Crux Mathematicorum is a problem-solving journal at the secondary and university undergraduate levels,

published online by the Canadian Mathematical Society. Its aim is primarily educational; it is not a research
journal. Online submission:
https://publications.cms.math.ca/cruxbox/

Crux Mathematicorum est une publication de résolution de problèmes de niveau secondaire et de premier
cycle universitaire publiée par la Société mathématique du Canada. Principalement de nature éducative,
le Crux n’est pas une revue scientifique. Soumission en ligne:
https://publications.cms.math.ca/cruxbox/

The Canadian Mathematical Society grants permission to individual readers of this publication to copy articles for
their own personal use.
c CANADIAN MATHEMATICAL SOCIETY 2020. ALL RIGHTS RESERVED.

ISSN 1496-4309 (Online)
La Société mathématique du Canada permet aux lecteurs de reproduire des articles de la présente publication à des
fins personnelles uniquement.

c SOCIÉTÉ MATHÉMATIQUE DU CANADA 2020 TOUS DROITS RÉSERVÉS.

ISSN 1496-4309 (électronique)

Supported by / Soutenu par :

• Intact Financial Corporation
• University of the Fraser Valley

Editorial Board

Editor-in-Chief Kseniya Garaschuk University of the Fraser Valley

MathemAttic Editors John McLoughlin University of New Brunswick

Shawn Godin Cairine Wilson Secondary School
Kelly Paton Quest University Canada

Olympiad Corner Editors Alessandro Ventullo University of Milan

Anamaria Savu University of Alberta

Articles Editor Robert Dawson Saint Mary’s University

Associate Editors Edward Barbeau University of Toronto

Chris Fisher University of Regina
Edward Wang Wilfrid Laurier University
Dennis D. A. Epple Berlin, Germany
Magdalena Georgescu BGU, Be’er Sheva, Israel
Chip Curtis Missouri Southern State University

Guest Editors Vasile Radu Birchmount Park Collegiate Institute

Aaron Slobodin University of Victoria
Andrew McEachern York University

Editor-at-Large Bill Sands University of Calgary

Managing Editor Denise Charron Canadian Mathematical Society
 IN THIS ISSUE / DANS CE NUMÉRO

143 Editorial Kseniya Garaschuk

144 In Memoriam Robert Dawson
145 MathemAttic: No. 14
145 Problems: MA66–MA70
149 Solutions: MA41–MA45
153 Problem Solving Vignettes: No. 11 Shawn Godin
159 Olympiad Corner: No. 382
159 Problems: OC476–OC480
161 Solutions: OC451–OC455
169 Michel Bataille’s Focus On. . . Index
175 Problems: 4531–4540
181 Bonus Problems: B1–B25
185 Solutions: 4481–4490

Crux Mathematicorum
Founding Editors / Rédacteurs-fondateurs: Léopold Sauvé & Frederick G.B. Maskell
Former Editors / Anciens Rédacteurs: G.W. Sands, R.E. Woodrow, Bruce L.R. Shawyer,
Shawn Godin

Crux Mathematicorum
with Mathematical Mayhem
Former Editors / Anciens Rédacteurs: Bruce L.R. Shawyer, James E. Totten, Václav Linek,
Shawn Godin
 Editorial /143

EDITORIAL
Math in the Time of Coronavirus. While we find ourselves adjusting many of our
habits, changing our everyday activities due to various necessary restrictions, for
me, engaging in mathematics offered a kind of repose and feeling of normalcy.
I now teach via videoconferencing with a toddler in the background and hold
research meetings over the phone while walking my dog. However, my Crux work
has changed very little, and I am indeed grateful for that. So if you too are looking
for a mathematical distraction from the pandemic, look no further.
This issue has a couple of non-standard features. First, Chris Fisher has put to-
gether a comprehensive index for Michel Bataille’s Focus On . . . column. There,
you will find columns arranged according to topics (Algebra, Geometry, Inequal-
ities, Calculus, Combinatorics, Trigonometry) with citations and short content
descriptions. Secondly, in this issue we have 25 Bonus Problems. While we have
high standards for problem acceptance, we simply receive too many good prob-
lems. As we try to balance each issue’s problem offerings between authors and
topics, we inevitably acquire a backlog. To ensure that no problem stays in the
waiting-to-be-published stage for too long, we will be occasionally publishing a
Bonus Problems list. This gives authors a chance to cite their problems, while
providing our readers with more materials. Please note that this material is truly
bonus: we will not be considering solutions to these problems.
Stay healthy and safe.
Kseniya Garaschuk

Copyright © Canadian Mathematical Society, 2020

144/ In Memoriam

IN MEMORIAM
Fans of recreational mathematics will mourn the recent death of John H. Con-
way. While many of his important discoveries in geometry and group theory can’t
be explained without many hours of lead-in, (even the grand antiprism is fairly
mindboggling) and even the many of our readers will have experimented with Con-
way’s “Game of Life.” Maybe you have read “On Numbers and Games”, or the
more popular “Winning Ways” which he coauthored with Elwyn Berlekamp and
Richard Guy (by sad coincidence, all three authors of this tour de force have died
within a little over a year of each other.) Maybe at some point you learned his
“Doomsday Rule” for finding the day of the week of any day in history, or how to
win at Nim or Hackenbush. However it happened, whatever it was: so many of us
are the richer for John’s time among us, and if you aren’t yet - it’s not too late!
Randall Munroe’s XKCD webcomic gave John the rare tribute of a memorial
cartoon. Here are some snapshots by Mr. Munroe’s generous permission. For the
full animation, see https://imgs.xkcd.com/comics/rip_john_conway.gif
Robert Dawson

Crux Mathematicorum, Vol. 46(4), April 2020

 MathemAttic /145

MATHEMATTIC
No. 14
The problems featured in this section are intended for students at the secondary school
level.

Click here to submit solutions, comments and generalizations to any

problem in this section.

To facilitate their consideration, solutions should be received by June 15, 2020.

MA66. The 16 small squares shown in the diagram each have a side length of
1 unit. How many pairs of vertices (intersections of lines) are there in the diagram
whose distance apart is an integer number of units?

