · + (an x − bn )2 , which can be
rearranged as
f (x) = (a21 + a22 + · · · + a2n )x2 − 2(a1 b1 + a2 b2 + · · · + an bn )x
+ (b21 + b22 + · · · + b2n ).
It is clear that f (x) ≥ 0. Hence its discriminant Δ ≤ 0, i.e.,
(a1 b1 + a2 b2 + · · · + an bn )2 ≤ (a21 + a22 + · · · + a2n )(b21 + b22 + · · · + b2n ),
and the equality holds if and only if there exists x0 such that
a1 x0 − b1 = a2 x0 − b2 = · · · = an x0 − bn = 0.
Moreover, the Cauchy–Schwarz inequality has several straightforward
corollaries:

Corollary 4.4. For any real numbers a1 , a2 , . . . , an and positive real num-
bers b1 , b2 , . . . , bn ,
a21 a2 a2 (a1 + a2 + · · · + an )2
+ 2 + ···+ n ≥ .
b1 b2 bn b1 + b2 + · · · + bn
218 IMO Problems, Theorems, and Methods: Algebra

Corollary 4.5. For any real numbers a1 , a2 , . . . , an and b1 , b2 , . . . , bn ,

 
a21 + b21 + a22 + b22 + · · · + a2n + b2n

≥ (a1 + a2 + · · · + an )2 + (b1 + b2 + · · · + bn )2 .

Corollary 4.6. For any real numbers a1 , a2 , . . . , an ,

(a1 + a2 + · · · + an )2 ≤ n(a21 + a22 + · · · + a2n ).

In general, problems involving applications of the Cauchy–Schwarz

inequality to find the minimum value are quite common, which sometimes
leads to difficulty in associating the Cauchy–Schwarz inequality with
problems seeking the maximum value.

Example 4.4. Given a, b, c ≥ 0, prove that

a2 − bc b2 − ca c2 − ab
+ 2 + 2 ≥ 0.
2a2 2
+b +c 2 2
2b + c + a 2 2c + a2 + b2
2 (b+c)2
Proof. It is known that 2a2a+b−bc 2 +c2 =
1
2 (1 − 2a2 +b2 +c2 ), and thus the
inequality is equivalent to proving
(b + c)2 (c + a)2 (a + b)2
+ 2 + 2 ≤ 3.
2a2 2
+b +c 2 2
2b + c + a 2 2c + a2 + b2
Using the Cauchy–Schwarz inequality, we see that
(b + c)2 b2 c2
≤ + ,
2a2 + b2 + c2 a2 + b 2 c2 + a2
and similarly for the other two inequalities. Summing up these three
inequalities completes the proof.
2
b2 c2
However, the last step 2a(b+c)
2 +b2 +c2 ≤ a2 +b2 + c2 +a2 is easily overlooked.

(5) Reverse of the Cauchy–Schwarz inequality

Given real numbers a1 , a2 , . . . , an , b1 , b2 , . . . , bn , and a, A, b, B satisfying

0 < a ≤ ai ≤ A, 0 < b ≤ bi ≤ B, i ∈ {1, 2, . . . , n},

n n   2 n 2
  1 AB ab 
a2i b2i ≤ + ai b i ,
i=1 i=1
4 ab AB i=1

and the equality holds if and only if a = ai = A and b = bi = B.

 Inequality Problems 219

Indeed, the Cauchy–Schwarz inequality is a special case of Hölder’s

inequality.

(6) Hölder’s inequality

(i) Given ai,j > 0 and wi > 0 for i ∈ {1, 2, . . . , n}, j ∈ {1, 2, . . . , m}, and
n
i=1 wi = 1,
⎛ ⎞wi
n m m
 n
⎝ ai,j ⎠ ≥ aw i
i,j ,
i=1 j=1 j=1 i=1

and the equality holds if and only if the following n sequences are
proportional:
(a11 , a12 , . . . , a1m ), (a21 , a22 , . . . , a2m ), . . . , (an1 , an2 , . . . , anm ).
1
(ii) Given ai , bi > 0 for i = 1, 2, . . . , n, and positive p, q satisfying p + 1q = 1,
1 1
a1 b1 + a2 b2 + · · · + an bn ≤ (ap1 + ap2 + · · · + apn ) p (bq1 + bq2 + · · · + bqn ) q ,
where the equality holds if and only if api = λbqi for i = 1, 2, . . . , n.
Furthermore, there are several common corollaries to Hölder’s
inequality:

Corollary 4.7. For positive real numbers a1 , a2 , . . . , an , b1 , b2 , . . . , bn and a

positive integer m,
am+1
1 am+1
2 am+1
n (a1 + a2 + · · · + an )m+1
+ + · · · + ≥ ,
bm1 bm2 bm
n (b1 + b2 + · · · + bn )m
and the equality holds if and only if ai = λbi for i = 1, 2, . . . , n.

Corollary 4.8. For positive real numbers a1 , a2 , a3 , b1 , b2 , b3 , c1 , c2 , c3 ,

(a31 + a32 + a33 )(b31 + b32 + b33 )(c31 + c32 + c33 ) ≥ (a1 b1 c1 + a2 b2 c2 + a3 b3 c3 )3 .

Corollary 4.9. For positive real numbers a1 , a2 , . . . , an ,

√
(1 + a1 )(1 + a2 ) · · · (1 + an ) ≥ (1 + n a1 a2 · · · an )n .

4.1.3 Other famous inequalities

(1) Triangle inequality
For real numbers a and b,
||a| − |b|| ≤ |a + b| ≤ |a| + |b|,
220 IMO Problems, Theorems, and Methods: Algebra

and the first equality holds if and only if ab ≤ 0 and the second equality
holds if and only if ab ≥ 0.
Theorem 4.1. For real numbers a1 , a2 , . . . , an ,

|a1 + a2 + · · · + an | ≤ |a1 | + |a2 | + · · · + |an |.

This inequality also holds for complex numbers.

(2) Rearrangement inequality

For real numbers a1 , a2 , . . . , an , b1 , b2 , . . . , bn satisfying

a1 ≤ a2 ≤ · · · ≤ an and b1 ≤ b2 ≤ · · · ≤ bn ,

it holds that

a1 bn + a2 bn−1 + · · · + an b1 ≤ a1 bt1 + a2 bt2 + · · · + an btn

≤ a1 b 1 + a2 b 2 + · · · + an b n ,

where {t1 , t2 , . . . , tn } = {1, 2, . . . , n}.

That is, “reverse order sum” ≤ “mixed order sum” ≤ “same order sum.”

(3) Chebyshev’s inequality

For real numbers a1 , a2 , . . . , an , b1 , b2 , . . . , bn satisfying a1 ≤ a2 ≤ · · · ≤ an
and b1 ≤ b2 ≤ · · · ≤ bn ,
n
 n
 n
 n

1
ai bn+1−i ≤ ai bi ≤ ai b i .
i=1
n i=1 i=1 i=1

Suppose it is required to prove an inequality expressed as a sum of fractions:

x1 x2 xn
+ + ··· + ≥ 0,
y1 y2 yn
where x1 , x2 , . . . , xn are real numbers, and y1 , y2 , . . . , yn are positive real
numbers. Similar to the method of undetermined coefficients, if there exist
positive real numbers α1 , α2 , . . . , αn such that α1 x1 ≤ α2 x2 ≤ · · · ≤ αn xn
and α1 y1 ≥ α2 y2 ≥ · · · ≥ αn yn , then we can employ Chebyshev’s inequality,
x1 x2 xn
+ + ··· +
y1 y2 yn
 
1 1 1 1
≥ (α1 x1 + α2 x2 + · · · + αn xn ) + + ···+ .
n α1 y1 α2 y2 αn yn
 Inequality Problems 221

Thus, it is only necessary to prove that α1 x1 + α2 x2 + · · · + αn xn ≥ 0.

The advantage of this method is that it eliminates the fractions in the
inequality, and the values of α1 , α2 , . . . , αn can even be chosen appropriately
such that α1 x1 + α2 x2 + · · · + αn xn = 0.
1 1 1
Example 4.5. Let a, b, c, d > 0 and satisfy a + b + c + d = a + b + c + 1d .
Prove that

2(a + b + c + d) ≥ a2 + 3 + b2 + 3 + c2 + 3 + d2 + 3.

Proof. Assume without loss of generality that a ≥ b ≥ c ≥ d > 0. From

the given condition,
a2 − 1 b2 − 1 c2 − 1 d2 − 1
+ + + =0
a b c d
√ 3(a2 − 1)
and 2a − a2 + 3 = √ .
2a + a2 + 3
It is evident that y = x − x1 and y = 2x+√xx2 +3 are monotonically
increasing functions for x > 0. Therefore,
 a2 − 1   a2 − 1 a

√ = · √
cyc 2a + a2 + 3 cyc
a 2a + a2 + 3

1  a2 − 1 a 
≥ √ = 0,
cyc
4 a
cyc 2a + a2 + 3
 
where cyc denotes the cyclic sum, as in this case cyc a = a + b + c + d.
This method of using Chebyshev’s inequality to prove other inequalities
is also known as the Chebyshev Associate Technique.

(4) Abel transformation

For positive integers m, n with m < n,
n
 n−1

(Ak − Ak−1 )bk = An bn − Am−1 bm + Ak (bk − bk+1 ),
k=m k=m

which is known as Abel’s summation by parts formula. In this formula,

k
setting A0 = 0 and Ak = i=1 ai for 1 ≤ k ≤ n, we obtain
n
 n
 n−1
 k

ak b k = b n ak + ai (bk − bk+1 ).
k=1 k=1 k=1 i=1
222 IMO Problems, Theorems, and Methods: Algebra

(5) Abel’s inequality

t
Given b1 ≥ b2 ≥ · · · ≥ bn > 0 and m ≤ i=1 ai ≤ M for t ∈ {1, 2, . . . , n},
n

b1 m ≤ ak bk ≤ b1 M.
k=1

Example 4.6. Let a1 , a2 , . . . , an be n non-negative real numbers and

k
denote Sk = i=1 ai for 1 ≤ k ≤ n. Prove that
⎛ ⎞
 n  n n

⎝ a i Si a2j ⎠ ≤ (ai Si )2 .
i=1 j=i i=1
n
Proof. Let bi = ai Si and ci = j=i a2j for i = 1, 2, . . . , n. The original
inequality is equivalent to
n
 n

b i ci ≤ b2i . (*)
i=1 i=1

Note that for 1 ≤ i ≤ n,

Bi = b 1 + b 2 + · · · + b i = a 1 S 1 + a 2 S 2 + · · · + a i S i
≤ (a1 + a2 + · · · + ai )Si = Si2 .
Thus, by Abel’s summation by parts formula,
n
 n−1

b i ci = Bi (ci − ci+1 )+Bn cn
i=1 i=1
n−1

≤ a2i Si2 + Bn cn
i=1
n
 n

≤ a2i Si2 = b2i .
i=1 i=1

Hence, the inequality (*) holds, and therefore, the original inequality holds.

(6) Lagrange’s identity

For real numbers a1 , a2 , . . . , an , b1 , b2 , . . . , bn ,
n n n 2
   
a2i b2i − ai b i = (ai bj − aj bi )2 .
i=1 i=1 i=1 1≤i<j≤n
 Inequality Problems 223

(7) Minkowski’s inequality

For positive real numbers a1 , a2 , . . . , an , b1 , b2 , . . . , bn and r > 1,

1 1 1
n
 r n
 r n
 r
r
(ai + bi ) ≤ ari + bri ,
i=1 i=1 i=1

and the equality holds if and only if ai = λbi for i = 1, 2, . . . , n.

This inequality reverses when r < 1.

(8) Bernoulli’s inequality

Let n ≥ 2 be an integer. Then for real numbers a1 , a2 , . . . , an , all greater
than −1 and of the same sign,

(1 + a1 )(1 + a2 ) · · · (1 + an ) > 1 + a1 + a2 + · · · + an .

In particular, if x > −1 and n ≥ 2, then (1 + x)n > 1 + nx.

Let f (x) be a continuous function defined on the interval [a, b]. If for
any x1 , x2 ∈ [a, b],
 
x1 + x2 f (x1 ) + f (x2 )
f ≤ (or ≥) ,
2 2

then f (x) is termed as a convex function (or concave function) on [a, b].
If the second derivative f  (x) exists and f  (x) ≥ 0(or ≤ 0) for x ∈ [a, b],
then f (x) is a convex function (or concave function) on [a, b].

(9) Jensen’s inequality

Let f (x) be a convex function (or concave function) on [a, b]. Then for any
x1 , x2 , . . . , xn ∈ [a, b], it holds that
 
x1 + x2 + · · · + xn f (x1 ) + f (x2 ) + · · · + f (xn )
f ≤ (or ≥) ,
n n

and the equality holds if and only if x1 = x2 = · · · = xn .

More generally, if f (x) is a convex function (or concave function) on
[a, b], then for any x1 , x2 , . . . , xn ∈ [a, b], and any non-negative real numbers
224 IMO Problems, Theorems, and Methods: Algebra

λ1 , λ2 , . . . , λn satisfying λ1 + λ2 + · · · + λn = 1,

f (λ1 x1 + λ2 x2 + · · · + λn xn )
≤ (or ≥)λ1 f (x1 ) + λ2 f (x2 ) + · · · + λn f (xn ).

Example 4.7. Let a, b, x, y, z be positive real numbers such that

x + y + z = 1. Prove that
   
b b b
a+ a+ a+ ≥ (a + 3b)3 .
x y z
2
Proof. Let f (x) = ln(a + xb) for x ∈ (0, +∞). Then f  (x) = (ax
2abx+b
2 +bx)2 > 0,

and hence f (x) is a convex function. By Jensen’s Inequality,

 
x+y+z
f (x) + f (y) + f (z) ≥ 3f ,
3

i.e., ln(a + xb ) + ln(a + yb ) + ln(a + zb ) ≥ 3 ln(a + 3b

x+y+z ), which implies
   
b b b
a+ a+ a+ ≥ (a + 3b)3 .
x y z

(10) Schur’s inequality

For non-negative real numbers x, y, z and r > 0,

xr (x − y)(x − z) ≥ 0.
cyc

When r = 1, Schur’s inequality becomes


x(x − y)(x − z) ≥ 0,
cyc

which is equivalent to
 
(i) x3 − cyc x2 (y + z) + 3xyz ≥ 0;
cyc  
(ii) ( cyc x)3 − 4 cyc x cyc yz + 9xyz ≥ 0.

Example 4.8. Let x, y, z be positive real numbers such that x+y+z = xyz.
Prove that

x2 + y 2 + z 2 − 2(xy + yz + zx) + 9 ≥ 0.
 Inequality Problems 225

Proof. Since x + y + z = xyz, the original inequality is equivalent to

(x2 + y 2 + z 2 − 2(xy + yz + zx))(x + y + z) + 9xyz ≥ 0
⇔ x3 + y 3 + z 3 − (x2 y + y 2 z + z 2 x + xy 2 + yz 2 + zx2 ) + 3xyz ≥ 0,
 
which further simplifies to cyc x3 − cyc x2 (y + z) + 3xyz ≥ 0, that is (i).
Hence, the original inequality holds.

4.1.4 Common methods for proving inequalities

(1) Comparison method
To prove A ≥ B, it suffices to prove A − B ≥ 0. If B > 0, then it also
A
suffices to prove B ≥ 1.

Example 4.9. Let a, b, c be positive real numbers. Prove that

a2 + bc b2 + ca c2 + ab
+ + ≥ a + b + c.
b+c c+a a+b
Proof. The inequality can be simplified to
a2 + bc b2 + ca c2 + ab
−a+ −b+ −c
b+c c+a a+b
(a2 − b2 )2 + (b2 − c2 )2 + (c2 − a2 )2
= ≥ 0.
2(a + b)(b + c)(c + a)
Example 4.10. Let a, b, c be positive real numbers. Prove that
a2a b2b c2c ≥ ab+c bc+a ca+b .

Proof. Since the inequality is symmetric in a, b, c, assume without loss

of generality that a ≥ b ≥ c. Thus,
a2a b2b c2c  a a−b  b b−c  a a−c
= ≥ 1.
ab+c bc+a ca+b b c c

(2) Analytical method

Assume the inequality to be proven is valid. Then derive a series of equiv-
alent inequalities from it until reaching a more easily provable one or an
obviously true one.

Example 4.11. Let n be a positive integer. Prove that

   
1 1 1 1 1 1 1
1 + + ···+ ≥ + + ···+ .
n+1 3 2n − 1 n 2 4 2n
226 IMO Problems, Theorems, and Methods: Algebra

Proof. To prove the inequality, it suffices to prove

   
1 1 1 1 1
n 1 + + ···+ ≥ (n + 1) + + ···+ ,
3 2n − 1 2 4 2n
which is reduced to proving
n 1 1 1
≥ + + ··· +
2 2 4 2n
   
1 1 1 1
and n + ···+ ≥n + ···+ .
3 2n − 1 4 2n

(3) Proof by contradiction

Assume the inequality to be proven is false. Then derive a contradiction,
thereby confirming that the inequality is indeed true. The proof by contra-
diction is often suitable when a direct proof is difficult or when there are
few conditions. By assuming the opposite, it effectively adds an additional
condition, making it easier to approach the problem.
Example 4.12. Let x, y, z be positive real numbers. Prove that
 
3 √ √
x + y + 4 z ≥ 32 xyz.

Proof. Using proof by contradiction, assume there exist x0 , y0 , z0 such

that
 
√ √
x0 + 3 y0 + 4 z0 < 32 x0 y0 z0 .
√ √ √ √ √ √
Then x0 < 32 x0 y0 z0 , 6 y0 < 32 x0 y0 z0 , and 24 z0 < 32 x0 y0 z0 , implying
⎧ 16
⎪
⎨x0 < x0 y0 z0 ,
y016 < (x0 y0 z0 )3 ,
⎪
⎩ 16
z0 < (x0 y0 z0 )12 .
Multiplying the above three inequalities yields x16 16 16 16
0 · y0 · z0 < (x0 y0 z0 ) ,
leading to a contradiction. Therefore, the original inequality holds.

(4) Mathematical induction

When proving propositions related to positive integers n, mathematical
induction is often used.
Example 4.13. Given a sequence of positive numbers {xn } satisfying Sn ≥
2Sn−1 for n ∈ {2, 3, . . .}, where Sn = x1 + x2 + · · · + xn , prove that there
exists a constant C > 0 such that xn ≥ C · 2n for n ∈ {1, 2, 3, . . .}.
 Inequality Problems 227

Proof. We use mathematical induction with C = 14 x1 .

The conclusion is evident when n ∈ {1, 2}, i.e., x2 ≥ x1 = C · 22 .
When n ≥ 3, assume xk ≥ C · 2k for k ∈ {1, 2, 3, . . . , n − 1}. From
Sn ≥ 2Sn−1 ,

xn ≥ x1 + (x2 + · · · + xn−1 )
≥ x1 + (C · 22 + · · · + C · 2n−1 )
≥ C(22 + 22 + 23 + · · · + 2n−1 )
= C · 2n .

Thus, by mathematical induction, the conclusion holds.

(5) Bounding method

To prove an inequality A ≤ B, one can introduce one or more intermediate
quantities C. If it can be shown that A ≤ C and C ≤ B, then A ≤ B.

Example 4.14. Let x1 , x2 , x3 be non-negative real numbers satisfying x1 +

x2 + x3 = 1. Prove that
 x2 x3  9
(x1 + 3x2 + 5x3 ) x1 + + ≤ .
3 5 5
Proof. By the AM-GM inequality,
 x2 x3 
(x1 + 3x2 + 5x3 ) x1 + +
3 5
 
1 5x2
= (x1 + 3x2 + 5x3 ) 5x1 + + x3
5 3
  2
1 5x2
≤ (x1 + 3x2 + 5x3 ) + 5x1 + + x3
20 3
 2
1 14
= 6x1 + x2 + 6x3
20 3
1 9
≤ (6x1 + 6x2 + 6x3 )2 = .
20 5

(6) Variable substitution method

Consider a complex expression as a whole and replace it with a single let-
ter to simplify the problem.
√ When conditions or conclusions involve forms
2 2 2
such as x + y = r , r − x2 , |x| ≤ 1, etc., consider substitutions like
2
228 IMO Problems, Theorems, and Methods: Algebra

x=√ r sin α and x =√r cos α. When conditions or conclusions involve forms
like r2 + x2 and x2 − r2 , consider substitutions like x = r tan α and
x = r sec α. When making substitutions, it is important to note that the
range of values for α is determined by the range of values for the original
variable x.
√
Example 4.15. Given 0◦ ≤ α ≤ 90◦ , prove that 2 ≤ 5 − 4 sin α+sin α ≤
9
4.
√ 2
Proof. Let x = 5√− 4 sin α. Then sin α = 5−x 4 . Since 0 ≤ sin α ≤ 1, it
follows that 1 ≤ x ≤ 5. Define
√
y = 5 − 4 sin α + sin α
5 − x2
= x+
4
1 9
= − (x − 2)2 + .
4 4
√
Given 1 ≤ x ≤ 5, it follows that 2 ≤ y ≤ 94 .

(7) Method of undetermined coefficients

This method involves introducing appropriate parameters, determining
their values based on the characteristics of the expression in the problem,
and using them to prove the inequality.

Example 4.16. Let x, y, z be three real numbers, not all zero. Find the
maximum value of x2xy+2yz
+y 2 +z 2 .

Solution. It is required to prove the existence of a constant c such that

1
x2 + y 2 + z 2 ≥(xy + 2yz),
c
with equality holding for some values of x, y, z. Since the terms on the right
side are xy and 2yz, the term y 2 on the left side should be split into two
terms αy 2 and (1 − α)y 2 . By using
√ √
x2 + αy 2 ≥ 2 αxy, (1 − α)y 2 + z 2 ≥ 2 1 − αyz,
√
2 √1−α
and from 2 α
= 2, we obtain α = 15 .
√
Thus, x + y + z 2 ≥ √25 (xy + 2yz), implying c = 25 .
2 2

When dealing with non-symmetric inequalities, the coefficients of similar

terms are often not the same. Introducing additional parameters to group
 Inequality Problems 229

terms can facilitate the use of common inequalities such as the AM-GM
inequality, Cauchy–Schwarz inequality, etc.

(8) Function method

Prove inequalities by introducing functions and utilizing their properties
such as monotonicity, convexity/concavity, boundedness, etc.

Example 4.17. Let a, b be non-negative real numbers. Prove that

3a3 + 7b3 ≥ 9ab2 .

Proof. If a = 0 or b = 0, then the conclusion holds. If a, b > 0, then let

x = ab and f (x) = 3x3 + 7 − 9x for x > 0. Clearly, f  (x) = 9(x2 − 1).
When x ∈ (0, 1), since f  (x) < 0, f (x) is strictly decreasing on (0, 1).
When x ∈ (1, +∞), from f  (x) > 0, we see that f (x) is strictly increasing on
(1, +∞). Therefore, f (x) attains its minimum value at x = 1, i.e., f (x) ≥
f (1) = 1 > 0.

(9) Construction method

We can prove inequalities by constructing identities, figures, functions, etc.
(i) Construct Identities

Example 4.18. Given a2 + b2 + c2 + d2 = 1, prove that

(a + b)4 + (a + c)4 + (a + d)4 + (b + c)4 + (b + d)4 + (c + d)4 ≤ 6.

Proof. Consider the sum

(a − b)4 + (a − c)4 + (a − d)4 + (b − c)4 + (b − d)4 + (c − d)4 ,
and construct the identity
6(a2 + b2 + c2 + d2 )2
= (a + b)4 + (a + c)4 + (a + d)4 + (b + c)4 + (b + d)4 + (c + d)4
+ (a − b)4 + (a − c)4 + (a − d)4 + (b − c)4 + (b − d)4 + (c − d)4 ,
from which the inequality follows.

(ii) Construct Figures

Example 4.19. Let a, b, c be positive real numbers. Prove that

a2 + ab + b2 ≤ a2 − ac + c2 + b2 − bc + c2 .
230 IMO Problems, Theorems, and Methods: Algebra

Proof. Construct XY Z such that XY = a, XZ = b, and ∠Y XZ =

120◦ .
Inside ∠Y XZ, take a point W such that ∠Y XW and ∠W XZ are both
60◦ , and XW = c. By the law of cosines, it is evident that

YZ = a2 + ab + b2 , YW = a2 − ac + c2 , WZ = b2 − bc + c2 .

Clearly, Y Z ≤ Y W + W Z, so the original inequality holds.

(iii) Construct Functions

Example 4.20. Given real numbers x1 , x2 , . . . , xn , y1 , y2 , . . . , yn with

x21 + x22 + · · · + x2n ≤ 1,

prove that

(x1 y1 + x2 y2 + · · · + xn yn − 1)2
≥ (x21 + x22 + · · · + x2n − 1)(y12 + y22 + · · · + yn2 − 1).

Proof. If x21 + x22 + · · · + x2n = 1, then the original inequality holds.

If x21 + x22 + · · · + x2n < 1, construct a quadratic function

f (x) = (x21 + x22 + · · · + x2n − 1)x2 − 2(x1 y1 + x2 y2 + · · · + xn yn − 1)x

+ (y12 + y22 + · · · + yn2 − 1)
= (x1 x − y1 )2 + (x2 x − y2 )2 + · · · + (xn x − yn )2 − (x − 1)2 ,

whose graph is a downward-opening parabola.

Since f (1) = (x1 − y1 )2 + (x2 − y2 )2 + · · · + (xn − yn )2 ≥ 0, its

discriminant must be non-negative, i.e.,

4(x1 y1 + x2 y2 + · · · + xn yn − 1)2
− 4(x21 + x22 + · · · + x2n − 1)(y12 + y22 + · · · + yn2 − 1) ≥ 0.

Thus, (x1 y1 + x2 y2 + · · · + xn yn − 1)2 ≥ (x21 + x22 + · · · + x2n − 1)

(y12 + y22 + · · · + yn2 − 1).
The tangent line method is also closely related to constructing functions.
Suppose a + b + c is a constant value, and we need to prove f (a) + f (b) +
f (c) ≥ 0. However, if the expression of the function f (x) is complex, and
 Inequality Problems 231

there exists a tangent line y = l(x) to f (x) at some point x0 such that the
graph of the function f (x) lies above the tangent line, and l(a) + l(b) +
l(c) ≥ 0, then

f (a) + f (b) + f (c) ≥ l(a) + l(b) + l(c) ≥ 0.

Example 4.21. Given real numbers a, b, c > 0, prove that

√ √ √ √ √ √ √
a+b+c+ a a+b+c+ b a+b+c+ c 9+3 3
+ + ≥ √ .
b+c c+a a+b 2 a+b+c

Proof. Without loss of generality, assume√a + b + c = 1. The inequality

is equivalent to 1−1√a + 1−1√b + 1−1√c ≥ 9+32 3 .

1√ 1
Define the function f (x) = 1− x
, and its tangent line at x = 3 is
√
9+6 3 3
y= 4 x + 4. When 0 < x < 1,
√
1 9+6 3 3
√ ≥ x+
1− x 4 4
√ √ √
is equivalent to ( 3x − 1)2 ((3 + 2 3) x + 1) ≥ 0. Therefore,
√ √
1 1 1 9+6 3 9 9+3 3
√ + √ + √ ≥ (a + b + c) + = .
1− a 1− b 1− c 4 4 2

The advantage of this method is that it transforms the calculation of com-

plex function values into the calculation of linear function values. This
method can also be used to solve the following problems:

• (Chinese Western Mathematic Invitational 2007, Problem 3).

Let a, b, c be real numbers with a + b + c = 3. Prove that

1 1 1 1
+ + ≤ .
5a2 − 4a + 11 5b2 − 4b + 11 5c2 − 4c + 11 4

• (Korean Mathematical Olympiad 2011, Final Round, Prob-

lem 4). Let a, b, c be non-negative with a+b+c = 1. Find the maximum
value of
1 1 1
+ + .
a2 − 4a + 9 b2 − 4b + 9 c2 − 4c + 9
232 IMO Problems, Theorems, and Methods: Algebra

(10) Adjustment method

When dealing with multivariable inequalities, adjust some of the variables
to “positions” that are easier to handle. Since lower-order symmetric or
cyclic inequalities often have equalities when two numbers are equal or
when one of the numbers is zero, these are ideal “positions” for using the
adjustment method.

Example 4.22. Let a, b, c be positive real numbers. Prove that

63 (a + b + c)(a2 + b2 + c2 ) 27(a + b + c)
+ ≥ √ .
2 abc 2 3 abc
Proof. Without loss of generality, assume abc = 1, and let
63 27
f (a, b, c) = (a + b + c)(a2 + b2 + c2 ) + − (a + b + c).
2 2
Then,
√ √
f (a, b, c) − f (a, bc, bc)
 
√ √ 2 √ 2
√ √ 2 2 27
= ( b − c) (b + bc + c) + bc + a( b + c) + a −
2
 
√ √ √ 27
≥ ( b − c)2 10bc + 4a bc + a2 −
2
 
√ √ 10 10 10 √ √ 27
= ( b − c)2 + + + 2 a + 2 a + a2 −
3a 3a 3a 2
⎛  ⎞
 3
√ √ 10 27
≥ ( b − c)2 ⎝6 4 − ⎠ ≥ 0.
6

3 2

Thus, it suffices to prove f ( t12 , t, t) ≥ 0, which is equivalent to

(t − 1)2 (8t7 + 16t6 − 30t5 − 9t4 + 12t3 + 6t2 + 4t + 2) ≥ 0,

and this inequality can be easily proven using the AM-GM inequality.
If the equality conditions of an inequality are known, using the adjust-
ment method can be convenient. However, sometimes the equality condi-
tions of an inequality are not readily apparent. Here, we introduce a method
to find the equality conditions: the Lagrange multiplier method.

Example 4.23. Let a, b, c be positive real numbers, and ab + bc + ca = 11.

Find the minimum value of 3a + 4b + 5c.
 Inequality Problems 233

Solution. Define f (a, b, c) = 3a + 4b + 5c − k(ab + bc + ca − 11), where

k is an undetermined coefficient. Set the derivatives with respect to each
variable to zero:

fa (a, b, c) = 3 − k(b + c) = 0,
fb (a, b, c) = 4 − k(a + c) = 0,
fc (a, b, c) = 5 − k(a + b) = 0,
3 4 5
yielding k = b+c = c+a = a+b . From ab + bc + ca = 11, we get a = 3,
b = 2, and c = 1, which should be the equality conditions for 3a + 4b + 5c
to attains its minimum value.
We return to the original problem. From c = 11−ab
a+b ,

55 − 5ab
3a + 4b + 5c = 3a + 4b +
a+b
55 + 2ab + 3a2 + 4b2 − 22(a + b)
= + 22
a+b
(a + b − 5)2 + 2(a − 3)2 + 3(b − 2)2
= + 22
a+b
≥ 22.

This method involves knowledge of advanced mathematics and is pro-

vided for reference only.

(11) Sum of squares method

If an expression concerning a, b, c can be transformed into

S = f (a, b, c) = Sc (a − b)2 + Sb (c − a)2 + Sa (b − c)2 ,

where Sa , Sb , Sc are expressions in terms of a, b, c, then the inequality S ≥ 0

holds if at least one of the following conditions is satisfied:

(i)Sa ≥ 0, Sb ≥ 0, and Sc ≥ 0.
(ii)a ≥ b ≥ c, Sb ≥ 0, Sb + Sa ≥ 0, and Sb + Sc ≥ 0.
(iii)a ≥ b ≥ c, Sa ≥ 0, Sc ≥ 0, Sa + 2Sb ≥ 0, and Sc + 2Sb ≥ 0.
(iv) a ≥ b ≥ c, Sb ≥ 0, Sc ≥ 0, and a2 Sb + b2 Sa ≥ 0.
(v) a ≥ b ≥ c are the side lengths of a triangle, and Sa ≥ 0, Sb ≥ 0, and
b2 Sb + c2 Sc ≥ 0.
(vi) Sa + Sb + Sc ≥ 0 and Sa Sb + Sb Sc + Sc Sa ≥ 0.
234 IMO Problems, Theorems, and Methods: Algebra

The proofs for these conditions are not difficult, and conditions (ii) and
(iv) are illustrated here. For condition (ii), it is easy to see that (c − a)2 ≥
(a − b)2 + (b − c)2 , and thus,

Sc (a − b)2 + Sb (c − a)2 + Sa (b − c)2

≥ (Sc + Sb )(a − b)2 + (Sa + Sb )(b − c)2 ≥ 0.

a−c
For condition (iv), it is evident that b−c ≥ ab , from which,

Sc (a − b)2 + Sb (c − a)2 + Sa (b − c)2

≥ Sa (b − c)2 + Sb (c − a)2
a2 (b − c)2
≥ Sa (b − c)2 + Sb
b2
b 2 Sa + a 2 Sb
= (b − c)2
b2
≥ 0.

Example 4.24 (Iran Team Selection Test, 1996). Given non-

negative real numbers x, y, z, prove that

1 1 1 9
+ + ≥ .
(x + y)2 (y + z)2 (z + x)2 4(xy + yz + zx)

Proof. Let a = y + z, b = z + x, and c = x + y. Then the inequality

transforms into
 
1 1 1
(2ab + 2bc + 2ca − a2 − b2 − c2 ) + + ≥ 9,
a2 b2 c2
 2a 2b  a2 +b2 −2ab
which is equivalent to cyc ( b + a − 4) − cyc c2 ≥ 0, i.e.,
     
2 1 2 2 1 2 2 1
− 2 (a − b) + − 2 (b − c) + − 2 (c − a)2 ≥ 0.
ab c bc a ca b
2
Therefore, Sa = bc − a12 , Sb = ca
2
− b12 , and Sc = ab2
− c12 . Assume
a ≥ b ≥ c without loss of generality, and since b + c ≥ a, it is easy to see
that Sa ≥ 0, Sb ≥ 0, and Sa ≥ Sb ≥ Sc . By condition (v), it is sufficient
to prove b2 Sb + c2 Sc ≥ 0, which is equivalent to b3 + c3 ≥ abc, and this is
obviously true.
 Inequality Problems 235

However, when dealing with cyclic but non-symmetric inequalities, the

SOS method may not easily construct a satisfactory form. In such cases,
consider transforming the expression involving a, b, c into
S = f (a, b, c) = M (a − b)2 + N (c − a)(c − b),
where M, N are expressions in terms of a, b, c. If M, N are non-negative real
numbers, and c is the maximum or minimum among a, b, c, then S ≥ 0.
For instance, Schur’s inequality can be represented as

a(a − b)(a − c) = (a + b − c)(a − b)2 + c(c − a)(c − b) ≥ 0.
cyc

(12) The pqr method

The pqr method, also known as the elementary symmetric polynomial
method, defines p = a + b + c, q = ab + bc + ca, and r = abc. Then any sym-
metric polynomial in three variables concerning a, b, c can be expressed in
terms of p, q, r, and it can be shown that this representation is unique. For
non-negative real numbers a, b, c common forms of trivariate inequalities in
the pqr format include:
√ √
(i) a + b + c ≥ 3 3 abc: p ≥ 3 3 r;
(ii) (a + b + c)2 ≥ 3(ab + bc + ca): p2 ≥ 3q;
(iii) a(a − b)(a − c)+b(b − c)(b − a)+c(c − a)(c − b) ≥ 0: p3 −4pq+9r ≥ 0.
This allows us to use the relationships among p, q, r to simplify the prob-
lems.

Example 4.25 (Chinese High School Mathematics League 2014,

Problem 1). Let real numbers a, b, c satisfy a + b + c = 1 and abc > 0.
Prove that
√
abc 1
ab + bc + ca < + .
2 4
Proof. Without loss of generality, assume a ≥ b ≥ c. If a > 0 > b ≥ c,
then
(b + c)2
ab + bc + ca = bc + (1 − b − c)(b + c) ≤ + (b + c) − (b + c)2 < 0,
4
implying the original inequality holds.
Suppose a, b, c are all positive real numbers, since a + b + c = 1,√i.e.,
p = 1, by the pqr form of Schur’s inequality, q ≤ 14 + 94 r, and 94 r < 2r is
4
equivalent to r < 81 .
236 IMO Problems, Theorems, and Methods: Algebra

1
By the pqr form of the AM-GM inequality, r ≤ 27 , and hence
√
1 9 1 r
q≤ + r< + .
4 4 4 2

(13) Local inequality method

For symmetric sum-type inequalities, if it’s challenging to approach the
problem globally, one can start locally, utilizing local properties to complete
the overall proof.

