Vol 31 No 2 2018

M AT H E M AT I C S
C O MP E T I T I O NS

journal of the

WORLD FEDERATION OF NATIONAL

MATHEMATICS COMPETITIONS
Vol 31 No 2 2018

M AT H E M AT I C S
C O MP E T I T I O NS

journal of the

WORLD FEDERATION OF NATIONAL

MATHEMATICS COMPETITIONS
MATHEMATICS COMPETITIONS
J ournal of the W orld F ederation of N ational M athematics C ompetitions

(ISSN 1031 – 7503)

Published biannually by

Australian Maths Trust

170 Haydon Drive
Bruce ACT 2617
Australia

Articles (in English) are welcome.

Please send articles to:

Professor Alexander Soifer

University of Colorado
1420 Austin Bluffs Parkway
Colorado Springs, CO 80918
USA
[email protected]

©2019 Australian Mathematics Trust, AMTT Limited ACN 083 950 341
MATHEMATICS COMPETITIONS
Vol 31 No 2 2018

CONTENTSPAGE

WFNMC Committee  1

From the President 4

From the Editor 6

A History of the Federation through My Eyes 8

Alexander Soifer

The Cuban Mathematics Olympiad: a fragmentary journey 25

Frank Gamboa de la Paz & Jorge Macrhena Menéndez

Problems: The Treasure of Mathematics Competitions 43

Kiril Bankov

The Julia Robinson Mathematics Festival

A Complement and Alternative to Competitions 54
Mark Saul

An Exploratory Problem 75
Ryan Fang, Caleb Ji & Evan Liang

International Mathematics Tournament of Towns

Selected Problems from the Fall 2017 78
Andy Liu
 Mathematics Competitions Vol 31 No 2 2018

World Federation of National Mathematics

Competitions

Executive

President: Professor Kiril Bankov

Sofia University St. Kliment Ohridski
Sofia
BULGARIA

Senior Vice President: Dr. Robert Geretschläger

BRG Kepler
Keplerstrasse 1
8020 Graz
AUSTRIA

Vice Presidents: Sergey Dorichenko

School 179
Moscow
RUSSIA

Krzysztof Ciesielski
Jagiellonian University
Mathematics Institute
Krakow
POLAND

Publications Officer: Professor Alexander Soifer

University of Colorado
College of Visual Arts and Sciences
P.O. Box 7150 Colorado Springs
CO 80933-7150
USA

1
 Mathematics Competitions Vol 31 No 2 2018

Secretary: David Crawford

Leicester
UK
Professor Alexander Soifer
Immediate
University of Colorado
Past President &
College of Visual Arts and Sciences
Chairman,
P.O. Box 7150 Colorado Springs
Awards Committee:
CO 80933-7150
USA
Treasurer: Emeritus Professor Peter Taylor
PO Box 6165
O’Connor ACT 2601
AUSTRALIA

Regional Representatives
Africa: Liam Baker
University of Stellenboch
SOUTH AFRICA

Asia: M. Suhaimi Ramli

Kuala Lumpur
MALAYSIA

Europe: Professor Francisco Bellot-Rosado

Royal Spanish Mathematical Society
Dos De Mayo 16-8#DCHA
E-47004 Valladolid
SPAIN
Dr. Jaroslav Švrček
Dept. of Algebra and Geometry
Palacký University, Olomouc
CZECH REPUBLIC

2
 Mathematics Competitions Vol 31 No 2 2018

North America: vacant

Oceania: Peter Taylor

Canberra
AUSTRALIA
South America: vacant

The aims of the Federation are:

1. to promote excellence in, and research associated with,
mathematics education through the use of school math-
ematics competitions;
2. to promote meetings and conferences where persons inter-
ested in mathematics contests can exchange and develop
ideas for use in their countries;
3. to provide opportunities for the exchanging of information
for mathematics education through published material, no-
tably through the Journal of the Federation;
4. to recognize through the WFNMC Awards system persons
who have made notable contributions to mathematics edu-
cation through mathematical challenge around the world;
5. to organize assistance provided by countries with devel-
oped systems for competitions for countries attempting to
develop competitions;
6. to promote mathematics and to encourage young mathe-
maticians.

3
 Mathematics Competitions Vol 31 No 2 2018

From the President

Dear Federalists,
It has been a high honor and responsibility to be your elected President
for six years, 2012–2018, from ICME in South Korea to our Austrian
Congress. We have had wonderful WFNMC mini-conferences in 2012
and 2016, our great Congresses in 2014 and 2018, organized by Maria
Falk de Losada and Ricardo Losada and by Robert Geretschläger respec-
tively. We produced an impressive book 400-page compendium Compe-
titions for Young Mathematicians: A Perspective from Five Continents,
Springer, Germany, July 2017. A new book of 2018—Graz proceedings
is being compiled by Robert, and will be published in 2019. Finally, we
were joined by new fine colleagues from around the world, and returned
mathematicians from the United Kingdom and Poland to an active par-
ticipation in the Federation.

Our gratitude goes to the outgoing long-term Editor of Mathematics

Competitions (MC) Jaroslav Švrček. Starting with the next issue, I will
serve as the MC Editor. My Assistant Editor will be Sergey Dorichenko
of Russia. So, starting now, please send your articles for Mathematics
Competitions to me at [email protected]. You can find the formatting
instructions on the web pages of the Federation: http://www.wfnmc.
org/journalinvandformat.html.

The future of this journal, founded by Peter O’Halloran in the 1980s, is

in your hands—and minds. Make it a success by writing for MC well
and often!

Beyond scholarly exchanges, our Congresses and Competitions Topic

Groups and mini-conferences at ICME’s have become reunions of old
friends, and opportunities to make new ones. I hope to see you all during

4
 Mathematics Competitions Vol 31 No 2 2018

ICME in Shanghai-2020 and at WFNMC Congress in Sofia, Bulgaria, in

2022.

I am passing the presidential baton—or is it an Olympic flaming torch?—

to the new President Kiril Bankov. I have no doubt Kiril will continue
a fine tradition of leadership and transparency established by the Fed-
eration’s past presidents.

My best wishes to you and yours on a Healthy and Prosperous New Year!

Alexander Soifer
President of WFNMC
Editor of Mathematics Competitions
November 2018

5
 Mathematics Competitions Vol 31 No 2 2018

From the Editor

Welcome to Mathematics Competitions Vol 31, No 2.

First of all I would like to thank again the Australian Mathematics Trust
for continued support, without which each issue of the journal could not
be published, and in particular Bernadette Webster, Alexandra Carvajal
and Pavel Calábek for their assistance in the preparation of this issue.

Submission of articles:

The journal Mathematics Competitions is interested in receiving articles

dealing with mathematics competitions, not only at national and inter-
national level, but also at regional and primary school level. There are
many readers in different countries interested in these different levels of
competitions.

• The journal traditionally contains many different kinds of arti-

cles, including reports, analyses of competition problems and the
presentation of interesting mathematics arising from competition
problems. Potential authors are encouraged to submit articles of
all kinds.

• To maintain and improve the quality of the journal and its use-
fulness to those involved in mathematics competitions, all articles
are subject to review and comment by one or more competent ref-
erees. The precise criteria used will depend on the type of article,
but can be summarised by saying that an article accepted must
be correct and appropriate, the content accurate and interesting,
and, where the focus is mathematical, the mathematics fresh and
well presented. This editorial and refereeing process is designed to
help improve those articles which deserve to be published.

At the outset, the most important thing is that if you have anything
to contribute on any aspect of mathematics competitions at any level,
local, regional or national, we would welcome your contribution.

6
 Mathematics Competitions Vol 31 No 2 2018

Articles should be submitted in English, with a black and white photo-

graph and a short profile of the author. Alternatively, the article can
be submitted on an IBM PC compatible disk or a Macintosh disk. We
prefer LATEX or TEX format of contributions, but any text file will be
helpful.

Articles, and correspondence, can also be forwarded to the editor by mail

to

Professor Alexander Soifer

University of Colorado
College of Visual Arts and Sciences
P.O. Box 7150 Colorado Springs
CO 80933-7150
USA
[email protected]

Jaroslav Švrček
February 2019

7
 Mathematics Competitions Vol 31 No 2 2018

A History of the Federation

through My Eyes

Alexander Soifer

Born and educated in Moscow, Alexan-

der Soifer has for over 38 years been
a Professor at the University of Col-
orado, teaching math, and art and film
history. He has published over 300 ar-
ticles, and a good number of books.
In the past several years, 7 of his
books have appeared in Springer: The
Scholar and the State: In the search of
Van der Waerden; The Mathematical
Coloring Book: Mathematics of Color-
ing and the Colorful Life of Its Creators; Mathematics as Problem Solving;
How Does One Cut a Triangle?; Geometric Etudes in Combinatorial Math-
ematics; Ramsey Theory Yesterday, Today, and Tomorrow; and Colorado
Mathematical Olympiad and Further Explorations: From the Mountains of
Colorado to the Peaks of Mathematics. He has founded and for 32 years ran
the Colorado Mathematical Olympiad. Soifer has also served on the Soviet
Union Math Olympiad (1970–1973) and USA Math Olympiad (1996–2005).
He has been Secretary of the World Federation of National Mathematics Com-
petitions (WFNMC) (1996–2008), and Senior Vice President of the WFNMC
(2008–2012); from 2012 he has been the president of the WFNMC. He is a
recipient of the Federation’s Paul Erdős Award (2006). Soifer’s Erdős number
is 1.

Professor Peter O’Halloran of Australia, Executive Director of the Aus-

tralian Mathematics Trust, envisioned the World Federation of National
Mathematics Competitions.

O’Halloran’s singular vision, energy and will made the Federation pos-
sible. It was born in 1984 in Adelaide, Australia, during the 5th Inter-
national Congress on Mathematical Education (ICME-5).

O’Halloran led the Federation for 10 years, almost until his passing in
1994. Academician Blagovest H. Sendov of Bulgaria was handpicked

8
 Mathematics Competitions Vol 31 No 2 2018

Figure 1: Peter O’Halloran receiving his David Hilbert Award from the
Chairman of the AMT Board Professor Don Aitkin, at his home on 31
August 1994. Professor Bernard H. Neumann looks on.

by O’Halloran as his successor, but the successor was not into games
of mathematical kind. He aspired to the Presidency of Bulgaria, and
in 1995 became the second highest politician of his country, Chairman
of Bulgaria’s National Assembly. The steering wheel of the Federation
stood unattended.

Then there came Ronald G. Dunkley of Canada. In 1996 he was asked

to become the President of the Federation. A minute before the General
Membership Meeting of the Federation during ICME-8 in Seville, Spain,
Ron asked me whether I was willing to help. “Sure,” I replied, not
knowing what Ron had in mind. At the meeting, Ron commenced
a democratisation of the Federation. He nominated ‘critical’ thinkers
to the Executive Committee: Tony Gardiner of Great Britain to Vice-
President, and me to Secretary and member of the Program Committee.
Ron made the first Constitution of the Federation possible. Six years
earlier, in 1990, Ron Dunkley and his Canadian colleagues Ron Scoins,

9
 Mathematics Competitions Vol 31 No 2 2018

Dr. Oakland, and others, invented and hosted the unforgettable First
Congress of the Federation in their native Waterloo, Canada.

Figure 2: From left: Dimiter Dimitrov, Kiril Bankov, Alexander Soifer,

Ronald Dunkley, Peter Taylor & Konstantin Tsiskartidze, 3rd Congress
WFNMC, Zhongshan, China, 1998.

In 2000 in Japan, the Federation elected the new President, Peter J. Tay-
lor of Australia. Peter continued the democratisation of the Federation
by proposing a Constitutional Amendment limiting President’s service
to one term, which first applied to Peter himself. Peter also organized
the lovely 4th Congress in 2002 in Melbourne, Australia.

Figure 3: Three Presidents: from left, Peter Taylor, Alexander Soifer &
Petar Kenderov, 4th Congress of WFNMC, Melbourne, 2002.

10
 Mathematics Competitions Vol 31 No 2 2018

In 2004, academician Petar S. Kenderov of Bulgaria was elected Presi-

dent. For the first time the Federation earned its history, written by Pres-
ident Kenderov. Earlier, in 1994, Petar spearheaded the 2nd Congress of
the Federation in Pravetz, Bulgaria. At about the same time, Bulgarian
colleagues created the logo for the Federation.

The year 2008 saw the birth of the first Madam President, Maria Falk de
Losada of Colombia. A few days earlier, Maria organized, funded, and
ran the first Mini-Conference of the Federation a day before ICME-11 in
Monterrey, Mexico. The dinner in the Ethnographic Museum, among the
masterpieces of Pre-Columbian Art, was unforgettable. Maria continued
this most important undertaking by organizing and running the Mini-
Conference in 2012, before ICME-12 in Seoul, South Korea. In 2016, she
kindly took upon herself the logistics and funding of the Mini-Conference
in Hamburg, Germany, right before ICME-13. Maria also organized the
Federation’s 7th Congress in 2014 in Barranquilla, Colombia.

In 2012, at the General Meeting of the Federation in Seoul, during ICME-

12, the membership elected me to the Presidency. Two years earlier,
in 2010, I created the International Program of the Federation’s 6th
Congress in Riga, Latvia, on request of the person, who envisioned that
Congress but fell ill, Professor Agnis Andžāns. Ever since joined the
Federation 32 years ago, my main goal has been to bring Mathematical
Olympiads closer to research mathematics, to build two-way bridges
between the two. After all, according to Professor Boris Delone of Russia,
the only difference between an Olympiad problem and a research problem
is that the former takes 5 hours while the latter 5,000 hours to solve.

In 2018, during the 8th Congress of the Federation in Semriach, Austria,

the membership elected Kiril Bankov of Bulgaria to serve as the next
President. Robert Geretschläger became Senior Vice-President. Sergey
Dorichenko of Russia and Krzysztof Ciesielski of Poland became Vice
Presidents. David Crawford of the United Kingdom was elected Secre-
tary. Peter Taylor will continue as Treasurer (and our web master). I
will occupy two seats on the Executive, Immediate Past President and
Publications Officer, but of course will have just one vote :). We elected
fine members for our Standing Committees: Awards, Program, and Re-
gional Representatives. You can find the Committees membership on
our website: http://www.wfnmc.org/about.html.

11
 Mathematics Competitions Vol 31 No 2 2018

Figure 4: Madam President Maria Falk de Losada and Senator Ricardo

Losada, Canberra, 2000. Photo by A. Soifer

1 Publications of the Federation

The founding visionary Peter O’Halloran realized the importance of pub-
lications, and already in 1985 commenced the publication of the newslet-
ter of WFNMC. In 1988 the newsletter was upgraded to a biannual
journal Mathematics Competitions. For many years, until his retirement,
Warren Atkins of Australia was the editor of Mathematics Competitions.
Since 2004, the editor has been Jaroslav Švrček of Czech Republic. As
an editor and publisher of a journal myself, I can appreciate the great
job Warren and Jaroslav have done.

As the next editor of Mathematics Competitions, with Sergey Dorichenko

serving as the Assistant Editor, responsible for having accepted by me
articles typeset by the Moscow Center for Continuing Mathematics Ed-
ucation (MCCME).

12
 Mathematics Competitions Vol 31 No 2 2018

2 Awards of the Federation

In 1991, Peter O’Halloran introduced two international awards of the
Federation: the David Hilbert Award and the Paul Erdős Award, for
mathematicians prominent on an international or national scale in math-
ematical enrichment activities. Both Hilbert and Erdős were mathemati-
cians of genius, but I feared that having two awards would inspire an
unintended interpretation of one award being higher than another. When
in 1998 I expressed this concern, it did not create support of my fellow
Executives. However, a few years later, the Executive reached a consen-
sus on keeping just one award. We chose Paul Erdős, as the person of
our time, whom some of us knew well personally. The Federation now
presents up to four Paul Erdős awards at Congresses of the Federation
and at ICME’s.