MA67. Consider numbers of the form 10n + 1, where n is a positive inte-

ger. We shall call such a number grime if it cannot be expressed as the product
of two smaller numbers, possibly equal, both of which are of the form 10k + 1,
where k is a positive integer. How many grime numbers are there in the sequence
11, 21, 31, 41, . . . , 981, 991?

MA68. P QRS is a square. The points T and U are the midpoints of QR and
RS respectively. The line QS cuts P T and P U at W and V respectively. What
fraction of the area of the square P QRS is the area of the pentagon RT W V U ?

Copyright © Canadian Mathematical Society, 2020

146/ MathemAttic

MA69. The diagram shows two straight lines P R and QS crossing at O.

What is the value of x?

MA70. Challengeborough’s underground train network consists of six lines,

p, q, r, s, t, u, as shown. Wherever two lines meet, there is a station which enables
passengers to change lines. On each line, each train stops at every station. Jessica
wants to travel from station X to station Y . She does not want to use any line
more than once, nor return to station X after leaving it, nor leave station Y having
reached it. How many different routes, satisfying these conditions, can she choose?

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

Crux Mathematicorum, Vol. 46(4), April 2020

 MathemAttic /147

Les problémes proposés dans cette section sont appropriés aux étudiants de l’école sec-
ondaire.

Cliquez ici afin de soumettre vos solutions, commentaires ou

généralisations aux problèmes proposés dans cette section.

Pour faciliter l’examen des solutions, nous demandons aux lecteurs de les faire parvenir
au plus tard le 15 juin 2020.

La rédaction souhaite remercier Rolland Gaudet, professeur titulaire à la retraite à

l’Université de Saint-Boniface, d’avoir traduit les problèmes.

MA66. Les 16 petits carrés illustrés ci-bas sont tous de côtés 1 unité. Combien
de paires de sommets se trouvent à une distance entière d’unités?

MA67. Soient les entiers de la forme 10n + 1, où n est entier positif.
Un tel nombre est dit remier s’il n’est pas possible de le représenter comme
produit de deux plus petits entiers possiblement égaux, toujours de la forme
10k + 1 où k serait entier positif. Combien de nombres remiers y a-t-il parmi
11, 21, 31, 41, . . . , 981, 991?

MA68. P QRS est un carré. Les points T et U sont les mi points de QR

et RS respectivement. La ligne QS intersecte P T et P U en W et V respective-
ment. Quelle fraction de la surface du carré P QRS est occupée par le pentagone
RT W V U ?

Copyright © Canadian Mathematical Society, 2020

148/ MathemAttic

MA69. Le diagramme ci-bas montre deux lignes P R et QS intersectant en

O. Quelle est la valeur de x?

MA70. Le métro de Winnibourg consiste de six lignes, p, q, r, s, t, u, telles

qu’indiquées ci-bas. Lorsque deux lignes se rencontrent, on y retrouve une station
permettant de changer de ligne. De plus, le métro s’arrête à toute station sur
sa ligne. Jéhane désire voyager de la station X à la station Y . Mais elle refuse
d’utiliser une quelconque ligne plus qu’une fois, en plus de ne jamais revenir une
deuxième fois à la station X, ni de quitter la station Y après y être arrivée.
Déterminer le nombre de telles routes différentes.

Crux Mathematicorum, Vol. 46(4), April 2020

 MathemAttic /149

MATHEMATTIC
SOLUTIONS
Statements of the problems in this section originally appear in 2019: 45(9), p. 495–496.

MA41. The diagram shows the densest packing of seven circles in an equi-
lateral triangle.

Determine the exact fraction of the area of the triangle that is covered by the
circles.

Originally from “Shaking Hands in Corner Brook and Other Math Problems” by
Peter Booth, Bruce Shawyer and John Grant McLoughlin.

We received 7 submissions, of which 6 were correct and complete. We present the

solution by Dominique Mouchet, modified by the editor.

Copyright © Canadian Mathematical Society, 2020

150/ MathemAttic

On prend 1 comme longueur du côté du grand triangle équilatéral. On notera R

le rayon des cercles. Exprimons la hauteur M H en fonction de R:
• le triangle M KE a pour angles 90 − 60 − 30 et EK = R. Donc M E = 2R.
• EI est la hauteur d’un triangle équilatéral de côté 2R. Donc
√
3 √
EI = 2R · = R 3.
2

• IJ = JH = 2R.
Donc
√ √

M H = M E + EI + IJ + IH = 2R + R 3 + 2R + 2R = R 6 + 3 .
√
3
Comme M H = , on obtient
2
√ √ √
√
3 3 6− 3 2 3−1
R= √
= = .
2 6+ 3 2 · 33 22

La fraction p de la surface du triangle couverte par les 7 triangles est donc:

√ å
7π Ä √
Ç
7πR2 28π 13 − 4 3 ä
p= √ = √ = 13 3 − 12 ≈ 0.6371.
3 3 484 363
4

a + bx
MA42. Find all functions of the form f (x) =
b+x
where a and b are
constants such that f (2) = 2f (5) and f (0) + 3f (−2) = 0.
Originally Question 2 of 1980 J.I.R. McKnight Mathematics Scholarship Paper.
We received 8 submissions, all of which were correct and complete. We present
the solution by José Luis Dı́az-Barrero, modified by the editor.
The condition gives us that

a + 2b a + 5b a − 2b
Å ã Å ã
a
=2 and +3 = 0.
b+2 b+5 b b−2

The above results in the nonlinear system of equations:

8b2 + (10 + a)b − a = 0,

2
(1)
6b − 4ab + 2a = 0.

The resultant of the above system is

−4a(a + 1)(19a − 300).

Crux Mathematicorum, Vol. 46(4), April 2020

 MathemAttic /151

Substituting the zeros of a in the above we see that (a, b) = (0, 0), (a, b) = (−1, −1)
and (a, b) = (300/19, 10/19) solve (1). Thus

1+x 300 + 10x

f (x) = 0, f (x) = , and f (x) = ,
1−x 10 + 19x
are the only functions which satisfy the stated condition.