Example 4.26. Let x, y, z, w be positive real numbers. Prove that

x3 + y 3 + z 3 y 3 + z 3 + w3 z 3 + w3 + x3 w3 + x3 + y 3
+ + +
x+y+z y+z+w z+w+x w+x+y
≥ x2 + y 2 + z 2 + w2 .
x3 +y 3 +z 3 x2 +y 2 +z 2
Proof. Consider the local inequality x+y+z ≥ 3 . Indeed, by
the Cauchy–Schwarz inequality,

(x + y + z)(x3 + y 3 + z 3 ) ≥ (x2 + y 2 + z 2 )2
(x + y + z)2
≥ (x2 + y 2 + z 2 ) .
3
Similarly, the remaining three expressions can be derived, and adding these
four inequalities together suffices.

Sometimes the “local” part can be a single term or several terms.

Example 4.27 (Chinese Team Selection Test 2005, Problem 12).

Let a, b, c, d be positive real numbers such that abcd = 1. Prove that
1 1 1 1
2
+ 2
+ 2
+ ≥ 13.
(1 + a) (1 + b) (1 + c) (1 + d)2
1 1 1 1
Proof. Since 1+x + 1+y = 1 when xy = 1, 1 = 1+ab + 1+cd .
1 1 1
Consider the local inequality (1+a)2 + (1+b)2 ≥ 1+ab . It can be shown
that
1 1 1 ab(a − b)2 + (ab − 1)2
+ − = ≥ 0.
(1 + a)2 (1 + b)2 1 + ab (1 + a)2 (1 + b)2 (1 + ab)
1 1 1
Similarly, (1+c) 2 + (1+d)2 ≥ 1+cd , and summing these two inequalities com-
pletes the proof.
 Inequality Problems 237

4.2 Problems and Solutions

4.2.1 Solving inequalities
Problem 4.1 (IMO 2-2, proposed by Hungary). For what values of
the variable x is the following inequality
4x2
√ < 2x + 9
(1 − 1 + 2x)2
valid?

Solution 1. For the inequality to be valid, it is necessary that


1 + 2x ≥ 0,
√
1 − 1 + 2x = 0,

and thus x ≥ − 21 and x = 0. The original inequality is equivalent to

√
4x2 (1 + 1 + 2x)2
√ √ < 2x + 9,
(1 − 1 + 2x)2 (1 + 1 + 2x)2
√
4x2 (1 + 1 + 2x)2
< 2x + 9,
(1 − (1 + 2x))2
√
(1 + 1 + 2x)2 < 2x + 9,
√
which implies 2 1 + 2x < 7, yielding x < 458 . Therefore, the given inequal-
1 45
ity holds when − 2 ≤ x < 0 or 0 < x < 8 .
Solution 2. From

1 + 2x ≥ 0,
√
1 − 1 + 2x = 0,

we have x ≥ − 12 and x = 0. Eliminating the denominator gives

√
4x2 < (2x + 9)(1 − 1 + 2x)2 ,
√
which leads to (2x + 9) 1 + 2x < 11x + 9. Squaring both sides and
simplifying yield x2 (8x − 45) < 0. Solving this inequality, we find x < 45
8 .
Therefore, the given inequality holds when − 21 ≤ x < 0 or 0 < x < 458 .
Score Situation This particular problem saw the following distribution of scores among
contestants: 2 contestants scored 6 points, 1 contestant scored 5 points, no contestant
scored 4 points, 2 contestants scored 3 points, 1 contestant scored 2 points, 1 contestant
238 IMO Problems, Theorems, and Methods: Algebra

scored 1 point, and 3 contestants scored 0 point. The average score of this problem is 2.600,
indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the Czechoslovakia team achieved a total
score of 257 points, the Hungary team achieved a total score of 248 points, the Romania
team achieved a total score of 248 points, the Bulgaria team achieved a total score of 175
points, and the German Democratic Republic team achieved a total score of 38 points.
The gold medal cutoff for this IMO was set at 40 points (with 4 contestants earning gold
medals), the silver medal cutoff was 37 points (with 4 contestants earning silver medals), and
the bronze medal cutoff was 33 points (with 4 contestants earning bronze medals).
In this IMO, no contestant achieved a perfect score of 44 points.

Problem 4.2 (IMO 4-2, proposed by Hungary). Determine all real

numbers x which satisfy the inequality:
√ √ 1
3−x− x+1> .
2

Solution. From

3 − x ≥ 0,
x + 1 ≥ 0,
√ √
we obtain −1 ≤ x ≤ 3. When −1 ≤ x ≤ 3, the value of 3−x− x+1
decreases as x increases. Solve the equation
√ √
3 − x − x + 1 = 12 , (1)
7
√
4 − 2x = x + 1,
64x2 − 128x + 33 = 0,
√ √
yielding x = 8±8 31 . After verification, x = 8−8 31 is a solution to (1). That
√ √ √
is, when x = 8−8 31 , the value of 3 − x − x + 1 equals 12 . Therefore, the
√
solution to the inequality is −1 ≤ x < 8−8 31 .
Score Situation This particular problem saw the following distribution of scores among
contestants: 7 contestants scored 6 points, 3 contestants scored 5 points, 2 contestants
scored 4 points, 3 contestants scored 3 points, 1 contestant scored 2 points, 1 contestant
scored 1 point, and no contestant scored 0 point. The average score of this problem is 4.529,
indicating that it was simple.
Among the top five teams in the team scores, the Hungary team achieved a total score
of 289 points, the Soviet Union team achieved a total score of 263 points, the Romania team
 Inequality Problems 239

achieved a total score of 257 points, the Czechoslovakia team achieved a total score of 212
points, and the Poland team achieved a total score of 212 points.
The gold medal cutoff for this IMO was set at 41 points (with 4 contestants earning gold
medals), the silver medal cutoff was 34 points (with 12 contestants earning silver medals),
and the bronze medal cutoff was 29 points (with 15 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 46 points, namely Iosif
Bernstein from the Soviet Union.

Problem 4.3 (IMO 7-1, proposed by Yugoslavia). Determine all

values x in the interval 0 ≤ x ≤ 2π which satisfy the inequality
√ √ √
2 cos x ≤ | 1 + sin 2x − 1 − sin 2x| ≤ 2.

Solution. It is evident that

√ √
| 1 + sin 2x − 1 − sin 2x|
= | (sin x + cos x)2 − (sin x − cos x)2 |
= ||sin x + cos x| − |sin x − cos x||

2|cos x|, π4 ≤ x ≤ 34 π or 54 π ≤ x ≤ 74 π,
=
2|sin x|, in other cases.

When 34 π < x < 54 π, we obtain 2 cos x < 0 ≤ 2|sin x|. When 0 ≤ x < π
4
7
or 4π< x ≤ 2π, there holds 2 cos x > 2|sin x| ≥ 0.
The right inequality,
√ √ √ √ √
| 1 + sin 2x − 1 − sin 2x| ≤ max{ 1 + sin 2x, 1 − sin 2x} ≤ 2,

π 7π
is obvious. Therefore, the solution to the inequality is 4 ≤x≤ 4 .

Note. This problem is the same as the 6th problem in 1967 British Math-
ematical Olympiad Round 1.

Score Situation This particular problem saw the following distribution of scores among
contestants: 37 contestants scored 4 points, 10 contestants scored 3 points, 14 contestants
scored 2 points, 13 contestants scored 1 point, and 6 contestants scored 0 point. The average
score of this problem is 2.738, indicating that it had a certain level of difficulty.
240 IMO Problems, Theorems, and Methods: Algebra

Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 25 points (with a total team score of 281 points), the Hungary
team scored 29 points (with a total team score of 244 points), the Romania team scored 27
points (with a total team score of 222 points), the Poland team scored 28 points (with a
total team score of 178 points), and the German Democratic Republic team scored 23 points
(with a total team score of 175 points).
The gold medal cutoff for this IMO was set at 38 points (with 8 contestants earning gold
medals), the silver medal cutoff was 30 points (with 12 contestants earning silver medals),
and the bronze medal cutoff was 20 points (with 17 contestants earning bronze medals).
In this IMO, only two contestants achieved a perfect score of 40 points, namely László
Lovász from Hungary and Pavel Bleher from the Soviet Union.

Problem 4.4 (IMO 14-4, proposed by the Netherlands). Find all

solutions (x1 , x2 , x3 , x4 , x5 ) of the system of inequalities

⎧ 2
⎪(x1 − x3 x5 )(x22 − x3 x5 ) ≤ 0,
⎪
⎪ 2
⎪
⎪ 2
⎨(x2 − x4 x1 )(x3 − x4 x1 ) ≤ 0,
⎪
(x23 − x5 x2 )(x24 − x5 x2 ) ≤ 0,
⎪
⎪
⎪
⎪(x24 − x1 x3 )(x25 − x1 x3 ) ≤ 0,
⎪
⎪
⎩ 2
(x5 − x2 x4 )(x21 − x2 x4 ) ≤ 0,

where x1 , x2 , x3 , x4 , x5 are positive real numbers.

Solution 1. Obviously x1 = x2 = x3 = x4 = x5 is a solution to the system,

where xi (i = 1, 2, 3, 4, 5) are positive real numbers. On the other hand, let
x5+i = xi , and then the system of inequalities can be written as

(x2i − xi+2 xi+4 )(x2i+1 − xi+2 xi+4 ) ≤ 0, i ∈ {1, 2, 3, 4, 5}.

Summing up these five inequalities, we get

5

(x2i − xi+2 xi+4 )(x2i+1 − xi+2 xi+4 )
i=1
5

= (x2i x2i+1 − x2i xi+2 xi+4 − x2i+1 xi+2 xi+4 + x2i+2 x2i+4 )
i=1
 Inequality Problems 241

5
1 2
= (x x2 + x2i+1 x2i+2 + x2i x2i+2 + x2i+4 x2i+1
2 i=1 i+4 i

− 2x2i xi+2 xi+4 − 2x2i+1 xi+2 xi+4 )

5
1 2 2
= (x (x − 2xi+2 xi+4 + x2i+4 )
2 i=1 i i+2

+ x2i+1 (x2i+2 − 2xi+2 xi+4 + x2i+4 ))

5
1 2
= (x + x2i+1 )(xi+2 − xi+4 )2
2 i=1 i
≤ 0.

Since (x2i + x2i+1 )(xi+2 − xi+4 )2 ≥ 0 for i ∈ {1, 2, 3, 4, 5}, the above
inequality holds when x1 = x2 = x3 = x4 = x5 .
Therefore, the solution to the system is x1 = x2 = x3 = x4 = x5 .
Solution 2. Let (x1 , x2 , x3 , x4 , x5 ) be a solution to the system of inequal-
ities. Consider the case where

x1 = min{x1 , x2 , x3 , x4 , x5 } and x2 = min{x2 , x3 , x4 , x5 }.

Since x21 ≤ x3 x5 and x22 ≤ x3 x5 , along with (x21 − x3 x5 )(x22 − x3 x5 ) ≤ 0,

we obtain

x1 ≤ x2 = x3 = x5 ≤ x4 .

Therefore, x24 ≥ x1 x3 and x25 ≥ x1 x3 . From (x25 − x2 x4 )(x21 − x2 x4 ) ≤ 0, it

follows that

x1 = x2 = x3 = x4 = x5 . (1)

Similar reasoning applies to other cases, confirming the validity of (1).

On the other hand, for any positive real number x, it is evident that
(x, x, x, x, x) satisfies the given system of inequalities. Therefore, the solu-
tion to the system is x1 = x2 = x3 = x4 = x5 .
Score Situation This particular problem saw the following distribution of scores among
contestants: 21 contestants scored 7 points, no contestant scored 6 points, 2 contestants
scored 5 points, no contestant scored 4 points, no contestant scored 3 points, no contestant
scored 2 points, 5 contestants scored 1 point, and 5 contestants scored 0 point. The average
score of this problem is 4.909, indicating that it was simple.
242 IMO Problems, Theorems, and Methods: Algebra

Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 48 points (with a total team score of 270 points), the Hungary
team scored 52 points (with a total team score of 263 points), the German Democratic
Republic team scored 43 points (with a total team score of 239 points), the Romania team
scored 34 points (with a total team score of 208 points), and the United Kingdom team
scored 30 points (with a total team score of 179 points).
The gold medal cutoff for this IMO was set at 40 points (with 8 contestants earning gold
medals), the silver medal cutoff was 30 points (with 16 contestants earning silver medals),
and the bronze medal cutoff was 19 points (with 30 contestants earning bronze medals).
In this IMO, a total of eight contestants achieved a perfect score of 40 points.

Problem 4.5 (IMO 29-4, proposed by Ireland). Show that the set of
real numbers x which satisfy the inequality

70
 k 5
≥
x−k 4
k=1

is a union of disjoint intervals, the sum of whose lengths is 1988.

1 2 3 70
Solution. Consider the function y = x−1 + x−2 + x−3 + · · · + x−70 and its
5
graph intersecting with the line y = 4 . Let the solutions of the equation

1 2 3 70 5
+ + + ···+ =
x−1 x−2 x−3 x − 70 4

be denoted by x1 < x2 < · · · < x70 . The solution set for the inequality is
given by

S = (1, x1 ) ∪ (2, x2 ) ∪ · · · ∪ (70, x70 ).

Thus, the sum of the lengths of all solution intervals is

L = (x1 − 1) + (x2 − 2) + · · · + (x70 − 70)

= x1 + x2 + · · · + x70 − (1 + 2 + · · · + 70).

1 2 3 70
The graphs of the function y = x−1 + x−2 + x−3 +···+ x−70 and the line
5
y = 4 are illustrated in Figure 4.1.
 Inequality Problems 243

1 2 3 70
Figure 4.1 The Graphs of the Function y = x−1
+ x−2
+ x−3
+···+ x−70
and
5
the Line y = 4

Calculate x1 + x2 + · · · + x70 , and it is evident that the equation

1 2 3 70 5
+ + + ···+ = (1)
x−1 x−2 x−3 x − 70 4
is equivalent to
(x − x1 )(x − x2 ) · · · (x − x70 ) = 0. (2)
Equation (1) can be expressed as
(x − 2)(x − 3) · · · (x − 70) + 2(x − 1)(x − 3) · · · (x − 70) + · · ·

+ 70(x − 1)(x − 2) · · · (x − 69)

5
= (x − 1)(x − 2)(x − 3) · · · (x − 70),
4
which simplifies to x70 − 95 (1 + 2 + · · · + 70)x69 + · · · = 0.
Equation (2) can be expressed as
x70 − (x1 + x2 + · · · + x70 )x69 + · · · = 0.
Therefore,
9
x1 + x2 + · · · + x70 = (1 + 2 + · · · + 70),
5
yielding
9
L = (1 + 2 + · · · + 70) − (1 + 2 + · · · + 70) = 1988.
5
244 IMO Problems, Theorems, and Methods: Algebra

Note. Although the solution to this problem is not complicated, it encom-

passes a considerable range of knowledge, including the Fundamental The-
orem of Algebra and the relationship between roots and coefficients.
For any positive integer m, the total length of the solution set to

1 2 m
+ + ···+ ≥ r (r > 0)
x−1 x−2 x−m

m(m+1)
is 2r .

Score Situation This particular problem saw the following distribution of scores among
contestants: 66 contestants scored 7 points, 6 contestants scored 6 points, 3 contestants
scored 5 points, 3 contestants scored 4 points, 17 contestants scored 3 points, 12 contestants
scored 2 points, 25 contestants scored 1 point, and 136 contestants scored 0 point. The
average score of this problem is 2.332, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 38 points (with a total team score of 217 points), the China
team scored 42 points (with a total team score of 201 points), the Romania team scored
42 points (with a total team score of 201 points), the Germany team scored 31 points (with
a total team score of 174 points), and the Vietnam team scored 40 points (with a total team
score of 166 points).
The gold medal cutoff for this IMO was set at 32 points (with 17 contestants earning gold
medals), the silver medal cutoff was 23 points (with 48 contestants earning silver medals),
and the bronze medal cutoff was 14 points (with 65 contestants earning bronze medals).
In this IMO, a total of five contestants achieved a perfect score of 42 points.

4.2.2 Proving inequalities

Problem 4.6 (IMO 3-2, proposed by Poland). Let a, b, c be the side
lengths of ABC, and S its area. Prove that

√
a2 + b2 + c2 ≥ 4 3S.

In what case does the equality hold?

 Inequality Problems 245

Proof 1. By the law of cosine and the area formulas,

√
a2 + b2 + c2 − 4 3S
√
= 2a2 + 2c2 − 2ac cos B − 2 3ac sin B
  π 
= 2 a2 + c2 − 2ac sin B +
6
≥ 2(a2 + c2 − 2ac)
= 2(a − c)2 ≥ 0.

Thus, the inequality holds, and the equality holds if and only if
a = b = c.

Proof 2. By Heron’s formula, where p is the semi-perimeter of ABC,

the original inequality is equivalent to

(a2 + b2 + c2 )2 ≥ 48p(p − a)(p − b)(p − c)

⇔ (a2 + b2 + c2 )2 ≥ 3(a + b + c)(a + b − c)(b + c − a)(c + a − b)
⇔ (a2 + b2 + c2 )2 ≥ 3((b + c)2 − a2 )(a2 − (b − c)2 )
⇔ (a2 + b2 + c2 )2 ≥ 6(a2 b2 + b2 c2 + c2 a2 ) − 3(a4 + b4 + c4 )
⇔ a4 + b 4 + c4 ≥ a 2 b 2 + b 2 c2 + c2 a 2
⇔ (a2 − b2 )2 + (b2 − c2 )2 + (c2 − a2 )2 ≥ 0.

Thus, the inequality holds, and the equality holds if and only if
a = b = c.

Note. This problem has a variety of solutions, such as using the AM-GM
inequality, or discussing cases based on the size of angle A. Additionally,
this inequality is known as Weitzenböck’s inequality, and it has several
generalizations:

Generalization 1. Let a, b, c be the side lengths of ABC, and S its

area. Then

√
a2 + b2 + c2 ≥ 4 3S + (a − b)2 + (b − c)2 + (c − a)2 .
246 IMO Problems, Theorems, and Methods: Algebra

Proof. Assume without loss of generality that a ≥ b ≥ c. Then

(a − c)2 = ((a − b) + (b − c))2

= (a − b)2 + (b − c)2 + 2(a − b)(b − c)

≥ (a − b)2 + (b − c)2 .

Therefore,
√ √
a2 + b2 + c2 − 4 3S = 2a2 + 2c2 − 2ac cos B − 2 3ac sin B
  π 
= 2 a2 + c2 − 2ac sin B +
6
≥ 2(a2 + c2 − 2ac) = 2(a − c)2

≥ (a − b)2 + (b − c)2 + (c − a)2 .

This inequality is called the Finsler–Hadwiger inequality, named after Paul

Finsler and Hugo Hadwiger. We can also prove that:
√
(i) a2 + b2 + c2 ≥ 4 3S + 2(a − b)2 ;
√ 2
−b2 )2
(ii) a2 + b2 + c2 ≥ 4 3S + (a 2c 2 .

Generalization 2. Let a, b, c be the side lengths of ABC, and S its

area. For α, β, γ ∈ R+ ,

αa2 + βb2 + γc2 ≥ 4S αβ + βγ + γα,

2 2 2
a b c
with equality if and only if β+γ = γ+α = α+β .
This is a weighted generalization of Weitzenböck’s inequality. Setting
α = β = γ yields Weitzenböck’s inequality. The proof requires the following
lemmas:

Lemma 1. 16S 2 = 2(a2 b2 + b2 c2 + c2 a2 ) − (a4 + b4 + c4 ).

Lemma 2. For α, β, γ ∈ R+ and x, y, z ∈ R,

(αx + βy + γz)2 ≥ (αβ + βγ + γα)(2xy + 2yz + 2zx − x2 − y 2 − z 2 ),

x y z
with equality if and only if β+γ = γ+α = α+β .
 Inequality Problems 247

Proof of Lemma 2. The inequality follows from

(αx + βy + γz)2 − (αβ + βγ + γα)(2xy + 2yz + 2zx − x2 − y 2 − z 2 )
= (α + β)(γ + α)x2 + (β + γ)(α + β)y 2
+(γ + α)(β + γ)z 2 − 2γ(α + β)xy
− 2β(γ + α)zx − 2α(β + γ)yz
 2
x y
= (α + β)(β + γ)(γ + α) γ −
β+γ γ+α
 2  2
z x y z
+β − +α −
α+β β+γ γ+α α+β
≥ 0.
Using Lemmas 1 and 2, and setting x = a2 , y = b2 , and z = c2 , we get
Generalization 2.
Generalization 3. Let a, b, c and a , b , c be the side lengths of ABC
and A B  C  , respectively, with areas S and S  . Then
a2 (b2 + c2 − a2 ) + b2 (c2 + a2 − b2 ) + c2 (a2 + b2 − c2 ) ≥ 16SS  ,
with equality if and only if the two triangles are similar.
This inequality, initially found by J. Neuberg in 1897, was rediscovered
and proved by the American geometer D. Pedoe in 1942. It also attracted
many Chinese scholars with its elegant symmetry. Notably, in 1981, Pro-
fessors Lu Yang and Jingzhong Zhang established a similar inequality for
simplices in higher-dimensional spaces.
Furthermore, there are several similar problems:
• (British Mathematical Olympiad 2006, 1st Round, Problem 5).
For positive real numbers a, b, c, prove that
(a2 + b2 )2 ≥ (a + b + c)(a + b − c)(b + c − a)(c + a − b).
• (William Lowell Putnam Mathematical Competition 1998, A6).
Let A, B, C denote distinct points with integer coordinates in R2 . Prove
that if (|AB| + |BC|)2 < 8 · [ABC] + 1, then A, B, C are three vertices
of a square.
Here, |XY | is the length of segment XY and [ABC] is the area of ABC.
248 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 3 contestants scored 7 points, no contestant scored 6 points, 1 contestant
scored 5 points, no contestant scored 4 points, no contestant scored 3 points, 1 contestant
scored 2 points, no contestant scored 1 point, and 4 contestants scored 0 point. The average
score of this problem is 3.111, indicating that it was relatively straightforward.
Among the top five teams in the team scores, the Hungary team achieved a total score of
270 points, the Poland team achieved a total score of 203 points, the Romania team achieved
a total score of 197 points, the Czechoslovakia team achieved a total score of 159 points,
and the German Democratic Republic team achieved a total score of 146 points.
The gold medal cutoff for this IMO was set at 37 points (with 3 contestants earning gold
medals), the silver medal cutoff was 34 points (with 4 contestants earning silver medals), and
the bronze medal cutoff was 30 points (with 4 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Béla
Bollobás from Hungary.

Problem 4.7 (IMO 6-2, proposed by Hungary). Suppose a, b, c are

the side lengths of ABC. Prove that

a2 (b + c − a) + b2 (c + a − b) + c2 (a + b − c) ≤ 3abc.

Proof 1. Let a = y + z, b = z + x, and c = x + y, where x, y, z > 0. The

original inequality is equivalent to

2x(y + z)2 + 2y(z + x)2 + 2z(x + y)2 ≤ 3(x + y)(y + z)(z + x),
6xyz ≤ x2 y + y 2 z + z 2 x + x2 z + y 2 x + z 2 y. (1)

By the AM-GM inequality,

x2 y + y 2 z + z 2 x + x2 z + y 2 x + z 2 y
≥ 6 6 x2 y · y 2 z · z 2 x · x2 z · y 2 x · z 2 y
= 6xyz,

which proves (1).

Thus, the original inequality holds, and the equality is valid if and only
if a = b = c.
 Inequality Problems 249

Proof 2. Since a, b, c are the side lengths of a triangle,

(a − b)2 (a + b − c) ≥ 0,
(b − c)2 (b + c − a) ≥ 0,
(c − a)2 (c + a − b) ≥ 0.
Summing these three inequalities, we obtain
a2 ((a + b − c) + (c + a − b) − 2b − 2c)
+ b2 ((a + b − c) + (b + c − a) − 2a − 2c)
+ c2 ((b + c − a) + (c + a − b) − 2b − 2a) + 6abc ≥ 0,
which simplifies to
−2a2 (b + c − a) − 2b2 (c + a − b) − 2c2 (a + b − c) + 6abc ≥ 0.
Therefore, a2 (b + c − a) + b2 (c + a − b) + c2 (a + b − c) ≤ 3abc, and the
equality holds if and only if a = b = c.
Note. The substitution used in Proof 1 is often called the incircle substi-
tution. As illustrated in Figure 4.2, consider the incircle of ABC, with
D, E, F being the tangency points on the three sides, respectively. Let
AE = AF = x, BD = BF = y, and CD = CE = z. Then a = y + z,
b = z + x, and c = x + y.

A
x
x
E
F
z
y
B y z C
D

Figure 4.2 The Incircle of ABC

This substitution is very useful in proving inequalities related to the

side lengths of a triangle.
Similar to Proof 1, this problem can also be approached by using
1 √ √ √
(x + y)(y + z)(z + x) ≥ xy · yz · zx = xyz,
8
which implies abc ≥ (b + c − a)(c + a − b)(a + b − c). Expanding this leads
to the inequality to be proved.
250 IMO Problems, Theorems, and Methods: Algebra

Furthermore, there are several similar problems:

• (Austrian–Polish Mathematical Competition 2001, Problem 3).

Let a, b, c be side lengths of a triangle. Prove that

a + b b + c c + a a 3 + b 3 + c3
2< + + − ≤ 3.
c a b abc

• (Swedish Mathematical Competition 1982, Problem 2). Show

that abc ≥ (a + b − c)(b + c − a)(c + a − b) for positive reals a, b, c.
• (Proposed by Adrian Andreescu). Prove that for any non-negative
real numbers a, b, c,

(a − 2b + 4c)(c − 2a + 4b)(b − 2c + 4a) ≤ 27abc.

Score Situation This particular problem saw the following distribution of scores among
contestants: 7 contestants scored 7 points, no contestant scored 6 points, 1 contestant
scored 5 points, no contestant scored 4 points, no contestant scored 3 points, no contestant
scored 2 points, 2 contestants scored 1 point, and 7 contestants scored 0 point. The average
score of this problem is 3.294, indicating that it was relatively straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 54 points (with a total team score of 269 points), the Hungary
team scored 37 points (with a total team score of 253 points), the Romania team scored
35 points (with a total team score of 213 points), the Poland team scored 33 points (with a
total team score of 209 points), and the Bulgaria team scored 29 points (with a total team
score of 198 points).
The gold medal cutoff for this IMO was set at 38 points (with 7 contestants earning gold
medals), the silver medal cutoff was 31 points (with 9 contestants earning silver medals), and
the bronze medal cutoff was 27 points (with 19 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 42 points, namely David
Bernstein from the Soviet Union.

Problem 4.8 (IMO 11-6, proposed by the Soviet Union). Prove

that for all real numbers x1 , x2 , y1 , y2 , z1 , z2 , with x1 > 0, x2 > 0,
x1 y1 − z12 > 0, and x2 y2 − z22 > 0, the inequality

8 1 1
≤ 2 +
(x1 + x2 )(y1 + y2 ) − (z1 + z2 )2 x1 y1 − z1 x2 y2 − z22

is satisfied. Give necessary and sufficient conditions for equality.

 Inequality Problems 251

Proof 1. From x1 y1 > z12 ≥ 0 and x1 > 0, it follows that y1 > 0. Similarly,
y2 > 0. Therefore,
√
x1 y1 z22 + x2 y2 z12 ≥ 2z1 z2 x1 x2 y1 y2 ,
√ √
( x1 x2 y1 y2 − z1 z2 )2 = x1 x2 y1 y2 − 2z1 z2 x1 x2 y1 y2 + z12 z22

≥ x1 x2 y1 y2 + z12 z22 − (x1 y1 z22 + x2 y2 z12 )

= (x1 y1 − z12 )(x2 y2 − z22 ). (1)

Also

(x1 + x2 )(y1 + y2 ) − (z1 + z2 )2

= (x1 y1 − z12 ) + (x2 y2 − z22 ) + (x1 y2 + x2 y1 − 2z1 z2 )

√
≥ (x1 y1 − z12 ) + (x2 y2 − z22 ) + 2( x1 x2 y1 y2 − z1 z2 )

≥ (x1 y1 − z12 ) + (x2 y2 − z22 ) + 2 (x1 y1 − z12 )(x2 y2 − z22 ) (Using (1))

≥4 (x1 y1 − z12 )(x2 y2 − z22 ),

which implies
8 2
≤
(x1 + x2 )(y1 + y2 ) − (z1 + z2 )2 (x1 y1 − z12 )(x2 y2 − z22 )
1 1
≤ 2 + .
x1 y1 − z1 x2 y2 − z22
The equality holds if and only if x1 = x2 , y1 = y2 , and z1 = z2 .
Proof 2. Let A1 = x1 y1 − z12 and A2 = x2 y2 − z22 , where A1 , A2 > 0. The
original inequality becomes
 
1 1
+ · (A1 + A2 + x1 y2 + x2 y1 − 2z1 z2 ) ≥ 8.
A1 A2
Also
 
1 1
+ · (A1 + A2 + x1 y2 + x2 y1 − 2z1 z2 )
A1 A2
 
1 1 √
≥ + · (A1 + A2 + 2 x1 y2 x2 y1 − 2z1 z2 )
A1 A2
252 IMO Problems, Theorems, and Methods: Algebra

  
1 1
= + · (A1 + A2 + 2 (A1 + z12 )(A2 + z22 ) − 2z1 z2 )
A1 A2
 
1 1
≥ + · (A1 + A2 + 2( A1 A2 + z1 z2 ) − 2z1 z2 )
A1 A2
(using the Cauchy–Schwarz inequality)
 
1 1
= + · (A1 + A2 + 2 A1 A2 )
A1 A2
 
1 1 A1 + A2
= + · (A1 + A2 ) + 2 √
A1 A2 A1 A2
  2 √
1 1 2 A1 A2
≥ A1 + A2 + 2 √
A1 A2 A1 A2
= 4 + 2 · 2 = 8.
Hence, the original inequality holds. The equality holds if and only if x1 =
x2 , y1 = y2 , and z1 = z2 .
Note. The A1 and A2 in the problem can be represented as the determi-
nants of the following matrices M 1 and M 2 respectively, namely
   
x1 z1 x2 z2
M1 = and M 2 = ,
z1 y1 z2 y2
and the inequality to be proved can be written as
8 1 1
≤ + .
det(M 1 + M 2 ) det M 1 det M 2
If M 1 and M 2 are n×n positive definite matrices, then the above inequality
still holds.
Furthermore, there are several similar problems:
• (Austrian–Polish Mathematical Competition 1998, Problem 1).
Let x1 , x2 , y1 , y2 be real numbers such that x21 + x22 ≤ 1. Prove that
(x1 y1 + x2 y2 − 1)2 ≥ (x21 + x22 − 1)(y12 + y22 − 1).
• (From the Chinese journal High-School Mathematics, 1992(03):
26–27). Let a1 , a2 , b1 , b2 , c1 , c2 be real numbers such that a1 a2 > 0,
a1 c1 ≥ b21 , and a2 c2 ≥ b22 . Prove that

(a1 + a2 )(c1 + c2 ) ≥ (b1 + b2 )2 + 2 (a1 c1 − b21 )(a2 c2 − b22 ).
 Inequality Problems 253

• (Eötvös Mathematics Competition 1939, Problem 1). Let

a1 , a2 , b1 , b2 , c1 , c2 be real numbers such that a1 a2 > 0, a1 c1 ≥ b21 , and
a2 c2 ≥ b22 . Prove that

(a1 + a2 )(c1 + c2 ) ≥ (b1 + b2 )2 .

Score Situation This particular problem saw the following distribution of scores among con-
testants: 7 contestants scored 8 points, 2 contestants scored 7 points, 3 contestants scored
6 points, 9 contestants scored 5 points, 13 contestants scored 4 points, 13 contestants scored
3 points, 17 contestants scored 2 points, 15 contestants scored 1 point, and 33 contestants
scored 0 point. The average score of this problem is 2.438, indicating that it had a certain
level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Hungary team scored 30 points (with a total team score of 247 points), the German
Democratic Republic team scored 29 points (with a total team score of 240 points), the
Soviet Union team scored 29 points (with a total team score of 231 points), the Romania
team scored 33 points (with a total team score of 219 points), and the United Kingdom team
scored 23 points (with a total team score of 193 points).
The gold medal cutoff for this IMO was set at 40 points (with 3 contestants earning gold
medals), the silver medal cutoff was 30 points (with 20 contestants earning silver medals),
and the bronze medal cutoff was 24 points (with 21 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 40 points, namely Tibor
Fiala from Hungary, Vladimir Drinfeld from the Soviet Union, and Simon Phillips Norton from
the United Kingdom.

Problem 4.9 (IMO 13-1, proposed by Hungary). Prove that the

following assertion is true for n = 3 and n = 5, and that it is false for
every other natural number n > 2:
If a1 , a2 , . . . , an are arbitrary real numbers, then

(a1 − a2 )(a1 − a3 ) · · · (a1 − an ) + (a2 − a1 )(a2 − a3 ) · · · (a2 − an )

+ · · · + (an − a1 )(an − a2 ) · · · (an − an−1 ) ≥ 0.

Proof. Without loss of generality, assume a1 ≤ a2 ≤ · · · ≤ an . Let An

denote the given expression.
If n is even, let a1 < a2 = a3 = · · · = an , then

An = (−1)n−1 (a2 − a1 )n−1 < 0,

and therefore, the inequality does not hold.

254 IMO Problems, Theorems, and Methods: Algebra

If n = 3, then

A3 = (a1 − a2 )(a1 − a3 ) + (a2 − a1 )(a2 − a3 ) + (a3 − a1 )(a3 − a2 )

= a21 + a22 + a23 − a1 a2 − a2 a3 − a3 a1

1
= ((a1 − a2 )2 + (a2 − a3 )2 + (a3 − a1 )2 ) ≥ 0.
2

If n = 5, then

A5 = (a1 − a2 )(a1 − a3 )(a1 − a4 )(a1 − a5 )

+ (a2 − a1 )(a2 − a3 )(a2 − a4 )(a2 − a5 )

+ (a3 − a1 )(a3 − a2 )(a3 − a4 )(a3 − a5 )

+ (a4 − a1 )(a4 − a2 )(a4 − a3 )(a4 − a5 )

+ (a5 − a1 )(a5 − a2 )(a5 − a3 )(a5 − a4 ),

where the sum of the first two terms is

(a2 − a1 )((a3 − a1 )(a4 − a1 )(a5 − a1 ) − (a3 − a2 )(a4 − a2 )(a5 − a2 )) ≥ 0,

and similarly, the sum of the last two terms is also non-negative.
Additionally, the third term is non-negative, so A5 ≥ 0.
If n is odd and n ≥ 7, let a1 = a2 = a3 < a4 < a5 = a6 = · · · = an , then

An = (a4 − a1 )(a4 − a2 ) · · · (a4 − an ) < 0.

In conclusion, An ≥ 0 for any real numbers a1 , a2 , . . . , an if and only if

n = 3 or n = 5.

Note. The homogeneous polynomial of degree 4,

A5 = (a1 − aj ) + (a2 − aj ) + (a3 − aj ) + (a4 − aj )

j=1 j=2 j=3 j=4

+ (a5 − aj ),
j=5
 Inequality Problems 255

cannot be written as a sum of squares of quadratic forms in a1 , a2 , a3 , a4 , a5 ,

and we can prove it by the following lemma.

Lemma. Let Q be a quadratic form in a1 , a2 , a3 , a4 , a5 , which is zero when-

ever the ai satisfy

a1 = a2 , a3 = a4 = a5

or some permutation. Then Q = 0 for all values of the ai .

In fact, the general theory of these forms goes back to Hilbert, who
showed that every non-negative quartic form in three variables can be writ-
ten as a sum of squares, but gave an example of a quartic form in four
variables which cannot be so represented. In his famous list of 23 prob-
lems, Hilbert posed the following problem:
Suppose that f ∈ R(x1 , x2 , . . . , xn ) is non-negative at all points of Rn
where f is defined. Is f a finite sum of squares of rational functions?
This is indeed always possible and was proved by Emil Artin in 1927.
Furthermore, there is another problem related to Hilbert’s 17th
Problem:

• (Chinese Team Selection Test 2017, Problem 29). For any

x, y, z ∈ R, a three-variable real-coefficient polynomial

f (x, y, z) = f2 (x, y)z 2 + 2f3 (x, y)z + f4 (x, y) ≥ 0,

where fk (x, y) are homogeneous polynomials of degree k(k = 2, 3, 4). If

there is a real-coefficient polynomial r(x, y) such that

f2 (x, y)f4 (x, y) − f32 (x, y) = (r(x, y))2 ,

prove that there must be two real-coefficient polynomials g(x, y, z) and

h(x, y, z) such that

f (x, y, z) = g 2 (x, y, z) + h2 (x, y, z).