The great Paul Erdős attended the 2nd Congress of the Federation in
1994 in Pravetz, Bulgaria. The organizer Petar Kenderov asked me to
invite Paul, which is what I did.

Figure 5: Doing Math on Excursion: Jaroslav Švrček (left), Paul Erdős,

and Josef Molnár, 2nd WFNMC Congress, Pravetz, Bulgaria, 1994

The 1994 winners of the Paul Erdős Award had the distinct honor
of receiving it from Paul Erdős. They included Qiu Zonghu, China:

13
 Mathematics Competitions Vol 31 No 2 2018

Figure 6: Even dinner cannot stop our doing Math: Paul Erdős, Alexan-
der Soifer, and Kiril Bankov, 2nd WFNMC Congress, Pravetz, Bulgaria,
1994

Urgengtserengiin Sanjmyatav, Mongolia: Jordan Tabov, Bulgaria: and

Peter Taylor, Australia.

3 Congresses of the Federation

Starting in 1990, the Federation has been holding its International Con-
gresses every four years. I have already mentioned several of them above.
Let me list all of them here, so that you can see their amazing geography:
– WFNMC-1: Waterloo, Canada (1990)
– WFNMC-2: Pravets, Bulgaria (1994)
– WFNMC-3: Zhongshan, China (1998)
– WFNMC-4: Melbourne, Australia (2002)
– WFNMC-5: Cambridge, United Kingdom (2006)
– WFNMC-6: Riga, Latvia (2010)
– WFNMC-7: Barranquilla, Colombia (2014)
– WFNMC-8: Semriach, Austria (2018)

14
 Mathematics Competitions Vol 31 No 2 2018

– WFNMC-9: to be held in Sofia, Bulgaria (2022)

Permit me to add some information about the last three Congresses of

the Federation.

6th WFNMC Congress, Riga, Latvia, 2010

The 6th Congress of the Federation took place in the historic Riga,
Latvia. Local Organizing Committee was headed by Dr. Dace Bonka
and included Dr. Liiga Ramaana and other dedicated Latvian Mathe-
maticians. I created an elaborate program that included talks, work-
shops, problem hour, etc. We were treated to fine concerts of classical
and Latvian folk music, some performed by professional musicians and
others by students majoring in mathematics. Bach and Vivaldi per-
formed on Europe’s largest organ in the Riga Dom were unforgettable.
The following plenary talks were presented:
– Some Olympiad Problems in Combinatorics and their Generaliza-
tions, by Andris Ambainis (Latvia)
– New Bridges between Research and Olympiad Problems, by Alexan-
der Soifer (USA)
– Eyewitnessing the IMO—Decades of Stability and Change, by Matti
Lehtinen (Finland)
– The Application of Mathematics as the Source of New Ideas, by
Andris Buiķis (Latvia)

7th WFNMC Congress, Barranquilla, Colombia, 2014

The 7th Congress of the Federation was magnificently organized by
Maria Falk de Losada in Barranquilla, Colombia, in July-2014. All
delegates lived together in a resort-hotel that allowed for much inter-
action. The Congress featured the Opening Keynote Address Comput-
ers and Mathematics: Problems and Prospects by the world-renowned
mathematician and inspiring speaker Ronald L. Graham. The following
Plenary Talks were given:
– The organization of the International Mathematical Olympiad—
IMO with particular emphasis on IMO 2013 by Maria Falk de
Losada (Colombia);
– A Few Thoughts on the Putnam by Mark Krusemeyer (USA);

15
 Mathematics Competitions Vol 31 No 2 2018

Figure 7: Delegates of the 6th Congress of the Federation, July 2010,

Riga, Latvia

– Math Olympiads for the Public Schools in Brazil II by Michel Spira

(Brazil);
– Is there an impact of mathematical competitions on the development
of mathematical research? The Romanian experience by Radu Golo-
gan (Romania);
– A Paper about Pentagon Problems by Robert Geretschläger (Aus-
tria).

The Closing Address Predicting the Future: Four Classic Conjectures of

Mathematics was presented by Alexander Soifer. The Congress included
many other talks, workshops, and exhibits. A tour of colorful Cartagena
served as the icing on this delicious Congress. The following few photos
may enliven your perception of the event.

At the General Meeting, the delegates decided to hold Federation’s elec-

tion during our own Congresses, which have been much better attended
than our meetings during massive ICME Congresses. For that purpose,
the terms of the Federation officers have been extended by two years.

16
 Mathematics Competitions Vol 31 No 2 2018

Figure 8: The Russians are coming: Sergei Dorichenko (left) and Nikolai
Konstantinov, Barranquilla, 2014

Figure 9: Colorful Cartagena welcomes you!

17
 Mathematics Competitions Vol 31 No 2 2018

Figure 10: Opening and Closing Speakers: Ronald L. Graham (right) &
Alexander Soifer, Barranquilla, 2014

8th Congress, Semriach (Graz), July 2018

All local arrangements, registration, lodging, meals, excursions, etc.

were beautifully organized by our Vice-President Robert Geretschläger.
Zita Hauptmann Geretschläger created a lovely ambiance at the con-
gress. The Program was created by Senior Vice-President and Chair
of the Program Committee Kiril Bankov and me. There will be a book
assembling the talks and the workshop of the Congress, edited by Robert
Geretschläger, and published by World Scientific. It will also include a
chapter on competition problems proposed by the participants of the
Congress.

During the Congress three Constitutional Amendments were approved,

limiting service on Standing Committees and most members of the Ex-
ecutive to two consecutive 4-year terms, and allowing past presidents to
remain active voting members of the Executive. The following plenary
talks were presented:
– From the Lifting-the-Exponent-Lemma to Elliptic curves with iso-

18
 Mathematics Competitions Vol 31 No 2 2018

Figure 11: Robert and Zita Geretschläger in Riga, 2010

morphic groups of points—how Olympiad Mathematics influences

Mathematical Research, by Clemens Heuberger (Austria);
– Beyond the Rainbow—Thoughts on the Potential of Mathematics
Competition Problems in the Classroom, Opening the section “Work
with Students and Teachers”, by Robert Geretschlaeger, (Austria);
– The Impact of Mathematical Olympiads on the Mathematics Com-
munity of Colombia, Opening the section “Competitions around the
World”, by Maria Falk de Losada (Colombia);
– Building Bridges Between Olympiads & Mathematics: Three Long-
Distance Trains of Thought, Opening the section “Building Bridges
between Problems of Mathematical Research and Competitions”,
by Alexander Soifer (USA);
– Problems: The Treasure of The Mathematics Competitions, Open-
ing the section “Creating Problems and Problem Solving”, by Kiril
Bankov (Bulgaria).

19
 Mathematics Competitions Vol 31 No 2 2018

4 Our book Competitions for Young Mathemati-

cians: Perspectives from Five Continents
Springer International Publishing, 2017, was “dedicated to all those
people around the world, who are passing baton to next generations
of mathematicians.” Permit me to list its contents:

Part I Goals of Mathematics Instruction

Goals of Mathematics Instruction: Seven Thoughts and Seven
Illustrations of Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Alexander Soifer
Part II Geometry for Competitions
From a Mathematical Situation to a Problem . . . . . . . . . . . . . . . . 27
Francisco Bellot-Rosado
Techniques for Solving Problems of Plane Geometry . . . . . . . . . . . 55
K. P. Shum
Part III Combinatorics for Competitions
Arrangements and Transformations of Numbers on a Circle: An
Essay Inspired by Problems of Mathematics Competitions . . . . . . 101
Kiril Bankov

20
 Mathematics Competitions Vol 31 No 2 2018

Combinatorial Problems in the Mathematical Olympiad of Cen-

tral America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . 123
L. F. Cáceres-Duque, J. H. Nieto Said and R. J. Sánchez Lamoneda
Part IV Role of Competitions in the Classroom
The Rainbow of Mathematics—Teaching the Complete Spectrum
and the Role Mathematics Competitions Can Play . . . . . . . . . . . 145
Robert Geretschläger
Competition Aims to Develop Flexibility in the Classroom . . . . . 171
Ingrid Semanišinová, Matúš Harminc and Martina Jesenská
Discovering, Development, and Manifestation of Mathematical
Talent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Iliana Tsvetkova
Part V Original Competitions
International Mathematical Tournament of Towns . . . . . . . . . . . 205
Nikolay Konstantinov and Sergey Dorichenko
Madhava Mathematics Competition—A Recent Initiative in In-
dia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
V. M. Sholapurkar
Part VI Thoughts About Competitions’ Present and Future
From the Lifetime Experience of a Seasoned Math Educator—
Thoughts, Hopes, Views and Impressions . . . . . . . . . . . . . . . . . . 271
Romualdas Kašuba
Future Directions for Research in Mathematics Competitions . . . . 303
Peter James Taylor
Part VII A Bridge Between Competitions
and ‘Real’ Mathematics
Are Mathematics Competitions Changing the Mathematics that
Is Being Done and the Way Mathematics Is Done? . . . . . . . . . . . 329
Marı́a Falk de Losada
Classic Conjectures Allow Young Mathematicians to Commence
Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Alexander Soifer
About the Editor and Contributors . . . . . . . . . . . . . . . . . . 373
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385

21
 Mathematics Competitions Vol 31 No 2 2018

The Convenor and Chair of International Program Committee of ICME-

13, wrote a most complimentary foreword for our book. Here is a
quotation from her text:
Mathematical competitions are a chance for mathematically talented
young scholars to experience mathematics as a research-oriented
discipline. These competitions offer the chance to get insight into
the beauty of mathematical structures at a high level, which many
of these young mathematicians usually will not experience at home.
Furthermore, these competitions allow to meet other talented young
mathematicians, exchange their ideas with them and experience
that they are not singular and isolated youngsters, but part of an
important community.

Despite this high importance of mathematical competitions, either

as mathematical Olympiad or as mathematical tournament of towns
or other kinds of mathematical competitions, there exists hardly
any scientific research about mathematical competitions. This is
surprising, because these mathematical competitions have a long
tradition and a high influence on the promotion of young talented
mathematicians.

At the occasion of the 13th International Congress on Mathemat-

ical Education (ICME-13) a Topic Study Group on Mathematics
Competitions took place, at which famous researchers working in
this field met and exchanged about the state-of-the-art in this field.
This intensive work together with papers from related groups forms
the basis of this book. The book provides an excellent overview
about the current discussion, topical themes and experiences with
mathematical competitions. It starts with reflections on goals of
mathematics education, problems coming from geometry or combi-
natorics being used in mathematical competitions. The next parts
reflect on the role of competitions in the classroom, this theme is
hardly researched so far. Then two examples of mathematical com-
petitions are analyzed. The last two parts focus on the present state
of mathematical competitions and its future and a bridge between
competitions and ‘real’ mathematics.

To summarize, this book is more than overdue and reflects from an

academic perspective on the potential of mathematical competitions
for mathematics education in general.

22
 Mathematics Competitions Vol 31 No 2 2018

I wish to congratulate the editor—Alexander Soifer—and the con-

tributors to this timely and excellent book.

Hamburg, Germany
Gabriele Kaiser
Convenor of the 13th International Congress
on Mathematical Education, University of Hamburg

5 The State of the Federation: Summing Up

In this historical essay, I listed many talks to show what we do. In
essence, we are what we do!

Through the 34 years, the Federation has expanded its interests, publica-
tions, and activities. It was the 4th organization to become an Affiliate
of the International Commission of Mathematical Instruction (ICMI) of
the International Mathematics Union (IMU). Our members Peter Taylor
and Ed Barbeau led the way in an ICMI Study, resulting in a book. Our
Riga 2010 Congress produced two books. The Springer “Perspective
from Five Continents” book was primarily authored by our members.
Robert Geretschläger is assembling a book of Graz-2018 Congress.

However, compared to the earlier years, the number of national IMO

leaders, participating in the Federation activities, seems to have declined.
I hope their number will grow, and in part for that we are scheduling our
9th Congress of the Federation in 2022 in Europe to immediately follow
the International Mathematical Olympiad (IMO) that will be held in
Norway.

The so-called “Giftedness” Group, attracted away some of our regular

participants, such as our long term member Emilia Velikova and local
organizers of our Riga-2010 Congress. It is not surprising considering
that their circle of interests is broader than ours.

I encourage all my colleagues to become more active in our Congresses,

our topic study groups at ICME Congresses, and in writing for our
journal Mathematics Competitions (send your submissions to the Editor,
i.e., to me at [email protected]). We define ourselves: we are what we
do.

23
 Mathematics Competitions Vol 31 No 2 2018

The Federation is 34 years old. I hope it will reach the Golden An-
niversary in 2084 and continue beyond! I am grateful for the trust and
support of the membership and the opportunity to be your President
from 2012 through 2018.

Alexander Soifer
University of Colorado at Colorado Springs
1420 Austin Bluffs Parkway, Colorado Springs, CO 80918
USA
E-mail: [email protected]
http://www.uccs.edu/˜asoifer/

24
 Mathematics Competitions Vol 31 No 2 2018

The Cuban Mathematics Olympiad:

a fragmentary journey

Frank Gamboa de la Paz & Jorge Marchena Menéndez

Frank Gamboa is the student of Math-

ematics Sciences in University of Ha-
vana. He was member of the National
Mathematics Preselection in Cuba dur-
ing three courses, winner of bronze
and gold medals in the Iberoamerican
Collegiate Mathematics Olympiad in
2016 and 2017. Since 2016, he became
trainer for the Cuban teams to the
IMO, the Iberoamerican Mathematics
Olympiad and the Centro American
Mathematics Olympiad. Olympiad prob-
lems building and inequalities are his
passions.

Jorge Marchena is the student of Math-

ematics Sciences in University of Ha-
vana. He was member of the Na-
tional Mathematics Preselection and
the National Programming Preselec-
tion in Cuba. He also was winner
of bronze medals in the Iberoameri-
can Collegiate Mathematics Olympiad
in 2017. Since 2016, he is competitor
of the ACM-ICP developing a big in-
terest in topics such as Computational
Geometry and Combinatorics. He was
Problem Coordinator of the XX Cen-
troamerican and Caribbean Mathemat-
ics Olympiad in 2018, Havana, Cuba.

25
 Mathematics Competitions Vol 31 No 2 2018

1 A quick overview to the Selection Process in Cuba

The whole country was witness to the Cuban Mathematics Olympiad for
high school students in the second half of the February in 2018. This time
149 contestants from the three grades took part and finally, 28 of them
were chosen to participate in the National Mathematics Preselection,
in a preparative process that started in March 26th and will end in
September along with last event of the Olympic year, the Iberoamerican
Mathematics Olympiad. During this period was realized in Cuba from
the 15th to 23rd of June the Centro-American Mathematics Olympiad,
and also took place in the International Mathematics Olympiad at Cluj-
Napoca, Romania, in July.

The academic course in Cuba starts in September first week and the
selective process has its beginnings in the contests to a base level in
the educational centers. The next phase is composed by the municipal
contests (municipal phase) to which those students with high notable
results in the first phase are invited. Finally in January, the Cuban
Mathematics Olympiad (OMCC by its Spanish acronym) takes place in
every single one of the 16 provinces.

As expected, the difficulty of the contest gets necessarily higher from one
phase to another, and has as main aim the National Olympiad. Once
the international Olympiad training is started, near to the end of March
or the beginning of April, competitors of the National Preselection are
subjected to the team selection tests with a high difficulty and very
diverse problems.