MA43. If n is not divisible by 4, prove that 1n + 2n + 3n + 4n is divisible by

5 for any positive integer n.
Adapted from Problem 2 of the 1901 Competition in Hungarian Problem Book 1
(1963).
We received 13 submissions, all of which were correct and complete. We present the
generalized solution by the Problem Solving Group from Missouri State University,
modified by the editor.
We will show, more generally, that if p is any prime number and n is a positive
integer then
1n + 2n + 3n + . . . + (p − 1)n
is a multiple of p if and only if n is not divisible by p − 1.
It is well known that since p is prime, there is an element α ∈ Zp (a primitive root)
such that for all i ∈ Zp , i = αk for some integer k, with 0 ≤ k ≤ p − 2. Therefore
p−1 p−2 p−2
X X
n X k
in ≡ αk ≡ (αn ) .
i=1 k=0 k=0

Note that αn ≡ 1 mod p if and only if n is a multiple of p − 1.

We prove both directions:
⇒ (Contrapositive) If n is a multiple of p − 1, αn ≡ 1 mod p and
p−1
X p−2
X
in ≡ (αn )k ≡ p − 1 6≡ 0 mod p.
i=1 k=0

Thus our sum is not a multiple of p.

⇐ If n is not a multiple of p − 1, αn − 1 6≡ 0 mod p, so αn − 1 ∈ Zp . Using the
formula for finite geometric series (with b = αn ), we have
p−1
X p−2
X
n
(αn )k mod p ≡ bp−1 − 1 (b − 1)−1 mod p ≡ 0 mod p.


i mod p ≡
i=1 k=0

Thus our sum is a multiple of p.

We observe the original problem considers the case when p = 5.

Copyright © Canadian Mathematical Society, 2020

152/ MathemAttic

MA44. Find the largest positive integer which divides all expressions of the
form n5 − n3 where n is a positive integer. Justify your answer.
Proposed by John McLoughlin.
We received 11 submissions, all of which were correct and complete. We present
the joint solution by the Problem Solving Group from Missouri State University
and Tianqi Jiang (solved independently), modified by the editor.
First note that when n = 2, we have that 25 − 23 = 24. Thus the number we seek
must be a factor of 24.
We show 3 | n5 −n3 . As n5 −n3 = n3 (n+1)(n−1) is divisible by three consecutive
integers, it follows one of these numbers is a multiple of 3. Thus 3 | n5 − n3 .
We show 8 | n5 − n3 by considering cases. If n = 2k, then

n5 − n3 = (2k)3 (2k + 1)(2k − 1) = 8k 3 (2k + 1)(2k − 1).

Thus 8 | n5 − n3 . If n = 2k + 1, then

n5 − n3 = (2k + 1)3 (2k + 2)(2k) = (2k + 1)3 · 22 · k · (k + 1).

As one of k or k + 1 is even, it follows 8 | n5 − n3 .

Since 3 and 8 are relatively prime, 24 | n5 − n3 . As we established 24 as an upper
bound, our proof is complete.

MA45. A sequence s1 , s2 , . . . , sn is harmonic if the reciprocals of the terms are

in arithmetic sequence. Suppose s1 , s2 , . . . , s10 are in harmonic sequence. Given
s1 = 1.2 and s10 = 3.68, find s1 + s2 + · · · + s10 .
Originally Question 11 of 1988 Illinois CTM, State Finals AA.
We received 2 submissions, both correct and complete. We present the solution by
Doddy Kastanya.
The arithmetic sequence of interest is a1 , a2 , . . . , a10 where
1 1 1
s1 = , s2 = , . . . , s10 = .
a1 a2 a10
1 5 1 100
From the problem statement, we know that a1 = 1.2 = 6 and a10 = 3.68 = 368 .

For the arithmetic sequence, there are eight items in between a1 and a10 with equal
spacing. Turning the denominator for these two values to 9936, we get a1 = 8280 9936
and a10 = 2700 620
9936 . The spacing between two numbers is 9936 . With this knowledge,
7660
the other items can be determined: a2 = 9936 , a3 = 7040 3320
9936 , up to a9 = 9936 .

The corresponding values of s1 through s10 can be easily calculated. Finally, the
sum of s1 through s10 is calculated as 20.46.

Crux Mathematicorum, Vol. 46(4), April 2020

 Shawn Godin /153

PROBLEM SOLVING
VIGNETTES
No. 11
Shawn Godin
Picking Representations
In many cases, a problem’s solution is aided by thinking about the problem in
a different way than it was originally presented. This may be by looking at a
different, but related problem whose solution leads back to the original. We can
also think about a problem differently by choosing some other way to represent
it. Analytic geometry is an example, where we can think of geometric problems
algebraically or algebraic problems geometrically.
When I thought about this topic the following problem came to mind:
A gas powered go-cart is empty and on a track. Around the track are
a number of gas cans. The total amount of gas in all the cans is equal
to the amount of gas needed to go around the track once. Show that,
no matter how the gas and cans are distributed, you can find a place
to start so that you can make it all the way around the track.

I was introduced to this problem by Crux Editorial Board member Ed Barbeau

at a workshop he did for teachers over 20 years ago. It’s one of those problems
you can convince yourself must work, but coming up with an airtight argument
that convinces others is another thing. The key to the insightful solution that
was given by Ed was to imagine that we are allowed to have a “negative” amount
of gas in our tank. Then the graph of the gas in our tank versus the distance
driven will be a piecewise linear function where all the pieces of the graph will
have equal, negative slopes; there will be a step discontinuity at the location of
each gas can; and when we have finished one trip around the track our tank will,
again, be empty.
Gas in tank

B C D End
(A)
A Position on track

Copyright © Canadian Mathematical Society, 2020

154/ Problem Solving Vignettes

Thus if we draw the graph and find the lowest point, this will be the place that
we should start.

Gas in tank

D A B C End
Position on track (D)

For example, in the graphs above, we are assuming that there are four gas cans A,
B, C, and D. If we start at can A, we get the first graph above on the previous
page. Thus, we see that we should have started at can D, which would have given
us the second graph above.