Score Situation This particular problem saw the following distribution of scores among con-
testants: 4 contestants scored 5 points, 2 contestants scored 4 points, 3 contestants scored
3 points, 11 contestants scored 2 points, 6 contestants scored 1 point, and 2 contestants
scored 0 point. The average score of this problem is 2.321, indicating that it had a certain
level of difficulty.
256 IMO Problems, Theorems, and Methods: Algebra

Among the top five teams in the team scores, the scores of this problem are as follows:
the Hungary team scored 35 points (with a total team score of 255 points), the Soviet Union
team scored 34 points (with a total team score of 205 points), the German Democratic
Republic team scored 20 points (with a total team score of 142 points), the Poland team
scored 29 points (with a total team score of 118 points), the Romania team scored 22 points
(with a total team score of 110 points), and the United Kingdom team scored 21 points (with
a total team score of 110 points).
The gold medal cutoff for this IMO was set at 35 points (with 7 contestants earning gold
medals), the silver medal cutoff was 23 points (with 12 contestants earning silver medals),
and the bronze medal cutoff was 11 points (with 29 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 42 points, namely Imre
Ruzsa from Hungary.

Problem 4.10 (IMO 17-1, proposed by Czechoslovakia). Let xi ,

yi (i = 1, 2, . . . , n) be real numbers such that

x1 ≥ x2 ≥ · · · ≥ xn and y1 ≥ y2 ≥ · · · ≥ yn .

Prove that, if z1 , z2 , . . . , zn is any permutation of y1 , y2 , . . . , yn , then

n
 n

(xi − yi )2 ≤ (xi − zi )2 .
i=1 i=1

Proof. The original inequality is equivalent to

n
 n
 n
 n
 n
 n

x2i −2 xi yi + yi2 ≤ x2i −2 xi zi + zi2
i=1 i=1 i=1 i=1 i=1 i=1
n
 n
 n
 n

⇔ xi zi ≤ xi yi yi2 = zi2 ,
i=1 i=1 i=1 i=1

which is true according to the “mixed order sum” and “same order sum”
from the rearrangement inequality. Therefore, the original inequality is
valid.

Score Situation This particular problem saw the following distribution of scores among
contestants: 76 contestants scored 6 points, 12 contestants scored 5 points, 11 contestants
scored 4 points, 5 contestants scored 3 points, 9 contestants scored 2 points, 12 contestants
 Inequality Problems 257

scored 1 point, and 3 contestants scored 0 point. The average score of this problem is 4.727,
indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Hungary team scored 47 points (with a total team score of 258 points), the German
Democratic Republic team scored 46 points (with a total team score of 249 points), the
United States team scored 43 points (with a total team score of 247 points), the Soviet Union
team scored 48 points (with a total team score of 246 points), and the United Kingdom team
scored 45 points (with a total team score of 241 points).
The gold medal cutoff for this IMO was set at 38 points (with 8 contestants earning gold
medals), the silver medal cutoff was 32 points (with 25 contestants earning silver medals),
and the bronze medal cutoff was 23 points (with 36 contestants earning bronze medals).
In this IMO, a total of six contestants achieved a perfect score of 40 points.

Problem 4.11 (IMO 19-4, proposed by the United Kingdom).

Four real constants a, b, A, B are given, and

f (θ) = 1 − a cos θ − b sin θ − A cos 2θ − B sin 2θ.

Prove that if f (θ) ≥ 0 for all real θ, then

a2 + b 2 ≤ 2 and A2 + B 2 ≤ 1.

Proof. Since for any real number θ,

f (θ) = 1 − a cos θ − b sin θ − A cos 2θ − B sin 2θ ≥ 0, (1)

so it follows that
 π
f θ+ = 1 + a sin θ − b cos θ + A cos 2θ + B sin 2θ ≥ 0, (2)
2
f (θ + π) = 1 + a cos θ + b sin θ − A cos 2θ − B sin 2θ ≥ 0. (3)

Adding (1) and (2),

2 + (a − b) sin θ − (a + b) cos θ ≥ 0,
2+ (a − b)2 + (a + b)2 sin(θ − ϕ1 ) ≥ 0,
258 IMO Problems, Theorems, and Methods: Algebra

a−b a+b
where cos ϕ1 = √ and sin ϕ1 = √ . Therefore,
(a−b)2 +(a+b)2 (a−b)2 +(a+b)2

2− (a − b)2 + (a + b)2 ≥ 0,

which implies a2 + b2 ≤ 2.
By (1) and (3),

1 − A cos 2θ − B sin 2θ ≥ 0,
1− A2 + B 2 cos(2θ − ϕ2 ) ≥ 0,

where cos ϕ2 = √A2A+B 2 and sin ϕ2 = √ B

A2 +B 2
. Consequently, 1 −
√
A2 + B 2 ≥ 0, which implies A2 + B 2 ≤ 1.

Note. The expression of f (θ) can also be transformed into

f (θ) = 1 − r cos(θ − α) − R cos 2(θ − β),

√ √
where r = a2 + b2 and R = A2 + B 2 . Alternatively, one can employ
proof by contradiction to solve this problem.
In fact, the problem is a special case of the theorem: If
n

f (θ) = 1 + (aj cos jθ + bj sin jθ) ≥ 0,
j=1

then a2j + b2j ≤ 2 for j = 1, 2, . . . , n − 1 and a2n + b2n ≤ 1.

This is an important property of Fourier series of positive functions.
Score Situation This particular problem saw the following distribution of scores among
contestants: 16 contestants scored 6 points, 1 contestant scored 5 points, 1 contestant
scored 4 points, 1 contestant scored 3 points, 2 contestants scored 2 points, 7 contestants
scored 1 point, and 9 contestants scored 0 point. The average score of this problem is 3.216,
indicating that it was relatively straightforward.
Among the top five teams in the team scores, the United States team achieved a total
score of 202 points, the Soviet Union team achieved a total score of 192 points, the Hungary
team achieved a total score of 190 points, the United Kingdom team achieved a total score
of 190 points, and the Netherlands team achieved a total score of 185 points.
The gold medal cutoff for this IMO was set at 34 points (with 13 contestants earning gold
medals), the silver medal cutoff was 24 points (with 29 contestants earning silver medals),
and the bronze medal cutoff was 17 points (with 35 contestants earning bronze medals).
 Inequality Problems 259

In this IMO, a total of five contestants achieved a perfect score of 40 points.

Problem 4.12 (IMO 20-5, proposed by France).

Let {ak }(k = 1, 2, 3, . . . , n, . . .) be a sequence of distinct positive integers.
Prove that for all positive integers n,

a2 a3 an 1 1 1
a1 + + + ···+ 2 ≥ 1 + + + ···+ .
4 9 n 2 3 n

Proof 1. For any positive integer n, let b1 , b2 , . . . , bn be a permutation

of a1 , a2 , . . . , an such that b1 < b2 < · · · < bn , implying bi ≥ i for i =
1, 2, . . . , n.
Since b1 , b2 , . . . , bn and 1, 212 , . . . , n12 are in reverse order, by the rear-
rangement inequality,

n
 n
 n
 n

ak bk k 1
≥ ≥ = .
k2 k 2 k 2 k
k=1 k=1 k=1 k=1

Proof 2. Since the positive integers a1 , a2 , . . . , ak are distinct,

a1 + a2 + · · · + ak ≥ 1 + 2 + · · · + k.

Consequently,

a1 − 1 a2 − 2 an − n
2
+ 2
+ ··· +
1 2 n2
a1 + a2 − (1 + 2) a3 − 3 an − n
≥ 2
+ 2
+ ···+
2 3 n2
a1 + a2 + a3 − (1 + 2 + 3) a4 − 4 an − n
≥ 2
+ 2
+ ···+
3 4 n2
........................................................................
a1 + a2 + · · · + an − (1 + 2 + · · · + n)
≥ ≥ 0,
n2
a2 a3 an 1 1
and thus, a1 + 4 + 9 + ···+ n2 ≥1+ 2 + 3 + · · · + n1 .
260 IMO Problems, Theorems, and Methods: Algebra

Note. There are several similar problems:

• (Japan Team Selection Test 2022, Problem 4; IMO Shortlist

2021, Algebra, Problem 3). Given a positive integer n, find the small-
est value of
a  a  a 
1 2 n
+ + ···+
1 2 n
over all permutations (a1 , a2 , . . . , an ) of (1, 2, . . . , n).
• (Polish Mathematical Olympiad Finals 2008, Problem 1). The
numbers 1, 2, . . . , n2 are arranged in the cells of an n × n board in such
a way that the numbers 1, 2, . . . , n are in the first row (in this order),
n + 1, n + 2, . . . , 2n in the second, etc. We choose n cells of the board,
no two of which are in the same row or column. Let ai be the chosen
number in the ith row. Prove that
12 22 n2 n+2 1
+ + ···+ ≥ − 2 .
a1 a2 an 2 n +1
• (Polish Mathematical Olympiad Finals 2007, Problem 6). A
sequence of real numbers a0 , a1 , a2 , . . . is defined by a0 = −1 and
an−1 an−2 a1 a0
an + + + ···+ + = 0 for n ≥ 1.
2 3 n n+1
Show that an > 0 for n ≥ 1.
• (Polish Mathematical Olympiad Finals 1995, Problem 4). Let n
be a positive integer. Find the smallest possible value of
x22 x3 xn
x1 + + 3 + ···+ n,
2 3 n
where x1 , x2 , . . . , xn are positive reals such that x11 + x12 + · · · + x1n = n.
• (All Soviet Union Mathematical Olympiad 1979). For every n, a
decreasing sequence {xk } satisfies a condition
x4 x9 xn2
x1 + + + ···+ ≤ 1.
2 3 n
x2 x3 xn
Prove that for every n, it also satisfies x1 + 2 + 3 + ···+ n ≤ 3.

Score Situation This particular problem saw the following distribution of scores among
contestants: 39 contestants scored 6 points, 1 contestant scored 5 points, no contestant
scored 4 points, 2 contestants scored 3 points, no contestant scored 2 points, 3 contestants
scored 1 point, and 3 contestants scored 0 point. The average score of this problem is 5.167,
indicating that it was simple.
 Inequality Problems 261

Among the top five teams in the team scores, the scores of this problem are as follows:
the Romania team scored 48 points (with a total team score of 237 points), the United States
team scored 48 points (with a total team score of 225 points), the United Kingdom team
scored 45 points (with a total team score of 201 points), the Vietnam team scored 45 points
(with a total team score of 200 points), and the Czechoslovakia team scored 45 points (with
a total team score of 195 points).
The gold medal cutoff for this IMO was set at 35 points (with 5 contestants earning gold
medals), the silver medal cutoff was 27 points (with 20 contestants earning silver medals),
and the bronze medal cutoff was 22 points (with 38 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 40 points, namely Mark
Kleiman from the United States.

Problem 4.13 (IMO 24-6, proposed by the United States). Let a,

b, and c be the lengths of the sides of a triangle. Prove that

a2 b(a − b) + b2 c(b − c) + c2 a(c − a) ≥ 0.

Determine when the equality occurs.

Proof 1. Let a = y + z, b = z + x, and c = x + y, where x, y, z > 0. The

original inequality then becomes:

The left-hand side

= (y + z)2 (z + x)(y − x) + (z + x)2 (x+y)(z − y)
+ (x + y)2 (y + z)(x − z)
= 2(xy 3 + yz 3 + zx3 − x2 yz − xy 2 z − xyz 2 )
= 2(xy(y − z)2 + yz(z − x)2 + zx(x − y)2 )
≥ 0,

with the equality holding when x = y = z.

Proof 2. Without loss of generality, assume a = max{a, b, c}. The original

inequality then becomes:

The left-hand side = a(b − c)2 (b + c − a) + b(a − b)(a − c)(a + b − c) ≥ 0.

Note. In this problem, the inequality to be proven is not symmetric, so

one cannot assume a ≥ b ≥ c without loss of generality. Additionally, in
262 IMO Problems, Theorems, and Methods: Algebra

Proof 1, the Cauchy–Schwarz inequality can also be employed, as shown

below:

(xy 3 + yz 3 + zx3 )(z + x + y)

√ √ √
≥ (y xyz + z xyz + x xyz)2 = xyz(x + y + z)2 ,

which might be more straightforward than completing the square.

Furthermore, there is a similar problem:

• (From Oles Dobosevych’s paper “On a Method of Proving Sym-

metric Inequalities,” 2009: 3). Prove that

a3 b + b3 c + c3 a ≥ abc(a + b + c)

for all non-negative real numbers a, b, c.

Score Situation This particular problem saw the following distribution of scores among
contestants: 6 contestants scored 7 points, no contestant scored 6 points, 1 contestant
scored 5 points, 7 contestants scored 4 points, 2 contestants scored 3 points, no contestant
scored 2 points, 4 contestants scored 1 point, and 20 contestants scored 0 point. The average
score of this problem is 2.125, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the Germany team achieved a total score
of 212 points, the United States team achieved a total score of 171 points, the Hungary team
achieved a total score of 170 points, the Soviet Union team achieved a total score of 169
points, and the Romania team achieved a total score of 161 points.
The gold medal cutoff for this IMO was set at 38 points (with 9 contestants earning gold
medals), the silver medal cutoff was 26 points (with 27 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 57 contestants earning bronze medals).
In this IMO, a total of four contestants achieved a perfect score of 42 points.

Problem 4.14 (IMO 25-1, proposed by Germany). Prove that 0 ≤

7
yz + zx + xy − 2xyz ≤ 27 , where x, y, and z are non-negative real numbers
for which x + y + z = 1.

Proof 1. Without loss of generality, assume x ≥ y ≥ z. Then z ≤ 13 , and

we have

xy + yz + zx − 2xyz = xy(1 − 2z) + yz + zx ≥ 0,

with equality when x = 1 and y = z = 0.

 Inequality Problems 263

1
Let z = 3 − k, where 0 ≤ k ≤ 13 . Hence x + y = 2
3 + k. Further,
xy + yz + zx − 2xyz = xy(1 − 2z) + z(x + y)
    
1 1 2
= xy + 2k + −k +k
3 3 3
    
(x + y)2 1 1 2
≤ + 2k + −k +k
4 3 3 3
 2     
1 k 1 1 2
= + + 2k + −k +k
3 2 3 3 3
1 3 1 2 7
= k − k +
2 4 27
 
1 2 1 7 7
= k k− + ≤ ,
2 2 27 27
where the equality holds when k = 0 and x = y, i.e., when x = y = z = 13 .
Proof 2. Since 0 ≤ x, y, z ≤ 1,
xy + yz + zx − 2xyz ≥ xy + yz + zx − yz − zx = xy ≥ 0.
Also,
(1 − 2x)(1 − 2y)(1 − 2z) = 1 − 2(x + y + z) + 4(xy + yz + zx) − 8xyz
= −1 + 4(xy + yz + zx − 2xyz).
When 1 − 2x, 1 − 2y, and 1 − 2z are all non-negative real numbers,
 3
(1 − 2x) + (1 − 2y) + (1 − 2z) 1
(1 − 2x)(1 − 2y)(1 − 2z) ≤ = . (1)
3 27
Since x + y + z = 1, at most one of 1 − 2x, 1 − 2y, and 1 − 2z is negative.
In this case, inequality (1) still holds. Therefore,
1
−1 + 4(xy + yz + zx − 2xyz) ≤ ,
27
7
implying xy + yz + zx − 2xyz ≤ 27 .

Note. The original inequality can also be proven by using the adjustment
method or pqr method. Furthermore, if xi (i = 1, 2, . . . , n) are non-negative
real numbers satisfying x1 + x2 + · · · + xn = 1, then
    
n−1 1 1 1
0≤ xi xj − xi xj xk ≤ 1− 2+ .
n−2 6 n n
1≤i<j≤n 1≤i<j<k≤n
264 IMO Problems, Theorems, and Methods: Algebra

Furthermore, there are several similar problems:

• (From the book Problems and Solutions in Mathematical

Olympiad High School 2 , 2022: 31). Let x, y, z be non-negative real
numbers such that x + y + z = 1. Prove that
1
≤ x2 + y 2 + z 2 + 6xyz ≤ 1.
2
• (Chinese Team Selection Test 2019, Problem 13). Let x, y, z be
complex numbers satisfying |x|2 + |y|2 + |z|2 = 1. Prove that

|x3 + y 3 + z 3 − 3xyz| ≤ 1.

• (British Mathematical Olympiad 2010, 2nd Round, Problem 4).

Prove that for all positive real numbers x, y, and z,

4(x + y + z)3 > 27(x2 y + y 2 z + z 2 x).

• (From Mildorf ’s paper “Olympiad Inequalities,” 2005: 20). Let

0 ≤ a, b, c ≤ 12 be real numbers with a + b + c = 1. Show that

9
a3 + b3 + c3 + 4abc ≤ .
32
• (British Mathematical Olympiad 2002, 1st Round, Problem 3).
Let x, y, z be positive real numbers such that x2 + y 2 + z 2 = 1. Prove
that
1
x2 yz + xy 2 z + xyz 2 ≤ .
3
• (United States of America Mathematical Olympiad 2001, Prob-
lem 3). Let a, b, c be non-negative real numbers such that a2 + b2 + c2 +
abc = 4. Show that

0 ≤ ab + bc + ca − abc ≤ 2.

• (Canadian Mathematical Olympiad 1999, Problem 5). Let x, y,

and z be non-negative real numbers satisfying x + y + z = 1. Show that
4
x2 y + y 2 z + z 2 x ≤ ,
27
and find when the equality occurs.
 Inequality Problems 265

• (British Mathematical Olympiad 1999, 2nd Round, Problem 3).

Non-negative real numbers p, q, and r satisfy p + q + r = 1. Prove that

7(pq + qr + rp) ≤ 2 + 9pqr.

• (All Soviet Union Mathematical Olympiad 1989). Prove that if

a, b, c are the lengths of the sides of a triangle and a + b + c = 1, then the
following inequality holds:
1
a2 + b2 + c2 + 4abc < .
2

Score Situation This particular problem saw the following distribution of scores among
contestants: 86 contestants scored 7 points, 10 contestants scored 6 points, 7 contestants
scored 5 points, 9 contestants scored 4 points, 23 contestants scored 3 points, 35 contestants
scored 2 points, 3 contestants scored 1 point, and 19 contestants scored 0 point. The average
score of this problem is 4.557, indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 42 points (with a total team score of 235 points), the Bulgaria
team scored 42 points (with a total team score of 203 points), the Romania team scored 42
points (with a total team score of 199 points), the Hungary team scored 37 points (with a
total team score of 195 points), and the United States team scored 42 points (with a total
team score of 195 points).
The gold medal cutoff for this IMO was set at 40 points (with 14 contestants earning gold
medals), the silver medal cutoff was 26 points (with 35 contestants earning silver medals),
and the bronze medal cutoff was 17 points (with 49 contestants earning bronze medals).
In this IMO, a total of eight contestants achieved a perfect score of 42 points.

Problem 4.15 (IMO 28-3, proposed by Germany). Let

x1 , x2 , . . . , xn be real numbers satisfying x21 + x22 + · · · + x2n = 1. Prove
that for every integer k ≥ 2 there are integers a1 , a2 , . . . , an , not all 0, such
that |ai | ≤ k − 1 for all i and
√
(k − 1) n
|a1 x1 + a2 x2 + · · · + an xn | ≤ .
kn − 1
Proof. Since changing the sign of xi does not affect the condition x21 +
x22 + · · · + x2n = 1, and simultaneously changing the signs of both ai and
xi does not affect the sum a1 x1 + a2 x2 + · · · + an xn , we can assume that
x1 , x2 , . . . , xn are all non-negative real numbers.
Consider the number e1 x1 + e2 x2 + · · · + en xn , where ei ∈
{0, 1, 2, . . . , k − 1}. If there exist two different arrays (e1 , e2 , . . . , en ) and
266 IMO Problems, Theorems, and Methods: Algebra

n  n 
(e1 , e2 , . . . , en ) such that i=1 ei xi =
 
i=1 ei xi , let ai = ei − ei for
i = 1, 2, . . . , n. Then ai are integers, not all zero, and |ai | ≤ k − 1, yielding
√
(k − 1) n
|a1 x1 + a2 x2 + · · · + an xn | = 0 < .
kn − 1
Thus, the original inequality holds.
If e1 x1 + e2 x2 + · · · + en xn are all distinct, then the set
 n 

A= ei xi |ei ∈ {0, 1, 2, . . . , k − 1}
i=1
n
contains k distinct numbers. By the Cauchy–Schwarz inequality,

n n  n
   √ √
0≤ ei xi ≤ (k − 1) xi ≤ (k − 1) x2i · n = (k − 1) n,
i=1 i=1 i=1

which implies that all k n numbers in A fall within the interval

√
[0, (k − 1) n].
Dividing
√ this interval into k n − 1 smaller intervals, each with a length
(k−1) n
of kn −1 , we use the Pigeonhole Principle, implying that there must be
n n
two distinct numbers i=1 ei xi and i=1 ei xi in the same smaller interval
(including endpoints).
Let ai = ei − ei for i = 1, 2, . . . , n. Then |ai | ≤ k − 1 and there is the
inequality
√
(k − 1) n
|a1 x1 + a2 x2 + · · · + an xn | ≤ .
kn − 1
Thus, the original inequality holds.

Note. There is a similar problem:

• (Polish Mathematical Olympiad Finals 2009, Problem 4). Let
x1 , x2 , . . . , xn be non-negative numbers whose sum is 1. Show that there
exist numbers a1 , a2 , . . . , an chosen from amongst 0,1,2,3,4 such that
a1 , a2 , . . . , an are different from 2, 2, . . . , 2 and
2
2 ≤ a1 x1 + a2 x2 + · · · + an xn ≤ 2 + .
3n − 1

Score Situation This particular problem saw the following distribution of scores among
contestants: 59 contestants scored 7 points, 6 contestants scored 6 points, 2 contestants
scored 5 points, no contestant scored 4 points, 1 contestant scored 3 points, 5 contestants
 Inequality Problems 267

scored 2 points, 39 contestants scored 1 point, and 125 contestants scored 0 point. The
average score of this problem is 2.156, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Romania team scored 40 points (with a total team score of 250 points), the Germany
team scored 42 points (with a total team score of 248 points), the Soviet Union team scored
28 points (with a total team score of 235 points), the German Democratic Republic team
scored 42 points (with a total team score of 231 points), and the United States team scored
35 points (with a total team score of 220 points).
The gold medal cutoff for this IMO was set at 42 points (with 22 contestants earning gold
medals), the silver medal cutoff was 32 points (with 42 contestants earning silver medals),
and the bronze medal cutoff was 18 points (with 56 contestants earning bronze medals).
In this IMO, a total of 22 contestants achieved a perfect score of 42 points.

Problem 4.16 (IMO 35-1, proposed by France). Let m and n be

positive integers. Let a1 , a2 , . . . , am be distinct elements of {1, 2, . . . , n}
such that whenever ai + aj ≤ n for some i, j with 1 ≤ i < j ≤ m, there
exists k with 1 ≤ k ≤ m such that ai + aj = ak . Prove that
a1 + a2 + · · · + am n+1
≥ .
m 2
Proof. Without loss of generality, assume a1 > a2 > · · · > am . For any
positive integer i (1 ≤ i ≤ m),

ai + am+1−i ≥ n + 1. (1)

Indeed, if there exists i ∈ {1, 2, . . . , m} such that (1) is not satisfied,

then for this i, we have ai + am+1−i ≤ n. From a1 > a2 > · · · > am > 0, it
follows that

ai < ai + am < ai + am−1 < · · · < ai + am+1−i ≤ n.

Thus, {ai + am , ai + am−1 , . . . , ai + am+1−i } ∈ {a1 , a2 , . . . , ai−1 }, but

|{ai + am , ai + am−1 , . . . , ai + am+1−i }| = i > i − 1 = |{a1 , a2 , . . . , ai−1 }|,

which leads to a contradiction. Therefore, the inequality (1) holds, and

2(a1 + a2 + · · · + am )
= (a1 + am ) + (a2 + am−1 ) + · · · + (am + a1 ) ≥ m(n + 1),
a1 +a2 +···+am n+1
implying m ≥ 2 .
268 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 92 contestants scored 7 points, 11 contestants scored 6 points, 13 contestants
scored 5 points, 10 contestants scored 4 points, 15 contestants scored 3 points, 33 contestants
scored 2 points, 60 contestants scored 1 point, and 151 contestants scored 0 point. The
average score of this problem is 2.561, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the United States team scored 40 points (with a total team score of 252 points), the China
team scored 42 points (with a total team score of 229 points), the Russia team scored 29
points (with a total team score of 224 points), the Bulgaria team scored 36 points (with a
total team score of 223 points), and the Hungary team scored 35 points (with a total team
score of 221 points).
The gold medal cutoff for this IMO was set at 40 points (with 30 contestants earning gold
medals), the silver medal cutoff was 30 points (with 64 contestants earning silver medals),
and the bronze medal cutoff was 19 points (with 98 contestants earning bronze medals).
In this IMO, a total of 22 contestants achieved a perfect score of 42 points.

Problem 4.17 (IMO 36-2, proposed by Russia). Let a, b, c be posi-

tive real numbers such that abc = 1. Prove that

1 1 1 3
+ + ≥ .
a3 (b + c) b3 (c + a) c3 (a + b) 2

1 1 1
Proof 1. Since abc = 1, it follows that a + b + c = bc + ca + ab. By the
Cauchy–Schwarz inequality,

 
1 1 1
+ + (a(b + c) + b(a + c) + c(a + b))
a3 (b + c) b3 (a + c) c3 (a + b)
 2
1 1 1
≥ + + ,
a b c


1 1 1
and thus, ( a3 (b+c) + b3 (a+c) + c3 (a+b) ) ≥ 12 ( a1 + 1
b + 1c ) ≥ 3
2 · 3 1
abc = 32 .

Proof 2. Without loss of generality, assume a ≤ b ≤ c. Then,

bc ca ab
ab ≤ ac ≤ bc, ≥ ≥ .
a(b + c) b(c + a) c(a + b)
 Inequality Problems 269

By the rearrangement inequality,

1 1 1
+ 3 + 3
a3 (b + c) b (c + a) c (a + b)
(abc)2 (abc)2 (abc)2
= + +
a3 (b + c) b3 (c + a) c3 (a + b)
bc ca ab
= · bc + · ca + · ab
a(b + c) b(c + a) c(a + b)
bc ca ab
≥ · ca + · ab + · bc,
a(b + c) b(c + a) c(a + b)
and also
bc ca ab
· bc + · ca + · ab
a(b + c) b(c + a) c(a + b)
bc ca ab
≥ · ab + · bc + · ca.
a(b + c) b(c + a) c(a + b)
Summing up these two inequalities, we get
  
1 1 1 1 1 1 3 1 1 1
2 3
+ 3 + 3 ≥ + + ≥3 · · = 3.
a (b + c) b (c + a) c (a + b) a b c a b c
1 1 1
Hence a3 (b+c) + b3 (c+a) + c3 (a+b) ≥ 32 .
Proof 3. By the AM-GM inequality,

1 b+c b+c 1
+ ≥2 = .
a3 (b + c) 4bc 4a3 bc(b + c) a
1 c+a 1 1 a+b
Similarly, b3 (c+a) + 4ca ≥ b and c3 (a+b) + 4ab ≥ 1c . Thus,
 
1 1 1 1 b+c c+a a+b
3
+ 3 + 3 + + +
a (b + c) b (a + c) c (a + b) 4 bc ca ab
1 1 1
≥ + + ,
a b c
implying
1 1 1
+ 3 + 3
a3 (b + c) b (a + c) c (a + b)
  
1 1 1 1 1 3 1 1 1 3
≥ + + ≥ ·3 · · = .
2 a b c 2 a b c 2
270 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 90 contestants scored 7 points, 5 contestants scored 6 points, 4 contestants
scored 5 points, 1 contestant scored 4 points, 1 contestant scored 3 points, 5 contestants
scored 2 points, 7 contestants scored 1 point, and 299 contestants scored 0 point. The
average score of this problem is 1.709, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 236 points), the Romania team
scored 35 points (with a total team score of 230 points), the Russia team scored 42 points
(with a total team score of 227 points), the Vietnam team scored 42 points (with a total team
score of 220 points), and the Hungary team scored 21 points (with a total team score of
210 points).
The gold medal cutoff for this IMO was set at 37 points (with 30 contestants earning gold
medals), the silver medal cutoff was 29 points (with 71 contestants earning silver medals),
and the bronze medal cutoff was 19 points (with 100 contestants earning bronze medals).
In this IMO, a total of 14 contestants achieved a perfect score of 42 points.

Problem 4.18 (IMO 38-3, proposed by Russia). Let x1 , x2 , . . . , xn

be real numbers satisfying the conditions
n+1
|x1 + x2 + · · · + xn | = 1 and |xi | ≤ for i = 1, 2, . . . , n.
2
Show that there exists a permutation y1 , y2 , . . . , yn of x1 , x2 , . . . , xn such
that
n+1
|y1 + 2y2 + · · · + nyn | ≤ .
2
Proof. For any permutation π = (y1 , y2 , . . . , yn ) of x1 , x2 , . . . , xn , let
S(π) denote the value of the sum y1 + 2y2 + · · · + nyn .
Let r = n+1 2 , π0 = (x1 , x2 , . . . , xn ), and π = (xn , xn−1 , . . . , x1 ). If
|S(π0 )| ≤ r or |S(π)| ≤ r, then the original inequality holds. Otherwise,
|S(π0 )| > r and |S(π)| > r. In this case,

|S(π0 ) + S(π)| = |(n + 1)(x1 + x2 + · · · + xn )| = 2r.

Since the absolute values of S(π0 ) and S(π) both exceed r, their signs
must be opposite, with one greater than r and the other less than −r.
It is observed that starting from π0 , one can reach any permutation
by performing several exchanges of adjacent elements in the permutation
π0 . Thus, there exists a sequence of permutations π0 , π1 , . . . , πm such that
πm = π, and for each i ∈ {0, 1, 2, . . . , m − 1}, the permutation πi+1 is
obtained by swapping two adjacent elements in πi .
 Inequality Problems 271

This implies that if πi = (a1 , a2 , . . . , an ) and πi+1 = (b1 , b2 , . . . , bn ),

then there exists some index k ∈ {1, 2, . . . , n − 1} such that bk = ak+1 ,
bk+1 = ak , and bj = aj for j = k, k + 1. Since |xi | ≤ n+12 for i = 1, 2, . . . , n,
it follows that

|S(πi+1 ) − S(πi )| = |kbk + (k + 1)bk+1 − kak − (k + 1)ak+1 |

= |ak − ak+1 |
≤ |ak | + |ak+1 |
≤ 2r.

This indicates that in the sequence S(π0 ), S(π1 ), . . . , S(πm ), the dis-
tance between any two consecutive numbers on the number line does not
exceed 2r. Furthermore, S(π0 ) and S(πm ) are on opposite sides of the inter-
val [−r, r], so there must be at least one S(πi ) within this interval, implying
the existence of a permutation πi such that |S(πi )| ≤ r. This concludes the
proof.

Note. There are several similar problems:

• (Germany Team Selection Test 2022, AIMO6, Problem 3).
Determine all integers n ≥ 2 with the following property: every n dis-
tinct integers whose sum is not divisible by n can be arranged in some
order a1 , a2 , . . . , an such that n divides a1 + 2a2 + 3a3 + · · · + nan .
• (Polish Mathematical Olympiad 2020, 2nd Round, Problem 5).
Let p > 2 be a prime number and S be a set of p + 1 integers. Prove that
there exist distinct numbers a1 , a2 , . . . , ap−1 ∈ S such that
a1 + 2a2 + 3a3 + · · · + (p − 1)ap−1

is divisible by p.
• (William Lowell Putnam Mathematical Competition 2017,
B6). Find the number of ordered 64-tuples (x0 , x1 , . . . , x63 ) such that
x0 , x1 , . . . , x63 are distinct elements of {1, 2, . . . , 2017} and
x0 + x1 + 2x2 + 3x3 + · · · + 63x63

is divisible by 2017.

Score Situation This particular problem saw the following distribution of scores among
contestants: 88 contestants scored 7 points, 6 contestants scored 6 points, 2 contestants
scored 5 points, 4 contestants scored 4 points, 22 contestants scored 3 points, 5 contestants
272 IMO Problems, Theorems, and Methods: Algebra

scored 2 points, 64 contestants scored 1 point, and 269 contestants scored 0 point. The
average score of this problem is 1.778, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 38 points (with a total team score of 223 points), the Hungary team
scored 35 points (with a total team score of 219 points), the Iran team scored 42 points
(with a total team score of 217 points), the United States team scored 32 points (with a
total team score of 202 points), and the Russia team scored 29 points (with a total team
score of 202 points).
The gold medal cutoff for this IMO was set at 35 points (with 39 contestants earning gold
medals), the silver medal cutoff was 25 points (with 70 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 122 contestants earning bronze medals).
In this IMO, a total of four contestants achieved a perfect score of 42 points.

Problem 4.19 (IMO 41-2, proposed by the United States).

Suppose a, b, c are positive reals with product 1. Prove that
   
1 1 1
a−1+ b−1+ c−1+ ≤ 1.
b c a

Proof 1. From abc = 1, let a = xy , b = yz , and c = xz , where x, y, z ∈ R+ .

The original inequality becomes
(x − y + z) · (y − z + x) · (z − x + y) ≤ xyz.
If two of x − y + z, y − z + x, and z − x + y are negative, without loss
of generality, assume x − y + z < 0 and y − z + x < 0, then
2x = (x − y + z) + (y − z + x) < 0,
leading to a contradiction.
If only one of x − y + z, y − z + x, and z − x + y is negative, then the
inequality obviously holds.
If x − y + z, y − z + x, and z − x + y are all positve, then by the AM-GM
inequality,
(x − y + z) + (y − z + x)
(x − y + z) · (y − z + x) ≤ = x,
2
(y − z + x) + (z − x + y)
(y − z + x) · (z − x + y) ≤ = y,
2
(z − x + y) + (x − y + z)
(z − x + y) · (x − y + z) ≤ = z.
2
Multiplying these three inequalities yields
(x − y + z) · (y − z + x) · (z − x + y) ≤ xyz,
and thus, the original inequality holds.
 Inequality Problems 273

1 1 1
Proof 2. Since b − 1 + c = b(1 − b + bc ) = b(1 + a − 1b ),
   
1 1
a−1+ · b−1+
b c
    2
1 1 2 1
= b a−1+ a+1− =b a − 1− ≤ ba2 .
b b b

Similarly, (b − 1 + 1c ) · (c − 1 + a1 ) ≤ cb2 , (c − 1 + a1 ) · (a − 1 + 1b ) ≤ ac2 .

If a − 1 + 1b , b − 1 + 1c , and c − 1 + a1 are not all positive, assume without
loss of generality that a − 1 + 1b ≤ 0, then a ≤ 1 − 1b < 1, and b ≥ 1. Thus

1 1
b−1+ > 0, c−1+ > 0,
c a
indicating that the original inequality holds in this case.
If a − 1 + 1b , b − 1 + 1c , and c − 1 + a1 are all positive, then from the above
three inequalities,
 2  2  2
1 1 1
a−1+ · b−1+ · c−1+ ≤ a3 b3 c3 = 1,
b c a

which implies (a − 1 + 1b ) · (b − 1 + 1c ) · (c − 1 + a1 ) ≤ 1.

Proof 3. Since abc = 1,

   
1 1 1
b−1+ +a c−1+ = 2,
b c a
   
1 1 1
c−1+ +b a−1+ = 2,
c a b
   
1 1 1
a−1+ +c b−1+ = 2.
a b c

Let u = a − 1 + 1b , v = b − 1 + 1c , and w = c − 1 + a1 . If u, v, w are not all

positive, assume without loss of generality that u ≤ 0, then a ≤ 1 − 1b < 1,
and b ≥ 1. Hence,

1 1
v =b−1+ > 0, w =c−1+ > 0,
c a
implying uvw ≤ 0 < 1.
274 IMO Problems, Theorems, and Methods: Algebra

If u, v, w are all positive, then by the AM-GM inequality,


1 c
2 = u + cv ≥ 2 uv,
a a

1 a
2 = v + aw ≥ 2 vw,
b b

1 b
2 = w + bu ≥ 2 wu,
c c
so uv ≤ ac , vw ≤ ab , and wu ≤ cb . Consequently, (uvw)2 ≤ a
c · ab · cb = 1, and
uvw ≤ 1.
In conclusion, the original inequality holds.
Note. In Proof 1, the inequality (x − y + z)·(y − z + x)·(z − x + y) ≤ xyz
is equivalent to
x3 + y 3 + z 3 + 3xyz ≥ xy(x + y) + yz(y + z) + zx(z + x),
which is known as Schur’s inequality.
Score Situation This particular problem saw the following distribution of scores among
contestants: 108 contestants scored 7 points, 12 contestants scored 6 points, 18 contestants
scored 5 points, 6 contestants scored 4 points, 13 contestants scored 3 points, 41 contestants
scored 2 points, 213 contestants scored 1 point, and 50 contestants scored 0 point. The
average score of this problem is 2.768, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 218 points), the Russia team
scored 36 points (with a total team score of 215 points), the United States team scored 40
points (with a total team score of 184 points), the South Korea team scored 40 points (with
a total team score of 172 points), the Vietnam team scored 38 points (with a total team
score of 169 points), and the Bulgaria team scored 35 points (with a total team score of 169
points).
The gold medal cutoff for this IMO was set at 30 points (with 39 contestants earning gold
medals), the silver medal cutoff was 21 points (with 71 contestants earning silver medals),
and the bronze medal cutoff was 11 points (with 119 contestants earning bronze medals).
In this IMO, a total of four contestants achieved a perfect score of 42 points.