That way, competitors with the four highest scores of 10th grade are in-
vited to be part of the National Team to OMCC (the Centro American
teams to OMCC can be confirmed up to four members since 2018 accord-
ing to the international judge members). In addition, the four highest
scorers of 11th and 12th grade do also enter into the national team.
The highest scorer (Cub 1) gets to represent Cuba in the IMO in the
meantime the other four students represent Cuba on the Iberoamerican
Mathematics Olympiad. Occasionally only the best four positions are
classified in case that student (Cub 1) could also be allowed to partici-
pate in the Iberoamerican Mathematics Olympiad. This fact is given due
to since 11 years ago, the Cuban teams to IMO contain a single member,

26
 Mathematics Competitions Vol 31 No 2 2018

excluding Mexico 2005, in which, a team of four participated and the

other five places had been in vacancy caused by economic problems that
have stopped Cuba full participation in these competitions since then.
Between all the Cuban incursions to these Olympiads till 2017, we have
reached and achieved 24 mentions, 37 bronze medals, 7 silver medals and
a unique gold medal (IMO 2001), according to statistics from [1].

2 Cuban Mathematics Olympiad 2017: most repre-

sentative problems
In the Cuban Mathematics Olympiad, celebrated in 2017, on the second
day can be found a problem for 12th grade as question #3:

Problem Let p be a prime number. Show that the equation

m2 + n2 = (m, n, p) · [m, n, p],

has positive integer solutions if and only if p can be expressed as sum of

two perfect squares.

Solution 1 : We will first prove that if p = a2 + b2 for some integers a, b

there exists a positive integer solution for the above equation. Notice
that if d = (a, b), then d2 | p, by which d = 1, i.e. a and b are
relative primes. Furthermore, p  a and p  b, due to in a contrary
way (a, b) ≥ p > 1. Let m = a2 b and n = ab2 . We already know that
(m, n, p) = 1 and [m, n, p] = a2 b2 (a2 + b2 ), and finally

m2 + n2 = a2 b2 (a2 + b2 ) = (m, n, p) · [m, n, p].

Let’s see the other implication. We will assume that the equation has
positive integer solutions. Let (m, n) be one of them. It is evident that
(m, n, p) | p, by which (m, n, p) = 1 or (m, n, p) = p. If (m, n, p) = p, it
follows that p | m and p | n, there exist integers a, b, such that m = ap
and n = ap. Note that [m, n, p] = p · [a, b]. Substituting in the equation
and simplifying we obtain

a2 + b2 = [a, b].

27
 Mathematics Competitions Vol 31 No 2 2018

Using the Arithmetic and Geometry Means inequality (AM−GM in-

equality), we conclude that

a2 + b2 ≥ 2ab > ab ≥ [a, b].

Therefore, we reach a contradiction. Then (m, n, p) = 1, which implies

p  (m, n). We can assume without the loss of generality that p  m. Note
that p | [m, n, p], by which p | m2 + n2 , this directly implies that p  n,
because (m, n, p) = 1. Let d = (m, n) and m1 n1 be coprime positive
integers such that m = dm1 and n = dn1 . Substituting in the equation
and simplifying we obtain the equality

d(m21 + n21 ) = m1 n1 p,

and given that 1 = (m, n, p) = (d, p), it follows that d | m1 n1 . On the

other hand, it’s easy to prove that (m21 + n21 , m1 n1 ) = 1, so m1 n1 | d.
This implies that d = m1 n1 . Finally
 m 2  n 2
p = m21 + n21 = + .
d d

This problem, solved by a small number of students in the Olympiad,

though it does not offer a very high difficulty, goes through a fact that
stayed out the light in some solutions: the searched primes are of the
form 4k + 1 (except 2).

It is well known that an odd prime p can be represented as sum of

two perfect squares if and only if p = 2 or p = 4k + 1, for some positive
integer k: this result is known as Fermat’s Theorem about the sum of two
perfect squares or the Fermat’s Christmas Theorem [2]. Another well-
known fact within the Elementary Number Theory is that if p = 4k + 3
is a prime and p | x2 + y 2 with x, y ∈ Z, then p | x and p | y. So that
way, the following solution to the same problem is a little bit shorter and
but also more complicated to figure out.

Solution 2 : We will only prove that if the equation has positive integer
solutions, then p is as desired. The other side of the proof is the same
as the above solution. In fact, let (m, n) be a solution. Let’s assume
that p = 4k + 3 for certain positive integer k. Because p | [m, n, p], then

28
 Mathematics Competitions Vol 31 No 2 2018

p | m2 + n2 , by which p | m and p | n, so (m, n, p) = p, hence

 
2 2 2 m n m n
m + n = p · [m, n, p] = p · , ≤ p2 · · = mn < 2mn,
p p p p

which is absurd. By this, p = 2 or p = 4k + 1, and the conclusion is

given by Fermat’s Christmas Theorem.

The Cuban Mathematics Olympiad is divided into two exams within two
days: the three grades common test (day #1) and the individual tests
for each grade (day #2).

The next problem, also attached to Olympiad of 2017, is the problem

#3 of the first day. The statistics gave prove that in the Olympiad, the
average scores for this exercise is smaller than 1, in a total of 7 points;
this fact makes it a high difficulty problem. Before showing you the
statement, it is important to emphasize that Mathland (the country of
the Mathematics) is a fictitious place used to give place to combinatorial
problems, with an elegant and familiar touch to the students in the
Cuban MO.

Problem In Mathland there exist provinces as positive integers. From

province p it is only possible to travel to provinces 2p and p+1. Find the
least amount of required moves to get from province 2017 to province
21997 .

Solution 1 : It could seem intuitive to start to multiply by 2, up to the

maximal province p, where if we multiply again by 2 we will get to a
province k > 21997 . Once in p, it would be enough to add 1 several times
up to get to 21997 . Apparently, this does not seem absurd, but note that
211 = 2048 > 2017 is the nearest power of two to 2017, by which the
amount of times that we should multiply by 2, according to the above
strategy is 1986, because 2017 · 21986 < 21997 and 2017 · 21987 > 21997 .
But the amount of times that we should then add 1, would be in order
of 21986 , which is a huge number. A most reasonable strategy would be
adding 1 up to the first power of 2, which is obtained in 2048−2017 = 31
steps, and then multiply 1986 times by 2 up to 21997 , with a total of 2017
moves. There is no doubt that, by this way it is possible to make the

29
 Mathematics Competitions Vol 31 No 2 2018

required journey with 2017 moves. We will prove then, that this cannot
be possible in less than 2017 moves.

Consider the function G : A → Z with A = {(a, b) ∈ Z>0 × Z>0 : b ≥ a},

defined by G(a, b) = 0 if a = b; if b ≥ 2a with b even, then
 
b
G(a, b) = 1 + G a, ,
2

and G(a, b) = 1 + G(a, b − 1) in any other case. Let’s see the following
claims.

Claim 1 For all pairs (a, b), holds G(a, b + 1) ≤ 1 + G(a, b).

Proof : We proceed by strong induction on b, taking a as fixed. Notice

that G(a, a + 1) = 1 and G(a, a + 2) ≤ 2; that way G(a, a + 2) ≤
1 + G(a, a + 1). This proves the claim for b = a + 1. We assume that for
all j with a < j ≤ k it holds that G(a, j + 1) ≤ 1 + G(a, j). If k is even,
then k = 2t for certain integer t; it follows that k + 1 = 2t + 1, and

G(a, k + 1) = G(a, 2t + 2) = G(a, 2t) + 1 = G(a, k) + 1

and in this case, the claim holds. Assume then that k is odd, meaning
that k = 2t + 1 for a certain integer t; that way k + 1 = 2t + 2. If
t + 1 > a, we would have

G(a, k + 1) = G(a, 2t + 2) = 1 + G(a, t + 1) ≤ 2 + G(a, t)

= 1 + G(a, 2t) = G(a, 2t + 1)
= G(a, k) < G(a, k) + 1

where the inequality is strict; now, if t + 1 = a, then

G(a, k + 1) = G(a, 2t + 2) = 1 + G(a, t + 1) = 1 ≤ G(a, k) + 1.

The final case is that t + 1 = a, but this situation is evident due to

G(a, k + 1) = G(a, k) + 1 and the claim have been proved.

We will prove that function G represents algorithmically the sequence

to follow steps to get from one province a to another b using the allowed
travels.

30
 Mathematics Competitions Vol 31 No 2 2018

Claim 2 We denote by M (a, b), the least amount of moves to get from
a to b using the maps x → 2x and x → x + 1. Then M (a, b) = G(a, b)
for all (a, b) ∈ A.

Proof : Let’s use strong induction one more time on b. Note that
M (a, a+1) = 1 = G(a, a+1) and the base case holds. Assume that for all
j with a < j ≤ k, it holds that M (a, j) = G(a, j). With a reproductive
way, we split the analysis in two cases. If k = 2t, then k + 1 = 2t + 1,
hence M (a, k + 1) = 1 + M (a, k), because to have k + 1 odd, the last map
added was 1; that way, as M (a, k) = G(a, k) by hypothesis of induction,
the conclusion is direct to be G(a, k + 1) = 1 + G(a, k). Then, we assume
that k = 2t + 1. If t + 1 > a, we should analyze two possible cases
because (remember the definition of M )
  
k+1
M (a, k + 1) = min 1 + M (a, k), 1 + M a, .
2

We analyze that M (a, k + 1) = 1 + M (a, k). In that case

M (a, k+1) = 1+M (a, 2t+1) = 1+G(a, 2t+1) = 2+G(a, 2t) = 3+G(a, t)

The other possibility implies that, according to Claim 1

M (a, k + 1) = 1 + M (a, t + 1) = 1 + G(a, t + 1) ≤ 2 + G(a, t) < 3 + G(a, t).

We can conclude that

   
k+1 k+1
M (a, k + 1) = 1 + M a, = 1 + G a, = G(a, k + 1).
2 2

If t+1 = a, then M (a, k+1) = 1, because by simply multiply a per 2 and

G(a, k + 1) = 1 + G(a, t + 1) = 1. Finally, if t + 1 < a, then k + 1 < 2a,
and by this reason, we can only get to k + 1 with the maps x + 1 in this
case; hence M (a, k + 1) = k − a + 1 and similar G(a, k + 1) = k − a + 1,
in virtue of definitions of G.
 
To finish this problem it is enough to compute M 2017, 21997
   
M 2017, 21997 = G 2017, 21997
 
= 1986 + G 2017, 211 = 1986 + 31 = 2017

31
 Mathematics Competitions Vol 31 No 2 2018

and the problem is solved.

The previous solution was the official one proposed to the Olympiad with
the above formulation of the problem; however, we could generalize and
make it lose its nice structure with the following statement.

Problem Let D be a digraph with N vertices v1 , v2 , ..., vN such that,

for every vertex vk there exist exactly two edges going from vk : one that
connects it with vk+1 and another with v2k . Find the least amount of
edges that we need for move up to the vertex vB from vA , where A, B
are positive integers such that 1 ≤ A ≤ B ≤ N .

We are not going to represent another explicit solution on this problem,

however we will comment on some ways to attack it. The first one
would be purely computational, by modeling the problem; in fact, we
could create a counting function of the amount of passed edges, and
optimize this function about the restrictions of the graph in the previous
formulation. We present here a possible model for the problem:


B 
B
min ci,j
i=A j=i
s.t. ck,2k + ck,k+1 = 1, A < k < B, .
c1,2 = 1,
ci,j = 0, i = [A, B], j = [B + 1, ...2B]
ci,j = {0, 1} i, j ∈ [A, B]

Here cij , represent the decision of choosing or not the edge that joins the
i, j vertices to define the path. This model can be solved through the
usage of an adequately implicit enumeration algorithm, i.e. DFS (Depth
First Search).

An alternative method of solving the problem would be a refined way

exposed in the website www.mathlinks.ro, which is a well-known online
mathematics forum. If we write the number 21997 in binary system, we
will obtain the number a = 100...00 with 1997 zeros exactly, and 2017
in binary system is b = 11111100001. Then, the statement is equivalent
to find the lowest amount of mappings of type x → x + 1 and x → 2x,
to obtain a coming from b.

32
 Mathematics Competitions Vol 31 No 2 2018

The mappings of type x → 2x applied to a number N adds one 0 at the

end of the binary representation of N . The ones of type x → x + 1, if the
last digit is a 0 then it is changed into a 1, and in the contrary case it is
changed into a 0 and all previous digits are updated following the same
idea. Then, each N = 2k −1 with k ≥ 1, this map adds one more digit to
the binary representation of N . Then, a non-refined conclusion would be
to use the previous maps to go from an 11-digit number to a 1998-digit
one; if we exclusively use the previous mappings of type x → 2x, we
would need 1998 − 11 = 1987 applications up to the 1998-digit number,
however, this number will be formed by several digits equal to 1. Now,
during the process, we could apply x → x + 1 at any moment, but it is
obvious that if we do this at the beginning, we will obtain the desired
minimum. The reason for this affirmation is that we should add 1s until
we find to the next power of 2; if we do maps of type x → 2x before
doing the ones of type x → x + 1, then, we will need more applications
to find the next power of 2.

The nearest power of 2 to 2017 is 211 = 2048, and so we would need

2048 − 2017 = 31 maps of type x → x + 1 and 1997 − 11 = 1986 maps
of type x → 2x, for a total of 31 + 1986 = 2017 mappings. The problem
is that this solution consists in supposing that the optimal strategy to
make movements from A to B is to get always to the power of 2 and
further than A, which does not seems evident.

3 Cuban Mathematics Olympiad 2018: three prob-

lems with beautiful solutions
In the last part of this article, we will debate three problems of the
Olympiad 2018 with their official and alternative solutions; some of these
where offered by students in the contest itself. The first one of these
problems corresponds to the #3 problem of day #1, that is, the common
exam, where, fortunately, more than one perfect score was reached. As
is usual with geometrical problems this exercise came out spontaneously
in the assistant Geogebra, meanwhile one of the authors played to create
circumferences over a square. The solution 1 was taken from the official
solution to the problem in the Olympiad, and it’s a contribution of Jorge
Estrada, professor of Mathematics at the University of Havana.

33
 Mathematics Competitions Vol 31 No 2 2018

Problem 1 Let ABCD be a square. Determine the geometrical place

of all points P inside the square such that the circumcenters of triangles
ABP , BCP , CDP and DAP are concyclic.

Solution 1 : Let’s prove that the square diagonals are the geometrical
place. In effect, if P is over the diagonals it is easy to check that the given
circumcenters form an isosceles trapezium, which is clearly cyclic. Later,
the diagonals accomplish the required property. Then, let us assume P
is such that all circumcenters are concyclic. If P is the center of ABCD,

the circumcenters make up a square. If it is not the case, let us assume

without loss of generality that P ∈
/ AC, and let’s prove that P ∈ BD.

34
 Mathematics Competitions Vol 31 No 2 2018

Let G, H, I and J be the circumcenters of triangles AP B, BP C, CP D

and DP A, respectively. Let P  be the image of translate P through
−−→
the vector AB. You shall notice that AP ⊥JG, BP ⊥GH, CP ⊥HI and
DP ⊥IJ, this implies ∠AP B = π − ∠JGH, and ∠CP D = π − ∠JIH.

By hypothesis, GHIJ is a cyclical quadrilateral, then ∠JGH + ∠JIH =

π, and hence ∠AP B + ∠CP D = π. By translation, ∠AP B = ∠CP  D,
and then CP  DP is cyclical. It follows that ∠P  DC = ∠P  P C. Due
to ∠P  DC = ∠P AB by translation properties and ∠P  P C = ∠P CB,
since they are alternate angles, we have ∠P AB = ∠P CB. Let’s look
now at the next lemma.

Lemma Consider the points X, Y, Z such that |XY | = |Y Z|. Within

∠XY Z let’s take one point T ∈/ ZX, such that the oriented angles
∠Y XT = ∠T ZY . Then, T lies on the bisector of ∠XY Z.

Proof : Let M be the intersection point of XT and Y Z; N the inter-

section point of ZT and XY ; F and G projections of T on XY and
Y Z. We have that the triangles Y XM and Y ZN are equal. Later,
|Y M | = |Y N |, and since |XY | = |Y Z| we also have |XN | = |ZM |.
Moreover, ∠Y N T = ∠Y M T , then ∠T N X = ∠T M Z. We conclude
that XN T and T M Z are equal triangles. It follows that |T F | = |T G|
are equal altitudes within equal triangles, i.e. T is on the bisector of
∠XY Z.