Choosing the graphical representation not only helped make our argument clearer,
it also gave way to the solution. Now, let’s consider Problem 3 from the 2019
Canadian Mathematical Olympiad:

Let m and n be positive integers. A 2m × 2n grid of squares is coloured

in the usual chessboard fashion. Find the number of ways of placing
mn counters on the white squares, at most one counter per square, so
that no two counters are on white squares that are diagonally adjacent.
An example of a way to place the counters when m = 2 and n = 3 is
shown below.

Since counters cannot be on diagonally adjacent squares, any 2 × 2 square drawn

on the grid can only contain at most one counter. This suggests partitioning the
grid into 2 × 2 squares, as in the diagram below. Since there are (2m)(2n)
(2)(2) = mn
2 × 2 squares on the grid, each of these squares will have exactly one counter.

Crux Mathematicorum, Vol. 46(4), April 2020

 Shawn Godin /155

Notice that there are two configurations that the 2 × 2 squares can be in: either
the counter can be in the upper corner (U ), or the lower corner (L). When we
look at two 2 × 2 squares beside each other we see that U U , U L and LL are all
valid configurations.

UU UL LL

On the other hand the configuration LU is not allowed.

LU

Similarly, going from top to bottom we can have U followed by U or L, but L can
only be followed by another L. Hence, our original 2m × 2n grid was replaced with
an m × n grid of 2 × 2 squares, which in turn can be replaced by an m × n grid
filled with the symbols U and L.
The original example can now be replaced with

U U L
L L L

Notice that in any row or column we have a number of U s (possibly none) followed
by a number of Ls. Once an L appears in a row or column, all entries to the right
and below it are also Ls. Thus starting from the top row and proceeding downward,
each new row has at least as many Ls as the row above it, and possibly more.
Considering the setup from the example in the problem statement, in which a 4×6
starting grid reduces to a 2 × 3 grid of U s and Ls, we can readily construct the 10
ways that the grid can be filled and count the number of Ls in each row (the case
in bold blue is the example from the problem statement).

Copyright © Canadian Mathematical Society, 2020

156/ Problem Solving Vignettes

U U U 0 U U U 0 U U U 0 U U U 0
U U U 0 U U L 1 U L L 2 L L L 3

U U L 1 U U L 1 U U L 1 U L L 2
U U L 1 U L L 2 L L L 3 U L L 2

U L L 2 L L L 3
L L L 3 L L L 3

Thus we can think of our problem in another way: a sequence of two non-negative
integers representing the number of Ls in each row. For the configuration consid-
ered above these would be:

0, 0; 0, 1; 0, 2; 0, 3; 1, 1; 1, 2; 1, 3; 2, 2; 2, 3; 3, 3.

Notice that, to satisfy the conditions of the problem, all of these sequences are non-
decreasing, containing numbers less than or equal to 3 (the number of columns).
Thus our problem is equivalent to finding the number of non-decreasing sequences
of m terms chosen (possibly with repetition) from the set {0, 1, 2, . . . , n}. In the
example that would be the number of non-decreasing sequences of 2 terms from
the set {0, 1, 2, 3}.

We will choose another representation to attack the sequence problem. Since my

numbers can be as large as 3, we will use 3 stars. Since we have 2 terms, we will
use 2 bars. We will arrange these 5 symbols in some order, like

∗| ∗ ∗|

This arrangement is interpreted by counting all the stars to the left of the leftmost
bar, 1, and all the stars to the left of the rightmost bar, 3. Hence the arrangement
above represents the non-decreasing sequence of 2 terms from the set {0, 1, 2, 3}:
1, 3. Each sequence is represented by a unique arrangement of stars and bars,
and each arrangement of stars and bars corresponds to a unique sequence. Hence
counting the number of arrangements of 3 stars and 2 bars gives us the solution
to the configuration in the problem statement.

We can count the number of ways to arrange the 3 stars and 2 bars in several ways
5!
to get = 10. This can be interpreted as arranging 5 things, 3 of one type
3!2!
(stars) and 2 of another (bars). We can also interpret this as we have 5 positions
to put our symbols and we must choose 3 of them to put the stars, leaving the
rest of the places for
bars.
Alternatively, we could have picked the places for the
bars first yielding 52 = 53 = 10.

Returning to the problem, in the general case we have a 2m × 2n grid of squares

filled with counters. We are representing this by a non-decreasing sequence of
m numbers, where each number is less than or equal to n, which represents the
number of L’s in our smaller grid. Converting this to m bars and n stars, we get

Crux Mathematicorum, Vol. 46(4), April 2020

 Shawn Godin /157

the total number of ways our task can be completed is

Ç å Ç å
m+n m+n (m + n)!
= = .
m n m!n!

It is worthwhile to go back to the problem and convince yourself that you would
get the same result with these slight variations on our technique:
• Counting the number of non-increasing sequences representing the number
of U s in each row.
• Counting the number of non-increasing sequences representing the number
of U s in each column.
• Counting the number of non-decreasing sequences representing the number
of Ls in each column.
The official solution to the problem uses a slightly different approach. It also
notices the differences between the U s and Ls and notes that the boundary sep-
arating these two types of 2 × 2 cells makes a path from the lower left corner of
the big grid to the upper right corner, travelling either to the right or up. The
original example with the boundary highlighted in red is in the diagram below.
As an exercise, you may want to solve the problem using this representation.

Keep in mind that sometimes changing your point of view through a different
representation of the problem may lead you to a solution. You may want to
check out the seventh number of this column, Counting Carefully [2019: 386-389],
where the stars and bars technique was used in a slightly different way. For your
enjoyment, here are the rest of the problems from the 2019 Canadian Mathematical
Olympiad.

The 2019 Canadian Mathematical Olympiad

1. Amy has drawn three points in a plane, A, B, and C, such that AB = BC =
CA = 6. Amy is allowed to draw a new point if it is the circumcenter of a
triangle whose vertices she has already drawn. For example, she can draw
the circumcenter O of triangle ABC, and then afterwards she can draw the
circumcenter of triangle ABO.