Problem 4.20 (IMO 42-2, proposed by South Korea). Prove that

a b c
√ +√ +√ ≥1
2
a + 8bc 2
b + 8ca 2
c + 8ab
for all positive real numbers a, b, and c.
 Inequality Problems 275

Proof 1. We introduce a parameter λ and set it such that

a aλ
√ ≥ λ .
a2 + 8bc a + b λ + cλ

By the AM-GM inequality,

 λ λ
2  λ λ

a2 (aλ + bλ + cλ )2 ≥ a2 aλ + 2b 2 c 2 = a2 a2λ + 4aλ b 2 c 2 + 4bλ cλ
 λ 3λ 3λ

≥ a2 a2λ + 8a 2 b 4 c 4
λ 3λ 3λ
= a2λ+2 + 8a 2 +2 b 4 c 4 .

Compared with a2λ (a2 + 8bc), we find that when λ = 43 ,

λ 3λ 3λ
a2λ+2 + 8a 2 +2 b 4 c 4 = a2λ (a2 + 8bc).

4
Thus, √ a ≥ 4
a3
4 4 . Similarly,
a2 +8bc a3 +b 3 +c 3

4 4
b b3 c c3
√ ≥ 4 4 4 and √ ≥ 4 4 4 .
b2 + 8ca a3 + b3 + c3 c2 + 8bc a3 + b3 + c3

Adding these three inequalities, we obtain

a b c
√ +√ +√ ≥ 1.
2
a + 8bc 2
b + 8ca 2
c + 8ab

Proof 2. Using Hölder’s inequality, we have

3
 a  a  
√ √ a(a2 + 8bc) ≥ a .
cyc a2 + 8bc cyc a2 + 8bc cyc cyc

 
Thus, we only need to prove ( cyc a)3 ≥ 2
cyc a(a + 8bc), which is
 2
equivalent to proving cyc a(b − c) ≥ 0.
Therefore, the original inequality holds.
276 IMO Problems, Theorems, and Methods: Algebra

Proof 3. Without loss of generality, let a + b + c = 1 and f (x) = √1 . The

x
function f (x) is convex, so by Jensen’s inequality,

af (a2 + 8bc) + bf (b2 + 8ca) + cf (c2 + 8ab)

≥ f (a(a2 + 8bc) + b(b2 + 8ca) + c(c2 + 8ab)).

Thus, we only need to prove a(a2 + 8bc) + b(b2 + 8ca) + c(c2 + 8ab) ≤ 1,
which is equivalent to

a(a2 + 8bc) + b(b2 + 8ca) + c(c2 + 8ab) ≤ (a + b + c)3

⇔ a(b − c)2 + b(c − a)2 + c(a − b)2 ≥ 0.

Therefore, the original inequality holds.

Note. In Proof 1, the left side of the inequality we wished to prove is a

fraction, and we transformed the denominator into a symmetric algebraic
expression in terms of a, b, c through the use of local inequalities. By the
method of undetermined coefficients, we then determined these coefficients
based on the inequality that needed to be proved. This is one of the common
methods used in proving inequalities.
Additionally, we can also employ proof by contradiction to solve this
problem. Assume

a b c
x= √ , y=√ , z= √ ,
2
a + 8bc 2
b + 8ca 2
c + 8ab

and then use ( x12 − 1)( y12 − 1)( z12 − 1) = 512 to derive a contradiction.
Next, we can prove a more general proposition:
Let a1 , a2 , . . . , an be positive real numbers with a1 a2 · · · an = 1. Then

1 1 1
+ + ···+ ≥ 1.
1 + (n2 − 1)a1 1 + (n2 − 1)a2 1 + (n2 − 1)an

Let xi = √ 1
for i ∈ {1, 2, . . . , n}. We prove that if x1 + x2 +
1+(n2 −1)ai
· · · + xn < 1, then a1 a2 · · · an > 1. Denote x1 x2 · · · xn = P . Since ai =
1−x2i
(n2 −1)x2
,
i

n n
(1 − x2i ) = (n2 − 1)n P 2 ai .
i=1 i=1
 Inequality Problems 277

From x1 +x2 +· · ·+xn < 1 and the AM-GM inequality, for j ∈ {1, 2, . . . , n},

P
1 − xj > x1 + · · · + xj−1 + xj+1 + · · · + xn ≥ (n − 1) n−1
,
xj

1 + xj > xj + x1 + x2 + · · · + xn ≥ (n + 1) n+1 xj P .

Thus,
2
− n2 −1
2n
1 − x2j > (n2 − 1)xj P n2 −1 ,
n
(1 − x2i ) > (n2 − 1)n P 2 .
i=1

n n
Hence, (n2 − 1)n P 2 ai =
i=1
2 2 n 2
i=1 (1 − xi ) > (n − 1) P , implying
a1 a2 · · · an > 1.
Furthermore, there are several related propositions:

(i) Let a, b, c, d ∈ R+ . Then

3 3 3 3
a2 b2 c2 d2
√ +√ +√ +√ ≥ 1.
a3 + 15bcd b3 + 15cda c3 + 15dab d3 + 15abc
p
(ii) Let xi ∈ R+ (i = 1, 2, . . . , p) with i=1 xi = 1 and λ ≥ pn − 1. Then
p
 1 p
√
n
≥ √
n
.
i=1
1 + λxi 1+λ

(iii) Let a, b, c ∈ R+ and λ ≥ 8. Then

a b c 3
√ +√ +√ ≥√ .
2
a + λbc 2
b + λca 2
c + λab 1+λ
There are several similar problems:

• (Chinese Team Selection Test 2018, Problem 11). Given positive

integers n and k satisfying n ≥ 4k, find the smallest real number λ =
λ(n, k) such that for any positive real numbers a1 , a2 , . . . , an ,
n
 ai
 ≤ λ,
i=1 a2i + a2i+1 + · · · + a2i+k

where an+j = aj for j = 1, 2, . . . , k.

278 IMO Problems, Theorems, and Methods: Algebra

• (From the book Secrets in Inequalities, 2007: 50, proposed by

Pham Kim Hung). Let a, b, c be positive real numbers with a + b + c =
1. Prove that

a b c
√
3
+ √
3
+ √
3
≥ 1.
a + 2b b + 2c c + 2a

• (From the book Secrets in Inequalities, 2007: 60). Let a, b, c be

positive real numbers. Prove that

a2 + 8bc + b2 + 8ca + c2 + 8ab ≤ 3(a + b + c).

• (Chinese Team Selection Test 2006, Problem 11). Let a, b, c be

positive real numbers such that a + b + c = 1. Prove that
√
ab bc ca 2
√ +√ +√ ≤ .
ab + bc bc + ca ca + ab 2

• (Mathematical Olympiad Program 2002). Let a, b, c be positive

reals. Prove that
  23   23   23
2a 2b 2c
+ + ≥ 3.
b+c c+a a+b

Score Situation This particular problem saw the following distribution of scores among
contestants: 77 contestants scored 7 points, 3 contestants scored 6 points, 7 contestants
scored 5 points, 11 contestants scored 4 points, 9 contestants scored 3 points, 15 contestants
scored 2 points, 40 contestants scored 1 point, and 311 contestants scored 0 point. The
average score of this problem is 1.550, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 40 points (with a total team score of 225 points), the United States
team scored 31 points (with a total team score of 196 points), the Russia team scored 28
points (with a total team score of 196 points), the South Korea team scored 42 points (with
a total team score of 185 points), and the Bulgaria team scored 19 points (with a total team
score of 185 points).
The gold medal cutoff for this IMO was set at 30 points (with 39 contestants earning gold
medals), the silver medal cutoff was 20 points (with 81 contestants earning silver medals),
and the bronze medal cutoff was 11 points (with 122 contestants earning bronze medals).
In this IMO, a total of four contestants achieved a perfect score of 42 points.
 Inequality Problems 279

Problem 4.21 (IMO 44-5, proposed by Ireland). Given n > 2 and

reals x1 ≤ x2 ≤ · · · ≤ xn , show that
⎛ ⎞2
 n  n n n
⎝ 2(n2 − 1)  
|xi − xj |⎠ ≤ (xi − xj )2 .
i=1 j=1
3 i=1 j=1

Also show that the equality is valid if and only if the sequence is an arith-
metic progression.

Proof. (i) Since subtracting the same number from all xi does not change
n
the inequality, without loss of generality, we assume i=1 xi = 0. From the
given condition,
n  n  n
|xi − xj | = 2 (xj − xi ) = 2 (2i − n − 1)xi ,
i=1 j=1 i<j i=1

and by the Cauchy–Schwarz inequality,

⎛ ⎞2
n  n n 2
 
⎝ |xi − xj |⎠ = 4 (2i − n − 1)xi
i=1 j=1 i=1

n
 n

≤4 (2i − n − 1)2 x2i
i=1 i=1
n
4n(n + 1)(n − 1)  2
= xi .
3 i=1
Furthermore,
n n n
 n
 n
 n
 n

(xi − xj )2 = n x2i − 2 xi xj + n x2j = 2n x2i .
i=1 j=1 i=1 i=1 j=1 j=1 i=1
n n 2 n n
Thus, ( i=1 j=1 |xi − xj |)2 ≤ 2(n 3−1) i=1 j=1 (xi − xj )2 .
(ii) From the conditions for the equality in the Cauchy–Schwarz inequal-
ity, the equality holds if and only if there exists a real number k such that
xi = k(2i − n − 1), i.e., xi = 2ki − k(n + 1). Thus, x1 , x2 , . . . , xn form an
arithmetic progression.
If x1 , x2 , . . . , xn form an arithmetic sequence with common difference d,
then xi = d2 (2i − n − 1) + x1 +x 2
n
. Subtracting x1 +x
2
n
from each xi yields
n

d
xi = (2i − n − 1), and xi = 0.
2 i=1
In this case, the equality holds in the given inequality.
280 IMO Problems, Theorems, and Methods: Algebra

Note. There are several similar problems:

• (From the book Problems and Solutions in Mathematical

Olympiad High School 2 , 2022: 17). Let x1 , x2 , . . . , xn be real num-
bers. Prove that
⎛ ⎞2
 
⎝ |xi − xj |⎠ ≥ (n − 1) |xi − xj |2 .
1≤i<j≤n 1≤i<j≤n

• (Iran Team Selection Test 2006, Problem 4). Let x1 , x2 , . . . , xn be

real numbers. Prove that
 n
n  n

|xi + xj | ≥ n |xi |.
i=1 j=1 i=1

• (Chinese Team Selection Test 2006, Problem 3). Given n real

numbers a1 , a2 , . . . , an , prove that there exist real numbers b1 , b2 , . . . , bn
satisfying:
(a) for any 1 ≤ i ≤ n, the number ai − bi is a positive integer;
 2 n2 −1
(b) 1≤i<j≤n (bi − bj ) ≤ 12 .

• (Canadian Mathematical Olympiad 1974, Problem 4). Let n be

a fixed positive integer. To any choice of n real numbers satisfying

0 ≤ xi ≤ 1, i = 1, 2, . . . , n,

there corresponds the sum


|xi − xj | = |x1 − x2 | + |x1 − x3 | + · · · + |x1 − xn | + |x2 − x3 |
1≤i<j≤n

+ |x2 − x4 | + · · · + |x2 − xn | + |x3 − x4 |

+ |x3 − x5 | + · · · + |x3 − xn | + · · · + |xn−1 − xn | .
(*)

Let S(n) denote the largest possible value of the sum (*). Find S(n).

Score Situation This particular problem saw the following distribution of scores among
contestants: 67 contestants scored 7 points, 4 contestants scored 6 points, 1 contestant
scored 5 points, 5 contestants scored 4 points, 6 contestants scored 3 points, 20 contestants
scored 2 points, 161 contestants scored 1 point, and 193 contestants scored 0 point. The
average score of this problem is 1.613, indicating that it was relatively challenging.
 Inequality Problems 281

Among the top five teams in the team scores, the scores of this problem are as follows:
the Bulgaria team scored 37 points (with a total team score of 227 points), the China team
scored 42 points (with a total team score of 211 points), the United States team scored 36
points (with a total team score of 188 points), the Vietnam team scored 36 points (with a
total team score of 172 points), and the Russia team scored 30 points (with a total team
score of 167 points).
The gold medal cutoff for this IMO was set at 29 points (with 37 contestants earning gold
medals), the silver medal cutoff was 19 points (with 69 contestants earning silver medals),
and the bronze medal cutoff was 13 points (with 104 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 42 points, namely
Bao Le Hùng Viet and Trong Canh Nguyen from Vietnam, and Yunhao Fu from China.

Problem 4.22 (IMO 45-4, proposed by South Korea). Let n ≥ 3

be an integer. Let t1 , t2 , . . . , tn be positive real numbers such that
 
1 1 1
n2 + 1 > (t1 + t2 + · · · + tn ) + + ···+ .
t1 t2 tn

Show that ti , tj , tk are side lengths of a triangle for all i, j, k with 1 ≤ i <
j < k ≤ n.

Proof 1. We employ proof by contradiction. Suppose there are three num-

bers among t1 , t2 , . . . , tn that cannot form the side lengths of a triangle.
Without loss of generality, assume these are t1 , t2 , t3 and t1 + t2 ≤ t3 .
Since
 
1 1 1
(t1 + t2 + · · · + tn ) + + ···+
t1 t2 tn
  ti tj

= + +n
tj ti
1≤i<j≤n

  
t1 t3 t2 t3 ti tj
= + + + + + +n
t3 t1 t3 t2 tj ti
1≤i<j≤n
(i,j)∈{(1,3),(2,3)}
/
  
t 1 + t2 1 1
≥ + t3 + + 2+n
t3 t1 t2
1≤i<j≤n
(i,j)∈{(1,3),(2,3)}
/
282 IMO Problems, Theorems, and Methods: Algebra

t 1 + t2 4t3
≥ + + 2(C2n − 2) + n
t3 t 1 + t2
t 1 + t2 4t3
= + + n2 − 4. (1)
t3 t 1 + t2
t3
Let x = t1 +t2 . Then x ≥ 1, and
1 (x − 1)(4x − 1)
4x + −5= ≥ 0.
x x
From (1),
 
1 1 1
(t1 + t2 + · · · + tn ) + + ···+ ≥ 5 + n2 − 4 = n2 + 1,
t1 t2 tn
which leads to a contradiction. Therefore, the original proposition holds.
Proof 2. We employ mathematical induction to prove that if there exist
ti , tj , tk such that ti + tj ≤ tk , then
 
1 1 1
(t1 + t2 + · · · + tn ) + + ···+ ≥ n2 + 1.
t1 t2 tn
For n = 3, assume t1 + t2 ≤ t3 , and it is easy to prove that
 
1 1 1
(t1 + t2 + t3 ) + + ≥ 10.
t1 t2 t3
Suppose when n = l ≥ 3 and t1 + t2 ≤ t3 ,
 
1 1 1
(t1 + t2 + · · · + tl ) + + ···+ ≥ l2 + 1.
t1 t2 tl
For n = l + 1,
 
1 1 1
(t1 + t2 + · · · + tl+1 ) + + ···+
t1 t2 tl+1
   
1 1 1 1
= (t1 + · · · + tl ) + ··· + + tl+1 + ···+
t1 tl t1 tl
1
+ (t1 + · · · + tl ) + 1
tl+1
  
2 1 1 1
≥ l + 1 + 2 tl+1 + ··· + (t1 + · · · + tl ) + 1
t1 tl tl+1

≥ l2 + 1 + 2 l2 + 1 + 1
> l2 + 1 + 2l + 1
= (l + 1)2 + 1.
 Inequality Problems 283

Therefore, for any integer n ≥ 3, ti , tj , tk are side lengths of a triangle

for all i, j, k with 1 ≤ i < j < k ≤ n.
Note. In Proof 1, from the Cauchy–Schwarz inequality,
 
1 1
(t1 + t2 ) + ≥ 4.
t1 t2

Thus, t11 + t12 ≥ t1 +t

4
2
.
Furthermore, we can also prove that
 
1 1 1 √
(t1 + t2 + · · · + tn ) + + ···+ ≤ (n + 10 − 3)2 .
t1 t2 tn
There are several similar problems:

• Problem 4.30 (IMO 64-4) in this chapter.

• (Chinese Girls’ Mathematical Olympiad 2023, Problem 6). Let
xi (i = 1, 2, . . . , 22) be reals such that xi ∈ [2i−1 , 2i ]. Find the maximum
possible value of
 
1 1 1
(x1 + x2 + · · · + x22 ) + + ···+ .
x1 x2 x22
• (Chinese Mathematical Olympiad 2016, Problem 6). Given an
integer n ≥ 2 and positive numbers a < b, let x1 , x2 , . . . , xn ∈ [a, b].
Find the maximum value of
x21 x22 x2n−1 x2n
x2 + x3 + ···+ xn + x1
.
x1 + x2 + · · · + xn−1 + xn
• (Chinese Mathematical Olympiad 2014, Problem 1). Given a real
number r ∈ (0, 1), prove that if n complex numbers z1 , z2 , . . . , zn satisfy
|zk − 1| ≤ r for k = 1, 2, . . . , n, then
! !
!1 1 1 !!
!
|z1 + z2 + · · · + zn | · ! + + · · · + ! ≥ n2 (1 − r2 ).
z1 z2 zn
• (United States of America Mathematical Olympiad 2012, Prob-
lem 1). Find all integers n ≥ 3 such that among any n positive real
numbers a1 , a2 , . . . , an with

max{a1 , a2 , . . . , an } ≤ n · min{a1 , a2 , . . . , an },

there exist three that are the side lengths of an acute triangle.
284 IMO Problems, Theorems, and Methods: Algebra

• (United States of America Mathematical Olympiad 2009, Prob-

lem 4). For n ≥ 2, let a1 , a2 , . . . , an be positive real numbers such that
   2
1 1 1 1
(a1 + a2 + · · · + an ) + + ··· + ≤ n+ .
a1 a2 an 2
Prove that max{a1 , a2 , . . . , an } ≤ 4 min{a1 , a2 , . . . , an }.
• (From the book Secrets in Inequalities, 2007: 77). Suppose that
p < q are positive constants and a1 , a2 , . . . , an ∈ [p, q]. Prove that
 
1 1 1 kn (p − q)2
(a1 + a2 + · · · + an ) + + ··· + ≤ n2 + ,
a1 a2 an 4pq
where kn = n2 if n is even and n2 − 1 if n is odd.
• (Vietnam Team Selection Test 2006, Problem 4). Prove that for
all real numbers a, b, c ∈ [1, 2],
   
1 1 1 a b c
(a + b + c) + + ≥6 + + .
a b c b+c c+a a+b
• (British Mathematical Olympiad 2005, 2nd Round, Problem 3).
Let a, b, c be positive real numbers. Prove that
 2  
a b c 1 1 1
+ + ≥ (a + b + c) + + .
b c a a b c
• (Chinese Mathematical Olympiad 1988, Problem 4).
(a) Let a, b, c be positive real numbers satisfying

(a2 + b2 + c2 )2 > 2(a4 + b4 + c4 ).

Prove that a, b, c can be the lengths of three sides of a triangle

respectively.
(b) Let a1 , a2 , . . . , an be n(n > 3) positive real numbers satisfying

(a21 + a22 + · · · + a2n )2 > (n − 1)(a41 + a42 + · · · + a4n ).

Prove that any three of a1 , a2 , . . . , an can be the lengths of three

sides of a triangle respectively.
• (All Soviet Union Mathematical Olympiad 1978). Given 0 < a ≤
x1 ≤ x2 ≤ · · · ≤ xn ≤ b, prove that
 
1 1 1 (a + b)2 2
(x1 + x2 + · · · + xn ) + + ···+ ≤ n .
x1 x2 xn 4ab
 Inequality Problems 285

• (United States of America Mathematical Olympiad 1977, Prob-

lem 5). Let a, b, c, d, e be positive numbers bounded by p and q, i.e., they
lie in [p, q] with 0 < p. Prove that
    2
1 1 1 1 1 p q
(a + b + c + d + e) + + + + ≤ 25 + 6 −
a b c d e q p
and determine when there is the equality.

Score Situation This particular problem saw the following distribution of scores among
contestants: 250 contestants scored 7 points, 6 contestants scored 6 points, 6 contestants
scored 5 points, 8 contestants scored 4 points, 16 contestants scored 3 points, 27 contestants
scored 2 points, 33 contestants scored 1 point, and 140 contestants scored 0 point. The
average score of this problem is 4.080, indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 220 points), the United States
team scored 42 points (with a total team score of 212 points), the Russia team scored 42
points (with a total team score of 205 points), the Vietnam team scored 42 points (with a
total team score of 196 points), and the Bulgaria team scored 42 points (with a total team
score of 194 points).
The gold medal cutoff for this IMO was set at 32 points (with 45 contestants earning gold
medals), the silver medal cutoff was 24 points (with 78 contestants earning silver medals),
and the bronze medal cutoff was 16 points (with 120 contestants earning bronze medals).
In this IMO, a total of four contestants achieved a perfect score of 42 points.

Problem 4.23 (IMO 46-3, proposed by South Korea). Let x, y, z

be three positive reals such that xyz ≥ 1. Prove that
x5 − x2 y5 − y2 z5 − z2
+ + ≥ 0.
x5 + y 2 + z 2 x2 + y 5 + z 2 x2 + y 2 + z 5

Proof 1. The original inequality can be transformed into

x2 + y 2 + z 2 x2 + y 2 + z 2 x2 + y 2 + z 2
+ + ≤ 3.
x5 + y 2 + z 2 y 5 + z 2 + x2 z 5 + x2 + y 2
Using the Cauchy–Schwarz inequality and the given condition xyz ≥ 1, we
have
1
(x5 + y 2 + z 2 )(yz + y 2 + z 2 ) ≥ (x2 (xyz) 2 + y 2 + z 2 )2
≥ (x2 + y 2 + z 2 )2 .
286 IMO Problems, Theorems, and Methods: Algebra

x2 +y 2 +z 2 yz+y 2 +z 2
Therefore, x5 +y 2 +z 2 ≤ x2 +y 2 +z 2 . Similarly,

x2 + y 2 + z 2 zx + z 2 + x2 x2 + y 2 + z 2 xy + x2 + y 2
≤ , ≤ .
y 5 + z 2 + x2 x2 + y 2 + z 2 z 5 + x2 + y 2 x2 + y 2 + z 2
Summing up these three inequalities and using x2 +y 2 +z 2 ≥ xy+yz+zx,
we obtain
x2 + y 2 + z 2 x2 + y 2 + z 2 x2 + y 2 + z 2 xy + yz + zx
+ + 5 ≤2+ 2 ≤ 3.
x5 + y 2 + z 2 y 5 + z 2 + x2 z + x2 + y 2 x + y2 + z 2

Proof 2. It suffices to prove:

 x5  x2
≥ 1 ≥ .
cyc
x5 + y 2 + z 2 cyc
x5 + y 2 + z 2

Given xyz ≥ 1, we have

 x5  x5
≥
cyc
x5 + y 2 + z 2 cyc
x5 + xyz(y 2 + z 2 )
 x4
= .
cyc
x4 + y 3 z + yz 3

Since (y − z)(y 3 − z 3 ) ≥ 0, it follows that y 4 + z 4 ≥ y 3 z + yz 3 . Hence,

 x4  x4
≥ = 1.
cyc
x4 + y 3 z + yz 3 cyc
x4 + y 4 + z 4

Furthermore, since
x5 + xyz(y 2 + z 2 ) ≤ xyz(x5 + y 2 + z 2 ),
 x2  x2 · xyz
≤
cyc
x5 + y 2 + z 2 cyc
x5 + xyz(y 2 + z 2 )
 x2 yz
= .
cyc
x4 + yz(y 2 + z 2 )

By the AM-GM inequality,

x4 + x4 + y 3 z + yz 3 ≥ 4x2 yz,
x4 + y 3 z + y 3 z + y 2 z 2 ≥ 4xy 2 z,
x4 + yz 3 + yz 3 + y 2 z 2 ≥ 4xyz 2 ,
y 3 z + yz 3 ≥ 2y 2 z 2 ,
 Inequality Problems 287

and summing up these four inequalities yields

x4 + yz(y 2 + z 2 ) ≥ x2 yz + xy 2 z + xyz 2 .
Hence,
 x2 yz  x2 yz
≤ = 1.
cyc
x4 2 2
+ yz(y + z ) cyc
x yz + xy 2 z + xyz 2
2

Thus, the original inequality holds.

Proof 3. Since
x5 − x2 x5 − x2 x2 (x3 − 1)2 (y 2 + z 2 )
− 3 2 = 3 5 ≥ 0,
x5 2
+y +z 2 2 2
x (x + y + z ) x (x + y 2 + z 2 )(x2 + y 2 + z 2 )
we have
 x5 − x2  x5 − x2 1  1

2
≥ = 2 x − .
cyc
x5 + y 2 + z 2 cyc
x3 (x2 + y 2 + z 2 ) x + y 2 + z 2 cyc x
1 1 1
Given xyz ≥ 1, we obtain x + y + z ≤ yz + zx + xy. Therefore,
 1  
(x2 − )≥ x2 − yz ≥ 0.
cyc
x cyc cyc

Thus, the original inequality holds.

Note. Proof 3 was provided by the Moldovan contestant Iurie Boreico, who
received a special prize for this solution. Furthermore, there are several
related propositions:
(i) Let n be a positive integer and x1 , x2 , . . . , xn be positive real numbers
satisfying x1 x2 . . . xn ≥ 1. Then
 x2n−1 − xn−1
1 1
≥ 0,
cyc
x2n−1
1 + xn−1
2 + · · · + xn−1
n

where cyc denotes the cyclic sum over x1 , x2 , . . . , xn , and the equality
holds if and only if x1 = x2 = · · · = xn = 1.
(ii) Let x, y, z be positive real numbers satisfying yz + zx + xy ≥ 3. Then
 x4 − x2
≥ 0,
cyc
x4 + y 2 + z 2

where cyc denotes the cyclic sum over x, y, z, and the equality holds
if and only if x = y = z = 1.
288 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 55 contestants scored 7 points, 9 contestants scored 6 points, no contestant
scored 5 points, no contestant scored 4 points, no contestant scored 3 points, 3 contestants
scored 2 points, 23 contestants scored 1 point, and 423 contestants scored 0 point. The
average score of this problem is 0.912, indicating that it was extremely difficult.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 35 points (with a total team score of 235 points), the United States
team scored 35 points (with a total team score of 213 points), the Russia team scored
20 points (with a total team score of 212 points), the Iran team scored 22 points (with a
total team score of 201 points), and the South Korea team scored 27 points (with a total
team score of 200 points).
The gold medal cutoff for this IMO was set at 35 points (with 42 contestants earning gold
medals), the silver medal cutoff was 23 points (with 79 contestants earning silver medals),
and the bronze medal cutoff was 12 points (with 128 contestants earning bronze medals).
In this IMO, a total of 16 contestants achieved a perfect score of 42 points.

Problem 4.24 (IMO 48-1, proposed by New Zealand). Real num-

bers a1 , a2 , . . . , an are given. For each i (1 ≤ i ≤ n), define
di = max{aj : 1 ≤ j ≤ i} − min{aj : i ≤ j ≤ n}
and let
d = max{di : 1 ≤ i ≤ n}.
(a) Prove that, for any real numbers x1 ≤ x2 ≤ · · · ≤ xn ,
d
max{|xi − ai | : 1 ≤ i ≤ n} ≥ . (∗)
2
(b) Show that there are real numbers x1 ≤ x2 ≤ · · · ≤ xn such that the
equality holds in (∗).

Proof 1. (a) Let d = dg (1 ≤ g ≤ n) and define

ap = max{aj : 1 ≤ j ≤ g} and ar = min{aj : g ≤ j ≤ n}.
Then 1 ≤ p ≤ g ≤ r ≤ n, and d = ap − ar . For any real numbers x1 ≤ x2 ≤
· · · ≤ xn , note that
(ap − xp ) + (xr − ar ) = (ap − ar ) + (xr − xp ) ≥ ap − ar = d,
d
implying either ap − xp ≥ 2 or xr − ar ≥ d2 . Thus,
max{|xi − ai | : 1 ≤ i ≤ n} ≥ max{|xp − ap |, |xr − ar |}
d
≥ max{ap − xp , xr − ar } ≥ .
2
 Inequality Problems 289

(b) Define a sequence {xk } as follows:

 "
d d
x1 = a1 − , xk = max xk−1 , ak − , 2 ≤ k ≤ n.
2 2
It is evident that the sequence {xk } is non-decreasing by definition, and
xk − ak ≥ − d2 for all 1 ≤ k ≤ n. Next, we prove that

d
xk − ak ≤ for all 1 ≤ k ≤ n. (1)
2
For any 1 ≤ k ≤ n, let l ≤ k be the smallest index such that xk = xl .
Then either l = 1 or l ≥ 2 and xl > xl−1 . In both cases,
d
xk = xl = al − . (2)
2
Since al − ak ≤ max{aj : 1 ≤ j ≤ k} − min{aj : k ≤ j ≤ n} = dk ≤ d,
by (2),
d d d
xk − ak = al − ak − ≤d− = ,
2 2 2
which is (1). This shows that − d2 ≤ xk − ak ≤ d
2 for all 1 ≤ k ≤ n. Hence,

d
max{|xi − ai | : 1 ≤ i ≤ n} ≤ .
2
By (a), the sequence {xk } indeed makes the equality in (*).
Proof 2. (a) The same as Proof 1.
(b) For each i(1 ≤ i ≤ n), let

Mi = max{aj : 1 ≤ j ≤ i} and mi = min{aj : i ≤ j ≤ n}.

Then

Mi = max{aj : 1 ≤ j ≤ i} ≤ max{a1 , . . . , ai , ai+1 } = Mi+1 ,

mi = min{aj : i ≤ j ≤ n} ≤ min{ai+1 , ai+2 , . . . , an } = mi+1 ,

indicating both sequences {Mi } and {mi } are non-decreasing, and by their
definition, mi ≤ ai ≤ Mi .
Let xi = Mi +m
2
i
, and from di = Mi − mi ,

di mi − M i M i − mi di
− = = xi − Mi ≤ xi − ai ≤ xi − mi = = .
2 2 2 2
290 IMO Problems, Theorems, and Methods: Algebra

Therefore,
 "
di d
max{|xi − ai | : 1 ≤ i ≤ n} ≤ max :1≤i≤n = .
2 2
By (a), the sequence {xk } indeed makes the equality in (*).
Note. There are several similar problems:

• (Turkey Team Selection Test 2002, Problem 3). Given a positive

integer n and real numbers a1 , a2 , . . . , an , show that there exist integers
 n
m and k such that | m i=1 ai − i=m+1 ai | ≤ |ak |.
• Let a1 , a2 , . . . , an be real numbers. Determine the real numbers
x1 , x2 , . . . , xn , x1 ≤ x2 ≤ · · · ≤ xn , such that max1≤k≤n |ak − xk | is
minimized.

Score Situation This particular problem saw the following distribution of scores among
contestants: 161 contestants scored 7 points, 21 contestants scored 6 points, 18 contestants
scored 5 points, 18 contestants scored 4 points, 105 contestants scored 3 points, 8 contestants
scored 2 points, 13 contestants scored 1 point, and 176 contestants scored 0 point. The
average score of this problem is 3.383, indicating that it was relatively straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Russia team scored 42 points (with a total team score of 184 points), the China team
scored 36 points (with a total team score of 181 points), the Vietnam team scored 41 points
(with a total team score of 168 points), the South Korea team scored 38 points (with a total
team score of 168 points), and the United States team scored 41 points (with a total team
score of 155 points).
The gold medal cutoff for this IMO was set at 29 points (with 39 contestants earning gold
medals), the silver medal cutoff was 21 points (with 83 contestants earning silver medals),
and the bronze medal cutoff was 14 points (with 131 contestants earning bronze medals).
In this IMO, no contestant achieved a perfect score of 42 points.

Problem 4.25 (IMO 49-2, proposed by Austria). (a) Prove that

x2 y2 z2
2
+ 2
+ ≥1
(x − 1) (y − 1) (z − 1)2
for all real numbers x, y, z, each different from 1, and satisfying xyz = 1.
(b) Prove that the equality holds above for infinitely many triples of
rational numbers x, y, z, each different from 1, and satisfying xyz = 1.
 Inequality Problems 291

x y z a b
Proof 1. (a) Let a = x−1 , b = y−1 , and c = z−1 . Then x = a−1 , y = b−1 ,
c
and z = c−1 . From xyz = 1, we have abc = (a − 1)(b − 1)(c − 1), i.e.,

a + b + c − 1 = ab + bc + ca.

Therefore,

a2 + b2 + c2 = (a + b + c)2 − 2(ab + bc + ca)

= (a + b + c)2 − 2(a + b + c − 1)

= (a + b + c − 1)2 + 1 ≥ 1.

x2 y2 z2
Hence, (x−1)2 + (y−1)2 + (z−1)2 ≥ 1.
k 2 k−1
(b) Let (x, y, z) = (− (k−1) 2 , k − k , k2 ),
where k is a positive integer.
Then (x, y, z) is a triple of rational numbers, and x, y, z are all not equal to 1.
For different positive integers k, the triples of rational numbers (x, y, z) are
distinct. In this case,

x2 y2 z2
2
+ 2
+
(x − 1) (y − 1) (z − 1)2
k2 (k − k 2 )2 (k − 1)2
= + 2 + 2
(k 2 − k + 1)2 (k − k + 1) 2 (k − k + 1)2
k 4 − 2k 3 + 3k 2 − 2k + 1
= = 1.
(k 2 − k + 1)2

Thus, the proposition holds.

p
Proof 2. (a) From xyz = 1, let p = x, q = 1, and r = y1 . Then x = q,
y = qr , and z = xy
1
= pr , where p, q, r are distinct. Therefore,

x2 y2 z2
+ + ≥1
(x − 1)2 (y − 1)2 (z − 1)2
p2 q2 r2
⇔ 2
+ 2
+ ≥ 1. (1)
(p − q) (q − r) (r − p)2
292 IMO Problems, Theorems, and Methods: Algebra

p q r
Let a = p−q ,b= q−r , and c = r−p . Then the inequality (1) becomes
 2
a ≥ 1. Since

a−1 q b−1 r c−1 p

= , = , and = ,
a p b q c r

a−1 b−1 c−1

it follows that a · b · c = 1, i.e.,
 
1− a+ ab = 0. (2)
cyc cyc

 
From (2), 1 − cyc a2 = −(a + b + c − 1)2 ≤ 0. Thus, cyc a2 ≥ 1, and
the inequality (1) holds.
2
(b) Let b = t2t+t+1+t
, c = t2t+1 bc
+t+1 , and a = − b+c , where t can take
any rational number except 0 and −1. Changing t results in an infinite
number of rational triples (a, b, c), where a, b, c are all not equal to 1 and
  2
cyc a = cyc a = 1.
a b c
Therefore, (x, y, z) = ( a−1 , b−1 , c−1 ) makes (b) hold.
√ √ √
3
√ 3 a2
3 2
x
Proof 3. (a) Let a = x, b = 3 y, and c = z. Then = √ =
bc 3 yz
2 2
√ x b c
3 xyz = x, and similarly ca = y and ab = z. Thus, the inequality to be
proved becomes

a4 b4 c4
+ 2 + 2 ≥ 1.
(a2 − bc)2 (b − ca)2 (c − ab)2

By the Cauchy–Schwarz inequality,

a4 b4 c4
+ 2 + 2
(a2 − bc)2 (b − ca)2 (c − ab)2
(a2 + b2 + c2 )2
≥ ,
(a2 − bc)2 + (b2 − ca)2 + (c2 − ab)2

and further,

(a2 + b2 + c2 )2 − ((a2 − bc)2 + (b2 − ca)2 + (c2 − ab)2 )

= (ab + bc + ca)2 ≥ 0.