Using the previous lemma, we conclude that P is on the bisector of

∠ABC. Following the fact that ABCD is a square, that bisector is BD,
i.e. P ∈ BD.

The key of all known solutions is to prove that ∠P AB = ∠P CB, or

similar relations with points P, A, B, C, D. From that point, there exist
other ways to conclude without using the previous lemma. The one
we will present now was offered by the 11th grade student Sofı́a Albizu
Campos, golden medalist in the XIX OMCC 2017 in El Salvador with
a perfect score and golden medalist with the best score in the Cuban
Mathematics Olympiad 2018. The conclusion uses the Law of Sines.

Solution 2 : Consider that Q, R, S and T are the cirumcenters of the

triangles P AB, P BC, P CD and P DA respectively. It is not difficult
to check that SR ⊥ P C and QT ⊥ P A. Since SQ and RT intersects

35
 Mathematics Competitions Vol 31 No 2 2018

perpendicularly, we can deduce that SQ ⊥ CD, and following it we

have ∠RSQ = ∠P CD. Analogously ∠P AD = ∠QT R. Since the
quadrilateral is cyclical, it follows that ∠P AD = ∠P CD, i.e. ∠P AB =
∠P CB. Applying the Law of Sines for the triangles P AB and P CB
respectively we have:
AB PB BC PB
= , =
sin ∠AP B sin ∠P AB sin ∠BP C sin ∠P CB

Following the previous conclusions and as AB = BC, it follows that

sin ∠AP B = sin ∠BP C. At this point, we have two possible situations:
∠AP B = ∠BP C or ∠AP B + ∠BP C = π. In the first case, we have
P ∈ BD. In the second one P ∈ AC, then P belongs to the diagonals of
ABCD. On the other side, is clear that if P belongs to the diagonals,
the quadrilateral QRST is an isosceles trapezium, and hence is cyclical.
In conclusion, the searched locus are the square diagonals.

It is not difficult to prove that if we don’t restrain P at the interior

of ABCD, and we just demand that P doesn’t belong to the straight
line which determines the square sides (in that case, triangles shall be
degenerated), then the locus are the diagonals of ABCD. However,
for avoiding the unnecessary analysis of some cases it was preferred to
state the problem in the way shown. Many students just proved that
the diagonals of ABCD accomplished the required property, and hence,
that they were a subset of the requested locus.

Along with this geometrical problem, two others problems were evalu-
ated in the first day of the Cuban Mathemathics Olympiad. The prob-
lem # 1, was without any doubts the easier and the one with higher
accumulated scores between the two exams, accomplishing the Problem
Selection Committee expectations. However, when this easy Number
Theory exercise was elaborated the very first time it wasn’t easy at all.
The original statement said:

Problem 2 For certain pairs of integer numbers (m, n), we say that a
rational number r is (m, n)-representable if there exist integers a and b
such that
a3 − m
k= 3 .
b +n
36
 Mathematics Competitions Vol 31 No 2 2018

If we can guarantee also that a, b > 0, then r is said (m, n)-megarepre-

sentable,

a) Prove that 2018 is not (3, 23)-representable.

b) Prove that 2 is not (1, 1)-megarepresentable.

c) Prove that for all rational numbers r > 0 there exist positive
integers m, n for which r is (−m3 , n3 )-megarepresentable.

In the end it was decided to just ask for case a) with the following for-
mulation, which minimized the problem difficulty almost in its totality.

Problem 2a Prove that there do not exist integer numbers x, y that

verify the equation
x3 − 3
= 2018.
y 3 + 23

Solution 1 : To prove it let’s use modular congruence choosing adequately

n that allows us to arrive to a contradiction. Let’s analyze n = 7. Let’s
prove that the number 2018y 3 +23·2018+3 is not a perfect cube modulo
7. By simple inspection we can notice that y 3 ≡ 0, ±1 (mod 7), (also
p−1
we can note that if p is an odd number, then a 2 is congruent with ±1
modulo p always that p  a), and hence

x3 ≡ 2018y 3 + 23 · 2018 + 3 ≡ 2y 3 + 4 + 3 ≡ 2y 3 (mod 7).

Necessarily x ≡ y ≡ 0 (mod 7), then 73 | (x3 − 2018y 3 ) = 23 · 2018 + 3 =

46417, but this is a contradiction.

What is more curious about this exercise is that just a few given answers
used modulo 7. The majority of students preferred modulo 9.

Solution 2 : Let’s analyze modulo 9. In effect, we have that y 3 ≡ 0, ±1

(congruence are taken modulo 9), since x3 ≡ 2018y 3 + 23 · 2018 + 3 ≡
2y 3 + 4. If y 3 ≡ 0, then x3 ≡ 4; if y 3 ≡ 1, then x3 ≡ 6; if y 3 ≡ −1,
then x3 ≡ 2. For any of these three cases we get to a contradiction, from
which we conclude that there exist not integer solutions.

37
 Mathematics Competitions Vol 31 No 2 2018

It’s worthy to comment about the other two exercises presented in the
original statement of the problem. Case b) is equivalent to proving that
there are no positive integer solutions for the equation 2m3 + 3 = n3
(even more, the only pair (m, n) of integers that actually accomplish
this is (−1, 1), info gathered with Wolfram Alpha). This problem
was posted on the website www.mathlinks.ro looking for an elemental
solution without the use of Ring Theory. The unique one arrived after
many tries uses the famous Liouville Theorem for Diophantine approxi-
mations, which states in one of its versions:

Liouville Theorem If α is a rational algebraic number with n degree

 > 0 such that
over the rational numbers, then there exists α
 
  
α − p  > α .
 q  qn

The another version of the original question is exactly the problem N2

of the Short List of IMO 1999, and it can be found in [3]. An excellent
generalization of the problem stated in the case a) would be to describe
the class M of all integers k that can be represented in the form

x3 − 3
k=
y 3 + 23

for some pair (x, y) of integer numbers. In particular, the Olympiad

problem asks proof that 2018 ∈/ M.

The last problem we want to introduce to you was the problem # 2

of the individual exam for 12th grade students of the Cuban Mathe-
matics Olympiad 2018, another about geometry. It was proposed by
Robert Bosch to the Problem Selection Committee, known Cuban prob-
lem solver and a person who gained our admiration due to his outstand-
ing contributions to the Olympiad world. Robert told us when he showed
us this problem that he was unable to reach a purely synthetic solution,
any of his solutions strongly required the use of trigonometry, or at least
of metrical calculations. It was also considered the possibility of giving
a special prize in the Olympiad to those students who were capable to
find a synthetic solution. And regarding the Number 6 problem of IMO

38
 Mathematics Competitions Vol 31 No 2 2018

1988, the capabilities of students for reaching elegant solutions can not
be underestimated.

Problem 3 Consider the circumferences Γ1 , Γ2 and Γ3 with their

centers on A, B and C and with radius lengths a, b and c respectively.
Let’s suppose that Γ2 is tangent to Γ1 in P and Γ3 is tangent to Γ1 in
Q. Let be r the exterior tangent segment to Γ2 and Γ3 . It is known that
the segments r and P Q are located in different half-planes with respect
to the straight line BC. Prove that
a2 |r|2
|P Q|2 = .
(a + b)(a + c)

Solution 1 : Let’s denote the angle ∠P AQ = α and d = |r|. Let us

consider a few auxiliary constructions: the segments AB, AC, BC, the
radius lines of C2 and C3 that are perpendicular to r, and lastly we trace
a segment starting from the point B and that is perpendicular to the
C3 radius line. It’s clear that this new segment has length d. Look at
the next figure. The solution uses the Law of Cosines and Pythagoras’
Theorem. The following relations are true:
P Q2 = 2a2 (1 − cos α), (1)
BC 2 = (a + b)2 + (a + c)2 − 2(a + b)(a + c) cos α, (2)
applying the Law of Cosines in the triangles P AQ and BAC respectively.
Note that we shall prove the following equality
d2 = 2(1 − cos α)(a + b)(a + c).
Now from Pythagoras’ Theorem we have
BC 2 = d2 + (c − b)2 ,
and if we equate with equation (2) it results
d2 = (a + b)2 + (a + c)2 − (c − b)2 − 2(a + b)(a + c) cos α,
= 2a2 + 2ab + 2bc + 2ca − 2(a + b)(a + c) cos α,
= 2(a + b)(a + c) − 2(a + b)(a + c) cos α,
= 2(1 − cos α)(a + b)(a + c).

39
 Mathematics Competitions Vol 31 No 2 2018

It was not a complicated problem in the Olympiad, but beautiful enough

for captivating the attention of the tribunal members when it came out
the first synthetic solution, offered by a girl student of 12th grade, mem-
ber of the National Mathematics Preselection 2018, and gold medalist
in this edition of the Olympiad.

Solution 2 : Let M and N be tangential points of r with Γ2 and Γ3

respectively. Let be α = ∠P AQ, β = ∠QCB, γ = ∠BCN , δ = ∠CBM
and  = ∠P BC. The angle ∠QN M is semi-inscribed on the arc N Q,
hence 2∠QN M = ∠QCN = β + γ. Analogously 2∠P M N = δ + . On
the other hand
α = 180◦ − 2∠AQP,
∠AQP + ∠P QN + ∠N QC = ∠AQP + ∠P QN + 90◦ − ∠QN M.

It implies that 2∠N QP = α+β+γ. In the same way, 2∠QP M = α+δ+.

Since P M N Q is a quadrilateral, we know that ∠N QP + ∠QP M +

40
 Mathematics Competitions Vol 31 No 2 2018

∠P M N + ∠QN M = 360◦ and hence

α + β + γ + δ +  = 360◦ .
So, noticing the opposite angles sum, we realize that this is actually a
cyclical quadrilateral, then, it follows that
PQ DP DQ
= = .
|r| DN DM

If we consider the sum of the triangle DM N within angles is clear that

we obtain 2∠M DN = ∠P AQ, then D ∈ Γ1 , and hence A is the triangle
DP Q circumcenter, later AD = AP = AQ = a. Angles ∠AP D and
∠BP M are equal. Since AP = AD = a and P B = BM = b, moreover
the triangles AP D and M P B are similar. Analogously, it is obtained
that the triangles DAQ and CQN are similar. This way we know that
MP b NQ c DM a+b DN a+c
= , = =⇒ = , = .
DP a DQ a DP a DQ a
41
 Mathematics Competitions Vol 31 No 2 2018

Then, we can conclude

|r|2 DN · DM (a + b)(a + c)
= = .
|P Q|2 DP · DQ a2

References
[1] Robert Bosch, Concursos Nacionales Preuniversitarios de
Matemática (2000–2015).
[2] Paul Yiu, Number Theory 2 Vol 2, Spring, 2007, 41–42.

[3] D. Djukić, V. Janković, I. Matić, N. Petrović, The IMO Compendium

Vol 2, Springer 2010.

Frank Gamboa de la Paz Jorge Marchena Menéndez

Fac. of Mathematics and Computer Fac. of Mathematics and Computer
Sciences Sciences
University of Havana University of Havana
Havana, Cuba Havana, Cuba
[email protected] [email protected]

42
 Mathematics Competitions Vol 31 No 2 2018

Problems: The Treasure of Mathematics

Competitions

Kiril Bankov
Kiril Bankov prepares future mathe-
matics teachers as a professor of math-
ematics education at the University of
Sofia and the Bulgarian Academy of
Sciences in Bulgaria. He graduated and
received his PhD in mathematics at the
same University. Prof. Bankov has
been involved in mathematics competi-
tions in Bulgaria for more than 20 years
as an author of contest problems and
as a member of juries. Kiril Bankov
was the Secretary of World Federation
of National Mathematics Competitions
(WFNMC) from 2008 till 2012. In 2012
he was elected as the Senior Vice Pres-
ident of WFNMC and in July 2018 he
became the President of the Federa-
tion.

1 Introduction
There are hundreds of mathematics competitions around the world. Ev-
ery year thousands of students participate and win awards. Their names
remain in the lists and build the history of the competitions. No doubt
this is an impressive number of people, many of whom continue their
career as professional mathematicians. This is one of the goals of math-
ematics competitions: to stimulate the development of mathematical
talents.

It is not only the participants that build the history of competitions.

Actually, for any competition there are two important sets: the par-
ticipants, and the problems. The problems are the other side of the
mathematics competitions. They are the intellectual product that re-
mains in the history. There are hundreds of thousands of ideas that

43
 Mathematics Competitions Vol 31 No 2 2018

are implemented in the problems presented on mathematics competi-

tions. It is impossible to review or consider all of them. Two types of
“unforgettable” problems need special attention:
(i) Problems that do not need particular knowledge from mathematics
school curriculum to be understood but their solutions require deep
thinking, mathematical reasoning, experience, and a lot of intuition.
Sometimes the solutions are quite unexpected. These problems are
one of the best examples of the beauty of mathematics.
(ii) Problems that lead to interesting generalizations. Sometimes they
are sources of unsolved problems.

This chapter presents examples of unsolved problems that are inspired

by problems presented on mathematics competitions.

2 Problem for Exploring of a Period

The base of the problem is the following:

Situation Let n > 2 cells be arranged into a circle. Each cell can be
occupied by 1 or 0. The following operation is admissible: draw another
n cells—one between any two of the existing cells; in these new cells
write 0 if the numbers in the neighboring existing cells are equal, and 1
if these numbers are different; then delete the existing cells.

The situation was used in a problem presented on a mathematics com-

petition in 1975 in former Yugoslavia:

Problem 2.1 In the given situation, let n = 9 and four of the cells be
occupied by 1, the other five be occupied by 0. Is it possible to obtain
0s in all nine cells in a finite number of admissible steps?

Solution. The answer is “No” and here is the argument. Assume that
in a finite number of admissible steps all nine cells contain 0s. Then in
the second to the last arrangement all nine cells contain 1s. Therefore,
in the previous arrangement any two neighboring cells contain different
numbers, which is impossible, having nine cells.

44
 Mathematics Competitions Vol 31 No 2 2018

The problem gives rise to variety of generalizations. Different variations

of the initial arrangements can be considered, depending on the number
n of the cells, and on the number and the positions of the initial 1s. An
interesting generalization is presented in the following problem.

Problem 2.2 In the given situation, initially there is a 1 in one cell

and 0s elsewhere. For which values of n is it possible to obtain 0s in all
cells in a finite number of admissible steps?

Let n be such a number that it is possible to obtain 0s in all cells in a

finite number of admissible steps. Because of the arguments presented
in the solution of Problem 2.1, n must be an even number. This is why
only even values of n are interesting.

I have presented this as an open problem in my talk at TSG-30 in ICME-

13, Hamburg, 2016. At that time I did not know the answer. Soon after
the end of ICME-13, I had to prepare my talk as a chapter for a book. I
tried to find a solution to the problem and to include it in the chapter.
My desk was full of sheet of papers with circles and numbers. I used
a computer to check for different values of n. The time was passing. . .
Just before the timeline for submission of the chapter, I realized that the
Sierpinski’s triangle may help and I proved that the answer to Problem
2.2 is that the required values of n are only all powers of 2 (Bankov,
2017).

After Problem 2.2 is solved, the question is what happens for the other
values of n.

Obviously, if n is not a power of 2 the execution of the admissible

operation will never end. On the other hand, there are a finite number
of arrangements of 0 and 1 in the cells of the circle. This means that
after a certain number of steps the arrangements of the numbers will
cyclically repeat. The least number of the steps in this repetition is
called a period. Here is an open problem.

Open problem Explore how the period depends on n.

45
 Mathematics Competitions Vol 31 No 2 2018

3 Large Subset of Disjoint Figures

This part presents an open problem as a generalization of a problem
given on two famous mathematics competitions. It is about a set of
disjoint figures on the plane. Before presenting the problems, let us
consider the one dimensional case.