Copyright © Canadian Mathematical Society, 2020

158/ Problem Solving Vignettes

(a) Prove that Amy can eventually draw a point whose distance from a
previously drawn point is greater than 7.
(b) Prove that Amy can eventually draw a point whose distance from a
previously drawn point is greater than 2019.

(Recall that the circumcenter of a triangle is the center of the circle

that passes through its three vertices.)
2. Let a and b be positive integers such that a+b3 is divisible by a2 +3ab+3b2 −1.
Prove that a2 + 3ab + 3b2 − 1 is divisible by the cube of an integer greater
than 1.
4. Let n be an integer greater than 1, and let a0 , a1 , . . . , an be real numbers
with a1 = an−1 = 0. Prove that for any real number k,
n−2
X
|a0 | − |an | ≤ |ai − kai+1 − ai+2 |.
i=0

5. David and Jacob are playing a game of connecting n ≥ 3 points drawn in a

plane. No three of the points are collinear. On each player’s turn, he chooses
two points to connect by a new line segment. The first player to complete
a cycle consisting of an odd number of line segments loses the game. (Both
endpoints of each line segment in the cycle must be among the n given points,
not points which arise later as intersections of segments.) Assuming David
goes first, determine all n for which he has a winning strategy.
The author would like to thank Ed Barbeau for reminding him about the details
of the go-cart problem and providing valuable feedback that greatly improved the
article.

Crux Mathematicorum, Vol. 46(4), April 2020

 OLYMPIAD CORNER /159

OLYMPIAD CORNER
No. 382
The problems in this section appeared in a regional or national mathematical Olympiad.

Click here to submit solutions, comments and generalizations to any

problem in this section

To facilitate their consideration, solutions should be received by June 15, 2020.

OC476. Let x be a real number such that both sums S = sin 64x + sin 65x
and C = cos 64x + cos 65x are rational numbers. Prove that in one of these sums,
both terms are rational.

OC477. Let A = {z ∈ C | |z| = 1}.

√ √
(a) Prove that (|z + 1| − 2)(|z − 1| − 2) ≤ 0 ∀z ∈ A.
(b) Prove that for any z1 , z2 , . . . , z12 ∈ A, there is a choice of signs “±” so that
12
X
|zk ± 1| < 17.
k=1

OC478. Consider two noncommuting matrices A, B ∈ M2 (R).

(a) Knowing that A3 = B 3 , prove that An and B n have the same trace for any
nonzero natural number n.
(b) Give an example of two noncommuting matrices A, B ∈ M2 (R) such that for
any nonzero n ∈ N, An 6= B n , and An and B n have different trace.

OC479. We say that the function f : Q∗+ → Q has the property P if

f (xy) = f (x) + f (y) ∀x, y ∈ Q∗+ .

(a) Prove that there do not exist injective functions with property P.
(b) Do there exist surjective functions with property P?

OC480. In the plane, there are points C and D on the same region with
respect to the line defined by the segment AB so that the circumcircles of triangles
ABC and ABD are the same. Let E be the incenter of triangle ABC, let F be the
incenter of triangle ABD and let G be the midpoint of the arc AB not containing
the points C and D. Prove that points A, B, E, F are on a circle with center G.

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

Copyright © Canadian Mathematical Society, 2020

160/ OLYMPIAD CORNER

Les problèmes présentés dans cette section ont déjà été présentés dans le cadre d’une
olympiade mathématique régionale ou nationale.

Cliquez ici afin de soumettre vos solutions, commentaires ou

généralisations aux problèmes proposés dans cette section.

Pour faciliter l’examen des solutions, nous demandons aux lecteurs de les faire parvenir
au plus tard le 15 juin 2020.
La rédaction souhaite remercier Rolland Gaudet, professeur titulaire à la retraite à
l’Université de Saint-Boniface, d’avoir traduit les problèmes.

OC476. Soit x un nombre réel tel que les deux sommes S = sin 64x + sin 65x
et C = cos 64x + cos 65x sont rationnelles. Démontrer que dans une des sommes,
les deux termes sont rationnels.

OC477. Soit A = {z ∈ C | |z| = 1}.

√ √
(a) Démontrer que (|z + 1| − 2)(|z − 1| − 2) ≤ 0 ∀z ∈ A.
(b) Démontrer que pour tout z1 , z2 , . . . , z12 ∈ A, il existe un choix de signes “±”
tels que
X12
|zk ± 1| < 17.
k=1

OC478. Soient deux matrices qui ne commutent pas, A, B ∈ M2 (R).

(a) Si A3 = B 3 , démontrer que An et B n ont la même trace ∀n ∈ N, n 6= 0.
(b) Donner une exemple de deux matrices qui ne commutent pas, A, B ∈ M2 (R),
telles que, pour tout non nul n ∈ N, An 6= B n , puis An et B n sont de
différentes traces.

OC479. La fonction f : Q∗+ → Q possède la propriété P si

f (xy) = f (x) + f (y) ∀x, y ∈ Q∗+ .

(a) Démontrer qu’il n’existe aucune fonction injective possédant la propriété P.

(b) Des fonctions surjectives avec la propriété P existent-elles?

OC480. Soient C et D deux points dans le même demi plan par rapport au
segment AB, de façon à ce que les cercles circonscrits des triangles ABC et ABD
soient les mêmes. Soit E le centre du cercle inscrit du triangle ABC et soit F le
centre du cercle inscrit du triangle ABD ; soit aussi G le mi point de l’arc AB
contenant ni C ni D. Démontrer que A, B, E, F se trouvent sur un cercle de
centre G.

Crux Mathematicorum, Vol. 46(4), April 2020

 OLYMPIAD CORNER /161

OLYMPIAD CORNER
SOLUTIONS
Statements of the problems in this section originally appear in 2019: 45(10), p. 504–505.