(a2 +b2 +c2 )2

Thus, (a2 −bc)2 +(b2 −ca)2 +(c2 −ab)2 ≥ 1.
 Inequality Problems 293

Consequently, the original inequality is true.

(b) The same as Proof 1 or Proof 2.
Note. There are several similar problems:

• (Iran Team Selection Test 2018, Problem 2). Determine the least
real number k such that the inequality
 2  2  2  
2a 2b 2c 2a 2b 2c
+ + +k ≥4 + +
a−b b−c c−a a−b b−c c−a
is satisfied for all real numbers a, b, c.
• (Canadian Mathematical Olympiad 2017, Problem 1). Let a, b,
and c be non-negative real numbers, no two of which are equal. Prove
that
a2 b2 c2
2
+ 2
+ > 2.
(b − c) (c − a) (a − b)2
• (German Mathematical Olympiad 2010, Problem 2). Let a, b, c
be distinct real numbers. Show that
 2  2  2
2a − b 2b − c 2c − a
+ + ≥ 5.
a−b b−c c−a
• (From the book Secrets in Inequalities, 2007: 151). Let x, y, z be
distinct real numbers. Prove that
x2 y2 z2
+ + ≥ 1.
(x − y)2 (y − z)2 (z − x)2
• (United Kingdom Team Selection Test 2005, NST2, Problem 3).
Let n ≥ 3 be an integer and a1 , a2 , . . . , an be positive real numbers such
that a1 a2 . . . an = 1. Prove that
a1 + 3 a2 + 3 an + 3
+ + ··· + ≥ 3.
(a1 + 1)2 (a2 + 1)2 (an + 1)2

Score Situation This particular problem saw the following distribution of scores among
contestants: 94 contestants scored 7 points, 7 contestants scored 6 points, 53 contestants
scored 5 points, 35 contestants scored 4 points, 3 contestants scored 3 points, 24 contestants
scored 2 points, 209 contestants scored 1 point, and 110 contestants scored 0 point. The
average score of this problem is 2.563, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 217 points), the Russia team
294 IMO Problems, Theorems, and Methods: Algebra

scored 33 points (with a total team score of 199 points), the United States team scored 30
points (with a total team score of 190 points), the South Korea team scored 42 points (with
a total team score of 188 points), and the Iran team scored 37 points (with a total team
score of 181 points).
The gold medal cutoff for this IMO was set at 31 points (with 47 contestants earning gold
medals), the silver medal cutoff was 22 points (with 100 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 120 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 42 points, namely
Xiaosheng Mu and Dongyi Wei from China, and Alex Zhai from the United States.

Problem 4.26 (IMO 53-2, proposed by Australia). Let n ≥ 3 be

an integer, and let a2 , a3 , . . . , an be positive real numbers such that
a2 a3 . . . an = 1. Prove that

(1 + a2 )2 (1 + a3 )3 · · · (1 + an )n > nn .

Proof. By the AM-GM inequality, for k ∈ {2, 3, . . . , n},

⎛ ⎞k
 k−1
⎜ 1 1 1 ⎟ 1
k
(1 + ak ) = ⎝⎜ + + ···+ ⎟
+ak ⎠ ≥ k k
ak .
k−1 k−1 k−1 k−1
$ %& '
k−1 times

Thus,

(1 + a2 )2 (1 + a3 )3 · · · (1 + an )n
3 3 a3 4 4 a4 nn an
≥ 2 2 a2 × × × · · · ×
22 33 (n − 1)n−1
= nn ,
1
and the equality holds when ak = k−1 for k ∈ {2, 3, . . . , n}.
However, this contradicts the condition a2 a3 · · · an = 1. Hence,

(1 + a2 )2 (1 + a3 )3 · · · (1 + an )n > nn .

Note. There are several similar problems:

• Let n ≥ 3 be an integer and a2 , a3 , . . . , an be positive real numbers such

that a2 a3 · · · an = 1. Prove that
1
(1 + a2 )2 (1 + a3 )3 · · · (1 + an )n > nn (n − 1)n−1 .
4n−1
 Inequality Problems 295

• (Chinese Beiyue University League Independent Recruitment

Test 2014, Problem 10). Let x1 , x2 , . . . , xn be positive real numbers
with x1 x2 · · · xn = 1. Prove that
√ √ √ √
( 2 + x1 )( 2 + x2 ) · · · ( 2 + xn ) ≥ ( 2 + 1)n .

• (All-Russian Mathematical Olympiad 2007, Regional Round,

Grade 11, Problem 8). Let x1 , x2 , . . . , xn be positive real numbers.
Prove that

(1 + x1 )(1 + x1 + x2 ) · · · (1 + x1 + x2 + · · · + xn )
√
≥ (n + 1)n+1 x1 x2 · · · xn .

• (From Mildorf ’s paper “Olympiad Inequalities,” 2005: 14). Let

x1 , x2 , . . . , xn be positive reals such that
1 1 1 1
+ + ··· + = .
x1 + 1998 x2 + 1998 xn + 1998 1998
√
n x x ···x
Prove that 1 2
n−1
n
≥ 1998.
• (Asian Pacific Mathematics Olympiad 2002, Problem 1). Let
a1 , a2 , . . . , an be a sequence of non-negative integers, where n is a positive
integer. Let An = a1 +a2 +···+an
n
. Prove that

a1 !a2 ! · · · an ! ≥ ( An !)n ,

where An is the greatest integer less than or equal to An , and a! =

1 × 2 × · · · × a for a ≥ 1 (and 0! = 1).
• (Open Mathematical Olympiad of 239 Presidential Physics and
Mathematics Lyceum in Saint-Petersburg 2000, Grade 10-11,
Problem 3). Let a1 , a2 , . . . , an be positive real numbers. Prove that
√
( 2)n (a1 + a2 )(a2 + a3 ) · · · (an + a1 )
≤ (a1 + a2 + a3 )(a2 + a3 + a4 ) · · · (an + a1 + a2 ).

• (Turkey Team Selection Test 1992, Problem 3).

Let x1 , x2 , . . . , xn+1 be positive real numbers satisfying
1 1 1
+ + ··· + = 1.
1 + x1 1 + x2 1 + xn+1

Prove that x1 x2 · · · xn+1 ≥ nn+1 .

296 IMO Problems, Theorems, and Methods: Algebra

Score Situation This particular problem saw the following distribution of scores among
contestants: 171 contestants scored 7 points, 7 contestants scored 6 points, 2 contestants
scored 5 points, 8 contestants scored 4 points, 5 contestants scored 3 points, 8 contestants
scored 2 points, 83 contestants scored 1 point, and 263 contestants scored 0 point. The
average score of this problem is 2.550, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the South Korea team scored 42 points (with a total team score of 209 points), the China
team scored 40 points (with a total team score of 195 points), the United States team scored
40 points (with a total team score of 194 points), the Russia team scored 35 points (with a
total team score of 177 points), the Thailand team scored 42 points (with a total team score
of 159 points), and the Canada team scored 32 points (with a total team score of 159 points).
The gold medal cutoff for this IMO was set at 28 points (with 51 contestants earning gold
medals), the silver medal cutoff was 21 points (with 88 contestants earning silver medals),
and the bronze medal cutoff was 14 points (with 137 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 42 points, namely Jeck Lim
from Singapore.

Problem 4.27 (IMO 55-1, proposed by Austria). Let a0 < a1 <

a2 < · · · be an infinite sequence of positive integers. Prove that there exists
a unique integer n ≥ 1 such that
a 0 + a1 + · · · + an
an < ≤ an+1 . (1)
n
Proof. For n = 1, 2, . . ., define
dn = (a0 + a1 + · · · + an ) − nan .
Then the first inequality in (1) holds if and only if dn > 0. Furthermore,
nan+1 − (a0 + a1 + · · · + an )
= (n + 1)an+1 − (a0 + a1 + · · · + an + an+1 ) = −dn+1 ,
implying that the second inequality in (1) holds if and only if dn+1 ≤ 0.
By definition, the sequence d1 , d2 , . . . consists of integers, and
d1 = (a0 + a1 ) − 1 · a1 = a0 > 0,
while
dn+1 − dn = ((a0 + a1 + · · · + an ) − nan+1 ) − ((a0 + a1 + · · · + an ) − nan )
= n(an − an+1 ) < 0,
implying dn+1 < dn .
 Inequality Problems 297

Therefore, the sequence d1 , d2 , . . . is a strictly decreasing integer

sequence with a positive first term. Hence, there exists a unique n such
that dn > 0 ≥ dn+1 .

Note. This problem is novel but not difficult to solve. Essentially, it is a

discrete form of the Intermediate Value Theorem.
Furthermore, there are several similar problems:

• (British Mathematical Olympiad 2023, 2nd Round, Problem 3).

For an integer n ≥ 3, we say that A = (a1 , a2 , . . . , an ) is an n-list if every
ak is an integer satisfying 1 ≤ ak ≤ n. For each k = 1, 2, . . . , n−1, let Mk
a +a +···+a
be the minimal possible non-zero value of | 1 2k+1 k+1 − a1 +a2 +···+a k
k
|
across all n-lists. We say that an n-list A is ideal if
! !
! a1 + a2 + · · · + ak+1 a1 + a2 + · · · + ak !!
! −
! k+1 k !
= Mk for each k = 1, 2, . . . , n − 1.

Find the number of ideal n-lists.

• (All-Russian Mathematical Olympiad 2000, Final Round,
Grade 11, Problem 4). Let a1 , a2 , . . . , an be a sequence of non-
negative integers. For k = 1, 2, . . . , n, denote
ak−l+1 + ak−l+2 + · · · + ak
mk = max .
1≤l≤k l
Prove that for every α > 0 the number of values of k for which mk > α
is less than a1 +a2 +···+aα
n
.
• (All Soviet Union Mathematical Olympiad 1978). Given
a1 , a2 , . . . , an , define bk = a1 +a2 +···+a
k
k
for 1 ≤ k ≤ n. Let

C = (a1 − b1 )2 + (a2 − b2 )2 + · · · + (an − bn )2 ,

D = (a1 − bn )2 + (a2 − bn )2 + · · · + (an − bn )2 .

Prove that C ≤ D ≤ 2C.

Score Situation This particular problem saw the following distribution of scores among
contestants: 370 contestants scored 7 points, 23 contestants scored 6 points, 18 contestants
scored 5 points, 15 contestants scored 4 points, 22 contestants scored 3 points, 14 contestants
scored 2 points, 23 contestants scored 1 point, and 75 contestants scored 0 point. The
average score of this problem is 5.348, indicating that it was simple.
298 IMO Problems, Theorems, and Methods: Algebra

Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 201 points), the United States
team scored 42 points (with a total team score of 193 points), the Chinese Taiwan team
scored 42 points (with a total team score of 192 points), the Russia team scored 42 points
(with a total team score of 191 points), and the Japan team scored 38 points (with a total
team score of 177 points).
The gold medal cutoff for this IMO was set at 29 points (with 49 contestants earning gold
medals), the silver medal cutoff was 22 points (with 113 contestants earning silver medals),
and the bronze medal cutoff was 16 points (with 133 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 42 points, namely Jiyang
Gao from China, Po-Sheng Wu from Chinese Taiwan, and Alexander Gunning from Australia.

Problem 4.28 (IMO 61-2, proposed by Belgium). Four real num-

bers a, b, c, d are such that a ≥ b ≥ c ≥ d > 0 and a + b + c + d = 1.
Prove that

(a + 2b + 3c + 4d)aa bb cc dd < 1.

Proof 1. Since a + b + c + d = 1, by the weighted AM-GM inequality,

aa bb cc dd ≤ a · a + b · b + c · c + d · d = a2 + b2 + c2 + d2 .

Thus, it suffices to prove that

(a + 2b + 3c + 4d)(a2 + b2 + c2 + d2 ) < 1.

Given a ≥ b ≥ c ≥ d > 0,
  
(a + b + c + d)3 = a3 + 3 a2 b + 6 abc

> a2 (a + 3b + 3c + 3d) + b2 (3a + b + 3c + 3d)

+ c2 (3a + 3b + c + 3d) + d2 (3a + 3b + 3c + d)

≥ (a + 2b + 3c + 4d)(a2 + b2 + c2 + d2 ),

implying (a + 2b + 3c + 4d)(a2 + b2 + c2 + d2 ) < 1.

Hence, the original inequality is valid.
 Inequality Problems 299

Proof 2. If a ≤ 12 , then
(a + 2b + 3c + 4d)aa bb cc dd ≤ (a + 3b + 3c + 3d)aa+b+c+d

= (3 − 2a)a

= 1 − (1 − a)(1 − 2a)

≤ 1.
1
The equality holds if and only if a = b = c = d = 2, which is a
contradiction.
If a > 12 , then,
(a + 2b + 3c + 4d)aa bb cc dd ≤ (3 − 2a)aa (1 − a)1−a .
By the weighted AM-GM inequality,
a 1−a
·a+ · (1 − a) ≥ aa (1 − a)1−a ,
a + (1 − a) a + (1 − a)
and thus,
(3 − 2a)aa (1 − a)1−a ≤ (3 − 2a)(a2 + (1 − a)2 )

= (3 − 2a)(1 − 2a + 2a2 )

= 1 + 2(1 − a)2 (1 − 2a)

< 1.
Hence, the original inequality is true.
Note. The key to Proof 1 is to use the weighted AM-GM inequality to
transform aa bb cc dd into a2 + b2 + c2 + d2 , and then to prove a four-variable
cubic inequality.
Score Situation This particular problem saw the following distribution of scores among
contestants: 138 contestants scored 7 points, 9 contestants scored 6 points, 7 contestants
scored 5 points, 4 contestants scored 4 points, 9 contestants scored 3 points, 129 contestants
scored 2 points, 29 contestants scored 1 point, and 291 contestants scored 0 point. The
average score of this problem is 2.248, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 38 points (with a total team score of 215 points), the Russia team
scored 31 points (with a total team score of 185 points), the United States team scored 36
points (with a total team score of 183 points), the South Korea team scored 32 points (with
a total team score of 175 points), and the Thailand team scored 37 points (with a total team
score of 174 points).
300 IMO Problems, Theorems, and Methods: Algebra

The gold medal cutoff for this IMO was set at 31 points (with 49 contestants earning gold
medals), the silver medal cutoff was 24 points (with 112 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 155 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 42 points, namely Jinmin Li
from China.

Problem 4.29 (IMO 62-2, proposed by Canada). Show that the

inequality
 n 
n  n 
n 

|xi − xj | ≤ |xi + xj | (*)
i=1 j=1 i=1 j=1

holds for all real numbers x1 , x2 , . . . , xn .

Proof. Apply mathematical induction on n. For n = 1 and n = 2, the

inequality (*) holds obviously.

Assume the inequality (*) holds for fewer than n real numbers. If there
is some xi = 0, then the terms containing xi on both sides of inequality
(*) are equal. By removing xi from {x1 , x2 , . . . , xn }, we reduce it to the
inductive hypothesis for n − 1 real numbers.
Similarly, if xi = −xj = 0, then all terms on both sides of inequality (*)
involving xi or xj are

2 2|xi | + ( |xk − xi | + |xk + xi |),
k = 1, . . . , n
k = i, j

and by removing xi and xj from {x1 , x2 , . . . , xn }, we reduce it to the induc-

tive hypothesis for n − 2 real numbers.
Assume xi + xj = 0 for all i, j ∈ {1, . . . , n}(allowing i = j). Apply
a translation to all variables xi → xi + a such that after the translation,
there exist two variables whose sum is 0. Note that this translation does
not change the value of the left side of inequaltiy (*).
If there exists a positive real number a such that xi +xj +a = 0 for some
i, j ∈ {1, . . . , n}, let a+ be the smallest such a; otherwise, let a+ = +∞.
If there exists a negative real number a such that xi + xj + a = 0 for some
i, j ∈ {1, . . . , n}, let a− be the largest such a; otherwise, let a− = −∞.
√ √
Since the functions x and −x are concave in their domains, for
√
any i, j ∈ {1, . . . , n}, xi + xj + a is a concave function of a in the interval
 √
[a− , a+ ]. Moreover, the sum ni,j=1 xi + xj + a is also a concave function
 Inequality Problems 301

of a in the interval [a− , a+ ]. Therefore, there exists a number b (either a+

or a− , but b = ±∞) such that
n  n 

|xi + xj | ≥ |xi + xj + b|.
i,j=1 i,j=1

(Note that a+ = +∞ and or a− = −∞ cannot be valid at the same

time. If a+ = +∞, then xi + xj > 0 for i, j ∈ {1, 2, . . . , n} and b = a− . If
a− = −∞, then xi + xj < 0 for i, j ∈ {1, 2, . . . , n} and b = a+ .)
Let yi = xi + 2b . Then there exist i, j such that yi + yj = 0.
Consequently,
 n n  n 
 n  n n 
|xi − xj | = |yi − yj | ≤ |yi + yj |
i=1 j=1 i=1 j=1 i=1 j=1
n 
n 

≤ |xi + xj |,
i=1 j=1

where the first inequality follows from the previous transformation to the
inductive hypothesis. This completes the inductive proof.
Note. If f is a non-decreasing concave function defined on [0, +∞) with
f (0) = 0, then for any real numbers x1 , x2 , . . . , xn ,
 n
n  n 
 n
f (|xi − xj |) ≤ f (|xi + xj |).
i=1 j=1 i=1 j=1

Bin Zhao provided the proof in his paper “A Generation of the 2021
IMO Problem 2 (2021 IMO ),” published on the
New Star Math Website (http://www.nsmath.cn/jszl).
Furthermore, there are several similar problems:

• (From Junpu Hu’s paper “Two Generations of a Problem

( ),” published on the New Star Math Web-
site (http://www.nsmath.cn/xszl)). Given a positive integer n and
n distinct real numbers x1 , x2 , . . . , xn , find the maximum value of λ =
λ(n) such that
! !
 ! xi + xj !
! !
! xi − xj ! ≥ λ(n).
1≤i<j≤n

• (From Junpu Hu’s paper “Two Generalizations of a Prob-

lem ( ),” published on the New Star Math
302 IMO Problems, Theorems, and Methods: Algebra

Website (http://www.nsmath.cn/xszl)). Given a positive integer

n and n distinct real numbers x1 , x2 , . . . , xn , find the maximum value of
λ = λ(n) such that

n !

!
! xi + xi+1 !
! !
! xi − xi+1 ! ≥ λ(n).
i=1

• (From a New Star Math Camp Test, China). Given n ≥ 2, find

the minimum value of λ = λ(n) such that

n
 ai − b i
≤λ
i=1
ai + b i

for any positive real numbers a1 , a2 , . . . , an , b1 , b2 , . . . , bn that satisfy

n n
i=1 ai = i=1 bi .
• (International Zhautykov Olympiad in Mathematics 2019,
Problem 2). Find the largest real number C such that for all distinct
positive real numbers a1 , a2 , . . . , a2019 , the following inequality holds:

a1 a2 a2018 a2019
+ + ···+ + > C.
|a2 − a3 | |a3 − a4 | |a2019 − a1 | |a1 − a2 |

Score Situation This particular problem saw the following distribution of scores among
contestants: 16 contestants scored 7 points, 2 contestants scored 6 points, 1 contestant
scored 5 points, 3 contestants scored 4 points, 2 contestants scored 3 points, 12 contestants
scored 2 points, 61 contestants scored 1 point, and 522 contestants scored 0 point. The
average score of this problem is 0.375, indicating that it was extremely difficult.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 16 points (with a total team score of 208 points), the Russia team
scored 17 points (with a total team score of 183 points), the South Korea team scored
9 points (with a total team score of 172 points), the United States team scored 2 points
(with a total team score of 165 points), and the Canada team scored 8 points (with a total
team score of 151 points).
The gold medal cutoff for this IMO was set at 24 points (with 52 contestants earning gold
medals), the silver medal cutoff was 19 points (with 103 contestants earning silver medals),
and the bronze medal cutoff was 12 points (with 148 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 42 points, namely Yichuan
Wang from China.
 Inequality Problems 303

Problem 4.30 (IMO 64-4, proposed by the Netherlands). Let

x1 , x2 , . . . , x2023 be different positive real numbers such that
  
1 1 1
an = (x1 + x2 + · · · + xn ) + + ··· +
x1 x2 xn

is an integer for every n = 1, 2, . . . , 2023. Prove that a2023 ≥ 3034.

Proof 1. By the Cauchy–Schwarz inequality,

 
 n−1 n−1  n−1 n−1
   1 1    1
an =  xi + xn + ≥ xi
i=1
x
i=1 i
xn i=1 i=1
xi

1
+ xn · = an−1 + 1,
xn
and the equality holds if and only if
n−1 n−1

2 n−1
i=1 xi 1 i=1 xi
x2n = n−1 = xi · ⇔ xn = .
1
i=1 xi i=1
a2n−1 an−1

Next, we prove that an+1 − an−1 ≥ 3 for all n ≥ 2. If it is not true,

then

an+1 − an = an − an−1 = 1,

implying
n−1 n−1
i=1 xi i=1 xi + xn an−1 xn + xn
xn = , xn+1 = = = xn .
an−1 an an−1 + 1
However, this contradicts the assumption that x1 , x2 , . . . , x2023 are all
distinct. Therefore, an+1 − an−1 ≥ 3, and consequently
1001

a2023 = a1 + (a2i+1 − a2i−1 ) ≥ 1 + 1001 × 3 = 3034.
i=1

Proof 2. It is sufficient to prove that an+2 ≥ an + 3 for every n in the set

{1, 2, . . . , 2021}. Then

a2023 ≥ a2021 + 3 ≥ a2019 + 6 ≥ · · · ≥ a1 + 3033 = 3034.

304 IMO Problems, Theorems, and Methods: Algebra

By the Cauchy–Schwarz inequality,


 n n
   1 1 1
an+2 =  xi + xn+1 + xn+2 + +
i=1 i=1
xi xn+1 xn+2

 
 n n  
   1 1 1
≥ xi + (xn+1 + xn+2 ) +
i=1
x
i=1 i
xn+1 xn+2
  
1 1
= an + (xn+1 + xn+2 ) + .
xn+1 xn+2
Since x1 , x2 , . . . , x2023 are all distinct and a1 , a2 , . . . , a2023 are integers,
1 1
(xn+1 + xn+2 )( xn+1 + xn+2 ) > 4, so an+2 > an + 2, implying an+2 ≥ an + 3
for every n in the set {1, 2, . . . , 2021}.
Therefore, a2023 ≥ 3034.
Note. There are several similar problems, which are presented in
Problem 4.22.
Score Situation This particular problem saw the following distribution of scores among
contestants: 384 contestants scored 7 points, 3 contestants scored 6 points, 1 contestant
scored 5 points, 4 contestants scored 4 points, 8 contestants scored 3 points, 32 contestants
scored 2 points, 100 contestants scored 1 point, and 86 contestants scored 0 point. The
average score of this problem is 4.717, indicating that it was simple.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 240 points), the United States
team scored 42 points (with a total team score of 222 points), the South Korea team scored
42 points (with a total team score of 215 points), the Romania team scored 42 points (with
a total team score of 208 points), and the Canada team scored 42 points (with a total team
score of 183 points).
The gold medal cutoff for this IMO was set at 32 points (with 54 contestants earning gold
medals), the silver medal cutoff was 25 points (with 90 contestants earning silver medals),
and the bronze medal cutoff was 18 points (with 170 contestants earning bronze medals).
In this IMO, a total of five contestants achieved a perfect score of 42 points.

4.2.3 Determining value ranges

Problem 4.31 (IMO 16-5, proposed by the Netherlands). Deter-
mine all possible values of
a b c d
S= + + + ,
d+a+b a+b+c b+c+d c+d+a
where a, b, c, d are arbitrary positive numbers.
 Inequality Problems 305

Solution 1. Since
a b a b
+ < + = 1,
d+a+b a+b+c a+b a+b
c d c d
+ < + = 1,
b+c+d c+d+a c+d c+d
it follows that S < 2. Furthermore,
a b c d
S> + + + = 1,
a+b+c+d a+b+c+d a+b+c+d a+b+c+d
implying 1 < S < 2.
For any value 1 + t(0 < t < 1) in the interval (1, 2), let a = c = 1. Then
S = 1 + t if
2 2 d 2 b 2
t=S−1= − + = + − . (1)
1+b+d b+2 d+2 1+b+d b+2 d+2
2
Choose b = b0 sufficiently small such that (b0 +1)(b0 +2)
> t > b0b+2
0
. Then
the function of d,
 
2 2 d
f (d) = (1 + b0 + d)(d + 2) − + −t
1 + b0 + d b0 + 2 d + 2
 
2 b0 2
= (1 + b0 + d)(d + 2) + − −t ,
1 + b0 + d b0 + 2 d + 2
has f (0) = 2( b02+2 − t(b0 + 1)) > 0, and f (d) becomes negative as d → +∞.
Therefore, there must exist d0 such that f (d0 ) = 0, i.e., there are b0 , d0
making (1) hold.
Thus, the range of S is the open interval (1, 2).
Solution 2. It is evident that the sum S is homogeneous in a, b, c, d, so we
can assume
x = a + c, y = b + d, and x + y = 1.
2
a c a c
Consider the sum S1 = +
d+a+b = 1−c
b+c+d + 1−a = 2ac+x−x
ac+1−x .
If x is fixed, then the product ac can take any real number from 0 to
x2 x2
( a+c 2
2 ) = 4 , i.e., 0 < ac ≤ 4 . Furthermore,

2ac + x − x2 3x − 2 − x2
S1 = =2+ ,
ac + 1 − x ac + 1 − x
2x
which is monotonic in ac. Thus, the range of S1 is (x, 2−x ].
b d 2y
Similarly, if y is fixed, then the range of S2 = a+b+c + c+d+a is (y, 2−y ].
2x 2y
Consequently, the range of the sum S is (x + y, 2−x + 2−y ], i.e.,
(1, 4−4xy
2+xy ].
306 IMO Problems, Theorems, and Methods: Algebra

4−4xy
Noting that the range of xy is (0, 14 ], we find the range of 2+xy to be
[ 43 , 2).
Therefore, the range of the sum S is (1, 2).

Note. There are several similar problems:

• Let a, b, c, d be positive real numbers satisfying ab + bc + cd + da = 1.

Prove that

a3 b3 c3 d3 1
+ + + ≥ .
b+c+d c+d+a d+a+b a+b+c 3

• Let a, b, c, d be positive real numbers. Prove that

a b c d 2
+ + + ≥ .
b + 2c + 3d c + 2d + 3a d + 2a + 3b a + 2b + 3c 3

• (From the book Secrets in Inequalities, 2007: 37, proposed by

Pham Kim Hung). Let a, b, c, d be positive real numbers. Prove that

a b c d 4
+ 2 + 2 + 2 ≥ .
b2 2 2 2 2 2 2
+c +d c +d +a d +a +b a +b +c2 2 a+b+c+d

• (From the book Secrets in Inequalities, 2007: 38). Let

a1 , a2 , . . . , an be positive real numbers. Prove that
a1 a2 an
+ 2 + ···+ 2
a22 + a23 + · · · + a2n a1 + a23 + · · · + a2n a1 + a22 + · · · + a2n−1
4
≥ .
a1 + a2 + · · · + an

• (Nesbitt’s inequality in six variables). Prove that for all positive

real numbers a, b, c, d, e, f ,

a b c d e f
+ + + + + ≥ 3.
b+c c+d d+e e+f f +a a+b

• (Turkey Team Selection Test 1997, Problem 6). Let n ≥ 2 be an

integer and x1 , x2 , . . . , xn be positive real numbers such that x21 + x22 +
· · · + x2n = 1. Determine the smallest possible value of

x51 x52 x5n

+ + ···+ .
x2 + x3 + · · · + xn x3 + · · · + xn + x1 x1 + x2 + · · · + xn−1
 Inequality Problems 307

• (From Mildorf ’s paper “Olympiad Inequalities,” 2005: 34). Let

n ≥ 2 be a positive integer and k ≥ n−1 n be a real number. Show that
for all positive reals a1 , a2 , . . . , an ,
 k  k
(n − 1)a1 (n − 1)a2
+ + ···
a2 + a3 + · · · + an a3 + · · · + an + a1
 k
(n − 1)an
+ ≥ n.
a1 + a2 + · · · + an−1

Score Situation This particular problem saw the following distribution of scores among
contestants: 27 contestants scored 7 points, 8 contestants scored 6 points, 15 contestants
scored 5 points, 13 contestants scored 4 points, 8 contestants scored 3 points, 12 contestants
scored 2 points, 22 contestants scored 1 point, and 35 contestants scored 0 point. The
average score of this problem is 3.100, indicating that it was relatively straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 44 points (with a total team score of 256 points), the United
States team scored 39 points (with a total team score of 243 points), the Hungary team
scored 33 points (with a total team score of 237 points), the German Democratic Republic
team scored 28 points (with a total team score of 236 points), and the Yugoslavia team
scored 22 points (with a total team score of 216 points).
The gold medal cutoff for this IMO was set at 38 points (with 10 contestants earning gold
medals), the silver medal cutoff was 30 points (with 24 contestants earning silver medals),
and the bronze medal cutoff was 23 points (with 37 contestants earning bronze medals).
In this IMO, a total of six contestants achieved a perfect score of 40 points.

Problem 4.32 (IMO 40-2, proposed by Poland). Let n be a fixed

integer with n ≥ 2.

(a) Determine the least constant C such that the inequality

n 4
 
xi xj (x2i + x2j ) ≤C xi
1≤i<j≤n i=1

holds for all real numbers x1 , x2 , . . . , xn ≥ 0.

(b) For this constant C, determine when the equality holds.

Solution 1. Since the inequality is symmetric and homogeneous, we may


assume x1 ≥ x2 ≥ · · · ≥ xn ≥ 0 and ni=1 xi = 1. Thus, it suffices to
308 IMO Problems, Theorems, and Methods: Algebra

consider the maximum value of


F (x1 , x2 , . . . , xn ) = xi xj (x2i + x2j ).
1≤i<j≤n

Let xk+1 (k ≥ 2) be the last non-zero number among x1 , x2 , . . . , xn and

adjust
x = (x1 , x2 , . . . , xk−1 , xk , xk+1 , 0, . . . , 0)
to
x = (x1 , x2 , . . . , xk−1 , xk + xk+1 , 0, 0, . . . , 0).
Then
k−1


F (x ) − F (x) = xk xk+1 3(xk + xk+1 ) xi − x2k − x2k+1
i=1

= xk xk+1 (3(xk + xk+1 )(1 − xk − xk+1 ) − x2k − x2k+1 )

= xk xk+1 ((xk + xk+1 )(3 − 4(xk + xk+1 )) + 2xk xk+1 ).
Since 1 ≥ x1 + xk + xk+1 ≥ 12 (xk + xk+1 ) + (xk + xk+1 ), xk + xk+1 ≤
2
3 < 34 .
Hence, F (x )−F (x) > 0. In other words, the function value of F strictly
increases when x is transformed into x .
For any x = (x1 , x2 , . . . , xn ), after several adjustments, one can eventu-
ally obtain
F (x) ≤ F (a, b, 0, . . . , 0) = ab(a2 + b2 )
1 1
= (2ab)(1 − 2ab) ≤ .
2 8
Therefore, the minimum value of C is 18 and the equality holds if and
only if two of the xi are equal and the others are all equal to 0.
Solution 2. When x1 = x2 = 1 and x3 = · · · = xn = 0, we have
1 1
C≥ × 1 × 1 × (12 + 12 ) = .
16 8
Next, we prove that the inequality
n 4
 1 
xi xj (x2i + x2j ) ≤ xi
8 i=1
1≤i<j≤n

is true for all non-negative real numbers x1 , x2 , . . . , xn .

 Inequality Problems 309

Since
⎛ ⎞2
n 4 n
  
xi =⎝ x2k + 2 xi xj ⎠
i=1 k=1 1≤i<j≤n
⎛ ⎞
n
 
≥4 x2k ⎝2 xi xj ⎠
k=1 1≤i<j≤n

 n

=8 xi xj x2k
1≤i<j≤n k=1

≥8 xi xj (x2i + x2j ),
1≤i<j≤n
it follows that
n 4
 1 
xi xj (x2i + x2j ) ≤ xi .
8 i=1
1≤i<j≤n

Thus, the minimum value of C is 18 , and the equality is true if and only if
two of the xi are equal and the others are all equal to 0.
Score Situation This particular problem saw the following distribution of scores among
contestants: 59 contestants scored 7 points, 11 contestants scored 6 points, 4 contestants
scored 5 points, 5 contestants scored 4 points, 10 contestants scored 3 points, 37 contestants
scored 2 points, 129 contestants scored 1 point, and 195 contestants scored 0 point. The
average score of this problem is 1.671, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 41 points (with a total team score of 182 points), the Russia team
scored 37 points (with a total team score of 182 points), the Vietnam team scored 37 points
(with a total team score of 177 points), the Romania team scored 42 points (with a total
team score of 173 points), and the Bulgaria team scored 23 points (with a total team score
of 170 points).
The gold medal cutoff for this IMO was set at 28 points (with 38 contestants earning gold
medals), the silver medal cutoff was 19 points (with 70 contestants earning silver medals),
and the bronze medal cutoff was 12 points (with 118 contestants earning bronze medals).
In this IMO, no contestant achieved a perfect score of 42 points.

Problem 4.33 (IMO 47-3, proposed by Ireland). Determine the

least real number M such that the inequality
|ab(a2 − b2 ) + bc(b2 − c2 ) + ca(c2 − a2 )| ≤ M (a2 + b2 + c2 )2
holds for all real numbers a, b, and c.
310 IMO Problems, Theorems, and Methods: Algebra

Solution. Consider P (t) = tb(t2 − b2 ) + bc(b2 − c2 ) + ct(c2 − t2 ). It is evi-

dent that P (b) = P (c) = P (−c − b) = 0, and thus,

|ab(a2 − b2 ) + bc(b2 − c2 ) + ca(c2 − a2 )|

= |P (a)| = |(b − c)(a − b)(a − c)(a + b + c)|.

The original inequality is equivalent to

|(b − c)(a − b)(a − c)(a + b + c)| ≤ M (a2 + b2 + c2 )2 .

Assuming without loss of generality that a ≤ b ≤ c, we have

 2
(b − a) + (c − b) (c − a)2
|(a − b)(b − c)| = (b − a)(c − b) ≤ = ,
2 4

with the equality if and only if b − a = c − b, i.e., 2b = a + c. Also,

 2
(c − b) + (b − a) (c − b)2 + (b − a)2
≤ ,
2 2

which is equivalent to

3(c − a)2 ≤ 2((b − a)2 + (c − b)2 + (c − a)2 ),

with the equality if and only if 2b = a + c. Thus,

|(b − c)(a − b)(a − c)(a + b + c)|

1
≤ |(c − a)3 (a + b + c)|
4
1
= (c − a)6 (a + b + c)2
4
 3
1 2((b − a)2 + (c − b)2 + (c − a)2 )
≤ (a + b + c)2
4 3
⎛ ⎞2
√  3
2 2
2 ⎝ 4 (b − a) + (c − b) + (c − a) 2
= (a + b + c)2 ⎠ .
2 3
 Inequality Problems 311

By the AM-GM inequality,

|(b − c)(a − b)(a − c)(a + b + c)|

√  2
2 (b − a)2 + (c − b)2 + (c − a)2 + (a + b + c)2
≤
2 4
√
9 2 2
= (a + b2 + c2 )2 .
32
√
9 2
Thus, M = 32 , with the equality if and only if 2b = a + c and

(b − a)2 + (c − b)2 + (c − a)2

= (a + b + c)2 ,
3
i.e., 2b = a + c and (c − a)2 = 18b2 . √ √
Taking b = 1, we obtain a = 1 − 32 2 and c = 1 + 32 2. Hence, when
 
3√ 3√
(a, b, c) = 1 − 2, 1, 1 + 2 ,
2 2
√
9 2
the equality in the original inequality is achieved, confirming that M = 32 .

Note. There are several similar problems:

• (Turkey Mathematical Olympiad 2013, 2nd Round, Problem 5).

Find the maximum value of M such that

a3 + b3 + c3 − 3abc ≥ M (ab2 + bc2 + ca2 − 3abc)

for all positive real numbers a, b, c.

• (Chinese Southeast Mathematical Olympiad 2006, Problem 6).
Find the minimum value of real numbers m such that

m(a3 + b3 + c3 ) ≥ 6(a2 + b2 + c2 ) + 1

for all positive real numbers a, b, c with a + b + c = 1.