Problem 3.1 Let M be a finite set of segments on the line, the length
of which union is equal to L. Then there is a disjoint subset of M , the
sum of the length of whose segments is not less then L2 .
A solution can be found in Schkljarskij at al., 1974. More interesting is
that this statement presents the best possible result, i.e. for every δ > 0
there is a covering of a given segment of length L by a finite set M of
segments, so that the sum of the length of the segments of any disjoint
subset is less than L2 + δ.
A possible two dimensional case presents the following problem that was
set on the Moscow Mathematical Olympiad 1979, and on the Austrian-
Polish Mathematics Competition 1983.

Problem 3.2 Let M be a finite set of circles in the plane, the area of
which union is equal to A. Prove that there is a disjoint subset of M ,
the sum of the area of whose circles is not less than A9 .
Solution. The proof uses the Principle of Mathematical Induction on the
number of the circles of M . The statement is obvious if M contains 1 or
2 circles. Let n be a natural number, n ≥ 3. Assume that the statement
is true if M contains k circles for every k < n. We will prove that the
statement is true if M contains n circles, M = {K1 , K2 , . . . , Kn }. Let
K be the circle of M having the largest radius R among all circles of
M . Denote by A(K) the area of K. If A(K) ≥ A9 , the required subset
consists of one circle, which is K. Otherwise, denote by 3K the circle
concentric with K with radius 3R. If a circle C of M has a common
point with K, then C ⊂ 3K, because R is the largest radius of the circles
of M , (Figure 3.1). Because A(3K) = 9A(K) < A, there are circles of
M that do not have a common point with K. Denote the set of these
circles with N . Obviously, the area of the union of the circles of N is

46
 Mathematics Competitions Vol 31 No 2 2018

not less than A − 9A(K). According to the inductive assumption, there

is a disjoint subset P of N the sum of the area of which circles is not less
than A−9A(K)
9 = A9 − A(K). Then the set P ∪ {K} is a disjoint subset
of M , the area of whose circles is not less than A9 .

Figure 1: Illustration of the solution of Problem 3.2

I have presented this solution to Problem 3.2 because the same method
can be used for the proof of a generalization of this problem. The
generalization considers not only circles, but any set of bounded figures.
First, we need to get familiar with the notion of a neighborhood.
Denote by d(X, Y ) the distance between the points X and Y in the
plane. Let K be a bounded figure in the plane. The number d(K) =
sup{d(X, Y ); X ∈ K, Y ∈ K} is called a diameter of K. Let Z be a
X,Y
point in the same plane. We call a distance between the point Z and
the figure K the number d(Z, K) = inf {d(Z, Y ); Y ∈ K}. For any ε > 0,
Y
the neighborhood Oε (K) of K with radius equal to ε, is the set of all
points X in the plane with distance not greater than ε apart from K,
i.e. Oε (K) = {X; d(X, K) ≤ ε}.
To visualize the notion of a neighborhood, think of the following experi-
ment. Throw figure K in a water and look how the waves are spreading.
Their shape has the form of the neighborhood of K.
For example, if K is a circle with radius R, the neighborhood of a circle
with radius ε is a concentric circle with radius R + ε. The neighborhood
of a square with side a is shown in Figure 2. Its area is equal to

47
 Mathematics Competitions Vol 31 No 2 2018

Figure 2: Neighborhood of a square

a2 + 4aε + πε2 . The neighborhood of an equilateral

√
triangle with side a
a 3
is shown in Figure 3. Its area is equal to 4 + 3aε + πε2 .

Figure 3: Neighborhood of an equilateral triangle

The next theorem is a generalization of Problem 3.2.

Theorem Let M = {K1 , K2 , . . . , Km }. be a finite set of bounded

figures on the plane, the area of which union is equal to A. For every
A(K
i )  ,
i = 1, 2, . . . , m let us denote by di the diameter of Ki , λi =
A(Odi Ki )
and λ = min{λ1 , λ2 , . . . , λm }. Then there is a disjoint subset of M , the
sum of the area of whose figures is not less than λA.

The proof of the theorem (Bankov, 1996) uses the method of the solution
of Problem 3.2.

Namely, let K be the figure of M having the largest diameter among all
figures of M . If A(K) ≥ λA, we are done. If not, consider the set N of
the figures of M that do not have a common point with K. Prove that

48
 Mathematics Competitions Vol 31 No 2 2018

N is not empty. Use the Principle of Mathematical Induction to select

a disjoint subset of N with “large” area, and add K to this subset.

Corollary Let M be a finite set of squares in the plane, the area of

which union is equal to A. Then there is a disjoint subset of M , the sum
A
of the area of whose squares is greater than 13 .

The statement follows from the fact that in the case of squares.
1 1
λ = λ1 = λ2 = · · · = λm = √ > .
1 + 4 2 + 2π 13

State a similar statement for a finite set of equilateral triangles in the

plane.

The question now is whether the numbers 19 (in Problem 3.2), λ (in
1
the Theorem), 13 (in the corollary) give the best possible results. This
means, whether it is true that if we increase any of these numbers (for
example, λ) by a “small” number δ > 0, there is a set of bounded figures
that cover the area A but any disjoint subset covers the area less than
λ + δ. The answer is negative. For example, in Shkljarski at al., 1974,
the following problem can be found.

Problem 3.3 A square K of area A is covered by a finite set M of

squares, whose sides are parallel to the sides of K. Prove that there is a
disjoint subset of M , the area sum of which squares is not less than A9 .

Certainly, Problem 3.3 gives a better result that can be obtained by the
Theorem.

Here is what is known about the best possible result.

Let Ω be a set of bounded figures in the plane. For example, Ω could

be the set of all squares with parallel sides, or the set of all squares,
or the set all circles, or the set of all regular polygons, etc. Let M =
{K1 , K2 , . . . , Km ; Ki ∈ Ω, i = 1, 2, . . . , m} be a finite set of figures on the
plane, each of which belongs to Ω. Denote by A the area of the union of
the figures of M , and D(M ) = {Ki1 , Ki2 , . . . , Kin } any disjoint subset

49
 Mathematics Competitions Vol 31 No 2 2018

 
of M , and A D(M ) = A(Ki1 ) + A(Ki2 ) + · · · + A(Kin ). The main
problem is to find the number
 
µ = inf sup A D(M ) .
M ∈Ω D(M )

The notation M ∈ Ω means that M is any finite set of figures of Ω.

Roughly speaking, to determine µ, consider all disjoint subsets of M ;

take the one that has the maximum area; then consider all finite sets of
figures of Ω and take the minimum across these maximum areas.

Certainly, the number µ depends on the set Ω. The determination of µ

is not an easy task. Here are some results.

Let Ω be the set of all squares with parallel sides. The Hungarian
mathematician T. Rado, 1928, conjectured that µ = A4 . The German
A
mathematician R. Rado, 1950, proved that µ > 8.75 . Ten years later,
A
the Russian mathematician Zalgaller, 1960, proved that µ > 8.6 . Until
1973 the conjecture of T. Rado seems to be true. But the Hungarian
mathematician Ajtai, 1973, constructed a set of squares with parallel
sides which disproves it. This made the problem much more attractive.

The next open problems focus on concrete sets Ω.

Open Problem 3.4 Find µ if Ω is the set of all squares with parallel
sides.

Open Problem 3.5 Find µ if Ω is the set of all circles.

Open Problem 3.6 Find µ if Ω is the set of all squares.

Open Problem 3.7 Find µ if Ω is the set of all equilateral triangles.

50
 Mathematics Competitions Vol 31 No 2 2018

Open Problem 3.8 Find µ if Ω is the set of all regular polygons.

Certainly, some other similar open problems arise, which the reader may
create themselves.

4 Problems on a Square Grid

This part is inspired by the “Chess 7×7” problem, presented by A. Soifer
on the 21st Colorado Mathematical Olympiad, April 16, 2004 (Soifer,
2016). Even though the problem is presented as a play between two
chess teams, it is equivalent to the following:

Problem 4.1
a) 22 squares are colored in a square grid 7 × 7. Prove that there are
4 colored squares that form a rectangle.
b) Prove that the statement in a) is not true if 21 squares are colored.

This problem gives the best possible results, i.e. it determines the
minimum number of squares that must be colored in a 7 × 7 square
grid such that there are 4 colored squares that form a rectangle.

Denote by C(n) the minimum number of squares that have to be colored

in a square grid n × n, such that there are 4 colored squares that form
a rectangle. Problem 4.1 shows that C(7) = 22.

Open Problem 4.2 Find C(n) for any n. (It is interesting to consider
values of n that are greater than 7.)

I am finishing this paper with one more problem for coloring squares on
a square grid that also gives rise to an interesting open problem.

Problem 4.3
a) 6 squares are colored in a square grid 4 × 4. Prove that there are
two rows and two columns that contain all colored squares.
b) Prove that the statement in a) is not true if 7 squares are colored.

51
 Mathematics Competitions Vol 31 No 2 2018

Solution.
a) Because of the Pigeonhole Principle, there is a row that contains
at least two colored squares. If this row contains exactly two
colored squares, there is one more row with at least two colored
squares; the remaining two colored squares may be chosen in two
columns. If this row contains more than two colored squares, we
choose one more row with a colored square; the remaining two
colored squares may be chosen in two columns.
b) An example is presented on Figure 4.

Figure 4: Example for part b)

Denote by D(n; k) 1 < k < n, the maximum number of squares that

have to be colored in a square grid n × n, such that there are k rows
and k columns that contain all colored squares. Problem 4.3 shows that
D(4; 2) = 6.

Open Problem 4.4 Find D(n; k) for any n > 4 and 1 < k < n.

This is a version of a plenary talk for the section “Creating Problems

and Problem Solving” of the 8th Congress of WFNMC, Graz, July 2018.

References
[1] Ajtai, M. (1973). The Solution of a Problem of T. Rado. Bull. Acad.
Pol. Sci., Ser. Sci., Math., Astron. Et Phys., 21.

[2] Bankov, K. (1996). Selection of a Large Subset of Disjoint Figures.

Geombinatorics, vol. VI, issue 2.

52
 Mathematics Competitions Vol 31 No 2 2018

[3] Bankov, K. (2017). Arrangements and Transformations of Numbers

on a Circle: An Essay Inspired by Problems of Mathematics Compe-
titions. Chapter from a book A. Soifer (Ed.) Competitions for Young
Mathematicians. Perspectives from Five Continents (pp. 101–122).
Springer.
[4] Rado, R. (1950). Some Covering Theorems. Proc. London Math.
Soc., 51.

[5] Rado, T. (1928). Sur une probleme relative a une teoreme de Vitali,
Fundamenta Math., 11.

[6] Soifer, A. (2016). Beyond Lǎozı̌: The Goals and Means of Mathe-
matics Instruction. Mathematics Competitions, Vol 29, No 1, 2016.

[7] Schljarskij, Chencov, Yaglom. (1974). Geometric Estimates and

Problems from Combinatorial Geometry (pp. 203–204), Nauka,
Moscow (in Russian).

[8] Zalgaller, V. (1960). A Note of Rado’s Theorem. Matematicheskoe

Prosveschenie, 5 (in Russian).

Kiril Bankov
Faculty of Mathematics and Informatics, University of Sofia, and
Institute of Mathematics and Informatics, Bulgarian Academy of Sciences
bul. Janes Baucher 5
1164 Sofia
BULGARIA
[email protected]

53
 Mathematics Competitions Vol 31 No 2 2018

The Julia Robinson Mathematics Festival

A Complement and Alternative to
Competitions

Mark Saul
Mark Saul is the Executive Director of
the Julia Robinson Mathematics Fes-
tival. Saul grew up in the Bronx,
got his BA from Columbia Univer-
sity and his Ph.D. from New York
University. He then spent 35 years
in and around New York, teaching
mathematics in classrooms from grades
3 through 12. He initiated a stu-
dent exchange program between Rus-
sian and American students, as well as
an “Intel/Westinghouse” style compe-
tition for students of mathematics in
China. He has given talks and led
workshops in more than 20 countries,
including Russia, Bulgaria, China, Saudi
Arabia, South Africa, and India. He
served as President of the American
Regions Mathematics League, mathe-
matics field editor of Quantum (the
English-language version of the Rus-
sian journal Kvant).

1 An Autobiographical Introduction
Like many students with an interest and ability in mathematics, I rarely
learned anything new in my standard math classes, until college. I always
read ahead in the book, always had fun with the new concepts I learned,
always got high grades on my tests—and rarely did much in class but
anticipate the teacher in my mind while politely waiting for a ‘stretch’
question, or looking for a new way to derive the result being discussed.

54
 Mathematics Competitions Vol 31 No 2 2018

This changed when I got onto the math team. There we did hard
problems. There I learned the joy of struggling with mathematics, and
made friends doing it. The only interesting mathematics I encountered
before college was in contests and contest preparation.

This was half a century ago, and things have certainly changed. Inter-
ested students now find a landscape of advanced coursework, after school
activities, summer programs, and web resources. These resources pro-
vide access to exciting and challenging mathematics that students can
work on at their own pace. And students of more modest ability, or lower
motivation, also find activities that stretch their idea of mathematics and
the role it can play in their lives.

The point of my autobiographical sketch is that competition is a vehi-

cle. We use it to stimulate challenging mathematical thought and for
motivating the study of advanced mathematics.

My most recent work is on another vehicle that takes us to the same

place: the Julia Robinson Mathematics Festival. We have developed,
test driven, and brought it to scale. In many ways, it builds on and
extends our experience with competitions, and offers mathematics to
students for whom competition may be less attractive than we might
wish.

2 What is a Julia Robinson Mathematics Festival?

A Julia Robinson Mathematics Festival (JRMF) is a non-competitive
mathematical event that occurs outside of the formal classroom. A
local host secures a space with a capacity of about ten tables, with
room to move around between the tables. Each table is dedicated to
a game, puzzle, problem set, or other mathematical activity, and each
table has a ‘facilitator’: a mathematician, professional, or older student,
who manages the activity.

The local host invites about 100 students (roughly 10 per table) into
the space for 2–3 hours. Students visit the tables, find an activity that
attracts them, and work on it for as long as they choose. They can
take the problems home for further study. The facilitator at the table
provides hints and materials, manages the social environment, and finds

55
 Mathematics Competitions Vol 31 No 2 2018

ways to prevent the students from getting frustrated with the sometimes
difficult problems. The facilitator does not teach or explain. Rather,
the learning is guided by the students themselves and their interactions
with the activity.

There is usually no schedule to be followed, nor is there any evaluation

of the students’ achievement (except to correct errors that students may
not see themselves). Students go as far as they want, using as much
time as they want. We, however, evaluate the quality of our activities
by collecting data on how long students work on each activity, rather
than how far they get in solving a problem. Students typically form
impromptu groups and make new friends as they work on the activities.
Their motivation comes from their natural sense of intrigue, and from
the social situation.

While the local host provides a venue, tables, and facilitators, the na-
tional organization, a not-for-profit program within the American Insti-
tute of Mathematics, lends support in various ways. First and foremost,
the national organization maintains a data bank of more than 100 ac-
tivities from which local hosts can select. We are also developing ‘facili-
tators’ guides’ for many of these activities. We are constantly adding to
and improving our activities, largely based on feedback from the field.
The chief criterion for selecting activities is their holding power: we col-
lect information on how long students are engaged in each activity at a
given age.

The national JRMF organization can also provide facilitator training.

Sometimes we arrange for a visit by an experienced Festival host, who
will work for an hour or so before the Festival to prepare facilitators
for the experience. Often this is done by pairing facilitators. First one
facilitator acts as a student, exploring the activity of the other facilitator
(which he or she has not yet seen). Then the two switch roles. This
process typically takes an hour or so. Since the facilitators will be doing
in the training exactly what they do with the students in the Festival,
they can segue directly into the Festival following the training.