OC451. Determine the least natural number a such that

n
X
a≥ ak cos(a1 + · · · + ak )
k=1

for any nonzero natural number n and for any positive real numbers a1 , a2 , . . . , an
whose sum is at most π.
We received 1 submission. We present the solution by Oliver Geupel.
We show that
( n n
)
X X
sup ak cos(a1 + · · · + ak ) : n ≥ 1; a1 , . . . , ak > 0; ak ≤ π = 1,
k=1 k=1

which implies that the least value of a is 1. For 1 ≤ k ≤ n, let

k
X
xk = aj .
j=1

Then, 0 = x0 < x1 < x2 < · · · < xn ≤ π, and the sum

n
X n
X
ak cos(a1 + · · · + ak ) = (xk − xk−1 ) cos xk (1)
k=1 k=1

is a right Riemann sum which underestimates the integral

Z xn
I(xn ) = cos x dx
0

of the decreasing function cos x on the interval [0, xn ]. Since I(xn ) = sin xn ≤ 1 for
every xn ≤ π, we obtain that a ≤ 1. However, if ak = π/(2n) and xk = kπ/(2n)
for any 1 ≤ k ≤ n then the sequence defined by the sums (1) converges towards
I(π/2) = 1 as n → ∞. Hence a = 1.
Editor’s Comment. The restrictions on an ’s can be changed. For example, if

a1 + · · · + an ≤ π/2,

then the value of the upper bound a is 1, as before. However, if

a1 + · · · + an = π,

then a = 0.

Copyright © Canadian Mathematical Society, 2020

162/ OLYMPIAD CORNER

OC452. Let ABCD be a square. Consider the points E ∈ AB, N ∈ CD

and F, M ∈ BC such that triangles AM N and DEF are equilateral. Prove that
P Q = F M , where {P } = AN ∩ DE and {Q} = AM ∩ EF .
We received 11 correct submissions. We present two solutions.
Solution 1, by Miguel Amengual Covas.
The right-angled triangles ABM and ADN have equal hypotenuses AM and AN ,
and the legs AB and AD are respectively equal. Thus 4ABM and 4ADN are
congruent with ∠M AB = ∠N AD.
Now, ∠DAB = ∠M AB + ∠N AM + ∠N AD. Next, since ∠DAB = 90◦ and
∠N AM = 60◦ , it follows that ∠M AB and ∠N AD are each 15◦ . Analogously,
∠CDF = ∠ADE = 15◦ .
Clearly, then, 4ABM , 4CDF , 4DAE, 4ADN are congruent (these are right-
angled triangles which have equal legs AB, CD and DA and contain another pair
of equal angles) with
AE = BM = F C = N D.
Consequently, AEN D is a rectangle, so that the segments AN and DE bisect
each other. Thus P is the midpoint of segment DE. Then, we have in equilateral
triangle DEF that ∠F P E is a right angle and ∠P F E = 30◦ .
D N C

Q
15◦
A E B

Subtracting AE = F C from both sides of AB = BC gives AB − AE = BC − F C.

This makes EB = BF and consequently 4EBF is an isosceles right-triangle with
∠EF B = 45◦ . Therefore,
∠P F M = ∠P F B = ∠P F E + ∠EF B = 30◦ + 45◦
= 75◦
= 90◦ − 15◦
= 90◦ − ∠M AB
= ∠AM B,

Crux Mathematicorum, Vol. 46(4), April 2020

 OLYMPIAD CORNER /163

implying that P F is parallel to AM , that is, P F k QM.

Next, P Q subtends 60◦ angles at A and E, making P AEQ cyclic and on chord
EQ we have
∠QP E = ∠QAE = ∠M AB = 15◦
and
∠F P Q = ∠F P E − ∠QP E = 90◦ − 15◦ = 75◦ .
That is to say, the exterior angle QM B in quadrilateral P QM F is equal to the
interior and opposite angle P . Thus P QM F is cyclic. Since P F k QM , P QM F
is an isosceles trapezium. The conclusion follows.

Solution 2, by Miguel Amengual Covas.

As in Solution 1, we conclude that ∠M AB = ∠N AD = ∠CDF = ∠ADE = 15◦ .

D N C

15◦

A E B

Suppose (wlog) the unity of measurement equal to the length of the side of the
given square. Then

DN = AE = BM = CF = tan 15◦ . (1)

Hence,
F M = BC − CF − M B = 1 − 2 tan 15◦ . (2)

Moreover, since AE k DN , AEN D is a rectangle, so that segments AN and DE

bisect each other. Therefore, P is the midpoint of DE and we have

1 1
PE = DE = . (3)
2 2 cos 15◦

Copyright © Canadian Mathematical Society, 2020

164/ OLYMPIAD CORNER

Observing that the equal segments F B and BE make EBF an isosceles right-
angled triangle, the exterior angle theorem, applied to 4AEQ at E, yields

∠EQA = ∠QEB − ∠QAE = ∠F EB − ∠M AB = 45◦ − 15◦ = 30◦ .

Now, the law of sines asserts that

QE AE
= ,
sin 15◦ sin 30◦
and therefore
QE
= 2 · AE,
sin 15◦
yielding (by (2))
2 sin2 15◦
QE = . (4)
cos 15◦

Applying the law of cosines to 4P EQ we get

P Q2 = P E 2 + QE 2 − 2 · P E · QE · cos 60◦ .

Substituting for P E and QE from (4) and (5),

1 4 sin4 15◦
P Q2 = ◦
+ − tan2 15◦ . (5)
2
4 cos 15 cos2 15◦

Now, we write the identity 2 sin 30◦ = 1 in the equivalent form 4 sin 15◦ cos 15◦ = 1,
multiply it by tan 15◦ and square, obtaining 16 sin4 15◦ = tan2 15◦ . Hence we can
rewrite (6) as
1 + tan2 15◦
P Q2 = − tan2 15◦ ,
4 cos2 15◦
or, equivalently,
1
P Q2 = − tan2 15◦ .
4 cos4 15◦

This, in turn, is equivalent to

1 − 4 sin2 15◦ cos2 15◦ 1 − sin2 30◦ 3

P Q2 = ◦
= ◦
= .
4
4 cos 15 4
4 cos 15 16 cos4 15◦

Thus √
3 √
PQ = ◦
= 3 tan 15◦ . (6)
4 cos2
15
√
Taking into account that tan 15◦ = 2 − 3, from (3) and (7) we conclude that
√
F M = 2 3 − 3 = P Q.