Score Situation This particular problem saw the following distribution of scores among
contestants: 28 contestants scored 7 points, 2 contestants scored 6 points, 2 contestants
scored 5 points, 1 contestant scored 4 points, 1 contestant scored 3 points, 4 contestants
scored 2 points, 95 contestants scored 1 point, and 365 contestants scored 0 point. The
average score of this problem is 0.659, indicating that it was extremely difficult.
312 IMO Problems, Theorems, and Methods: Algebra

Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 35 points (with a total team score of 214 points), the Russia team
scored 11 points (with a total team score of 174 points), the South Korea team scored 30
points (with a total team score of 170 points), the Germany team scored 9 points (with a
total team score of 157 points), and the United States team scored 7 points (with a total
team score of 154 points).
The gold medal cutoff for this IMO was set at 28 points (with 42 contestants earning gold
medals), the silver medal cutoff was 19 points (with 89 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 122 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 42 points, namely Zhiyu
Liu from China, Iurie Boreico from Moldova, and Alexander Magazinov from Russia.

4.3 Summary
Inequalities are pervasive across various branches of mathematics, includ-
ing analysis, algebra, number theory, and combinatorics, providing crucial
tools for comparison, estimation, and bounding of mathematical expres-
sions. However, more important and challenging than proving inequalities
is the ability to identify and establish relationships of inequality between
quantities in problems.
In the first 64 IMOs, there were a total of 33 inequality problems.
These problems can be broadly categorized into three types, as depicted in
Figure 4.3. The score details for these problems are presented in Table 4.2.
Due to the smaller number of participating teams and missing contestant
score information in early IMOs, there are several blanks in Table 4.2.
Problems 4.1–4.5 focus on “solving inequalities;” among these five prob-
lems, the one with the lowest average score is Problem 4.5 (IMO 29-4),
proposed by Ireland. Problems 4.6–4.30 deal with “proving inequalities;”
among these 25 problems, the one with the lowest average score is Prob-
lem 4.29 (IMO 62-2), proposed by Canada. Problems 4.31–4.33 are about
“determining value ranges;” among these three problems, the one with the
lowest average score is Problem 4.33 (IMO 47-3), proposed by Ireland.
These 33 problems were proposed by 18 countries, with Hungary con-
tributing the most, totaling four problems. Ireland, South Korea, and the
Netherlands each proposed three problems, while the United States, Poland,
France, Germany, Russia, and Austria each contributed two problems.
From Table 4.2, it can be observed that in the first 64 IMOs, there were
three inequality problems with an average score of 0–1 point; five problems
 Inequality Problems 313

Solving Inequalities Proving Inequalities

8
Determining Value Ranges
7

0
1–10 11–20 21–30 31–40 41–50 51–60 61–64

Figure 4.3 Numbers of Inequality Problems in the First 64 IMOs

with an average score of 1–2 points; 12 problems with an average score of 2–

3 points; five problems with an average score of 3–4 points; eight problems
with an average score above 4 points. Overall, the inequality problems were
of certain difficulty, with most problems having an average score in the
range of 2–3 points.
In the 24th–64th IMOs, there were a total of 21 inequality problems.
Among these, three had an average score of 0–1 point; five had an average
score of 1–2 points; eight had an average score of 2–3 points; one had
an average score of 3–4 points; four had an average score above 4 points.
Further analysis of the problem numbers of these 21 inequality problems,
as shown in Table 4.3, reveals that these problems frequently appeared as
the 1st/4th or 2nd/5th problem. The majority of these problems, totaling
18, were of the type proving inequalities. The other two types of inequality
problems were less frequent, with only three problems appearing.
Excluding Problem 4.13 (IMO 24-6), the inequality problems from the
25th–64th IMOs are arranged in order of their average scores, from left
to right, and a scatter plot of the score details is presented, as shown in
Figure. 4.4.
From Table 4.2 and Figure 4.4, it is observable that in the inequality
problems, the average score of the top five teams generally exceeds the
average score of the problem by 3.5 points, and the average score of the 6th–
15th teams typically surpasses the average score by 2 points. However, in
more difficult problems, the performance of the top 15 teams is particularly
outstanding, but the average score of the 16th–25th teams is very close to
 Table 4.2 Score Details of Inequality Problems in the First 64 IMOs

314
Problem 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8
Full points 6.000 6.000 4.000 7.000 7.000 7.000 7.000 8.000
Average score 2.600 4.529 2.738 4.909 2.332 3.111 3.294 2.438
Top five mean 3.300 5.175 6.433 4.700 3.600
6th–15th mean 3.661
16th–25th mean 2.293
Problem number in IMO 2-2 4-2 7-1 14-4 29-4 3-2 6-2 11-6

IMO Problems, Theorems, and Methods: Algebra

Proposing Hungary Hungary Yugoslavia The Ireland Poland Hungary The
country Netherlands Soviet Union
Problem 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16
Full points 5.000 6.000 6.000 6.000 7.000 7.000 7.000 7.000
Average score 2.321 4.727 3.216 5.167 2.125 4.557 2.156 2.561
Top five mean 3.354 5.725 5.775 6.833 6.233 6.133
6th–15th mean 1.836 4.797 5.224 5.767 3.600 4.133
16th–25th mean 4.167 1.200 3.467
Problem number in IMO 13-1 17-1 19-4 20-5 24-6 25-1 28-3 35-1
Proposing Hungary Czechoslovakia The France The Germany Germany France
country United United
Kingdom States
Problem 4.17 4.18 4.19 4.20 4.21 4.22 4.23 4.24
Full points 7.000 7.000 7.000 7.000 7.000 7.000 7.000 7.000
Average score 1.709 1.778 2.768 1.550 1.613 4.080 0.912 3.383
Top five mean 6.067 5.867 6.417 5.333 6.033 7.000 4.633 6.600
6th–15th mean 3.867 4.183 4.815 3.800 3.050 6.850 3.083 6.033
16th–25th mean 2.167 2.483 3.817 1.517 2.305 6.000 0.983 5.000
Problem number in IMO 36-2 38-3 41-2 42-2 44-5 45-4 46-3 48-1
Proposing Russia Russia The South Ireland South South New
country United Korea Korea Korea Zealand
States
(Continued )
 Table 4.2 (Continued)

Problem 4.25 4.26 4.27 4.28 4.29 4.30 4.31 4.32

Full points 7.000 7.000 7.000 7.000 7.000 7.000 7.000 7.000
Average score 2.563 2.550 5.348 2.248 0.375 4.717 3.100 1.671
Top five mean 6.133 6.417 6.867 5.800 1.733 7.000 4.150 6.000
6th–15th mean 5.400 4.852 6.850 4.500 1.283 6.850 3.052 3.883
16th–25th mean 3.950 3.467 6.917 3.783 0.683 6.700 1.833

Inequality Problems
Problem number in IMO 49-2 53-2 55-1 61-2 62-2 64-4 16-5 40-2
Proposing Austria Australia Austria Belgium Canada The The Poland
country Netherlands Netherlands

Problem 4.33

Full points 7.000

Average score 0.659
Top five mean 3.067
6th–15th mean 1.924
16th–25th mean 0.537
Problem number in IMO 47-3
Proposing country Ireland

Note. Top five mean = Total score of the top five teams ÷ Total number of contestants from the top five teams,
6th–15th mean = Total score of the 6th–15th teams ÷ Total number of contestants from the 6th–15th teams,
16th–25th mean = Total score of the 16th–25th teams ÷ Total number of contestants from the 16th–25th teams.

315
316 IMO Problems, Theorems, and Methods: Algebra

Table 4.3 Numbers of Inequality Problems in the 24th–64th IMOs.

Problem Number
Number of Problems
Inequality Problem 1, 4 2, 5 3, 6 in the First 64 IMOs

Solving inequalities 1 0 0 5
Proving inequalities 6 8 4 25
Determining value ranges 0 1 1 3

Total 7 9 5 33

0
0 5 10 15 20 25

Average Score Average Score of Top Five Teams

Average Score of 6th–15th Teams Average Score of 16th–25th Teams

Figure 4.4 Score Details of Inequality Problems in the 25th–64th IMOs

the average score of the problem, as seen in Problem 4.33 (IMO 47-3),
Problem 4.23 (IMO 46-3), and Problem 4.20 (IMO 42-2).
From Figure 4.4, it can also be observed that there are several inequal-
ity problems where the average score is higher than the average score
of the 16th–25th teams, such as Problem 4.15 (IMO 28-3), Problem 4.5
(IMO 29-4), and Problem 4.14 (IMO 25-1). This phenomenon is due to
the smaller number of participating teams in early IMOs. It was not until
the 30th IMO in 1989 that the number of participating teams exceeded 50.
Therefore, it is common to see situations where the average score is close
to or even higher than the average score of the 16th–25th teams during this
period.
Chapter 5

Other Algebra Problems

Algebra, deriving from the Arabic word “al-jabr,” evolved in ancient times.
As a plethora of problems concerning quantitative relationships arose,
the need for more general methods to solve these problems led to the
development of elementary algebra, with solving equations as its core
focus.
The basic components of elementary algebra include numbers, expres-
sions, and equations, which roughly align with the algebra curriculum cur-
rently established in secondary schools worldwide. However, the algebra
taught in secondary schools also encompasses additional topics such as func-
tions, sequences, inequalities, and polynomials.
In the first 64 IMOs, there had been a total of 10 other algebra prob-
lems, approximately accounting for 9.9% of all algebra problems. These
problems can be primarily categorized into three types: (1) proving trigono-
metric identities, totaling two problems; (2) finding polynomials, totaling
two problems; (3) proving properties of polynomials and sets, totaling six
problems. The statistical distribution of these three types of problems in
the previous IMOs is presented in Table 5.1.
It can be observed that other algebra problems were not frequent, with
only 1–3 appearing in every 10 IMOs.
Specifically, there were only two problems about proving trigonomet-
ric identities. On the one hand, students begin to engage with trigono-
metric functions from middle school, and have developed a comprehensive
understanding, so IMO problems seldom specifically focused on trigono-
metric concepts. Instead, these concepts were integrated into geometry,
where algebraic methods can also be applied. On the other hand, since

317
318 IMO Problems, Theorems, and Methods: Algebra

Table 5.1 Numbers of Other Algebra Problems in the First 64 IMOs

Session
Content 1–10 11–20 21–30 31–40 41–50 51–60 61–64 Total

Proving trigonometric 2 0 0 0 0 0 0 2
identities
Finding polynomials 0 1 0 0 1 0 0 2
Proving properties 0 1 1 1 2 0 1 6
Algebra problems 20 20 14 13 15 13 6 101
The percentage of 10.0% 10.0% 7.1% 7.7% 20.0% 0.0% 16.7% 9.9%
other algebra
problems among the
algebra problems

both sides of an identity are necessarily equal, it merely requires a specific

manipulation, making the difficulty relatively low. Therefore, it is not a
primary focus.
As for polynomial problems, the divisibility and congruence of poly-
nomials can be linked with number theory, while finding polynomials can
be associated with functional equations. Consequently, there are not many
polynomial problems presented in this chapter. Although polynomials are
part of “Advanced Algebra” in university mathematics, many polynomial
concepts and methods are quite elementary.
This chapter will be divided into three parts. The first part primarily
introduces some knowledge about trigonometry and polynomials, includ-
ing common trigonometric identities, common polynomials, and common
techniques for polynomials.
The second part revolves around three types of problems: “proving
trigonometric identities,” “finding polynomials,” and “proving properties
of polynomials and sets.” These problems are presented in chronological
order, and some problems include various solutions, generalizations, and
similar problems.
It is important to note that for each problem, the solutions are fol-
lowed by information on the scores, including the number of contestants in
each score range, the average score, and the scores of the top five teams.
However, early IMOs often lacked information on contestant scores, so the
number of contestants in each score range only represents the counted
number of contestants, and some problems lack scores of the top five
teams.
The third part provides a brief summary of this chapter.
 Other Algebra Problems 319

5.1 Common Theorems, Formulas, and Methods

5.1.1 Common trigonometric identities
When dealing with trigonometric problems, it is generally necessary to
simplify or transform expressions, highlighting the significance of formulas
such as sum and difference identities. Typically, multiplication forms are
more convenient than addition forms and allow for the elimination of certain
elements.

(i) sin 3α = 4 sin α sin(60◦ − α) sin(60◦ + α).

(ii) cos 3α = 4 cos α cos(60◦ − α) cos(60◦ + α).
(iii) tan 3α = tan α tan(60◦ − α) tan(60◦ + α).
tan α+tan β+tan γ−tan α tan β tan γ
(iv) tan(α + β + γ) = 1−tan α tan β−tan β tan γ−tan γ tan α .
sin(n+1)d·sin(α+nd)
(v) sin α + sin(α + 2d) + · · · + sin(α + 2nd) = sin d .
sin(n+1)d·cos(α+nd)
(vi) cos α + cos(α + 2d) + · · · + cos(α + 2nd) = sin d .

Furthermore, by utilizing the properties of unit roots, we can obtain:

(n−1)π
(vii) sin πn sin 2π
n · · · sin n
n
= 2n−1 .
 
 (n−1)π  1−(−1)n
(viii) cos πn cos 2π
n · · · cos n = 2n .

Sometimes, there exist certain quantitative relationships between angles.

For example, in the triangle ABC, the following relationships can be
derived:

(ix) sin2 A + sin2 B + sin2 C = 2 + 2 cos A cos B cos C.

(x) cos2 A + cos2 B + cos2 C = 1 − 2 cos A cos B cos C.
(xi) sin A + sin B + sin C = 4 cos A B C
2 cos 2 cos 2 .
(xii) cos A + cos B + cos C = 1 + 4 sin A B C
2 sin 2 sin 2 .
(xiii) sin A B C A+B B+C C+A
2 + sin 2 + sin 2 = 1 + 4 sin 4 sin 4 sin 4 .
(xiv) cos A B C A+B B+C C+A
2 + cos 2 + cos 2 = 4 cos 4 cos 4 cos 4 .
(xv) sin 2A + sin 2B + sin 2C = 4 sin A sin B sin C.
(xvi) cos 2A + cos 2B + cos 2C = −1 − 4 cos A cos B cos C.
(xvii) tan A B B C C A
2 tan 2 + tan 2 tan 2 + tan 2 tan 2 = 1.

Especially in a non-right-angled triangle ABC, the following relation-

ships hold:
320 IMO Problems, Theorems, and Methods: Algebra

(xviii) tan A + tan B + tan C = tan A tan B tan C.

(xix) cot A cot B + cot B cot C + cot C cot A = 1.
cot A+cot B cot B+cot C cot C+cot A
(xx) tan A+tan B + tan B+tan C + tan C+tan A = 1.
Mastering basic trigonometric transformations is crucial for solving
geometry problems. The technique can be employed to establish common
relationships between sides and angles in triangles.
Theorem 5.1 (Projection Formula). a = b cos C + c cos B.
a b c
Theorem 5.2 (Law of Sines). sin A = sin B = sin C = 2R.
Theorem 5.3 (Law of Cosines). c2 = a2 + b2 − 2ab cos C.
A−B
a+(−1)k b cos( kπ
2 − 2 )
Theorem 5.4 (Mollweide’s Formula). c = sin( kπ C , where
2 + 2 )
k ∈ {0, 1}.
A−B
a−b tan 2
Theorem 5.5 (Law of Tangents). a+b = tan A+B .
2

Theorem 5.6 (Half Angle Formulas).

 
A (p − b)(p − c) A p(p − a)
sin = , cos = ,
2 bc 2 bc

A 1 (p − a)(p − b) (p − c) r
tan = = ,
2 p−a p p−a
where p is the semi-perimeter and r is the inradius.
Additionally, trigonometric knowledge can be employed to characterize
the area of geometric figures, which will be presented in IMO Problems,
Theorems, and Methods: Geometry.

5.1.2 Common methods for proving trigonometric

identities
(1) Transforming to the same angle
When multiple distinct angles appear in the given conditions, one can con-
sider expressing them using a single angle based on their quantitative rela-
tionships, thereby simplifying the problem.

Example 5.1. Prove that

3
2 sin4 x + sin2 2x + 5 cos4 x − cos 3x cos x = 2(1 + cos2 x).
4
 Other Algebra Problems 321

Proof. Consider transforming angles x and 3x into 2x:

3
2 sin4 x + sin2 2x + 5 cos4 x − cos 3x cos x
4
 2  2
1 − cos 2x 3 2 1 + cos 2x 1
=2 + sin 2x + 5 − (cos 4x + cos 2x)
2 4 2 2
7 + 6 cos 2x + 7 cos2 2x 3   1
= + 1 − cos2 2x − (2 cos2 2x − 1 + cos 2x)
4 4 2
= 3 + cos 2x
= 2(1 + cos2 x).

(2) Transforming to the same trigonometric function

When there are multiple different trigonometric functions in the given con-
ditions, utilize basic trigonometric identities to transform them into the
same trigonometric functions.
2
Example 5.2. Given tan(A−B)
tan A + sin C
sin2 A
= 1, prove that tan A, tan C, and
tan B are in geometric progression.

Proof. The purpose is to prove that tan2 C = tan A tan B. Convert the
sine function in the condition into the tangent function:

1 tan2 C
sin2 C 1 − cos2 C 1− 1+tan2 C 1+tan2 C
2 = 2
= 1 = tan2 A
,
sin A 1 − cos A 1− 1+tan2 A 1+tan2 A
 
tan(A − B) tan A − tan B tan B 1 + tan2 A
1− = 1− = .
tan A tan A(1 + tan A tan B) (1 + tan A tan B) tan A
2
tan C tan A tan B
Combining these two expressions yields 1+tan 2 C = 1+tan A tan B , establish-

ing the conclusion.

Notably, this solution also used the identity 1 = sin2 x + cos2 x =
sec x − tan2 x for transformations, and such constant-containing formu-
2

las are crucial for solving extremum problems or proving inequalities.

Example 5.3. Given tan x1 tan x2 · · · tan x2022 = 1, find the maximum
value of

sin x1 sin x2 · · · sin x2022 .

322 IMO Problems, Theorems, and Methods: Algebra

Solution. From the given condition,

sin x1 sin x2 · · · sin x2022 = cos x1 cos x2 · · · cos x2022 .

Thus, we need to find the maximum value of cos x1 cos x2 · · · cos x2022 .
Since

1
= 1 + tan2 xi ≥ 2 |tan xi | ,
cos2 xi

1
we have cos2 x1 cos2 x2 ··· cos2 x2022 ≥ 22022 |tan x1 tan x2 · · · tan x2022 | = 22022 ,
so

1
sin x1 sin x2 · · · sin x2022 = cos x1 cos x2 · · · cos x2022 ≤ ,
21011
π
and the equality holds when x1 = x2 = · · · = x2022 = 4.

(3) Transforming to the same power

When trigonometric functions of varying or higher powers appear in the
conditions, consider raising or reducing their powers to transform them
into functions of the same power or into simpler forms.

x x+2π x+4π 3
Example 5.4. Prove that cos3 3 + cos3 3 + cos3 3 = 4 cos x.

Proof. Using the triple-angle formula cos 3θ = 4 cos3 θ − 3 cos θ for cosine,
we have

x 3 x 1
cos3 = cos + cos x,
3 4 3 4
x + 2π 3 x + 2π 1 3 x + 2π 1
cos3 = cos + cos(x + 2π) = cos + cos x,
3 4 3 4 4 3 4
x + 4π 3 x + 4π 1 3 x − 2π 1
cos3 = cos + cos(x + 4π) = cos + cos x.
3 4 3 4 4 3 4

It is evident that cos x3 +cos x+2π

3 +cos
x−2π
3 = cos x3 +2 cos x3 cos 2π
3 = 0.
Therefore, adding the both sides of the above three equalities, we have

x x + 2π x + 4π 3
cos3 + cos3 + cos3 = cos x.
3 3 3 4
 Other Algebra Problems 323

(4) Elimination method

When conditions include parameters, it is common to derive expres-
sions for these parameters and substitute them to eliminate the
parameters.

Example 5.5. Given

a cos2 α + b sin2 α = m cos2 β,

a sin2 α + b cos2 α = n sin2 β,
m tan2 α = n tan2 β,

where β = kπ, prove that (a + b)(m + n) = 2mn.

Proof. When m = 0, from m tan2 α = n tan2 β, we have n = 0, so the

conclusion holds.
2
α+b cos2 α
When m = 0, from the given condition, aa sin
cos2 α+b sin2 α
n
= m tan2 β.
a sin2 α+b cos2 α
Also, m tan2 α = n tan2 β, so a cos2 α+b sin2 α = tan2 α, from which
2
a tan α+b
a+b tan2 α = tan2 α.
Solving this, we can get tan2 α = 1. Hence sin2 α = cos2 α = 12 . Conse-
2
quently, cos2 β = a+b a+b a+b a+b
2m and sin β = 2n . Hence we obtain 2m + 2n = 1.

(5) Telescoping method

Similar to summation in sequences, when proving trigonometric identities
in the summation form, the Telescoping Method can be employed for elim-
inations.

Example 5.6. Prove that

1 1 1 cos 1◦
+ + · · · + = .
cos 0◦ cos 1◦ cos 1◦ cos 2◦ cos 88◦ cos 89◦ sin2 1◦
sin 1◦ sin((k+1)◦ −k◦ )
Proof. Since cos k◦ cos(k+1)◦ = cos k◦ cos(k+1)◦ = tan(k + 1)◦ − tan k ◦ ,

88 88
1
sin 1◦ = (tan(k + 1)◦ − tan k ◦ ) = tan 89◦ = cot 1◦ .
cos k ◦ cos(k + 1)◦
k=0 k=0

1 1 1 cos 1◦
Hence, cos 0◦ cos 1◦ + cos 1◦ cos 2◦ + ···+ cos 88◦ cos 89◦ = sin2 1◦
.
324 IMO Problems, Theorems, and Methods: Algebra

(6) Mathematical induction

For propositions related to positive integers n, mathematical induction can
be considered.

Example 5.7. Prove that

1 α 1 α 1 α 1 α
tan + tan + · · · + n tan n = n cot n − cot α.
2 2 4 4 2 2 2 2
cos α cos2 α 2 α
2 −sin 2 1−tan2 α 2
Proof. For n = 1, we have cot α = sin α = 2 sin α cos α = 2 tan α .
2 2 2
Therefore,
1 α 1 1 − tan2 α2 1 α
cot − cot α = α − α = tan .
2 2 2 tan 2 2 tan 2 2 2
1
Assume 2 tan α2 + 1
4 tan α4 + · · · + 1
2k tan 2αk = 1
2k cot 2αk − cot α. For
n = k + 1,
α
1 α 1 α 1 1 − tan2 2k+1 1 α
cot + tan = · α + k+1 tan k+1
2k 2k 2k+1 2k+1 2k 2 tan 2k+1 2 2
1 α
= cot k+1 .
2k+1 2
Thus,
 
1 α 1 α 1 α 1 α
tan + tan + · · · + k tan k + tan k+1
2 2 4 4 2 2 2k+1 2
1 α
= cot k+1 − cot α.
2k+1 2
Hence, the proposition holds for any positive integer n.

5.1.3 Common polynomials

(1) Univariate polynomials
Let x be an independent variable, an algebraic expression of the form
f (x) = an xn + an−1 xn−1 + · · · + a1 x + a0
is called a univariate polynomial in x, or simply a polynomial. And
a0 , a1 , . . . , an are known as the coefficients of the polynomial. If a0 = a1 =
· · · = an = 0, thenf (x) is termed the zero polynomial.
When an = 0, we call n as the degree of the polynomial, denoted as
deg f = n. If deg f = 0, then f (x) is a 0th degree polynomial. However,
the degree of the zero polynomial is undefined.
 Other Algebra Problems 325

Note. N+ , Z, Q, R, and C respectively represent the sets of positive inte-

gers, integers, rational numbers, real numbers, and complex numbers. The
notations Z[x], Q[x], R[x], and C[x] respectively denote the sets of univari-
ate polynomials in x with integer, rational, real, and complex coefficients.

Proposition 5.1. Two polynomials f (x) = an xn + · · · + a1 x + a0 and

g(x) = bm xm + · · · + b1 x + b0 are equal, where an bm = 0, if and only if
m = n and ai = bi for i = 0, 1, · · · , n.

Proposition 5.2. If f (x) and g(x) are non-zero polynomials, then

deg(f · g) = deg f + deg g,
and when f (x) ± g(x) is a non-zero polynomial,
deg(f ± g) ≤ max{deg f, deg g}.

(2) Integer polynomials

A polynomial with all coefficients being integers is called an integer poly-
nomial. Similarly, one can describe a polynomial with rational, real, and
complex coefficients as rational, real, and complex polynomial respectively.
Theorem 5.7 (Gauss’s Theorem). If an integer polynomial f (x) of
degree n(n > 0) can be factored into a product of nonconstant rational
polynomials, then it can also be factored into a product of nonconstant
integer polynomials.
Theorem 5.8 (Eisenstein’s Criterion). Let f (x) = an xn + · · · + a1 x +
a0 (an = 0) be an integer polynomial. If there exists a prime number p such
that:
(i) p  an ;
(ii) p|ai for i = 0, 1, . . . , n − 1;
(iii) p2  a0 ,
then f (x) is irreducible over the rational number field.

Corollary 5.1. Let f (x) = an xn + · · · + a1 x + a0 (an = 0) be an integer

polynomial. If there exist a prime number p and a positive number m ≤ n
such that:

(i) p  an ;
(ii) p|ai for i = 0, 1, . . . , m − 1;
(iii) p2  a0 ,
326 IMO Problems, Theorems, and Methods: Algebra

then f (x) has an irreducible factor with integer coefficients, whose degree is
not less than m.

Proposition 5.3. Let f (x) = an xn + · · · + a1 x + a0 (an = 0) be an integer

polynomial. If a rational number pq (in reduced form) is a root of f (x), then

(i) q|an and p|a0 ;

(ii) f (x) = x − pq g(x), where g(x) ∈ Z[x].

Corollary 5.2. If the leading coefficient of an integer polynomial f (x) is

1, then the rational roots of f (x) can only be integers.

Corollary 5.3. The integer roots of an integer polynomial are necessarily

divisors of its constant term.

Proposition 5.4. Let integer m be a divisor of the constant term of an

integer polynomial f (x) and k(k = m) is any integer. Then m is not a root
of f (x) if f (k) is not divisible by m − k.

Proposition 5.5. Let f (x) be an irreducible rational polynomial and g(x)

is a rational polynomial. If f (x) and g(x) have a common root, then
f (x)|g(x).
√
Corollary 5.4. If a rational √ polynomial f (x) has an irrational root a+ b,
where√ a, b are rational and b is irrational, then f (x) has the irrational root
a − b.
√
Corollary
√ 5.5. If a rational polynomial
√ f√(x) has an irrational root a b +
c d, where a, b, c, d are rational, b and d are irrational and√not of √ the
same√ type,
√ and ac 
= √ 0, then
√ f (x) has the irrational roots a b − c d,
−a b + c d, and −a b − c d.

(3) Integer-valued polynomials

A polynomial f (x) is called an integer-valued polynomial if, for any integer
value of x, the value of f (x) is an integer.

(4) Difference Polynomials

1
The polynomial Pk (x) = k! x(x − 1) · · · (x − k + 1), where k is a positive
integer and P0 (x) = 1, is referred to as the difference polynomial of degree
k and denoted as Ckx .
 Other Algebra Problems 327

If x is a positive integer n greater than or equal to k, then Pk (n) = Ckn

is an integer. If x is a negative integer −m, then Pk (−m) = (−1)k Ckm+k−1
is also an integer. If x ∈ {0, 1, . . . , k − 1}, then Pk (x) = 0. Therefore,
difference polynomials are integer-valued polynomials.
Similar to the properties of binomial coefficients, Pk (x + 1) − Pk (x) =
Pk−1 (x). Furthermore, we can define the first order difference of a polyno-
mial f (x) as
Δf (x) = f (x + 1) − f (x),
and (r + 1)th order difference of f (x) as
Δr+1 f (x) = Δ(Δr f (x)) = Δr f (x + 1) − Δr f (x).
n n
Proposition 5.6. Δn f (x) = i=0 (−1)
n−i i
Cn f (x + i) = i=0 (−1)
i

Cin f (x + n − i).

Proposition 5.7. If f (x) is a polynomial of degree n with the leading

coefficient an , then Δn f (x) = n!an .

Proposition 5.8. If Δn+1 f (x) = 0 for n ∈ N, then f (x) is a polynomial

in x of degree not greater than n.

Proposition 5.9. Any polynomial f (x) of degree n can be uniquely

expressed as
f (x) = bn Cnx + bn−1 Cn−1
x + · · · + b1 C1x + b0 ,
where bk (k = 0, 1, . . . , n) are constants.
Obviously, when bk (k = 0, 1, . . . , n) are all integers, f (x) is an integer-
valued polynomial. Conversely, if f (x) is an integer-valued polynomial, then
bk (k = 0, 1, . . . , n) are all integers.

Proposition 5.10. Let f (x) be a polynomial of degree n. Then

n n
f (k) = Δi f (0) · Ci+1
n+1 .
k=0 i=0

Proposition 5.11. A polynomial f (x) of degree n is an integer-valued

polynomial if and only if, when x takes n + 1 consecutive integer values, the
values of f (x) are all integers.

Proposition 5.12. For integer-valued polynomials

fk (x) = L(k) · Ckx (k = 1, 2, . . .)
328 IMO Problems, Theorems, and Methods: Algebra

and any two distinct integers a and b,

(a − b)|(fk (a) − fk (b)),

where L(k) is the least common multiple of 1, 2, . . . , k.

Proposition 5.13. Let f (x) = dk=0 ak Ckx be an integer-valued polynomial

of degree d. If for integers i, j satisfying 0 ≤ i = j ≤ d,

(i − j)|(f (i) − f (j)),

then L(k)|ak (0 ≤ k ≤ d), and for all distinct integers a and b,

(a − b)|(f (a) − f (b)),

where L(k) is the least common multiple of 1, 2, . . . , k.

(5) Multivariate polynomials

A polynomial of the form

f (x1 , x2 , . . . , xn ) = ak1 k2 ···kn xk11 xk22 · · · xknn

is termed an n-variable polynomial of degree m, where x1 , x2 , . . . , xn are

variables, and m = max{k1 + k2 + · · · + kn } with ki ∈ N(i = 1, 2, . . . , n).
If every monomial of f (x1 , x2 , . . . , xn ) is of degree m, then
f (x1 , x2 , . . . , xn ) is called a homogeneous polynomial of degree m, and in
this case,

f (tx1 , tx2 , . . . , txn ) = tm f (x1 , x2 , . . . , xn ).

If substituting x1 with x2 , x2 with x3 , . . . , xn−1 with xn , and xn with

x1 in f (x1 , x2 , . . . , xn−1 , xn ) always results in

f (x1 , x2 , . . . , xn−1 , xn ) = f (x2 , x3 , . . . , xn , x1 ),

then f (x1 , x2 , . . . , xn ) is called a cyclic polynomial.

If for any permutation i1 , i2 , . . . , in of 1, 2, . . . , n,

f (x1 , x2 , . . . , xn ) = f (xi1 , xi2 , . . . , xin ),

then f (x1 , x2 , . . . , xn ) is called a symmetric polynomial. Clearly, every sym-

metric polynomial is a cyclic polynomial, but not every cyclic polynomial
is symmetric.
 Other Algebra Problems 329

For n-variable symmetric polynomials, we introduce the following nota-

tions:
σ1 = x1 + x2 + · · · + xn ,
σ2 = xi1 xi2 ,
1≤i1 <i2 ≤n
························
σk = xi1 xi2 · · · xik ,
1≤i1 <i2 <···<ik ≤n
························
σn = x1 x2 · · · xn ,
where σ1 , σ2 , . . . , σn are called the elementary symmetric polynomials in
x1 , x2 , . . . , xn .
Theorem 5.9 (Fundamental Theorem of Symmetric Polynomials).
Any symmetric polynomial f (x1 , x2 , . . . , xn ) can be uniquely expressed as
a polynomial in σ1 , σ2 , . . . , σn , i.e.,
f (x1 , x2 , . . . , xn ) = ϕ(σ1 , σ2 , . . . , σn ).

Theorem 5.10 (Newton’s Identities). Let Sk = xk1 + xk2 + · · · + xkn (k ∈

N+ ), where x1 , x2 , . . . , xn are roots of the polynomial
f (x) = xn + a1 xn−1 + · · · + an−1 x + an .
Then,
(i) for k ≤ n, there is Sk + a1 Sk−1 + · · · + ak−1 S1 + kak = 0;
(ii) for k > n, there is Sk + a1 Sk−1 + · · · + an Sk−n = 0.

(6) Cyclotomic polynomials

2πi
Let n be a positive integer, and εn = e n be one of the nth roots of
unity. The polynomial Φn (x) = 1≤k<n,(k,n)=1 x − εkn is termed the nth
cyclotomic polynomial.

Proposition 5.14. For a positive integer n, xn − 1 = d|n Φd (x).

Proposition 5.15. For a positive integer n, Φn (x) is an irreducible integer

polynomial.
 
Proposition 5.16. Φn (x) = xϕ(n) Φn x1 , where ϕ(n) is the Euler’s
totient function.
330 IMO Problems, Theorems, and Methods: Algebra

Proposition 5.17. Let p be a prime number.

Φn (xp )
(i) If p  n, then Φnp (x) = Φn (x) .

(ii) If p|n, then Φnp (x) = Φn (xp ).

n
Proposition 5.18. Φn (x) = d|n (x
d
− 1)μ( d ) , where μ(n) is the Möbius
function.

(7) Chebyshev polynomials

Let x = cos θ. Then cos nθ can be expressed as a polynomial of degree n in
x, known as the Chebyshev polynomial, denoted as Tn (x) = cos(n arccos x),
where |x| ≤ 1. Alternatively, Tn (x) can be represented as
 n2 
Tn (x) = (−1)k C2k
n x
n−2k
(1 − x2 )k ,
k=0

where x represents the greatest integer less than or equal to x.

Chebyshev polynomials satisfy the recurrence relation
Tn+1 (x) = 2xTn (x) − Tn−1 (x).

Proposition 5.19. The polynomial Tn (x) is of degree n, with the leading

coefficient 2n−1 .

Proposition 5.20. For all x ∈ C and n ∈ N, Tn (−x) = (−1)n Tn (x), i.e.,

Tn (x) is an odd (even) function when n is odd (even).

Proposition 5.21. If |x| ≤ 1 with x ∈ R, then |Tn (x)| ≤ 1.

Proposition 5.22. For all m ∈ N, T2m+1 (0) = 0 and T2m (0) = (−1)m .

Proposition 5.23. The polynomial Tn (x) has n distinct real roots in the
interval [−1, 1], given by
(2k − 1) π
xk = cos (k = 1, 2, . . . , n).
2n
Proposition 5.24. The polynomial Tn (x) has n + 1 points in the inter-
val [−1, 1], xk = cos kπ
n (k = 0, 1, 2, . . . , n), where it alternates between the
maximum value 1 and the minimum value −1.

Proposition 5.25. Another recurrence relation for Tn (x) is Tn

y+y −1 n −n

2 = y +y
2 , for y ∈ C with y = 0.
 Other Algebra Problems 331

Proposition 5.26. The generating function for the sequence {Tn (x)} is
1 − xt
Tn (x)tn = , for t ∈ R with |t| ≤ 1.
1 − 2xt + t2
n≥0

5.1.4 Common techniques for polynomials

(1) Factorization
A polynomial can be completely or partially factorized into a product of
factors or decomposed into a sum of several parts to reveal its properties.
Especially when it is known that a polynomial f (x) has a root α, it can be
factored as

f (x) = (x − α)q(x).

Example 5.8. Let f (x) be an integer polynomial. Prove that if the abso-
lute values of f (x) are all 1 when x takes three distinct integers, then f (x)
has no integer roots.

Proof. Suppose f (x) has an integer root x0 . Then f (x) can be factored
as f (x) = (x − x0 )q(x), where q(x) is an integer polynomial. Let a, b, c be
three distinct integers such that |f (a)| = |f (b)| = |f (c)| = 1. Then

|a − x0 | = |b − x0 | = |c − x0 | = 1,

implying that a, b, c must be x0 ± 1. Consequently, at least two of these

values must be equal, which is a contradiction.

(2) Parity analysis

When dealing with problems related to integer polynomials, parity analysis
is often employed to analyze the parity of coefficients, integer roots, and
the values of the polynomial, so as to deduce relationships among these
quantities.

Example 5.9. Prove that if an integer coefficient equation ax2 + bx + c =

0(a = 0) has a rational root, then at least one of a, b, c must be even.