The national organization also offers additional materials (at cost) to lo-
cal hosts—tee shirts, table cloths, banners, posters, and press releases—
that can help with local organization. Often, we can put local hosts in

56
 Mathematics Competitions Vol 31 No 2 2018

touch with local volunteer facilitators, groups of student participants, or

even local funders.

3 The Audience
A JRMF can reach a wide variety of students, including groups not
often served by competitions. Among the almost 200 Festivals we have
held so far, many have been for students as advanced as high school and
others for children as young as age ten. By including their parents in the
activity, we have held Festivals serving even younger students. Festivals
have been held in which most of the students have had experience in
high-level national competitions, and also in schools and communities
whose students have not had access to competitions.

Notably, our Festivals attract and retain girls in equal numbers to boys,
continuing all the way through high school age. This is in sharp con-
trast to the experience of the competition community. (Some of the
best data on this phenomenon can be found in Glenn Ellison and Ashley
Swanson, “The Gender Gap in Secondary School Mathematics at High
Achievement Levels: Evidence from the American Mathematics Compe-
titions”, Journal of Economic Perspectives Volume 24, Number 2, Spring
2010, Pages 109–128 (see https://economics.mit.edu/files/7598
(accessed June 2018); see also https://theatlantic.com/education/
archive/2016/04/girls-math-international-competiton/478533/
(accessed June 2018)). In every Festival we have observed, girls are close
to 50 % of the population, working side by side with boys and in the
same roles. The exception is Festivals which are organized solely for
girls, which have been popular and successful.

Of particular interest is the success of Festivals in working class and

high-poverty areas. Schools in these areas tend not to participate in
competitions. Why would they want to be compared to schools whose
students have many more resources, and have had a tradition of training
and success in competition? But in a Julia Robinson Mathematics
Festival, no schools are being compared. So a Festival is a useful way to
get high-level mathematics into these schools and communities.

57
 Mathematics Competitions Vol 31 No 2 2018

4 The Content
Without the long history of mathematical competitions, it would be
difficult to find content suitable for a JRMF. Indeed, the founder and
originator of the idea, Nancy Blachman, began by working mathematics
problems with her father, problems that were meant to select students
for an on-site contest at St. Mary’s College in Moraga, California. She
enjoyed solving these qualifying problems with her father much more
than solving contest problems by herself.
So Blachman ‘detached’ the mathematics from the competition, using
social engagement rather than competition as motivation. She found
ways to provide opportunities for many students to work on problems
together with a facilitator, someone who asked more questions than
they answer—just as her father had done. The first Julia Robinson
Mathematics Festival was hosted by Google in their cafeteria on April
22, 2007.
The development of activities for a JRMF involves many sources, and not
just competitions. Researchers have contributed ‘slices’ of their work.
We have reshaped old puzzles and bits of mathematical folklore to fit
the genre. Some examples appear below.

Example 1 (Square puzzles) These are courtesy of Gordon Hamilton

of Math Pickle (http://mathpickle.com/). These puzzles may seem
simple, but for elementary school students, they require a certain amount
of insight. For middle school students, an algebraic version is a useful
extension. Older students can be invited to construct such puzzles, a task
that can be challenging. The example shown below involves arithmetic,
but there are also puzzles involving algebraic expressions for the sides of
the squares.
Note that the large rectangle containing the squares may not itself be a
square.

Example 2 Digit sums and graphs. This set of puzzles was written
by Josh Zucker, based on an idea of Erich Friedman (https://www2.
stetson.edu/~efriedma/mathmagic/1208.html accessed June 2018).

58
 Mathematics Competitions Vol 31 No 2 2018

Squaring Puzzles
Squaring Puzzles
by Gord Hamilton, Math Pickle

These abstract squaring puzzles give students addition and subtraction practice with numbers
usually below 100. They also link these numerical activities to geometry. What a beautiful way
​ Gord Hamilton, Founder of Math Pickle.
to practice subtraction! —

The number in each square represents the length of a side of that square. Determine the length
of a side of all the squares in this rectangle and the lengths of the sides of the rectangle.

Find more square and subtracting puzzles here:

mathpickle.com/project/squaring­the­square/​.

59

3
 Mathematics Competitions Vol 31 No 2 2018

Simply understanding what is required can be difficult even for high

school students. The construction of such a puzzle can require significant
analysis.

Example 3 Color Triangles. This example illustrates three types of

problems:
1. Part of the attraction is the manipulative used to state the problem.
Younger students are attracted by the bright colors, and stay to
work on the problem.
2. The problem has a low threshold and a high ceiling. Anyone can
understand it and explore the situation. But a full solution is quite
difficult.
3. The problem is not one which JRMF participants are expected to
solve—and certainly not in the time allotted to a Festival. Exploring
the problem is productive, whether or not a solution is achieved.
The activity: A row of five discs is laid out, each disc one of three colors
(say red, yellow, blue). A row of 4 discs is constructed below it, using
the following rules:
– If the two discs above the new disc are the same color, the new disc
is this color.
– If the two discs above the new disc are of different colors, the new
disc is the third color.
A new row of 3 discs is constructed below the four discs, then a row of
two discs, then a ‘row’ consisting of a single disc.
The problem is to determine the color of the last disc, given the colors
of the first five discs.
A general solution to the problem can be found at https://ijpam.
eu/contents/2013-85-1/6/6.pdf (accessed June 2018). Students typ-
ically go through several stages of deliberation in working with the prob-
lem. The first stage is simply understanding the construction of each new
row. Sometimes the students don’t see that the color of the final disc
is determined, and think they are being asked to guess it, as if it were
a game of chance. For some students, comprehending the deterministic
nature of the situation is itself a breakthrough.

60
Mathematics Competitions Vol 31 No 2 2018

Digit Sums & Graphs

 
 
61
7  
 
10 
 Mathematics Competitions Vol 31 No 2 2018

Students later are able to notice “local” patterns. For example, if a row
consists of discs of all one color, then all the discs beneath that row are
that color, and the last disc is also that color. The process of forming
and testing hypotheses is motivated by the complexity of the patterns
and by the social situation.

One pattern that students notice, which may not be particularly helpful,
is that we can ‘read’ the completed triangle from different directions:
if we rotate a completed triangle by 120 degrees (or simply consider
a different one of its sides as having generated the triangle), the new
triangle obeys the same rules as the original.

Another insight students have is that the initial row need not be of length
5. It can be of any length. Typically, this starts students on a path of
case-by-case solution, which is too complicated to succeed. However,
they learn a lot by experimenting.

A full solution rests on the observation that if we start a triangle with

four discs (rather than five), then the final disc can be predicted from just
the first and fourth disc. The middle two discs do not affect the outcome.
The same is true for a row of ten discs, or of 28, or of 3n + 1 discs.
Students often have trouble seeing this, and the pedagogical problem of
how to support them in the discovery is so far unsolved.

The Color Triangles is a clear example of a problem from which students

can learn without actually having achieved a solution.

For more on this problem, see: https://wordplay.blogs.nytimes.

com/2013/05/13/triangle-mysteries/?mtrref=www.google.com\&
gwh=E2DE0F4FD0E6999B0357F29804B6DDC5\&gwt=pay, or https://sta
tic01.nyt.com/images/blogs/wordplay/posts/Triangle_Mysterie
s_Behrends_Humble.pdf (Both accessed July 2018.) There has been
extensive online discussion of this problem, including a number of con-
nections to deeper mathematical topics.

62
 Mathematics Competitions Vol 31 No 2 2018

Example 5 Nim. This is of course an entire set of games, some very

difficult and some unsolved.
We usually start with “one row nim”: A row of N counters are set up
and players take turns. In one turn, they can remove {1, 2, 3, . . . , M }
counters, where M and N can vary. (A typical initial game is M = 5,
N = 12). The winner is the person who takes the last counter. (Or, the
loser is the person who cannot move.)
The general solution appears after much experimentation, in the form
of game playing. Festival attendees play against each other or against
the facilitator. The facilitator will not advise the students on strategy,
except to guide them on what to observe. And the facilitator does not
always arrange to win.
For most values of N , the first player has a winning strategy: a row of
counters which is a multiple of M + 1 is a losing position. From any
other position, a player can put her opponent in a losing position. So
unless the initial row is a multiple of M + 1, the first player wins.
A more detailed description of typical student experiences can be found
at https://cims.nyu.edu/cmt/assets/pdfs/AA\_Problems/
OneRowNim.pdf (accessed June 2018). (This reference discusses the
misère form, in which the player who takes the last counter loses, but
the analysis and description of the learning can be easily adapted to the
present problem.)
Another nim game often used is two-row nim, in which players can take
as many counters as they want from either of two rows. See https://
cims.nyu.edu/cmt/assets/pdfs/AA\_Problems/OnePieceChess.pdf,
accessed June 2018.
The best known version of nim involves three rows, and has a solution
using binary arithmetic. It is a difficult pedagogical problem to go from
the two-row version to the three-row version without simply teaching the
students the computations involved in the binary arithmetic solution.
This should not be done, and certainly not at a JRMF: the students will
not learn much more than how to win this particular game.
Another difficult, but interesting, variant of nim starts with N coun-
ters, but the players can take only 2 or 3 counters. A charming account

63
 Mathematics Competitions Vol 31 No 2 2018

of this game is given in А. Шень, Игриы и стратегии с точки зре-

ния математики, издательство МЦНМО, москва 2007, available at
http://math.ru/lib/files/pdf/shen/Shen-games.pdf (accessed June
2018.)

Example 6 Toys: We often use commercially available activities. One

of our favorites is Shapeometry, developed and marketed by ThinkFun
(https://www.thinkfun.com/products/shapeometry/, accessed June
2018).

Shapeometry uses a manipulative consisting of plastic tetrominos and

pentaminos. The player is give two sets of pairs of pieces, and is asked
to make the same shape with each pair. The target shape is not given.

For example, each of these pairs of blocks

can form the same shape:

64
 Mathematics Competitions Vol 31 No 2 2018

But it is not easy to see what that shape might be or how to arrange
the pieces.

There are about 60 problems, arranged by level of difficulty. Shapeom-

etry offers an intuitive introduction to geometric transformations and
an intuitive ideas of area. Because the study of geometry in schools has
recently been suffering—worldwide—we are particularly eager to include
geometric activities.

5 A comparison of JRMF and Competition Materi-

als
Here is an extended example of a problem situation that can be adapted
to various levels of competition as well as to a festival. We give some
examples of how the same mathematical content can appear in these
different educational situations.

Broken Calculators I’ve got a collection of calculators someone left

me. Unfortunately, all of them are broken, each in a different way. Can
I use them to do calculations?

Every calculator displays 0 at the beginning.

Calculator 1

There are only two buttons that do anything on this calculator. Button
A adds 3 and button B adds 7.

65
 Mathematics Competitions Vol 31 No 2 2018

Problem 1 (format for an elementary short answer contest): What

will I get if I press ABAB?

(That is, first I press button A, then button B, then button A again,
then button B again.)

Problem 1 (format for a JRMF): Show that I can make the calculator
show 20 by hitting ABAB. What do I get when I hit ABBA?

Discussion: Notice that the student can draw his or her own conclusions
in each case, but that the JRMF version invites the student to generalize.
The insight here is that the order in which we press the buttons will not
matter.

Problem 2 (format for a middle school contest): What is the largest

whole number that I can NOT make by pressing keys on this calculator?

Problem 2 (format for a JRMF or an elementary school Olympiad):

It’s pretty easy to see that I can’t get the calculator to show 8. What
other numbers are impossible? What is the largest positive integer you
can’t display on calculator 1?

Solution: First two general notes:

a) The order in which we press the buttons doesn’t make a difference,
since we are just adding (and addition is commutative).
b) Every time we press another button we increase the display on
this calculator. We can never make the display smaller.

From (b), it’s pretty clear that we cannot get 1 or 2.

We can get 3 simply by pressing button A.

Suppose we could get 4. We can never have pressed B, or we would

overshoot our goal right away. If we had 4, then the button press
before that must have shown a 1, and we have already seen that this
is impossible. Thus we cannot get 4.

66
 Mathematics Competitions Vol 31 No 2 2018

Similar reasoning shows that we cannot get 5.

We can get 6 by pressing AA.

We can get 7 by pressing B.

We can show that we cannot get 8 by direct computation: we cannot

have more than 2 presses of A, and not more than 1 press of B. No
combination works.

Or, we could reason as before: The order of our buttons doesn’t matter
for this calculator. If we got 8 as a sequence of button presses, we can
arrange their order to have all the A’s first, then all the B’s. The last
press cannot be an A, since then there would be no B’s, and we would
get a multiple of 8. And if the last press is a B, then the calculator must
have shown 1 before the B button was pressed, and we have already seen
that this is not possible.

9 = AAA, 10 = AB.

We can show that because we cannot get 8, we also cannot get 11.
Suppose 11 were showing. The last button pressed was either A or B.

If the last press was A, then before this the number shown was 11−3 = 8,
and we already know this is impossible. If the last button pressed was
B, then before this the number shown was 11 − 7 = 4, and we already
know this is impossible. So we cannot get 11.

12 = AAAA, 13 = BAA, 14 = BB.

And now we note that if we can get a number N , then we can get N + 3.
So from 12 we can get 15, 18, 21,. . . ; from 13 we can get 16, 19, 22,. . . ;
and from 14 we can get 17, 20, 23,. . . Every number greater than 11
belongs to one of these sequences.

So the largest impossible number is 11. (Note that we used no algebra

in writing this proof.)

Calculator 2

On calculator 2, button A adds 1, and button B multiplies by 3.

67
 Mathematics Competitions Vol 31 No 2 2018

Problem 3 (format for a middle school short answer contest): When

I press ABBAAAA, I will get 13. But there is a shorter sequence of
button presses that will get me 13. Can you find it?

Solution: We work backwards, and try to get 1 from 13. We cannot have
pressed B to get 13 because 13 is not divisible by 3. So the last button
pressed must have been A, and the last number shown on the display
must have been 12.

How can we get 12? Well, if we had pressed B to get 12, then the display
must have held 4. And it is not hard to see that we can get 4 by pressing
ABA. The complete sequence is ABABA.

(Extra credit. Or a discussion after the contest): Let us try to form all
sequences that give us 13. As in the solution above, the first and last
button pressed must be A.

Axx. . .xxA.

If the next to last button is B, we will get ABABA, by the reasoning

above. (Well, not quite. We never showed that there is any other such
sequence. But it’s not hard to add to the reasoning to show this.)

If the next to last button is A, then either we have all A’s (length 13,
the maximum possible), or we must keep subtracting 1 until we have
another multiple of 3. We then have Axxx. . .xAAAA. Before the four
A’s are pressed, we would have had 9. There are only a few ways to get
9.

How can we get 9? We continue to work backwards. If the last button

pressed was B, then we must have had 3, which forces us to have ABB.
This gives us the string in the problem. And if the last button pressed
was A, we need to get to a multiple of 3, and we have Ax. . .xAAA.
Repeating the reasoning above, we must have had 6 before the last 3 As.

How can we get 6? Well, we certainly can use AAAAAA. This again
gives us thirteen A’s to get 13.

But can we get 6 with B? Yes: AAB. This gives us AABAAAAAAA,

which exhausts the possibilities.

68
 Mathematics Competitions Vol 31 No 2 2018

In summary, 13 = AAA AAA AAA AAA A = AAB AAA AAA A =

ABB AAA A = ABABA.
A case by case analysis, which we will not give here, can show that four
presses cannot give us 13.
In a Festival, this question provides a flexible environment. Younger stu-
dents can explore case-by-case, as shown above. Older or more advanced
students can reason more succinctly.