Crux Mathematicorum, Vol. 46(4), April 2020

 OLYMPIAD CORNER /165

OC453. Let n ≥ 2 be an integer and let A, B ∈ Mn (C). If (AB)3 = On , is

it true that (BA)3 = On ? Justify your answer.
We received 4 correct submissions. We present two solutions.
Solution 1, by Oliver Geupel.
We prove that the deduction is correct if and only if n ≤ 3.
First, let n ≤ 3 and let λ be an eigenvalue of the matrix C = BA with eigenvector
v. Then, λv = Cv and

λ4 v = C · λ3 v = C 2 · λ2 v = C 3 · λv = C 4 v = B(AB)3 Av = BOn Av = On .

Hence λ = 0. So 0 is the only eigenvalue of C. The characteristic polynomial of

C is then λn . By the Cayley-Hamilton theorem, the matrix C satisfies its own
characteristic equation, so that C n = On and therefore (BA)3 = C 3 = On .
We now turn to the case where n ≥ 4. Let

   
0 0 0 1 0 0 1 0
 0 0 1 0   0 1 0 0 
A4 = 
 0
, B4 =  .
1 0 0   1 0 0 0 
0 0 0 0 0 0 0 0

Consider the n-by-n block matrices

ï ò ï ò
A4 O4×(n−4) B4 O4×(n−4)
A= , B= .
O(n−4)×4 O(n−4)×(n−4) O(n−4)×4 O(n−4)×(n−4)

Straightforward computations yield (A4 B4 )3 = O4 and

 
0 0 0 1
 0 0 0 0 
(B4 A4 )3 = 
 0 0 0 0 


0 0 0 0

Thus,
(A4 B4 )3
ï ò
3 O4×(n−4)
(AB) = = On
O(n−4)×4 O(n−4)×(n−4)
and

(B4 A4 )3
ï ò
O4×(n−4)
(BA)3 = 6= On .
O(n−4)×4 O(n−4)×(n−4)

This completes the proof.

Copyright © Canadian Mathematical Society, 2020

166/ OLYMPIAD CORNER

Solution 2, by Missouri State University Problem Solving Group.

We prove a more general result. Fix k ≥ 1. Then
(1) If n ≤ k then (AB)k = 0 implies (BA)k = 0.
(2) If n ≥ k + 1 then there exist n × n matrices A and B such that (AB)k = 0
but (BA)k 6= 0.
Suppose n ≤ k and (AB)k = 0. Every eigenvalue of AB is 0 and the characteristic
polynomial of AB is xn . But AB and BA have the same characteristic polynomial,
hence (BA)k = 0. This proves (1).

(2) Let n = k + 1, let A be the n × n matrix whose (i, j) entry is 1 if and only if
i = j > 1 and 0 otherwise, and let B be the n × n matrix whose (i, j) entry is 1
if and only if j = i + 1 and 0 otherwise. Then BA = B and AB = C where the
(i, j) entry of C is 1 if and only if j = i + 1 > 2 and 0 otherwise.
A direct calculation yields (AB)n−1 = 0, but (BA)n−1 is non-zero. By taking the
direct sum of A and B with the zero matrix, we can get counterexamples for all
larger n.

For the original problem with k = 3, we take n = 4 and get

Ü ê Ü ê
0 0 0 0 0 1 0 0
0 1 0 0 0 0 1 0
A= and B= .
0 0 1 0 0 0 0 1
0 0 0 1 0 0 0 0

Then Ü ê Ü ê
0 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0
AB = and (AB)3 = ,
0 0 0 1 0 0 0 0
0 0 0 0 0 0 0 0

but Ü ê Ü ê
0 1 0 0 0 0 0 1
0 0 1 0 0 0 0 0
BA = and (BA)3 = .
0 0 0 1 0 0 0 0
0 0 0 0 0 0 0 0

OC454. Find all the functions f : N → N having the following property for
each natural number m: if d1 , d2 , . . . , dn are all the divisors of the number m, then

f (d1 )f (d2 ) · . . . · f (dn ) = m.

We received 8 submissions. We present the solution by Oliver Geupel.

Crux Mathematicorum, Vol. 46(4), April 2020

 OLYMPIAD CORNER /167

It is readily checked that the following function is a solution to the problem:

®
p if m = pk where p is a prime number and k ≥ 1
f (m) =
1 otherwise.

We show that there are no other solutions. Suppose f is any solution.

Putting m = 1 in the given condition, we obtain f (1) = 1. Setting m = p where
p is a prime number in the given condition, we get f (p) = f (1)f (p) = p. A
straightforward induction shows that f (pk ) = p for k ≥ 1.
Finally, let m have at least two distinct prime divisors, say m = pk11 pk22 · · · pk` ` ,
where p1 , . . . , p` are distinct prime divisors (` ≥ 2) and kj ≥ 1 for 1 ≤ j ≤ `.
Let D denote the set of those divisors of m that have at least two distinct prime
divisors. Then,
Ü ê
! Ñ é !
Y Y Y k Y Y
m = f (1) f (pij ) f (d) = pj j f (d) = m f (d).
1≤j≤` d∈D 1≤j≤` d∈D d∈D
1≤i≤kj

Hence, Y
f (d) = 1.
d∈D

It follows f (d) = 1 for all d ∈ D. Since m ∈ D, we conclude that f (m) = 1.

OC455. Let D be a point on the base AB of an isosceles triangle ABC.

Select a point E so that ADEC is a parallelogram. On the line ED, take a point
F such that E ∈ DF and EB = EF . Prove that the length of the chord that
the line BE cuts on the circumcircle of triangle ABF is twice the length of the
segment AC.
We received 4 submissions. We present the solution by Ivko Dimitrić.
Without loss of generality, we may assume that the vertices are labeled counter-
clockwise and that |DB| ≤ |AD|. Let ω be the circumcircle of 4ABF, O its
center and K the point where the line BE meets ω again. Let M be the midpoint
←→ ←→
of BK, S the midpoint of BF and N = CE ∩ BF . Further, set

α = ∠CAB = ∠CBA = ∠ECB, θ = ∠F BE = ∠EF B and ϕ = ∠ECM.