Proof. The equation has a rational root if and only if its discriminant
is a perfect square. If a, b, c are all odd, assume b = 2m + 1, then Δ =
4(m(m + 1) − ac) + 1. Clearly, m(m + 1) − ac is odd, denoted as 2n + 1,
so Δ = 8n + 5, implying Δ is an odd number and leaves a remainder of 5
when divided by 8.
332 IMO Problems, Theorems, and Methods: Algebra

However, the square of any odd number 2k + 1 is 4k(k + 1) + 1, which is

an integer that leaves a remainder of 1 when divided by 8. Thus, Δ cannot
be a perfect square, indicating that at least one of a, b, c must be even.

(3) Factor analysis

When dealing with problems involving integer polynomials, if the polyno-
mial f (x) can be factored as f (x) = g(x)h(x), then taking a specific integer
value n for x implies that g(n)|f (n). A subsequent analysis can then be
conducted on the values of f (n) and g(n). This is typically applicable in
situations where f (x) equals a given integer for multiple integer values of
x.

Example 5.10. Let a, b, c, d be distinct integers, and the equation

(x − a)(x − b)(x − c)(x − d) = 25
has an integer root. Prove that 4|(a + b + c + d).

Proof. Assume x0 is an integer root. Then 25 is the product of four dis-

tinct factors x0 − a, x0 − b, x0 − c, x0 − d, implying that these four numbers
must be ±1 and ±5. Therefore,
(x0 − a) + (x0 − b) + (x0 − c) + (x0 − a) = 1 + 5 + (−1) + (−5) = 0,
which implies that a + b + c + d = 4x0 .

(4) Root analysis

By utilizing properties related to the roots of a polynomial, such as the
number of roots not exceeding the degree of the polynomial, an analysis
can be conducted on the number of roots and their multiplicities, thereby
deducing special properties of the polynomial or relationships between two
polynomials.

Example 5.11. Find the polynomial f (x) such that f (x2 + 1) = f 2 (x) + 1
with f (0) = 0.

Solution. From the given conditions,

f (1) = f (02 + 1) = f 2 (0) + 1 = 1,
f (2) = f (12 + 1) = f 2 (1) + 1 = 2,
f (5) = f (22 + 1) = f 2 (2) + 1 = 5,
·································
 Other Algebra Problems 333

Construct a sequence {xn } with x0 = 0 and xn = x2n−1 + 1 for n ∈

{1, 2, . . .}. By mathematical induction, we can prove f (xn ) = xn for all
n ∈ {1, 2, . . .}, so the polynomial f (x) − x has infinitely many roots, which
implies f (x) − x = 0, i.e., f (x) = x.
There is a similar problem:
• (Polish Mathematical Olympiad Finals 2000, Problem 6).
Suppose that P (x) is a polynomial of an odd degree satisfying

P (x2 − 1) = (P (x))2 − 1 for all x.

Prove that P (x) = x for all x.

(5) Mathematical Induction

When proving properties of polynomials or solving polynomial equations,
the characteristics of some polynomials are often given, allowing the use of
mathematical induction based on the degree of the polynomial, the number
of iterations of the polynomial, the number of roots, etc.

Example 5.12. Prove that for any positive integer n, the polynomial
(x + 1)2n+1 + xn+2 is divisible by the polynomial x2 + x + 1.

Proof. To prove this, use mathematical induction. When n = 1, the con-

clusion is obviously true. Assume the conclusion holds for n ∈ {1, 2, . . . ,
k − 1}.
When n = k,

(x + 1)2k+1 + xk+2 = (x + 1)2 (x + 1)2k−1 + x · xk+1

= (x2 + 2x + 1)(x + 1)2k−1 + x · xk+1
= (x2 + x + 1)(x + 1)2k−1 + x((x + 1)2k−1 + xk+1 ),

and thus, the conclusion is valid.

(6) Inequality analysis

If relationships among the coefficients, roots, and values of a polynomial are
known, an inequality analysis can be employed to deduce the polynomial’s
properties.

Example 5.13. Prove that if a real coefficient equation x3 + ax − 2 = 0

has three real roots, then at least one real root must be greater than or
equal to 2.
334 IMO Problems, Theorems, and Methods: Algebra

Proof. Let the three real roots be x1 , x2 , x3 . Then x1 + x2 + x3 = 0 and

x1 x2 x3 = 2, implying that the roots consist of two negatives and one posi-
tive.
√
Assume x1 > 0 and x2 , x3 < 0. Then x1 = (−x2 ) + (−x3 ) ≥ 2 x2 x3 .
 2
Since 2 = x1 x2 x3 ≤ x1 12 x1 , it follows that x1 ≥ 2.

(7) Polynomial construction

Based on the given conditions, construct an appropriate polynomial to
transform the original problem into a polynomial problem, and then use
properties related to polynomials for analysis.

Example 5.14. Let a, b, c be real numbers with absolute values less than 1.
Prove that ab + bc + ca + 1 > 0.

Proof. Construct a polynomial f (x) = (b + c)x + bc + 1 for |x| < 1.

Its graph is a line segment excluding the two endpoints (−1, f (−1)) and
(1, f (1)). If it can be proven that the function values at these two endpoints,
f (−1) and f (1), are both greater than 0, then for every point x in its
domain, f (x) is always greater than 0, thereby proving the proposition.
Since
f (−1) = −(b + c) + bc + 1 = (1 − b)(1 − c) > 0,
f (1) = (b + c) + bc + 1 = (1 + b)(1 + c) > 0,
it follows that f (a) = a(b + c) + bc + 1 > 0.

(8) Difference method

This method is typically suitable for dealing with situations where the
variable x in a polynomial f (x) takes several consecutive integer val-
ues. By using the difference formula, Δn f (x) can be represented by
f (x + 1), f (x + 2), . . . , f (x + n), and then analyzed and solved via rele-
vant properties.

Example 5.15. Let p(x) be a polynomial of degree 2n satisfying

p(0) = p(2) = · · · = p(2n) = 0,
p(1) = p(3) = · · · = p(2n − 1) = 2,
p(2n + 1) = −30.
Find n.
 Other Algebra Problems 335

2n+1 k
Solution. From Δ2n+1 p (0) = k=0 (−1) Ck2n+1 p (2n + 1 − k) = 0,
n
(−30) + C2j
2n+1 · 2 = 0,
j=1
1 
which simplifies to (−30) + 2 2 · 22n+1 − 1 = 0. Thus, n = 2.

5.1.5 Other important theorems

Theorem 5.11 (Division Theorem). For polynomials f (x) and g(x)
with g(x) = 0, there exist unique polynomials q(x) and r(x) such that

f (x) = g(x)q(x) + r(x),

where r(x) = 0 or deg r(x) < deg g(x). The polynomial q(x) is called the
quotient of dividing f (x) by g(x), and r(x) is the remainder.

Theorem 5.12 (Remainder Theorem). The remainder of dividing a

polynomial f (x) by x − a is f (a).

Theorem 5.13 (Factor Theorem). A polynomial f (x) has a factor x−a

if and only if f (a) = 0.

Theorem 5.14 (Bezout’s Identity). If the greatest common divisor of

polynomials f (x) and g(x) is d(x), then there exist polynomials u(x) and
v(x) such that f (x)u(x) + g(x)v(x) = d(x).

Corollary 5.6. Two polynomials f (x) and g(x) are coprime if and only if
there exist polynomials u(x) and v(x) such that f (x)u(x) + g(x)v(x) = 1.

Theorem 5.15 (Unique Factorization Theorem). A non-constant

polynomial f (x) over a field F can be uniquely factored into the follow-
ing form

f (x) = apk11 (x)pk22 (x) · · · pkt t (x),

where a is the leading coefficient of f (x) and p1 (x), p2 (x), . . . , pt (x) are dis-
tinct irreducible polynomials with leading coefficients of 1. Here, pi (x) (i =
1, 2, . . . , t) are the ki -fold factors of f (x).
Theorem 5.16 (Fundamental Theorem of Algebra). Any complex
polynomial of degree n(n > 0) has, counted with multiplicity, exactly n
roots in the complex field.
336 IMO Problems, Theorems, and Methods: Algebra

Corollary 5.7. Let f (x) be a complex polynomial of degree n(n > 0). Then
it can be uniquely factored in the complex field as
f (x) = a(x − α1 )m1 (x − α2 )m2 · · · (x − αt )mt ,
where αi (i = 1, 2, . . . , t) are distinct complex numbers and mi -fold roots of
t
f (x), with i=1 mi = n, and a is the leading coefficient of f (x).

Theorem 5.17 (Complex Conjugate Root Theorem). If a real poly-

nomial f (x) has a non-real complex root α, then its conjugate ᾱ is also a
root of f (x), and α and ᾱ have the same multiplicity. In other words, the
non-real roots of a real polynomial appear in conjugate pairs.

Corollary 5.8. Any real polynomial of odd degree has at least one real
root.

Theorem 5.18 (Lagrange Interpolating Polynomial). Let f (x) be

a single-valued function with x0 < x1 < · · · < xn , and f (xi ) = yi for
i = 0, 1, . . . , n. There exists a unique polynomial p(x) with deg p(x) ≤ n
such that p(xi ) = yi , where
n n
x − xj
p(x) = f (xi ) ri (x) and ri (x) = .
i=0
x
j=0 i
− xj
j=i

Corollary 5.9. If f (x) is a polynomial of degree n, then for any integer k,

n
n!an = (−1)n−i Cin f (k + i),
i=0

where an is the leading coefficient of f (x).

5.2 Problems and Solutions

5.2.1 Proving trigonometric identities
Problem 5.1 (IMO 5-5, proposed by the German Democratic.
Republic). Prove that cos π7 − cos 2π 3π 1
7 + cos 7 = 2 .

Proof 1. It is evident that

π
 
π 2π 3π 2 cos 14 cos π7 − cos 2π 3π
7 + cos 7
cos − cos + cos = π .
7 7 7 2 cos 14
 Other Algebra Problems 337

Using the product-to-sum formula for cosine, we have

π π 3π π
2 cos cos = cos + cos ,
14 7 14 14
π 2π 5π 3π
2 cos cos = cos + cos ,
14 7 14 14
π 3π 7π 5π
2 cos cos = cos + cos .
14 7 14 14
Thus,

π 2π 3π cos 7π π
14 + cos 14 1
cos − cos + cos = π = .
7 7 7 2 cos 14 2
π
Proof 2. As illustrated in Figure 5.1, construct ∠M ON = 7.

Figure 5.1 ∠M ON .

On the line segment OM , choose a point A such that OA = 1. On the

line segment ON , distinct from O, choose a point B such that AB = 1.
On OM , distinct from A, choose a point C such that BC = 1. On ON ,
distinct from B, choose a point D such that CD = 1. Then,
2π 3π
∠CAB = ∠ACB = and ∠CBD = ∠CDB = .
7 7
Thus, ∠OCD = 3π 7 and OC = OD.
Construct AH⊥OB. Then OB = 2OH = 2 cos π7 . Similarly, AC =
2 cos 2π 3π
7 and BD = 2 cos 7 . From OA + AC = OC = OD = OB + BD,

2π π 3π
1 + 2 cos = 2 cos + 2 cos ,
7 7 7
which implies cos π7 − cos 2π 3π 1
7 + cos 7 = 2 .
338 IMO Problems, Theorems, and Methods: Algebra

Proof 3. Consider the roots of unity from z 7 = 1:

2(k − 1)π 2(k − 1)π

zk = cos + i sin for k = 1, 2, . . . , 7.
7 7
It is evident that z1 + z2 + · · · + z7 = 0. Thus
2π 4π 6π 8π 10π 12π
0 = 1 + cos + cos + cos + cos + cos + cos
7 7 7 7 7 7
 
2π 4π 6π
= 1 + 2 cos + cos + cos .
7 7 7

Since cos 6π π 4π 3π
7 = − cos 7 and cos 7 = − cos 7 ,

π 2π 3π 1
cos − cos + cos = .
7 7 7 2

Note. When z n = 1, each nth root of 1 is called an nth root of unity.

There are n such nth roots of unity, denoted as ε0 , ε1 , . . . , εn−1 , where
εk = cos 2kπ 2kπ
n + i sin n for k = 0, 1, 2, . . . , n − 1. This indicates that the
product of two nth roots of unity is still an nth root of unity, and the
integer power of an nth root of unity is also an nth root of unity.

(i) Let m be an integer and n be a positive integer greater than 1. Then


n, if n|m,
1 + εm m m
1 + ε2 + · · · + εn−1 =
0, if n  m.

In particular, 1 + ε1 + ε2 + · · · + εn−1 = 0.
By (i), we can prove the following proposition.
(ii) Let n be a positive integer. Then

π 3π (2n − 1) π
cosk + cosk + · · · + cosk
2n + 1 2n + 1 2n + 1
1
= for k = 1, 3, . . . , 2n − 1.
2
(n−1)π
(iii) sin πn sin 2π
n · · · sin n
n
= 2n−1 .
 
 (n−1)π  1−(−1)n
(iv) cos πn cos 2πn · · · cos n = 2n .
√
π (n−1)π n
(v) sin 2n sin 2π
2n · · · sin 2n = 2n−1 .
 Other Algebra Problems 339

π 2π nπ 1
(vi) cos 2n+1 cos 2n+1 · · · cos 2n+1 = 2n .
√
π 2π nπ 2n+1
(vii) sin 2n+1 sin 2n+1 · · · sin 2n+1 = 2n .

Score Situation This particular problem saw the following distribution of scores among
contestants: 3 contestants scored 6 points, 3 contestants scored 5 points, no contestant
scored 4 points, 1 contestant scored 3 points, no contestant scored 2 points, 2 contestants
scored 1 point, and 7 contestants scored 0 point. The average score of this problem is 2.375,
indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the Soviet Union team achieved a total
score of 271 points, the Hungary team achieved a total score of 234 points, the Romania
team achieved a total score of 191 points, the Yugoslavia team achieved a total score of 162
points, and the Czechoslovakia team achieved a total score of 151 points.
The gold medal cutoff for this IMO was set at 35 points (with 7 contestants earning gold
medals), the silver medal cutoff was 28 points (with 11 contestants earning silver medals),
and the bronze medal cutoff was 21 points (with 17 contestants earning bronze medals).
In this IMO, no contestant achieved a perfect score of 40 points.

Problem 5.2 (IMO 8-4, proposed by Yugoslavia). Prove that for

every positive integer n and real number x = kπ
2t (t = 0, 1, . . . , n; k any
integer),

1 1 1
+ + ··· + = cot x − cot 2n x.
sin 2x sin 4x sin 2n x

Proof. Use mathematical induction.

For x = kπ t
2t , the number cot 2 x is well defined. When n = 1,

cos x cos 2x
cot x − cot 2x = −
sin x sin 2x
sin 2x cos x − cos 2x sin x
=
sin x sin 2x
sin x
=
sin x sin 2x
1
= .
sin 2x

Therefore, the conclusion holds for n = 1.

340 IMO Problems, Theorems, and Methods: Algebra

Assume the conclusion holds for n = i. When n = i + 1,

1 1 1 1
+ + ···+ +
sin 2x sin 4x sin 2i x sin 2i+1 x
1
= cot x − cot 2i x + .
sin 2i+1 x
From the case for n = 1, it is evident that
1
= cot 2i x − cot 2i+1 x.
sin 2i+1 x
Thus, the conclusion holds for any positive integer n.
Note. (i) tan α + 2 tan 2α + · · · + 2n tan 2n α = cot α − 2n+1 cot 2n+1 α.
(ii) tan α tan 2α + tan 2α tan 3α + · · · + tan(n − 1)α tan nα = tan nα
tan α − n.
Furthermore, there is a similar problem:

• (William Lowell Putnam Mathematical Competition 1973,

B6). On the domain 0 ≤ θ ≤ 2π:
(a) Prove that sin2 θ · sin 2θ takes its maximum at π3 and 4π 3 . (Hence,
2π 5π
its minimum at 3 and 3 .)
(b) Show that
 2 
sin θ · sin3 2θ · sin3 4θ · · · sin3 2n−1 θ · sin 2n θ
π
takes its maximum at θ = 3. (The maximum may also be
attained at other points.)
(c) Derive the inequality:
 n
3
sin2 θ · sin2 2θ · sin2 4θ · · · sin2 2n θ ≤ .
4

Score Situation This particular problem saw the following distribution of scores among con-
testants: 23 contestant scored 5 points, 3 contestants scored 4 points, no contestant scored
3 points, no contestants scored 2 points, no contestants scored 1 point, and 1 contestant
scored 0 point. The average score of this problem is 4.704, indicating that it was simple.
Among the top five teams in the team scores, the Soviet Union team achieved a total
score of 293 points, the Hungary team achieved a total score of 281 points, the German
Democratic Republic team achieved a total score of 280 points, the Poland team achieved a
total score of 269 points, and the Romania team achieved a total score of 257 points.
The gold medal cutoff for this IMO was set at 39 points (with 13 contestants earning gold
medals), the silver medal cutoff was 34 points (with 15 contestants earning silver medals),
and the bronze medal cutoff was 31 points (with 11 contestants earning bronze medals).
In this IMO, a total of 11 contestants achieved a perfect score of 40 points.
 Other Algebra Problems 341

5.2.2 Finding polynomials

Problem 5.3 (IMO 17-6, proposed by the United Kingdom). Find
all polynomials P in two variables with the following properties:

(i) for a positive integer n and all real t, x, and y,

P (tx, ty) = tn P (x, y) (that is, P is homogeneous of degree n);

(ii) for all real a, b, and c,

P (b + c, a) + P (c + a, b) + P (a + b, c) = 0;

(iii) P (1, 0) = 1.

Solution 1. From conditions (ii) and (iii), P (x, y) is not a constant. Set-
ting a = b and c = −2a in (ii), we obtain

0 = P (2a, −2a) + P (−a, a) + P (−a, a)

n
= ((−2) + 2) P (−a, a) .

For n > 1, we have P (−a, a) = 0, implying P (x, y) = 0 for any y = −x.

It follows that P (x, y) is divisible by x + y. Therefore, we can assume

P (x, y) = (x + y) P1 (x, y).

It is evident that P1 (x, y) is a homogeneous polynomial of degree n − 1,

satisfying conditions (i), (ii), and (iii). Consequently,

P (x, y) = (x + y) P1 (x, y)
= (x + y)2 P2 (x, y)
··················
n−1
= (x + y) Pn−1 (x, y).

Here, Pn−1 (x, y) = Ax + By is a homogeneous polynomial of degree 1, still

meeting conditions (i), (ii), and (iii).
Setting a = b = c = x, we get 3Pn−1 (2x, x) = 0, which implies
2Ax + Bx = 0. The arbitrariness of x indicates 2A + B = 0. Further, from
Pn−1 (1, 0) = 1, we deduce A = 1. Thus B = −2 and Pn−1 (x, y) = x − 2y.
342 IMO Problems, Theorems, and Methods: Algebra

Hence, P (x, y) = (x + y)n−1 (x − 2y), which includes the case n = 1.

Solution 2. Setting b = 1 − a and c = 0 in (ii), we have
P (1 − a, a) + P (a, 1 − a) + P (1, 0) = 0.
Since P (1, 0) = 1,
P (1 − a, a) = −1 − P (a, 1 − a). (1)
Setting c = 1 − a − b in (ii), we get
P (1 − a, a) + P (1 − b, b) + P (a + b, 1 − a − b) = 0,
and by (1), this implies that
−2 − P (a, 1 − a) − P (b, 1 − b) + P (a + b, 1 − a − b) = 0. (2)
Therefore, with f (x) = P (x, 1 − x) + 2, we see that (2) can take the
form
f (a + b) = f (a) + f (b).
Since P (x, y) is continuous, so is f (x). Obviously, f (x) is known as
Cauchy’s functional equation (which was presented in Chapter 2). Hence,
f (x) = f (1) · x = (P (1, 0) + 2)x = 3x.
By definition f (x) = P (x, 1 − x) + 2, so P (x, 1 − x) = 3x − 2. If
a + b = 0, then
   
n a b n 3a
P (a, b) = (a + b) P , = (a + b) −2
a+b a+b a+b
= (a + b)n−1 (a − 2b).
Since P (x, y) is continuous, it follows that P (x, y) = (x + y)n−1 (x − 2y)
even when x + y = 0. It is easily verified that the polynomial
P (x, y) = (x + y)n−1 (x − 2y)
meets (i), (ii), and (iii).
Score Situation This particular problem saw the following distribution of scores among
contestants: 19 contestants scored 8 points, 1 contestant scored 7 points, 5 contestants
scored 6 points, 5 contestants scored 5 points, 8 contestants scored 4 points, 13 contestants
scored 3 points, 27 contestants scored 2 points, 19 contestants scored 1 point, and 31
contestants scored 0 point. The average score for this problem is 2.797, indicating that it
had a certain level of difficulty.
 Other Algebra Problems 343

Among the top five teams in the team scores, the scores of this problem are as follows:
the Hungary team scored 33 points (with a total team score of 258 points), the German
Democratic Republic team scored 33 points (with a total team score of 249 points), the
United States team scored 39 points (with a total team score of 247 points), the Soviet Union
team scored 28 points (with a total team score of 246 points), and the United Kingdom team
scored 44 points (with a total team score of 241 points).
The gold medal cutoff for this IMO was set at 38 points (with 8 contestants earning gold
medals), the silver medal cutoff was 32 points (with 25 contestants earning silver medals),
and the bronze medal cutoff was 23 points (with 36 contestants earning bronze medals).
In this IMO, a total of six contestants achieved a perfect score of 40 points.

Problem 5.4 (IMO 45-2, proposed by South Korea). Find all poly-
nomials f with real coefficients such that for all reals a, b, c satisfying
ab + bc + ca = 0, we have the following relation:

f (a − b) + f (b − c) + f (c − a) = 2f (a + b + c).

Solution. For any given a, b, c ∈ R with ab + bc + ca = 0,

f (a − b) + f (b − c) + f (c − a) = 2f (a + b + c). (1)

Setting a = b = c = 0 in (1), we obtain f (0) = 0. Setting b = c = 0

in (1), we find f (−a) = f (a) for any real number a. Consequently, all
coefficients of the odd-degree terms of f (x) are 0. Assume

f (x) = an x2n + an−1 x2(n−1) + · · · + a1 x2 (an = 0).

Setting b = 2a and c = − 23 a in (1), we have

     
8 5 7
f (−a) + f a + f − a = 2f a ,
3 3 3
n  2i  5 2i  2i 2i
which implies i=1 ai 1 + 83 + 3 − 2 73 a = 0 for all a ∈ R.
Since an =0 for n ≥ 3, from 86 = 262144
 8 2n  8 2n > 235298
 = 2 × 76 , it follows
8 6 5 2n 2n
that 7 ≥ 7 > 2, and thus 1 + 3 + 3 − 2 73 > 0. This
contradicts
n
  2i  2i  2i 
8 5 7
ai 1 + + −2 a2i = 0
i=1
3 3 3

for all a ∈ R.
Therefore, n ≤ 2. Let f (x) = αx4 + βx2 , where α, β ∈ R. It can be
easily verified that f (x) = αx4 + βx2 satisfies the given conditions.
344 IMO Problems, Theorems, and Methods: Algebra

√ √
Note. We can also set a = (1 − 3)x, b = x, and c = (1 + 3)x. It follows
that
√ √ √
f (− 3x) + f (− 3x) + f (2 3x) = 2f (3x).
√
√Forn the coefficient of xn to be nonzero, we must have 2(− 3)n +
(2 3) = 2 · 3n , which only holds for n = 2, 4.
There is a similar problem:

• (United States of America Mathematical Olympiad 2019, Prob-

lem 6). Find all polynomials P with real coefficients such that

P (x) P (y) P (z)

+ + = P (x − y) + P (y − z) + P (z − x)
yz zx xy

for all nonzero real numbers x, y, z satisfying 2xyz = x + y + z.

Score Situation This particular problem saw the following distribution of scores among
contestants: 83 contestants scored 7 points, 31 contestants scored 6 points, 23 contestants
scored 5 points, 23 contestants scored 4 points, 32 contestants scored 3 points, 57 contestants
scored 2 points, 158 contestants scored 1 point, and 79 contestants scored 0 point. The
average score for this problem is 2.761, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 41 points (with a total team score of 220 points), the United States
team scored 41 points (with a total team score of 212 points), the Russia team scored
34 points (with a total team score of 205 points), the Vietnam team scored 28 points (with
a total team score of 196 points), and the Bulgaria team scored 36 points (with a total team
score of 194 points).
The gold medal cutoff for this IMO was set at 32 points (with 45 contestants earning gold
medals), the silver medal cutoff was 24 points (with 78 contestants earning silver medals),
and the bronze medal cutoff was 16 points (with 120 contestants earning bronze medals).
In this IMO, a total of four contestants achieved a perfect score of 42 points.

5.2.3 Proving properties

Problem 5.5 (IMO 16-6, proposed by Sweden). Let P be a non-
constant polynomial with integer coefficients. If n(P ) is the number of dis-
tinct integers k such that (P (k))2 = 1, prove that n(P )−deg(P ) ≤ 2, where
deg(P ) denotes the degree of the polynomial P .
 Other Algebra Problems 345

Proof 1. Since (P (x))2 − 1 = (P (x) − 1)(P (x) + 1), the integer roots of
(P (x))2 = 1 are all integer roots of either P (x) = 1 or P (x) = −1. We will
prove that one of the equations P (x) = 1 or P (x) = −1 must have at most
two integer roots, and by the Fundamental Theorem of Algebra, the other
equation has at most deg(P ) integer roots, thus proving the conclusion.
We use a proof by contradiction. Assume that both equations
P (x) + 1 = 0 and P (x)−1 = 0 have at least three distinct integer roots, and
these roots are different from each other. Among these six distinct integers,
let the smallest be a.
Without loss of generality, assume a is a root of P (x) + 1 = 0. Then
P (x) + 1 = (x − a)Q(x), where Q(x) is an integer polynomial.
Let b,c, and d be three distinct integer roots of P (x) − 1 = 0, all greater
than a. Since P (x) − 1 = (x − a) Q(x) − 2 and P (b) − 1 = P (c) − 1 =
P (d) − 1 = 0, it follows that
2 = (b − a)Q(b) = (c − a)Q(c) = (d − a)Q(d),
where b − a, c − a, and d − a are three distinct positive integers. Clearly,
at least one of them must be greater than 2 and cannot be a divisor of 2.
This leads to a contradiction.
Proof 2. First, we prove a lemma.
Lemma. If m is an integer root of the integer polynomial F (x). Then the
integer roots of the polynomial F (x) + p or F (x) − p can only be m − p,
m − 1, m + 1, or m + p, where p is a prime number.
Proof of Lemma. Suppose F (x) = (x − m)G(x). Then
F (x) ± p = (x − m) G(x) ± p,

where G(x) is an integer polynomial. If F (x0 ) ± p = 0 for an integer x0 ,

then (x0 − m)G(x0 ) = ∓p. Therefore, x0 − m divides p, and since p is a
prime number, x0 − m = ±1 or x0 − m = ±p. The lemma is proved.
Returning to the original problem, assume that (P (x))2 = 1 has integer
roots, with m being the smallest one, and without loss of generality, let it
be a root of P (x) − 1 = 0.
Since P (x) + 1 = P (x) − 1 + 2, which is the case with p = 2 in the
lemma, the integer roots of P (x) + 1 can only be at most two, m + 1 or
m + 2. The rest of the proof is similar to Proof 1.
Score Situation This particular problem saw the following distribution of scores among con-
testants: 35 contestants scored 8 points, 8 contestants scored 7 points, 3 contestants scored
346 IMO Problems, Theorems, and Methods: Algebra

6 points, 5 contestants scored 5 points, 1 contestant scored 4 points, 3 contestants scored

3 points, 7 contestants scored 2 points, 28 contestants scored 1 point, and 50 contestants
scored 0 point. The average score of this problem is 3.100, indicating that it was relatively
straightforward.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Soviet Union team scored 40 points (with a total team score of 256 points), the United
States team scored 58 points (with a total team score of 243 points), the Hungary team
scored 54 points (with a total team score of 237 points), the German Democratic Republic
team scored 42 points (with a total team score of 236 points), and the Yugoslavia team
scored 36 points (with a total team score of 216 points).
The gold medal cutoff for this IMO was set at 38 points (with 10 contestants earning gold
medals), the silver medal cutoff was 30 points (with 24 contestants earning silver medals),
and the bronze medal cutoff was 23 points (with 37 contestants earning bronze medals).
In this IMO, a total of six contestants achieved a perfect score of 40 points.

Problem 5.6 (IMO 26-3, proposed by the Netherlands). For any

polynomial P (x) = a0 + a1 x + · · · + ak xk with integer coefficients, the
number of coefficients which are odd is denoted by w(P ). For i = 0, 1, . . .,
let Qi (x) = (1 + x)i . Prove that if i1 , i2 , . . . , in are integers such that
0 ≤ i1 < i2 < · · · < in , then

w(Qi1 + Qi2 + · · · + Qin ) ≥ w(Qi1 ).

Proof. Let R and S be two polynomials with integer coefficients, and let
m be a positive integer. There are the following two lemmas:

Lemma 1. If k = 2m , then (1 + x)k ≡ 1 + xk (mod 2).

Lemma 2. If the degree of R is less than k, then

w(R + xk S) = w(R) + w(S).

These two lemmas are evidently true.

Define Q = Qi1 + Qi2 + · · · + Qin . When in ∈ {0, 1}, the conclusion
w(Q) ≥ w(Qi1 ) is obviously true. By mathematical induction, assume the
conclusion w(Q) ≥ w(Qi1 ) holds for in < 2m (m ≥ 1). When k = 2m ≤
in < 2m+1 , consider two cases:
 Other Algebra Problems 347

Case 1: i1 < k.
Suppose ir < k ≤ ir+1 . Then Q = R + (1 + x)k S, where
R = Qi1 + Qi2 + · · · + Qir , S = (1 + x)−k (Qir+1 + Qir+2 + · · · + Qin ).
The degrees of R and S are both less than k. By Lemma 1, in (1 + x)k ,
the terms that affect the number of odd coefficients are only 1 and xk .
Thus,
w(Q) = w(R + (1 + x)k S) = w(R + S + xk S)
= w(R + S) + w(S) ≥ w(R).
The last step follows from the triangle inequality. R has w (R) odd
coefficients, and after adding S, there are t of them that become even,
implying that S must have at least t odd coefficients, i.e.,
w(R + S) ≥ w(R) − t and w(S) ≥ t.
Adding these two inequalities yields w(R + S) + w(S) ≥ w(R).
By the induction hypothesis, w(R) ≥ w(Qi1 ), so w(Q) ≥ w(Qi1 ).
Case 2: i1 ≥ k.
Suppose Qi1 = (1 + x)k R and Q = (1 + x)k S, where R = (1 + x)−k Qi1 =
(1 + x)i1 −k and S = (1 + x)−k Q = (1 + x)i1 −k + (1 + x)i2 −k + · · · +
(1 + x)in −k .
Then, the degrees of R and S are both less than k. By Lemma 1 and
Lemma 2,
w(Q) = w((1 + x)k S) = w((1 + xk )S) = 2w(S),
w(Qi1 ) = w((1 + x)k R) = w((1 + xk )R) = 2w(R).
Furthermore, by the induction hypothesis, w(S) ≥ w(R), so w(Q) ≥
w(Qi1 ).
Score Situation This particular problem saw the following distribution of scores among
contestants: 12 contestants scored 7 points, 3 contestants scored 6 points, no contestant
scored 5 points, 1 contestant scored 4 points, 5 contestants scored 3 points, 8 contestants
scored 2 points, 27 contestants scored 1 point, and 153 contestants scored 0 point. The
average score for this problem is 0.785, indicating that it was extremely difficult.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Romania team scored 16 points (with a total team score of 201 points), the United
States team scored 34 points (with a total team score of 180 points), the Hungary team
scored 17 points (with a total team score of 168 points), the Bulgaria team scored 10 points
348 IMO Problems, Theorems, and Methods: Algebra

(with a total team score of 165 points), and the Vietnam team scored 2 points (with a total
team score of 144 points).
The gold medal cutoff for this IMO was set at 34 points (with 14 contestants earning gold
medals), the silver medal cutoff was 22 points (with 35 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 52 contestants earning bronze medals).
In this IMO, only two contestants achieved a perfect score of 42 points, namely Géza
Kós from Hungary and Daniel Tătaru from Romania.

Problem 5.7 (IMO 34-1, proposed by Ireland). Let f (x) = xn +

5xn−1 + 3, where n > 1 is an integer. Prove that f (x) cannot be expressed
as the product of two nonconstant polynomials with integer coefficients.

Proof 1. Suppose f (x) can be factored into the product of two integer
polynomials, i.e., f (x) = g(x)h(x), where

g(x) = xp + ap−1 xp−1 + · · · + a1 x + a0 ,

h(x) = xq + bq−1 xq−1 + · · · + b1 x + b0 ,

with ap = bq = 1, ai and bj are integers, and p and q are positive integers.

First, prove that both p and q are greater than or equal to 2. Otherwise,
assume p = 1. Then f (x) = (x + a0 )h(x), and from a0 b0 = 3, we have
a0 = ±1 or a0 = ±3, i.e., f (x) has roots ±1 or ±3. However, when x is
odd, f (x) ≡ 1(mod 2), so f (x) cannot be zero, implying p and q are both
greater than or equal to 2.
Let 2 ≤ p ≤ q ≤ n − 2, and from a0 b0 = 3, if a0 = ±3, since ap is not a
multiple of 3, then there must be a k such that 3|ai for 0 ≤ i ≤ k − 1 and
ak is not a multiple of 3.
Considering the coefficients of xk in the expansions of f (x) and g(x)h(x),
we have 0 = ak b0 + ak−1 b1 + · · · + a0 bk , i.e., ak b0 = −(ak−1 b1 + · · · + a0 bk ).
This implies 3|ak , a contradiction.
Similarly, if b0 = ±3, then the same conclusion can be drawn. Therefore,
f (x) cannot be expressed as the product of two nonconstant polynomials
with integer coefficients.

Proof 2. Similar to Proof 1, suppose f (x) can be factored into the product
of two polynomials with integer coefficients, both having leading coefficients
of 1, i.e., f (x) = g(x)h(x).
Since g(0)h(0) = f (0) = 3, either |g(0)| = 1 or |h(0)| = 1. With-
out loss of generality, assume |g(0)| = 1, and let g(x) have k roots
 Other Algebra Problems 349

a1 , a2 , . . . , ak (k < n). Then, by Vieta’s formulas,

|a1 a2 · · · ak | = 1. (1)
Since ai is a root of f (x), we see that ani + 5an−1
i + 3 = 0 for i ∈
{1, 2, . . . , k}. Therefore,
|(ai + 5)an−1
i | = 3. (2)
Thus,
k
 k
 k
 k

 
(ai + 5)an−1  = 3k = |ai + 5| · |ai |n−1 = |ai + 5|.
i
i=1 i=1 i=1 i=1

Assume without loss of generality that |ak +5| = min1≤i≤k |ai +5|. Then
|ak +5| ≤ 3. Also, since |−ak |+|ak +5| ≥ 5, we have |ak | ≥ 5−|ak +5| ≥ 2.
From (1), we know that k ≥ 2.
Assume without loss of generality that |a1 | = min1≤i≤k−1 |ai |. Then
1 1
|a1 |k−1 ≤ |a1 ||a2 | · · · |ak−1 | = ≤ .
|ak | 2
1 1 1
Hence, |a1 | ≤ √
k−1
2
, and |a1 + 5| ≤ 5 + √
k−1
2
< 6 and |a1 |n−1 ≤ n−1 <
2 k−1
1
2. Therefore, |a1 + 5| · |a1 |n−1 < 3, which contradicts (2).
In conclusion, f (x) cannot be factored into the product of two noncon-
stant polynomials with integer coefficients.
Note. A famous irreducibility criterion that requires no information on the
canonical decomposition of the coefficients of an integer polynomial is the
following result of Perron:
Let f (x) = xn + an−1 xn−1 + · · · + a1 x + a0 ∈ Z[x] with a0 = 0. If
|an−1 | > 1 + |an−2 | + · · · + |a1 | + |a0 |,
then f is irreducible in Z[x].
Furthermore, there are several similar problems:

• (Asian Pacific Mathematics Olympiad 2018, Problem 5). Find

all polynomials P (x) with integer coefficients such that for all real num-
bers s and t, if P (s) and P (t) are both integers, then P (st) is also an
integer.
• (Romania Team Selection Test 2001, Problem 2). (a) Let f, g :
Z → Z be injective maps. Show that the function h : Z → Z, defined by
h(x) = f (x)g(x) for all x ∈ Z cannot be surjective.
(b) Let f : Z → Z be a surjective map. Show that there exist surjective
functions g, h : Z → Z such that f (x) = g(x)h(x) for all x ∈ Z.
350 IMO Problems, Theorems, and Methods: Algebra

• (Estonia Team Selection Test 1999, Problem 2). A polynomial

an xn + an−1 xn−1 + · · · + a1 x + a0 is called alternating when n ≥ 1 and
the coefficients ai and ai−1 are nonzero real numbers with different signs
for i = 1, 2, . . . , n.
Let P (x) and Q(x) be arbitrary alternating polynomials. Prove that
the polynomial R(x) = P (x)Q(x) is alternating.
• (Japan Mathematical Olympiad 1999, Final Round, Prob-
lem 4). Prove that the polynomial
    
f (x) = x2 + 1 x2 + 22 · · · x2 + n2 + 1

cannot be expressed as a product of two polynomials with integer coeffi-

cients with degree greater than 1.
• (Korean Mathematical Olympiad 1995, 2nd Round, Prob-
lem 5). Let a, b be integers and p be a prime number such that:

(i) p is the greatest common divisor of a and b;

(ii) p2 divides a.