Problem 4 (In an Olympiad environment): Determine with proof the

smallest number of button presses it takes to get to 102. What about
511?
In a JRMF environment, we can simply ask:
What is the smallest number of button presses it takes to get to 102?
What about 511?
Then the facilitator can see how formal the discussion should be.
Here is a “naı̈ve” solution, which you can expect from a good middle
school student:
The number 102 is 3 × 34. So if we can get 34, one more press of button
B will give us 102. In turn, 34 is 3×11+1. So if we can get 11, two more
presses gives us 34. We can get 11 by pressing ABBAA. The complete
string that gives 102 is then ABBAABAB.
For 511: 510 = 3 × 170, so if we can get 170, one more press of button
B will give us 510, then pressing A gives 511. So we need to get 170,
after which BA gives 511. 170 − 2 = 168 = 3 × 56. So we need to get
to 56, after which BAA gives 170. 56 − 2 = 54 = 3 × 3 × 6, so if we get
6, then BBAA will give us 56. To get six we press AAB. The complete
sequence is AAB(6)BBAA(56)BAA(170)BA(511).
The ’backwards’ form of the algorithm seems to be: go back to the lowest
multiple of 3, then divide.
But to show how these problems have a ‘high ceiling’, we give a much
more sophisticated solution, due to Behzad Mehrdad:

69
 Mathematics Competitions Vol 31 No 2 2018

The operations that this calculator performs can be described in base 3

as A: adding 1 to the numeral, and B: shift the digits over to the left one
place and filling the rightmost place with a 0. Note that the calculator
doesn’t actually show base 3 computations. It’s something we have in
our heads to explain the workings of the calculator.

For example, if we want the display to show 49, we convert the number
we want to base 3: 49 = 1×27+2×9+1×3+1 = 12113 , and this numeral
actually gives us instructions to generate 49 on our broken calculator:

We press A to get a 1. This will end up as the leftmost digit 1. Then we

press B to shift this digit over. We get 1×3 (without simplifying!). Then
we press A twice to add 2. We get 1×3+2 = 510 = 123 . Then we press B
to shift these digits over. We get 3(1×3+2) = 1×9+2×3 = 1510 = 1203 .
Then we press A to add 1. We get 1×9+2×3+1 = 1610 = 1213 . Then we
press B to shift digits over. We get 1 × 27 + 2 × 9 + 1 × 3 = 4810 = 12103 .
Finally, we press A to add one, and get our result.

The complete string is then ABAABABA.

This algorithm gives us some bonuses. First, it shows that we don’t need
to press A more than twice in a row (because there is no need for digits
larger than 2 in ternary notation). Perhaps more importantly, it shows
that any shorter string of A’s and B’s would generate a smaller number,
so this must be the shortest length string.

In just the same way, the rest of the following JRMF problem set can
be adapted to furnish contest problems of various formats.

The complete JRMF problem set for the Broken Calculator Problem
appears in an appendix to this article. The reader is invited to solve the
problems, and also to think about how they might be re-formatted for a
competition.

6 Conclusion
Mathematics has lent itself to competition since ancient times (Archi-
medes’ “Cattle Problem”). And competition has sometimes stimulated
significant results (the solution of the general cubic and quartic equa-

70
 Mathematics Competitions Vol 31 No 2 2018

tions by the mathematical competitors of the Italian renaissance, Stokes’

Theorem).

But mathematics is flexible, and some competitions do not attract some

very important groups of students. A Julia Robinson Mathematics Festi-
val offers deep mathematics to a broad variety of students. Other places
students can learn informal mathematics include recreational journals,
math circles, and web pages. We are fortunate to be living in a time
when all these vehicles for mathematical activity are evolving rapidly.

7 Acknowledgments
The following people and organizations have given me invaluable help
in writing this article: Nancy Blachman, Dan Finkel, Gordon Hamil-
ton, Scott Kim, Alice Peters, Bill Ritchie, Thinkfun, Inc., Josh Zucker.
Activities due to others are used with permission.

References
[1] Antonick, Gary: Triale Mysteries, https://wordplay.blogs.
nytimes.com/2013/05/13/triangle-mysteries/?mtrref=www.
google.com\&gwh=E2DE0F4FD0E6999B0357F29804B6DDC5\&gwt=
pay (accessed July 2018).
[2] Behrends, Ehrhard, and Steve Humble, Triangle Mysteries. In
Mathematical Gems and Curiosities, https://static01.nyt.
com/images/blogs/wordplay/posts/Triangle\_Mysteries\
_Behrends\_Humble.pdf (accessed July 2018).

[3] Ellison, Glenn and Ashley Swanson, The Gender Gap in Secondary
School Mathematics at High Achievement Levels: Evidence from
the American Mathematics Competitions, Journal of Economic Per-
spectives Volume 24, Number 2, Spring 2010, Pages 109–128 (see
https://economics.mit.edu/files/7598 (accessed June 2018).

[4] Friedman, Erich. Problem of the Month (December 2008), https:

//www2.stetson.edu/~efriedma/mathmagic/1208.html (accessed
June 2018).

71
 Mathematics Competitions Vol 31 No 2 2018

[5] Nishiyama, Yutaka. The Three-Color Triangle Problem, Interna-

tional Journal of Pure and Applied Mathematics Volume 85 No. 1
2013, 69-81 ISSN: 1311-8080 (printed version); ISSN: 1314-3395 (on-
line version) url: http://www.ijpam.eu doi: http://dx.doi.org/
10.12732/ijpam.v85i1.6 (accessed June 2018).
[6] Saul, Mark: One Row Nim, https://cims.nyu.edu/cmt/assets/
pdfs/AA_Problems/OneRowNim.pdf (accessed June 2018).

[7] Saul, Mark: One Piece Chess, https://cims.nyu.edu/cmt/

assets/pdfs/AA_Problems/OnePieceChess.pdf (accessed June
2018).

[8] Shen, Alexander (Шень), Игриы и стратегии с точки зре-

ния математики, издательство МЦНМО, москва 2007,
http://math.ru/lib/files/pdf/shen/Shen-games.pdf (accessed June
2018.)

[9] Whitney, A .K. Math for Girls, Math For Boys, The Atlantic, April
18. 2016, https://theatlantic.com/education/archive/2016/
04/girls-math-international-competiton/478533/ (accessed
June 2018)).

Mark Saul
[email protected]

72
 Mathematics Competitions Vol 31 No 2 2018

APPENDIX: The ‘Broken Calculator’ Activity

as used in many Julia Robinson Mathematics Festivals

(Original problem by Josh Zucker. This version by M. Saul and B. Mehrdad, December
2016)

Broken Calculators

I’ve got a collection of calculators someone left me. Unfortunately, all of them are broken, each
in a different way. Can I use them to do calculations? Every calculator displays 0 at the
beginning, except calculator (5).

Calculator 1

There are only two buttons that do anything on this calculator. Button A adds 3 and button B
adds 7.

Problem 1: Show that I can make the calculator show 20 by hitting ABAB. What do I get when I
hit ABBA?

Problem 2: It's pretty easy to see that I can't get the calculator to get 8. What other numbers are
impossible? What is the largest positive integer you can’t display on calculator 1?

Calculator 2

On calculator 2, button A adds 1, and button B multiplies by 3.

Problem 3. When I press ABBAAAA, I will get 13. But there is a shorter sequence of button
presses that will get me 13. Can you find it?

Problem 4: What is the smallest number of button presses it takes to get to 102? What about 511?

Calculator 3

On calculator 3, button A adds 6, button B divides by 2, and button C divides by 3.

Problem 5: Show how you can get Calculator 3 to display any positive integer.

73
 Mathematics Competitions Vol 31 No 2 2018

Problem 6: What is the smallest number of button presses it takes to display 99 on calculator 3?

Calculator 4

On calculator 4, button A adds 5, button B adds 7, and button C takes the square root.

Problem 7. Prove that it is possible to display any positive integer greater than 1.

Calculator 5

Calculator 5 has two memory slots. Wherever you are in your sequence of button presses, there
is a number displayed on the calculator. At that point, memory slot Alpha holds the last number
you've generated, and memory slot Beta holds the number you generated before that. Both
memory slots start with 0.

On this calculator, button A adds the contents of Alpha to the current number, and button B adds
the contents of Alpha plus the contents of Beta to the current number.

This calculator always displays 1 at the start, with 0 in each of the memory-slots.

Problem 8: What do you get if you hit button A once?

Problem 9: What do you get if you hit button A for a second time?

Problem 10: ….and again?

Problem 11: What happens if you keep hitting button A? Write out the first ten numbers in the
sequence of numbers displayed.

Problem 12: What sequence is generated if you keep hitting button B? Write out the first ten
terms.

Problem 13 (Difficult): Can this calculator display any positive integer?

74
 Mathematics Competitions Vol 31 No 2 2018

An Exploratory Problem

Ryan Fang, Caleb Ji & Evan Liang

We recently attended an Exploratory Problem Session, during which

the participants made individual contributions to the collective effort in
solving the problems. We present one of them here.

Problem For which positive integers k is there a configuration of 100

arcs on a circle such that each arc intersects exactly k others?

It was immediately observed that k = 1 is possible. We could just have

50 disjoint stacks of 2. This idea yielded the solutions for the first group
of values k = 3, 4, 9, 19, 24, 49 and 99. We simply took 25, 20, 10, 5, 4,
2 and 1 disjoint stacks of 4, 5, 10, 20, 25, 50 and 100, respectively.

We worked on k = 2 for a while until someone came up with the

construction of 100 arcs “holding hands” around the circle. This led
to the solutions for the second group of values k = 5, 11, 14, 29, 59 and
74. We simply took 50, 25, 20, 10, 5 and 4 stacks of 2, 4, 5, 10, 20 and
25, respectively, “holding hands” around the circle.

We had some difficulty with the next case k = 6, since 100 arcs could not
be divided into equal stacks of 3 or 7. So we re-examined our construction
for second group. Here we marked on the circle a number of evenly
spaced points, and each arc started and ended at these points. We
called this a canonical representation. The representations in the first
group were also canonical, since each arc started and ended at the same
point.

The question was raised whether any valid construction could be repre-
sented canonically. We claimed that the answer was affirmative. What
was needed was the trimming off of redundant parts of the arcs. This
we would achieve in two stages.

Suppose one arc was entirely contained in another without being iden-
tical to it. Since both arcs intersected exactly k other arcs, we could

75
 Mathematics Competitions Vol 31 No 2 2018

shrink the longer arc so that it coincided with the shorter one without
disturbing the intersection pattern.

We now assigned the clockwise direction to the arcs and marked all
the starting points, making them evenly spaced. Suppose there was
an arc whose end point was unmarked. We could retract this end point
counterclockwise until it reached the first marked point. As we had seen,
this might result in a degenerate arc of length 0. Again the intersection
pattern was not disturbed. Thus the claim was justified.

We returned to tackle the case k = 6. We marked the 100 starting points

P1 , P2 , . . . , P100 . We considered the arc which started at Pm for some
integer m. If it were to intersect exactly 6 other arcs, they should be
the 3 which started at Pm−3 , Pm−2 and Pm−1 along with the 3 which
started at Pm+1 , Pm+2 and pm+3 . This would be the case if the arc
ran from Pm to Pm+3 , so that its length is 3. This new construction
allowed us to solve all the cases where k = 2t is even. We just made t
the common arc length.

Someone made the observation that in all of the solved cases, the length
of the arcs was constant. Would this always be true? After some
discussion, we came up with a simple proof.

We would divide the circle into n equal parts by the marked points
P1 , P2 , . . . , Pn . An arc starting from P2 must be at least as long as an
arc starting from P1 , since it was not contained in the latter. Similarly,
an arc starting from P3 must be at least as long as an arc starting from
P2 , and so on. Going around the circle, an arc starting from P1 must be
at least as long as an arc starting from Pn . It follows that all arcs have
a common length.

We were now ready for the final assault. We let am be the number of
arcs starting from Pm , 1 ≤ m ≤ n. Then the number of arcs intersecting
an arc which started from Pm would be

am−t + · · · + am−1 + (am − 1) + am+1 + · · · + am+t = k,

t being the common arc length. Summing from m = 1 to n, we obtained

100(2t + 1) = n(k + 1) since a1 + a2 + · · · + an = 100. We now observed
that 8 could not divide 100(2t + 1). It followed that we could not have

76
 Mathematics Competitions Vol 31 No 2 2018

k ≡ 7 (mod 8). All other values of k under 100 were possible. The
following chart showed values of t and n for possible odd values of k.

k 1 3 5 9 11 13 17 19 21 25 27 29 33
t 0 0 1 0 1 3 4 0 5 6 3 1 8
n 50 25 50 10 25 50 50 5 50 50 25 10 50
k 35 37 41 43 45 49 51 53 57 59 61 65 67
t 4 9 10 5 11 0 6 13 14 1 15 16 8
n 25 50 50 25 50 2 25 50 50 5 50 50 25
k 69 73 75 77 81 83 85 89 91 93 97 99
t 3 18 9 19 20 10 21 4 11 23 24 0
n 10 50 25 50 50 25 50 10 25 50 50 1

It was pointed out that 100 could be replaced by any positive integer.
However, it was not expected that such a generalization would require
any new ideas.

Caleb Ji Ryan Fang Evan Liang

Grade 9 student Grade 8 student Grade 7 student
Crestwood Junior Vernon Barford Junior McKernan Junior
High School High School High School
Edmonton Edmonton Edmonton
CANADA CANADA CANADA

77
 Mathematics Competitions Vol 31 No 2 2018

International Mathematics
Tournament of Towns
Selected Problems from the Fall 2017

Andy Liu

1. Among 100 coins in a row are 26 fake ones which form a consecutive
block. The other 74 coins are real, and they have the same weight.
All fake coins are lighter than real ones, but their weights are not
necessarily equal. Find at least one fake coin using a standard
two-pan balance only once.
Solution. Weigh #1 to #25 against #76 to #100. If there is
equilibrium, all these 50 coins are real. It follows that #50 and
#51 are both fake. If there is no equilibrium, we may assume by
symmetry that #76 to #100 are heavier. Then these 25 coins are
real, whereas at least one of #1 to #25 is fake. It follows that #25
and #26 are both fake.

2. Let M be the midpoint of the side AC of triangle ABC. Let

L be the point on AB such that CL bisects ∠BCA. The line
through M perpendicular to AC intersects CL at K. Prove that
the circumcircles of triangles ABC and AKL are tangent.
Solution. Construct the tangent the circumcircle of ABC at A,
as shown in the diagram below. Then ∠T AB = ∠ACB. Since
K lies on the perpendicular bisector of AC, ∠KCA = ∠KAC.
Since CL is the bisector of ∠BCA, ∠KCA = ∠KCB. Hence
∠KAC = ∠KCB. Now

∠AKL = ∠KAC + ∠KCA = ∠KCB + ∠KCA = ∠ACB = ∠T AL.

It follows that T A is also tangent to the circumcircle of AKL at

A, so that the two circumcircles are tangent to each other.

78
 Mathematics Competitions Vol 31 No 2 2018

.........
.........
.........
T .........
.........
.......................................................................
...................
. .......
..........
A
........................................
............
...
... ....... ....... ...............................
....... ..... .......
........
. ....... .......
... ... .... ......
... .... .... .......
..
.
.
....
.
.
.
....
....
. ........ ... ... .... .......
...
.... .
.. .
......... ... .... ... .......
.... . .......... ..... .....
.
...
.
.
.
...
..
L ..
.
.
.........
.................
...
...
...
.....
....
.....
.....
. ...
.. ... ...
... . .
..... .... ............. ... .. ..... ...
...
......
.......
. ...
........... ...
...
.......
.. ........
..
.
...
... .
.
.
.
.
. ....M
. ..........
...
...
...
. .
..... .
... ......... .
... .. ..... ... . ...
... ...
... ... . . . ........ . .
. ..
. . ...
... ......
..... . . .
. .
... ..........
. .
...
. ...
.... ..... ....... ......... ...
..
. ....
.. ..
..... . .. .
..
............ . ...
.. ....
. . . ...
..
.
... .
......
.. ...........
. ........ . . . . .........
.
.... .....
. ....... ..... .................. . . ... .......... .... ...
....
.. ......
.......
.
..... K . ..
. ....... . . .
....... ..... .....
... .......... ....... ... ..
.... . .
B ................................................................................................................................................................................................................................................................................. C
3. We have a faulty two-pan balance with which equilibrium may only
be obtained if the ratio of the total weights in the left pan and in
the right pan is 3:4. We have a token of weight 6 kg, a sufficient
supply of sugar and bags of negligible weight to hold them. In each
weighing, you may put the token or any bags of sugar of known
weight on the balance, and add a bag of sugar so that equilibrium
is obtained. Is it possible to obtain a bag of sugar of weight 1 kg?
Solution by Ryan Morrill. Put the token in the left pan and a
bag of sugar in the right pan to obtain equilibrium. The weight
of the bag is 8 kg. Replace the token by a bag of sugar to obtain
equilibrium. The weight of the bag is 6 kg. Put the token and the
6-kg bag in the right pan. Put the 8-kg bag in the left pan and add
a bag of sugar to the left pan to obtain equilibrium. The weight of
this bag is 1 kg.
4. One hundred doors and one hundred keys are numbered 1 to 100
respectively. Each door is opened by a unique key whose number
differs from the number of the door by at most one. Is it possible
to match the keys with the doors in n attempts, where
(a) n = 99;
(b) n = 75;
(c) n = 74?
Solution by Ryan Morrill.
(a) For 1 ≤ k ≤ 99, try door k with key k. If all attempts are
successful, we know everything. Suppose the kth attempt fails
for some k. Then key k must open door k + 1 while key k + 1
must open door k. This actually saves us one question.

79
 Mathematics Competitions Vol 31 No 2 2018

(b) We claim that 3 attempts can settle 4 doors with 4 keys. Try
door 3 with key 3. If the attempt is successful, one more
attempt will settle doors and keys 1 and 2. Even if the doors
and keys go beyond 4, one more attempt will settle door 4
and key 4. Suppose the attempt fails. Try door 2 with key 3.
If the attempt is successful, then key 1 must open door 1 and
key 2 must open door 3. One more attempt will settle door 4
and key 4. Suppose the attempt fails. Then key 3 must open
door 4 and key 4 must open door 3. One more attempt will
settle doors and keys 1 and 2. Iterating this process, we can
accomplish the task in 75 attempts.
(c) Suppose that key k opens door k for 1 ek ≤ 100 but we do
not know that. We claim that we need at least 75 attempts.
For doors and keys 1 to 4, there are at least five possible sce-
narios.
(1) Key k opens door k for 1 ≤ k ≤ 4.
(2) There is a switch between 1 and 2.
(3) There is a switch between 2 and 3.
(4) There is a switch between 3 and 4.
(5) There is a double switch between 1 and 2 as well as be-
tween 3 and 4.
Since two attempts can distinguish among at most four sce-
narios, we need at least three attempts to settle doors and
keys 1 to 4. The same argument may be applied to each
successive block of four doors and keys, justifying our claim.
5. The digits of two integers greater than 1 are in reverse order of
each other. Is it possible that every digit of their product is 1?
Solution by Central Jury. Suppose such a pair of positive integers
exist. Let the digits of one of them be a1 , a2 , . . . , ak−1 , ak in that
order. Since the product of the two numbers ends in a 1, we must
have (a1 , ak ) = (1, 1), (9,9), (3,7) or (7,3). Since the product also
starts with a 1, only the first case is possible. Now, the product
is less than (2 × 10k − 1)(2 × 10k − 1) = 4 × 102k − 2, so that it
is a 2k − 1 digits. In the second column from either side when
the multiplication is performed, we must have ak a2 + ak−1 a1 +
a2 + ak−1 = 1 since any carrying over will make the leftmost digit
of the product greater than 1. Using an analogous argument on

80
 Mathematics Competitions Vol 31 No 2 2018

successive pairs of columns towards the middle, we arrive at the

conclusion that middle digit must be a21 + a22 + · · · + a2k ≥ 2 = 1.
This is a contradiction.
6. The incircle of triangle ABC is tangent to BC, CA and AB
at K, M and N respectively. The extensions of M N and M K
intersect the exterior bisector of ∠ABC at R and S respectively.
Prove that RK and S intersect on the incircle of ABC.
...........................................
............... ..........
.......... ........
........ .......
... ........
. .......
......... ......
......
. ... .... .....
.. ... .....
... .
. .....
.... .....
... ...
.. .. . . ...
... A .......
... ...
...... .....
R .
..
.
.
.....
.
....................................................
...
...
. .
.......
...... ...
...
.
. . ... . . ........................ .... ..
.
... ... ............. ....................... ..... ...
.. .... ....................... ... ...... ........................ ...
....
. ...
...
.......
.......
.......
.. . . .. .
......................
..... .. ... .. . ..
N
.......................
..
... ......
. ..
...
... .................. ... . .. .. ........... .....
....... ...
...... ..
... .... . .. . .. .. . . .
. ... . .... .........
... ....... .......................................
.. ... ....... ............ ......

....
... ...
...
.......
.......
.. .
......... .. ...
............ .... ........ M
.......................
..............................
..........
... . . ........ . .
... .... ... ....... .. ...
... ... .......... ... .. . . ... .....
... ....... ...... ... .. .. ... ... ...........
... ...
... ....... . .......... .... .. .... ....
.
. .
. .. ......
... ....... ..... . . ... ....
...
... ..... ...... ... ... ... ..... ... ... ....
... .......
...... ... ... ... ... ...
...
...
...
...
.......
....... .......... ...
. .. ...
. ... ....
.
... ...
.. ...
... .......... . . ... ... . ...
... .
. .
... .
. .. .. ...
... ... .
....... ........... . . ...
. ...
. .
... .. ...
...
...
... .. .
. . ... . .......
..
... .. ...
.
.
.
...
. ...
. .
... ...
... ..... ....... ... . .. ... .. . ...
... . .... ... ........ ... .. .. . .. . . ... .. .
. ...
... ...
.. .
. .... . .
. . . . . ..
. .
.. ...
... ... .... . ..................................... . ... .. . ...
...
...
.....
...
...
... . .
.......
.
. .
. ... .
.............
. . . . . .
..
P . .. . .... ..
.
. . .
......... ...... .. ................... ..
.
..............
.. .... .
....
...
ω .
....
...
...
.
.
.
.
.. ...
...
...
..... . ...
. ... ... .... . . . . .. .. ...
..... ...
.. . ...... . .
... ....
. .. .......... .
...... .. ....... .............. ...
. ... .
...... ...
..... .. .... .. .......... . ... .. ..... .......... ... .. .... ...
...... .... ..
. .
. . .. . ... . ...
.... .... .. ....... .. . .
......... .. .. . .. .... ...
...... ... ........... ........................... ..... .. . .
.. ...
........ .......... .. . . . . .. .
..... ... ...
....... .. .................... .... . ...... .
.................................................................................................................................................................................................................................................................................................................................................
. . .. . .
.. .
. . . ..... . ..... .
. .
. .........
. .. . .
. . . .
..... . .
. .
. .. ... ................ ...
. ....... . .
.. ..
.
. .. ...
.
. ................................................ . .
....
.. .
.
..
... C
B ..
..
. ...
...
... ..
..
K ...
.
...
. . ..
..
.. ... .. ...
..
..
. .. ... ... .... ..
.. ... ... ... ..
. .
... .. ..
.. .
. .
..
. ... . .. ..
.. .
... . .. ..
. . ..
.... ..
. .. ...
... ..
.... ..
... ... ... ..
. ..
.. ... .. .
... ..
.. ... .. .... ..
.. ... . . . ..
... . .. . ..
.. ... ... ..
.. ... .. .... ..
... ... .. . ... ..
..
... .
. ... ..
..
...
... ... . ... ..
.. ... .. .... ...
.. ... . .
. .. ..
.. ... .. .. .
.. ... .. ... ..
...
.. ... ..
.. ...
.. ... ..
.. ... . ...
. . .
.. ... .. ..
.. ... ... ..
.. ... .. ... ..
.. ... .. ... ..
.. ... ... .... ...
... ... .
.. .. ..
.. ...
.. ... ..
.. ...
.. ... ..
.. ... ... ..
.. ... ... .... .
.. ... . .. ..
.. ... .. .... ..
.. ... .. ... ..
.. ... ..
..
... ... ... ..... . ..
... . .. ..
...
... ... ..... ...
...
... . ... ...
...
... ..... ...
...
.... . . . . . . ...
.. ..... .
......
............. . .....
.. ...................
S

81
 Mathematics Competitions Vol 31 No 2 2018

Solution. Let RK cut the incirle ω at P . Let ∠N KM =

α, ∠KM N = β and ∠M N K = γ. Then α + β + γ = 180◦ .
Now ∠RP N = β since it is an exterior angle of KM N P . We
also have ∠BN K = β since BN is tangent to ω. Finally, since
N K and RS are both perpendicular to the bisector of ∠ABC,
∠RBN = β = ∠RP N , so that BP N R is cyclic. Note that
∠M SR = α and ∠M RS = γ. Hence ∠N P K = 180◦ − α and
∠N P B = 180◦ − γ. It follows that ∠BP K = 180◦ − α, so that
BP KS is also cyclic. We have ∠KP S = ∠KBS = β = ∠RP N .
Since P lies on RK, it also lies on N S.
7. Let a positive number be given. A piece of cheese is cut so that
a new part is created with each cut. Moreover, after every cut,
the ratio of the weight of any piece to the weight to any other one
must be greater than r.
(a) Prove that if r = 0.5, we can cut the cheese so that the process
will never stop.
(b) Prove that if r > 0.5, then at some point we will have to stop
cutting.
(c) What is the greatest number of parts we can obtain if r = 0.6?
Solution by Howard Halim.
(a) Let the initial piece be of weight 1. We can cut it into two
pieces of weight 12 . We then cut each into two pieces of weight
1
4 . Continuing this way, we will always have pieces of only two
different weights, one being half of the other.
(b) Let r > 0.5. Suppose to the contrary that cutting can con-
tinue forever. Consider any two pieces A and B with respec-
tive weights a and b where a ≥ b. Then A must be cut before
B, as otherwise one of the pieces resulting from cutting B
has weight at most 2b and hence less than ra. Suppose the
pieces resulting from cutting A have weights c and d with
c ≥ d and c + d = a. Then d ≤ a2 , so that r < db < 2b a
.
a
It follows that b > 2r > 1. Suppose at some stage we
have n pieces with respective weights a1 > a2 > · · · > an .
Then a1 > 2ra2 > (2r)2 a3 > · · · > (2r)n−1 an . Hence
1 n−1
( 2r ) > aan1 > r. This cannot hold forever as 2r
1
< 1.

82
 Mathematics Competitions Vol 31 No 2 2018

1
(c) From (b), n pieces may be obtained for r = 0.6 if ( 2×0.6 )n−1 >
0.6. This holds for n − 1 = 3 but not for n − 1 = 4. It follows
that at most 4 pieces may be obtained. We now show that
4 pieces may in fact be obtained. We may take the weight
of the initial piece of cheese to be 32. We first cut it into
two pieces with respective weights 18 and 14, observing that
14
18 > 0.6. Next, we cut the heavier piece into two, each with
9
weight 9, and observe that 14 > 0.6. Finally, we cut the piece
with weight 14 into two, each with weight 7.
8. The excircles of triangle ABC opposite A and B are tangent to BC
and CA at D and E, respectively. K is the point of intersection of
AD and BE. Prove that the circumcircle of triangle AKE passes
through the midpoint of JC, where J is the excentre of ABC
opposite C.
Solution by Central Jury. Let the excircle with centre J touch the
extension of CA at P , the extension of CB at Q, and the side AB
at F . Then CA + AF = CA + AP = CP = CQ = CB + BQ =
CB + BF . Hence CF bisects the perimeter of ABC. Thus AD
and BE also bisect the perimeter of ABC, so that AE = BD.
Similarly, we can prove that BQ = CE and DC = AP . Let M
be the midpoint of CJ. Then CM bisects ∠BCA. Since M is the
circumcentre of the cyclic quadrilateral CP JQ, M P = M C = M Q
so that ∠M P C = ∠M CP = ∠M CQ = ∠M QC. Perform a
rotation about M equal to ∠CM P = ∠CM Q. Then the line BC
lands on the line CA, with Q landing on C, B on E, D on A and C
on P . It follows that M A = M D, M B = M E, and both ∠AM D
and ∠BM E are equal to the angle of rotation. Hence AM D and
BM E are similar isosceles triangles. Thus ∠DAM = ∠BEM , so
that AEKM is a cyclic quadrilateral.

83
 Mathematics Competitions Vol 31 No 2 2018

..
...
...
......
. ........
.
......
... ...
.. ..
... ....
. ... ....
.
... ...
... ....
..
P . ........... . ....
... ..
...
..
. ..
... ................... ... . .... . . . . . . . . . .
.... ... ............. . . .. . . . . .... ..
... ... ... ...... . . ..
.. ... .... ...... .... . ... ..
... ... ... ...... .... .. ..
... ... .. ...... . ....... .
... ..
. ... .. . .
.
.... ... ... ........... . . ...
.
...
....
..
. ... ... ...
.....
......
....... . . ....
... ..
.... ...... ...... ..... . .

. ...
. ... A ....
......
. .
......
...... ...
.
.. .
.
...
..
...
.
...
.
.
.
....... . ..
....... .
... .
. ... . ...
. . .. .. .
...
... ..
. .. .. ................... ...
. .
. ... ... . .
. . .. . . . .... .. .... ..... .......... ...
.
.
.
. ... ... .. . ... .. ...... ...
............... . ... ... . .. ... ...... . .
J
... ..............
... ..........
. . ............ .
..
.
..
..
......
...... .
.
.. ...
... ..
. ..
......
......
...
...
....
.
.
.
.
... . . ... .. ... .... . ... ... ......
..
...
. .
.
... ............. . . . . . .... .
...... ...
... .......... .
... .... ... . ..
. . . ..
. . .
.......... ... . ... ... ...... ... .
... .......... ... ..... ... . .. ... ...... . .
... ............. .... ... ... ...... .... .
... .. ................... .... .. ... ...... ... .
... .. ..........
.......... .........
. . ...... ... .
..... ......... ...............
...
..
...... ....
.. . .
.
... ...........
..... ...... ..................... ........... .
F .... ..
............. ...................................................................... ........... ..
.. ... ................. ......... ..
... .... ................. ........ ............ .......
... ............. .................................... ...........
.... ... ..... M . . . ............................. .....
. ... . .. . . . . .......... . . . .. ..... .
..
......... . ..
.. .... .......... ....... .....
..... .....
. ..... E .........................
.... ....... .......... .............. K ...... . ... ........................................................ . . . . ............
..................
............ ....... .. ..............
.............................. ..... . . . . . . . . . . . . ....................
...... ............
.... ............. .. ....... .......................
. ..
.......... ...........................
... . . . . ........................... ..... .............. .... ..... ........ . .
.. . ..... ...
.. . .. ........ .... ....... .......... ..... ..........
... .......... ........ ....................... . .. . .
.................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................
. . . . . . . . . . .....
Q B ....
...
..
..... .... ..............
.
...................
D
............... . .........................
................
C .............
...
..
......
......
......
.... ...........
... ........... .......... ...
......... ..........
.. .........
... ......... ......... .....
.... . . .. . . ........ ........ ......
........ ......
. ... . . . . .. ...... .............
. ... ............
... ....... .............
. ... ..... ...... ...........
..... .......
. ... .. . .. ......
. ... .. . ...... ......
......
.
. ... ........... .
. ... ........
.... ........
...........
.
. .
... .....
........
.......
.......
..........
.
.
.....
...
...
...
..

Andy Liu
University of Alberta
CANADA
email: [email protected]

84
Australian Maths Trust
170 Haydon Drive, Bruce ACT 2617
Australia
Tel: +61 2 6201 5136