Prove that the polynomial xn+2 + axn+1 + bxn + a + b cannot be decom-

posed into the product of two polynomials with integer coefficients and
degree greater than 1.

Score Situation This particular problem saw the following distribution of scores among
contestants: 92 contestants scored 7 points, 13 contestants scored 6 points, 1 contestant
scored 5 points, 1 contestant scored 4 points, 1 contestant scored 3 points, 6 contestants
scored 2 points, 91 contestants scored 1 point, and 208 contestants scored 0 point. The
average score for this problem is 2,027, indicating that it had a certain level of difficulty.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 35 points (with a total team score of 215 points), the Germany team
scored 29 points (with a total team score of 189 points), the Bulgaria team scored 24 points
(with a total team score of 178 points), the Russia team scored 28 points (with a total
team score of 177 points), and the Chinese Taiwan team scored 42 points (with a total team
score of 162 points).
The gold medal cutoff for this IMO was set at 30 points (with 35 contestants earning gold
medals), the silver medal cutoff was 20 points (with 66 contestants earning silver medals),
and the bronze medal cutoff was 11 points (with 97 contestants earning bronze medals).
In this IMO, only two contestants achieved a perfect score of 42 points, namely Hong
Zhou from China and Hung-Wu Wu from Chinese Taiwan.
 Other Algebra Problems 351

Problem 5.8 (IMO 47-5, proposed by Romania). Let P (x) be a

polynomial of degree n > 1 with integer coefficients and let k be a positive
integer. Consider the polynomial Q(x) = P (P (· · · P (P (x)) · · · )), where P
occurs k times. Prove that there are at most n integers t such that Q(t) = t.

Proof. If every integer fixed point of Q is also a fixed point of P and the
polynomial P (x) − x has at most n roots, then the conclusion holds.
If not, then there exists an integer x0 such that Q(x0 ) = x0 but
P (x0 ) = x0 . Define xi+1 = P (xi ) for i = 0, 1, 2, . . .. Then xk = x0 and
xi+1 = xi for i = 1, 2, . . . , k. Clearly, (u − v)|(P (u) − P (v)) for distinct
integers u and v. Thus, for the following (non-zero) differences, the previ-
ous term divides the subsequent one:

x0 − x1 , x1 − x2 , . . . , xk−1 − xk , xk − xk+1 .

Since xk − xk+1 = x0 − x1 , the absolute values of the above differences

are equal.
Consider xm = min(x1 , x2 , . . . , xk ). Then xm−1 − xm = −(xm − xm+1 ),
so xm−1 = xm+1 = xm , implying that successive differences have opposite
signs, and x0 , x1 , . . . take two distinct values. Thus, x0 is a fixed point of
the polynomial P (P (x)).
(It is evident that the integer fixed points of P are fixed points of the
polynomial P (P (x)). Thus the integer fixed points of Q are all fixed points
of the polynomial P (P (x)).)
Assume a = x0 , and let b = P (a) = a. Then a = P (b). If P (P (x)) has
only two fixed points a and b, since n ≥ 2, then the conclusion holds.
If not, for any integer fixed point α of P (P (x)), let β = P (α).
Then α = P (β), where α and β can be the same, but distinct from a
and b. From the previous proof, we can know that, for the four pairs
(α, a), (β, b), (α, b), (β, a), the numbers α − a and β − b divide each other,
and α − b and β − a also divide each other, yielding

α − b = ±(β − a), α − a = ±(β − b).

If the plus sign is taken in both equations, then α − b = β − a and

α − a = β − b, implying a − b = b − a, which contradicts a = b. Thus, at
least one of the equations takes the minus sign, resulting in α + β = a + b,
i.e., P (α) + α − (a + b) = 0.
Let C represent a + b. We have shown that every fixed point of Q, not
equal to a or b, is a root of the polynomial F (x) = P (x) + x − C, and a
and b are also the roots of F (x).
352 IMO Problems, Theorems, and Methods: Algebra

Since the polynomial F (x) has the same degree as P (x), both being
polynomials of degree n, there are at most n distinct integer roots.

Note. There are also several similar problems:

• (All-Russian Mathematical Olympiad 2021, Grade 10, Prob-

lem 6). Given a real-coefficient polynomial P (x) of degree n > 1, it
is known that the equation P (P (P (x))) = P (x) has exactly n3 dis-
tinct real roots. Prove that these n3 roots can be divided into two
groups such that the arithmetic means of the roots in each group are
equal.
• (Japan Team Selection Test 2019, Problem 9). Let P (x) be a
rational coefficient polynomial and suppose P (P (x)) and P (P (P (x)))
are both integer coefficient polynomials. Prove that P (x) is an integer
coefficient polynomial.
• (All-Russian Mathematical Olympiad 2018, Grade 11, Prob-
lem 1). Given a polynomial P (x) such that P (P (x)) and P (P (P (x)))
are strictly monotone on the whole real axis, prove that P (x) is also
strictly monotone on the whole real axis.
• (All-Russian Mathematical Olympiad 2002, 4th Round, Grade
11, Problem 5). Let P (x) be a polynomial of an odd degree. Prove
that the equation P (P (x)) = 0 has at least as many different real roots
as the equation P (x) = 0 does.
• (Romania Team Selection Test 2000, Problem 11). Suppose P, Q
are monic complex polynomials such that P (P (x)) = Q(Q(x)). Prove
that P = Q.
• (William Lowell Putnam Mathematical Competition 2000, A6).
Let f (x) be a polynomial with integer coefficients. Define a sequence
a0 , a1 , . . . of integers such that a0 = 0 and an+1 = f (an ) for all n ≥ 0.
Prove that if there exists a positive integer m for which am = 0, then
either a1 = 0 or a2 = 0.
• (Turkey Mathematical Olympiad 2000, 2nd Round, Problem 6).
Find all continuous functions f : [0, 1] → [0, 1] for which there exists a
positive integer n such that f n (x) = x for x ∈ [0, 1], where f 0 (x) = x
and f k+1 (x) = f (f k (x)) for every positive integer k.
• (United States of America Mathematical Olympiad 1974, Prob-
lem 1). Let a, b, c denote three distinct integers and P denote a polyno-
mial with integer coefficients. Show that it is impossible that P (a) = b,
P (b) = c, and P (c) = a.
 Other Algebra Problems 353

Score Situation This particular problem saw the following distribution of scores among
contestants: 48 contestants scored 7 points, 2 contestants scored 6 points, 5 contestants
scored 5 points, 6 contestants scored 4 points, 25 contestants scored 3 points, 8 contestants
scored 2 points, 101 contestants scored 1 point, and 303 contestants scored 0 point. The
average score for this problem is 1.183, indicating that it was relatively challenging.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 38 points (with a total team score of 214 points), the Russia team
scored 28 points (with a total team score of 174 points), the South Korea team scored 29
points (with a total team score of 170 points), the Germany team scored 32 points (with a
total team score of 157 points), and the United States team scored 20 points (with a total
team score of 154 points).
The gold medal cutoff for this IMO was set at 28 points (with 42 contestants earning gold
medals), the silver medal cutoff was 19 points (with 89 contestants earning silver medals),
and the bronze medal cutoff was 15 points (with 122 contestants earning bronze medals).
In this IMO, only three contestants achieved a perfect score of 42 points, namely Zhiyu
Liu from China, Iurie Boreico from Moldova, and Alexander Magazinov from Russia.

Problem 5.9 (IMO 48-6, proposed by the Netherlands). Let n be

a positive integer. Consider
S = {(x, y, z) : x, y, z ∈ {0, 1, . . . , n}, x + y + z > 0}
as a set of (n + 1)3 − 1 points in a three-dimensional space. Determine the
smallest possible number of planes, the union of which contains S but does
not include (0, 0, 0).
Solution. It is easy to see that 3n planes satisfy the conditions. Take the
planes x = i, y = i, and z = i for i = 1, 2, . . . , n.
Suppose there are m planes ai x + bi y + ci z − di = 0 satisfying the
conditions, where 1 ≤ i ≤ m and di = 0. Then the polynomial of degree m
m

f (x, y, z) = (ai x + bi y + ci z − di )
i=1
is zero at every point in S but not at the origin, i.e., f (0, 0, 0) = 0. Define
the difference operator Δ ∈ {Δx , Δy , Δz } as follows:
Δx f (x, y, z) = f (x + 1, y, z) − f (x, y, z),
Δy f (x, y, z) = f (x, y + 1, z) − f (x, y, z),
Δz f (x, y, z) = f (x, y, z + 1) − f (x, y, z).
For every positive integer k, let Δk represent the mth order difference of
f (x, y, z).
354 IMO Problems, Theorems, and Methods: Algebra

If m < 3n, then Δnx Δny Δnz f (x, y, z) ≡ 0. Using the difference formula
for univariate polynomials, we have:
n
Δn P (x) = (−1)n−i Cin P (x + i),
i=0

so Δnx Δny Δnz f (x, y, z) = 

(i,j,k)∈(S (0,0,0)) (−1)
3n−i−j−k i j k
Cn Cn Cn f (x +
i, y + j, z + k) ≡ 0.
Setting x = y = z = 0, and knowing that f (i, j, k) = 0 for i, j, k ∈
{0, 1, . . . , n} with i + j + k > 0, we get

f (0, 0, 0) = (−1)i+j+k+1 Cin Cjn Ckn f (i, j, k) = 0,

(i,j,k)∈S

which contradicts f (0, 0, 0) = 0. Therefore, m ≥ 3n.

Note. For a non-zero polynomial F (x1 , x2 , . . . , xk ) with F (0, 0, . . . , 0) = 0,

if all points (x1 , x2 , . . . , xk ) satisfying x1 + x2 + · · · + xk > 0 are zeros of
F (x1 , x2 , . . . , xk ), where x1 , x2 , . . . , xk ∈ {0, 1, . . . , n}, then deg F ≥ kn.
Furthermore, this problem is related to Alon’s papers from 1993 and
1999. Theorem 4 in the 1993 paper “Covering the Cube by Affine Hyper-
planes” is:
Let F be an arbitrary field, let S1 , S2 , . . . , Sn be non-empty subsets of
F, |Si | = si , and let B be the set S1 × S2 × · · · × Sn . If m hyperplanes do
not cover B completely, then they miss at least
  n  
M s, si −m
i=1

points of B.
For a sequence of positive integers s = (s1 , s2 , . . . , sn ), let M (s, l) denote
the minimum of the product of n positive integers yi ≤ si the sum of which
n
is at least l. For l ≤ n, we have M = 1, and for l ≥ i=1 si we define
n
M = i=1 si .
Theorem 1.2 in the 1999 paper “Combinatorial Nullstellensatz” is:
Let F be an arbitrary field, and let f = f (x1 , x2 , . . . , xn ) be a polynomial
in F[x1 , x2 , . . . , xn ]. Suppose the degree deg(f ) of f is ni=1 ti , where each
ti is a non-negative integer, and suppose the coefficient of ni=1 xtii in f is
nonzero. If S1 , S2 , . . . , Sn are subsets of F with |Si | > ti , then there are
s1 ∈ S1 , s2 ∈ S2 , . . . , sn ∈ Sn such that f (s1 , s2 , . . . , sn ) = 0.
 Other Algebra Problems 355

There is a similar problem:

• (All-Russian Mathematical Olympiad 2010, Grade 11, Prob-

lem 4). Given an integer n ≥ 3, what is the smallest possible value of k
if the following statements are true?
For any n points Ai (xi , yi )(1 ≤ i ≤ n) on a plane, where no three points
are colinear, and for any n real numbers ci (1 ≤ i ≤ n), there exists a
polynomial P (x, y) of degree no higher than k such that
P (xi , yi ) = ci for i = 1, 2, . . . , n.

Note. A function in the following form is called a bivariate

polynomial:
P (x, y) = a0,0 + a1,0 x + a0,1 y + a2,0 x2 + a1,1 xy + a0,2 y 2 + · · ·
+ ak,0 xk + ak−1,1 xk−1 y + · · · + a0,k y k ,
where the degree of non-zero monomial ai,j xi y j is i + j, and the degree of
polynomial P (x, y) is the greatest degree of the degrees of its monomials.

Score Situation This particular problem saw the following distribution of scores among
contestants: 5 contestants scored 7 points, no contestant scored 6 points, no contestant
scored 5 points, no contestant scored 4 points, no contestant scored 3 points, 2 contestants
scored 2 points, 40 contestants scored 1 point, and 473 contestants scored 0 point. The
average score for this problem is 0.152, indicating that it was extremely difficult.
Among the top five teams in the team scores, the scores of this problem are as follows:
the Russia team scored 9 points (with a total team score of 184 points), the China team
scored 3 points (with a total team score of 181 points), the South Korea team scored 2
points (with a total team score of 168 points), the Vietnam team scored 0 point (with a total
team score of 168 points), and the United States team scored 4 points (with a total team
score of 155 points).
The gold medal cutoff for this IMO was set at 29 points (with 39 contestants earning gold
medals), the silver medal cutoff was 21 points (with 83 contestants earning silver medals),
and the bronze medal cutoff was 14 points (with 131 contestants earning bronze medals).
In this IMO, no contestant achieved a perfect score of 42 points.

Problem 5.10 (IMO 62-6, proposed by Austria). Let m ≥ 2 be an

integer, A be a finite set of (not necessarily positive) integers, and
B1 , B2 , B3 , . . . , Bm be subsets of A. Assume that for each k = 1, 2, . . . , m
the sum of the elements of Bk is mk . Prove that A contains at least m 2
elements.
356 IMO Problems, Theorems, and Methods: Algebra

m
Proof. Denote A = {a1 , . . . , an }. By contradiction, assume n = |A| < 2.
Note that the following mm sums are distinct:

f (c1 , . . . , cm ) := c1 m + c2 m2 + · · · + cm mm ,

where cj ∈ {0, 1, . . . , m − 1} for j = 1, 2, . . . , m. (Similar to the expansion

in base m.)
However, by replacing each mj (1 ≤ j ≤ m) with the sum of elements
in Bj , we can rewrite f (c1 , . . . , cm ) as the following form of the sum of
elements:

α1 a1 + α2 a2 · · · + αn an ,

where each αi ∈ {0, 1, . . . , m(m − 1)}. There are only

(m(m − 1) + 1)n < m2n < mm

such distinct expressions, which contradicts the earlier statement that all
f (c1 , . . . , cm ) are distinct. Therefore, n ≥ m
2.

Note. The problem statement also holds if A is a set of real numbers.

Score Situation This particular problem saw the following distribution of scores among
contestants: 37 contestants scored 7 points, no contestant scored 6 points, 2 contestants
scored 5 points, 1 contestant scored 4 points, 3 contestants scored 3 points, 2 contestants
scored 2 points, 12 contestants scored 1 point, and 562 contestants scored 0 point. The
average score for this problem is 0.481, indicating that it was extremely difficult.
Among the top five teams in the team scores, the scores of this problem are as follows:
the China team scored 42 points (with a total team score of 208 points), the Russia team
scored 28 points (with a total team score of 183 points), the South Korea team scored 36
points (with a total team score of 172 points), the United States team scored 29 points (with
a total team score of 165 points), and the Canada team scored 14 points (with a total team
score of 151 points).
The gold medal cutoff for this IMO was set at 24 points (with 52 contestants earning gold
medals), the silver medal cutoff was 19 points (with 103 contestants earning silver medals),
and the bronze medal cutoff was 12 points (with 148 contestants earning bronze medals).
In this IMO, only one contestant achieved a perfect score of 42 points, namely Yichuan
Wang from China.
 Other Algebra Problems 357

5.3 Summary
Trigonometry originated from the needs of navigation, astronomy, and sim-
ilar fields, and later evolved into an independent branch of mathematics.
Polynomial, on the other hand, is a classical subject in algebra. Therefore,
this chapter also introduces some knowledge of trigonometric identities and
polynomials.
In the first 64 IMOs, there were a total of 10 other algebra problems.
These problems can be broadly categorized into three types, as depicted in
Figure 5.2. The score details for these problems are presented in Table 5.2.
Due to the smaller number of participating teams and missing contestant
score information in early IMOs, there are several blanks in Table 5.2.

2.5
Proving Trigonometric Identities Finding Polynomials
Proving Properties
2

1.5

0.5

0
1–10 11–20 21–30 31–40 41–50 51–60 61–64

Figure 5.2 Numbers of Other Algebra Problems in the First 64 IMOs.

Problems 5.1–5.2 focus on “proving trigonometric identities;” among

these two problems, the one with the lowest average score is Problem 5.1
(IMO 5-5), proposed by the German Democratic Republic. Problems
5.3–5.4 deal with “finding polynomials;” among these two problems, the
one with the lowest average score is Problem 5.4 (IMO 45-2), proposed by
South Korea. Problems 5.5–5.10 are about “proving properties of polyno-
mials and sets;” among these six problems, the one with the lowest average
score is Problem 5.9 (IMO 48-6), proposed by the Netherlands.
These 10 problems were proposed by nine countries, and the Netherlands
contributed two problems.
 358
Table 5.2 Score Details of Other Algebra Problems in the First 64 IMOs

Problem 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8

Full points 6.000 5.000 8.000 7.000 8.000 7.000 7.000 7.000

IMO Problems, Theorems, and Methods: Algebra

Average score 2.375 4.704 2.797 2.761 3.100 0.785 2.027 1.183
Top five mean 4.425 6.000 5.750 2.633 5.267 4.900
6th–15th mean 2.329 5.350 2.416 1.150 4.636 2.682
16th–25th mean 3.667 0.167 2.463 2.296
Problem number in IMO 5-5 8-4 17-6 45-2 16-6 26-3 34-1 47-5
Proposing The German Yugoslavia The United South Sweden The Ireland Romania
country Democratic Republic Kingdom Korea Netherlands

Problem 5.9 5.10

Full points 7.000 7.000

Average score 0.152 0.481
Top five mean 0.600 4.967
6th–15th mean 0.467 1.317
16th–25th mean 0.367 0.600
Problem number in IMO 48-6 62-6
Proposing The Austria
country Netherlands

Note. Top five mean = Total score of the top five teams ÷ Total number of contestants from the top five teams,
6th–15th mean = Total score of the 6th–15th teams ÷ Total number of contestants from the 6th–15th teams,
16th–25th mean = Total score of the 16th–25th teams ÷ Total number of contestants from the 16th–25th teams.
 Other Algebra Problems 359

From Table 5.2, it can be observed that in the first 64 IMOs, there
were three other algebra problems with an average score of 0–1 point; one
problem with an average score of 1–2 points; four problems with an average
score of 2–3 points; one problem with an average score of 3–4 points; one
problem with an average score above 4 points. Overall, the other algebra
problems were relatively difficult.
In the 24th–64th IMOs, there were a total of six other algebra problems.
Among these, three had an average score of 0–1 point; one had an average
score of 1–2 points; two had an average score of 2–3 points; no problem
had an average score of 3–4 points; no problem had an average score above
4 points. Further analysis of the problem numbers of these six other algebra
problems, as shown in Table 5.3, reveals that these problems frequently
appeared as the 3rd/6th problem. The majority of these problems, totaling
five, were of the type proving properties of polynomials and sets.

Table 5.3 Numbers of Other Algebra Problems in the 24th–64th IMOs

Other Algebra Problem Number Number of Problems

Problems 1, 4 2, 5 3, 6 in the First 64 IMOs

Proving trigonometric 0 0 0 2
identities
Finding polynomials 0 1 0 2
Proving properties 1 1 3 6
Total 1 2 3 10

From Table 5.2, it can be observed that in other algebra problems,

the average score of the top five teams is generally 2.5 points higher than
the average score of the problem. Meanwhile, the average scores of the
6th–15th teams and the 16th–25th teams are both within 1 point of the
average score of the problem.
This page intentionally left blank
 Appendix A

IMO General Information

Number Gold Silver Bronze

of Partici- Number  
Cutoffs/Numbers
pating of Contes-
of Medalists
Session Year Host Teams tants

IMO 1 1959 Romania 7 52 37/3 36/3 33/5

IMO 2 1960 Romania 5 39 40/4 37/4 33/4
IMO 3 1961 Hungary 6 48 37/3 34/4 30/4
IMO 4 1962 Czechoslovakia 7 56 41/4 34/12 29/15
IMO 5 1963 Poland 8 64 35/7 28/11 21/17
IMO 6 1964 The Union of 9 72 38/7 31/9 27/19
Soviet
Socialist
Republics
IMO 7 1965 The German 10 80 38/8 30/12 20/17
Demo-
cratic
Republic
IMO 8 1966 Bulgaria 9 72 39/13 34/15 31/11
IMO 9 1967 Yugoslavia 13 99 38/11 30/14 22/26
IMO 10 1968 The Union of 12 96 39/22 33/22 26/20
Soviet
Socialist
Republics
IMO 11 1969 Romania 14 112 40/3 30/20 24/21
IMO 12 1970 Hungary 14 112 37/7 30/11 19/40

(Continued)

361
362 IMO Problems, Theorems, and Methods: Algebra

(Continued)

Number Gold Silver Bronze

of Partici- Number  
Cutoffs/Numbers
pating of Contes-
of Medalists
Session Year Host Teams tants

IMO 13 1971 Czechoslovakia 15 115 35/7 23/12 11/29

IMO 14 1972 Poland 14 107 40/8 30/16 19/30
IMO 15 1973 The Union of 16 125 35/5 27/15 17/48
Soviet
Socialist
Republics
IMO 16 1974 The German 18 140 38/10 30/24 23/37
Demo-
cratic
Republic
IMO 17 1975 Bulgaria 17 135 38/8 32/25 23/36
IMO 18 1976 Austria 18 139 34/9 23/28 15/45
IMO 19 1977 Yugoslavia 21 155 34/13 24/29 17/35
IMO 20 1978 Romania 17 132 35/5 27/20 22/38
IMO 21 1979 The United 23 166 37/8 29/32 20/42
Kingdom
IMO 22 1981 The United 27 185 41/36 34/37 26/30
States of
America
IMO 23 1982 Hungary 30 119 37/10 30/20 21/31
IMO 24 1983 France 32 186 38/9 26/27 15/57
IMO 25 1984 Czechoslovakia 34 192 40/14 26/35 17/49
IMO 26 1985 Finland 38 209 34/14 22/35 15/52
IMO 27 1986 Poland 37 210 34/18 26/41 17/48
IMO 28 1987 Cuba 42 237 42/22 32/42 18/56
IMO 29 1988 Australia 49 268 32/17 23/48 14/65
IMO 30 1989 Germany 50 291 38/20 30/55 18/72
IMO 31 1990 The People’s 54 308 34/23 23/56 16/76
Republic of
China
IMO 32 1991 Sweden 56 318 39/20 31/51 19/84
IMO 33 1992 The Russian 56 322 32/26 24/55 14/74
Federation
 Appendix A: IMO General Information 363

(Continued)

Number Gold Silver Bronze

of Partici- Number  
Cutoffs/Numbers
pating of Contes-
of Medalists
Session Year Host Teams tants

IMO 34 1993 Turkey 73 413 30/35 20/66 11/97

IMO 35 1994 Chinese 69 385 40/30 30/64 19/98
Hong Kong
IMO 36 1995 Canada 73 412 37/30 29/71 19/100
IMO 37 1996 India 75 424 28/35 20/66 12/99
IMO 38 1997 Argentina 82 460 35/39 25/70 15/122
IMO 39 1998 Chinese 76 419 31/37 24/66 14/102
Taiwan
IMO 40 1999 Romania 81 450 28/38 19/70 12/118
IMO 41 2000 Republic of 82 461 30/39 21/71 11/119
Korea
IMO 42 2001 The United 83 473 30/39 20/81 11/122
States of
America
IMO 43 2002 The United 84 479 29/39 23/73 14/120
Kingdom
IMO 44 2003 Japan 82 457 29/37 19/69 13/104
IMO 45 2004 Greece 85 486 32/45 24/78 16/120
IMO 46 2005 Mexico 91 513 35/42 23/79 12/128
IMO 47 2006 Slovenia 90 498 28/42 19/89 15/122
IMO 48 2007 Vietnam 93 520 29/39 21/83 14/131
IMO 49 2008 Spain 97 535 31/47 22/100 15/120
IMO 50 2009 Germany 104 565 32/49 24/98 14/135
IMO 51 2010 Kazakhstan 95 522 27/47 21/103 15/115
IMO 52 2011 The 101 563 28/54 22/90 16/137
Netherlands
IMO 53 2012 Argentina 100 547 28/51 21/88 14/137
IMO 54 2013 Colombia 97 527 31/45 24/92 15/141
IMO 55 2014 South Africa 101 560 29/49 22/113 16/133
IMO 56 2015 Thailand 104 577 26/39 19/100 14/143
IMO 57 2016 Chinese 109 602 29/44 22/101 16/135
Hong Kong

(Continued)
364 IMO Problems, Theorems, and Methods: Algebra

(Continued)

Number Gold Silver Bronze

of Partici- Number  
Cutoffs/Numbers
pating of Contes-
of Medalists
Session Year Host Teams tants

IMO 58 2017 Brazil 111 615 25/48 19/90 16/153

IMO 59 2018 Romania 107 594 31/48 25/98 16/143
IMO 60 2019 The United 112 621 31/52 24/94 17/156
Kingdom
IMO 61 2020 The Russian 105 616 31/49 24/112 15/155
Federation
IMO 62 2021 The Russian 107 619 24/52 19/103 12/148
Federation
IMO 63 2022 Norway 104 589 34/44 29/101 23/140
IMO 64 2023 Japan 112 618 32/54 25/90 18/170
 Appendix B

IMO Algebra Problem Index

Problem Problem
Number in Proposing Number in Page
the IMO Country Category the Book Number

IMO 1-2 Romania Finding Solutions, Problem 1.1 31

Equations
IMO 1-3 Hungary Investigating Problem 1.12 53
Conditions,
Equations
IMO 2-2 Hungary Solving Problem 4.1 237
Inequalities,
Inequalities
IMO 3-1 Hungary Investigating Problem 1.13 54
Conditions,
Equations
IMO 3-2 Poland Proving Problem 4.6 244
Inequalities,
Inequalities
IMO 3-3 Bulgaria Finding Solutions, Problem 1.4 38
Equations
IMO 4-2 Hungary Solving Problem 4.2 238
Inequalities,
Inequalities
IMO 4-4 Romania Finding Solutions, Problem 1.2 32
Equations
IMO 5-1 Czechoslovakia Finding Solutions, Problem 1.5 39
Equations
IMO 5-4 The Soviet Finding Solutions, Problem 1.6 40
Union Equations
(Continued)

365
366 IMO Problems, Theorems, and Methods: Algebra

(Continued)

Problem Problem
Number in Proposing Number in Page
the IMO Country Category the Book Number

IMO 5-5 The German Proving Problem 5.1 336

Democratic Trigonometric
Republic Identities,
Others
IMO 6-2 Hungary Proving Problem 4.7 248
Inequalities,
Inequalities
IMO 7-1 Yugoslavia Solving Problem 4.3 239
Inequalities,
Inequalities
IMO 7-2 Poland Proving Problem 1.8 43
Relationships,
Equations
IMO 7-4 The Soviet Finding Solutions, Problem 1.3 33
Union Equations
IMO 8-4 Yugoslavia Proving Problem 5.2 339
Trigonometric
Identities,
Others
IMO 8-5 Czechoslovakia Finding Solutions, Problem 1.7 42
Equations
IMO 9-5 The Soviet Determining Problem 3.1 171
Union Values,
Sequences
IMO 10-3 Bulgaria Proving Problem 1.9 45
Relationships,
Equations
IMO 10-5 The German Proving Properties, Problem 2.1 92
Democratic Functions
Republic
IMO 11-2 Hungary Proving Properties, Problem 2.2 94
Functions
IMO 11-6 The Soviet Proving Problem 4.8 250
Union Inequalities,
Inequalities
IMO 12-3 Sweden Existence Problem 3.5 180
Problems,
Sequences
IMO 13-1 Hungary Proving Problem 4.9 253
Inequalities,
Inequalities
IMO 14-4 The Solving Problem 4.4 240
Netherlands Inequalities,
Inequalities
(Continued)
 Appendix B: IMO Algebra Problem Index 367

(Continued)

Problem Problem
Number in Proposing Number in Page
the IMO Country Category the Book Number

IMO 14-5 Bulgaria Proving Properties, Problem 2.3 96

Functions
IMO 15-3 Sweden Investigating Problem 1.14 57
Conditions,
Equations
IMO 15-5 Poland Proving Properties, Problem 2.4 98
Functions
IMO 15-6 Sweden Existence Problem 3.6 182
Problems,
Sequences
IMO 16-5 The Determining Value Problem 4.31 304
Netherlands Ranges,
Inequalities
IMO 16-6 Sweden Proving Properties, Problem 5.5 344
Others
IMO 17-1 Czechoslovakia Proving Problem 4.10 256
Inequalities,
Inequalities
IMO 17-6 The United Finding Problem 5.3 341
Kingdom Polynomials,
Others
IMO 18-2 Finland Proving Problem 1.10 48
Relationships,
Equations
IMO 18-5 The Proving Problem 1.11 51
Netherlands Relationships,
Equations
IMO 18-6 The United Proving Problem 3.10 194
Kingdom Quantitative
Relationships,
Sequences
IMO 19-4 The United Proving Problem 4.11 257
Kingdom Inequalities,
Inequalities
IMO 19-6 Bulgaria Proving Properties, Problem 2.5 100
Functions
IMO 20-3 The United Determining Problem 2.10 111
Kingdom Values,
Functions
IMO 20-5 France Proving Problem 4.12 259
Inequalities,
Inequalities
(Continued)
368 IMO Problems, Theorems, and Methods: Algebra

(Continued)

Problem Problem
Number in Proposing Number in Page
the IMO Country Category the Book Number

IMO 21-5 Israel Investigating Problem 1.15 59

Conditions,
Equations
IMO 22-6 Finland Determining Problem 2.11 114
Values,
Functions
IMO 23-1 The United Determining Problem 2.12 116
Kingdom Values,
Functions
IMO 23-3 The Soviet Existence Problem 3.7 184
Union Problems,
Sequences
IMO 24-1 The United Deriving Problem 2.14 120
Kingdom Expressions,
Functions
IMO 24-6 The United Proving Problem 4.13 261
States Inequalities,
Inequalities
IMO 25-1 Germany Proving Problem 4.14 262
Inequalities,
Inequalities
IMO 26-3 The Proving Properties, Problem 5.6 346
Netherlands Others
IMO 26-6 Sweden Proving Problem 3.11 196
Quantitative
Relationships,
Sequences
IMO 27-5 The United Deriving Problem 2.15 122
Kingdom Expressions,
Functions
IMO 28-3 Germany Proving Problem 4.15 265
Inequalities,
Inequalities
IMO 28-4 Vietnam Proving Properties, Problem 2.6 101
Functions
IMO 29-3 The United Determining Problem 2.13 118
Kingdom Values,
Functions
IMO 29-4 Ireland Solving Problem 4.5 242
Inequalities,
Inequalities
(Continued)
 Appendix B: IMO Algebra Problem Index 369

(Continued)

Problem Problem
Number in Proposing Number in Page
the IMO Country Category the Book Number

IMO 31-4 Turkey Deriving Problem 2.16 123

Expressions,
Functions
IMO 32-6 The Existence Problem 3.8 187
Netherlands Problems,
Sequences
IMO 33-2 India Deriving Problem 2.17 125
Expressions,
Functions
IMO 34-1 Ireland Proving Properties, Problem 5.7 348
Others
IMO 34-5 Germany Proving Properties, Problem 2.7 104
Functions
IMO 35-1 France Proving Problem 4.16 267
Inequalities,
Inequalities
IMO 35-5 The United Deriving Problem 2.18 128
Kingdom Expressions,
Functions
IMO 36-2 Russia Proving Problem 4.17 268
Inequalities,
Inequalities
IMO 36-4 Poland Determining Problem 3.2 173
Values,
Sequences
IMO 37-6 France Proving Problem 3.12 199
Quantitative
Relationships,
Sequences
IMO 38-3 Russia Proving Problem 4.18 270
Inequalities,
Inequalities
IMO 40-2 Poland Determining Value Problem 4.32 307
Ranges,
Inequalities
IMO 40-6 Japan Deriving Problem 2.19 129
Expressions,
Functions
IMO 41-2 The United Proving Problem 4.19 272
States Inequalities,
Inequalities
(Continued)
370 IMO Problems, Theorems, and Methods: Algebra

(Continued)

Problem Problem
Number in Proposing Number in Page
the IMO Country Category the Book Number

IMO 42-2 South Korea Proving Problem 4.20 274

Inequalities,
Inequalities
IMO 43-5 India Deriving Problem 2.20 130
Expressions,
Functions
IMO 44-5 Ireland Proving Problem 4.21 279
Inequalities,
Inequalities
IMO 45-2 South Korea Finding Problem 5.4 343
Polynomials,
Others
IMO 45-4 South Korea Proving Problem 4.22 281
Inequalities,
Inequalities
IMO 46-3 South Korea Proving Problem 4.23 285
Inequalities,
Inequalities
IMO 47-3 Ireland Determining Value Problem 4.33 309
Ranges,
Inequalities
IMO 47-5 Romania Proving Properties, Problem 5.8 351
Others
IMO 48-1 New Zealand Proving Problem 4.24 288
Inequalities,
Inequalities
IMO 48-6 The Proving Properties, Problem 5.9 353
Netherlands Others
IMO 49-2 Austria Proving Problem 4.25 290
Inequalities,
Inequalities
IMO 49-4 South Korea Deriving Problem 2.21 133
Expressions,
Functions
IMO 50-3 The United Proving Problem 3.13 200
States Quantitative
Relationships,
Sequences
IMO 50-5 France Deriving Problem 2.22 134
Expressions,
Functions
(Continued)
 Appendix B: IMO Algebra Problem Index 371

(Continued)

Problem Problem
Number in Proposing Number in Page
the IMO Country Category the Book Number

IMO 51-1 France Deriving Problem 2.23 137

Expressions,
Functions
IMO 51-6 Iran Proving Problem 3.14 202
Quantitative
Relationships,
Sequences
IMO 52-3 Belarus Proving Properties, Problem 2.8 107
Functions
IMO 53-2 Australia Proving Problem 4.26 294
Inequalities,
Inequalities
IMO 53-4 South Africa Deriving Problem 2.24 138
Expressions,
Functions
IMO 54-5 Bulgaria Proving Properties, Problem 2.9 109
Functions
IMO 55-1 Austria Proving Problem 4.27 296
Inequalities,
Inequalities
IMO 56-5 Albania Deriving Problem 2.25 140
Expressions,
Functions
IMO 57-5 Russia Investigating Problem 1.16 61
Conditions,
Equations
IMO 58-1 South Africa Determining Problem 3.3 174
Values,
Sequences
IMO 58-2 Albania Deriving Problem 2.26 142
Expressions,
Functions
IMO 59-2 Slovakia Determining Problem 3.4 177
Values,
Sequences
IMO 60-1 South Africa Deriving Problem 2.27 146
Expressions,
Functions
IMO 61-2 Belgium Proving Problem 4.28 298
Inequalities,
Inequalities
(Continued)
372 IMO Problems, Theorems, and Methods: Algebra

(Continued)

Problem Problem
Number in Proposing Number in Page
the IMO Country Category the Book Number

IMO 62-2 Canada Proving Problem 4.29 300

Inequalities,
Inequalities
IMO 62-6 Austria Proving Properties, Problem 5.10 355
Others
IMO 63-2 The Deriving Problem 2.28 147
Netherlands Expressions,
Functions
IMO 64-3 Malaysia Existence Problem 3.9 191
Problems,
Sequences
IMO 64-4 The Proving Problem 4.30 303
Netherlands Inequalities,
Inequalities