To prove the claim, it suffices to show that 4CBM is isosceles.

From parallelogram ADEC we have ∠CED = α and from isosceles 4BF E we
get ∠DEB = 2θ, so that ∠CEB = α + 2θ. The points O, E and S are collinear,
because BE = EF and E, O belong to the perpendicular bisector of BF at S.
Since |DB| ≤ |AD| the foot of the perpendicular from B to CN belongs to the
segment CE just as the foot of the perpendicular from C to AB belongs to AD.

Copyright © Canadian Mathematical Society, 2020

168/ OLYMPIAD CORNER

That implies that ∠BEN ≥ 90◦ and ∠M EO = ∠BES < 90◦ . Hence, S is between
B and N and M is between E and K.
Since OM ⊥ BK and O and C belong to the perpendicular bisector of AB we
have ∠ECO = ∠OM E = 90◦ and the quadrilateral CEM O is cyclic so that

∠ECM = ∠EOM = ϕ and ∠OEC = ∠OM C.

Since the triangles ESN and OEM are right-angled we have

∠CN B = ∠EN S = 90◦ − ∠SEN

= 90◦ − ∠OEC
= 90◦ − ∠OM C
= ∠CM E = ∠CM B.

Therefore, quadrilateral BCM N is cyclic so that ϕ = ∠N CM = ∠N BM = θ.

Now, we have
∠BCM = ∠BCE + ∠ECM = α + ϕ
and from 4CM E we get

∠CM B = ∠CEB − ∠ECM = (α + 2θ) − ϕ = α + 2ϕ − ϕ = α + ϕ.

Therefore, we conclude that ∠BCM = ∠CM B, so that

1
AC = CB = M B = BK,
2
proving the claim.

Crux Mathematicorum, Vol. 46(4), April 2020

 Michel Bataille’s Focus On . . . Index /169

Focus On . . . Index
Algebra

1 Integer Part and Periodicity

38:3 (Mar 2012) 99-100
A periodic function that vanishes over a period is the zero function. This
obvious property can lead to elegant proofs for some identities involving the
integer part function.
3 From Linear Recurrences to a Polynomial Identity
38:7 (Sep 2012) 276-277
To determine the set of all sequences Un satisfying the linear recurrence
Un+2 = xUn+1 − yUn , a direct approach (instead of the classical method)
leads to a general polynomial identity.

7 Decomposition into Partial Fractions

39:5 (May 2013) 218-221
Some examples involving minimal calculations to display the usefulness of
this algebraic tool.

15 A Formula of Euler
41:1 (Jan 2015) 16-20
Pn n−k n
k m equals 0 if


Euler observed that the sum S(n, m) = k=0 (−1) k
m = 0, 1, . . . , n − 1 and equals n! when m = n.

17 Congruences (I)
41:5 (May 2015) 202-205
In many problems involving integers, an ingenious appeal to congruences
can be most helpful. Choosing the appropriate modulus is often the key to
a short solution.

18 Congruences (II)
41:7 (Sep 2015) 295-298
Here, the focus is on congruences modulo a prime number, with emphasis
on the consequences of three simple, but useful, theorems.

23 Vieta’s Formulas
42:7 (Sep 2016) 303-307
The formulas that relate the coefficients of a polynomial to sums and prod-
ucts of its roots can help solve algebra problems and establish inequalities.

Copyright © Canadian Mathematical Society, 2020

170/ Michel Bataille’s Focus On . . . Index

25 The Long Division of Polynomials

43:1 (Jan 2017) 16-20
The division algorithm for polynomials can provide the key to solving a
variety of algebra problems.
26 Degree and Roots of a Polynomial
43:5 (May 2017) 205-209
The focus here is on the links between the degree of a polynomial and the
number of its roots.
30 Cauchy’s Functional Equation
44:3 (Mar 2018) 106-109
Solutions to problems involving functional equations often come down to
an application of known results about Cauchy’s equation. A selection of
problems illustrate the various properties under consideration.

Inequalities

5 Inequalities via Lagrange Multipliers

39:1 (Jan 2013) 24-26
A few examples of problems requiring the proof of a constrained inequality
where, with some care, the method of Lagrange Multipliers leads to a simple
solution.

20 Inequalities via Complex Numbers

42:1 (Jan 2016) 20-23
Geometric and algebraic inequalities follow from familiar properties of the
modulus of a complex number.

23 Vieta’s Formulas
42:7 (Sep 2016) 303-307
(See under Algebra.)

40 Inequalities via Auxiliary Functions (I)

46:3 (Mar 2020) 117-122
We illustrate through examples how to verify a given inequality by choosing
an appropriate auxiliary function, then using calculus.

Crux Mathematicorum, Vol. 46(4), April 2020

 Michel Bataille’s Focus On . . . Index /171

Geometry

2 The Geometry Behind the Scene

38:5 (May 2012) 183-185
An algebra problem can sometimes be simplified if a connection to a corre-
sponding geometry problem can be found.

4 The Barycentric Equation of a Line

38:9 (Nov 2012) 367-368
A geometric look at the coefficients of the barycentric equation of a line in
the Euclidean plane, and some applications.

6 Glide Reflections in the Plane

39:3 (Mar 2013) 133-135
A few situations where glide reflections provide insight.

8 Generalized Inversion in the Plane

39:7 (Sep 2013) 307-310
We extend the notion of inversion to include the commutative product of a
classical inversion in a circle and the half turn about its center. This provides
a unique inversion with center O exchanging points A and B no matter the
relative position of the three collinear points.

12 Intersecting Circles and Spiral Similarities

40:5 (May 2014) 203-206
Let the circles C1 , C2 intersect at points U, V . Among the spiral similarities
transforming C1 into C2 , the one with center U provides a simple way to
obtain the image of any point P of C1 — it is the second point of intersection
of the circle C2 with the line through P and V .

13 The Dot Product

40:7 (Sep 2014) 289-292
The purpose of this essay is to show the dot product at work through alter-
native solutions to several past geometry problems.

16 Leibniz’s and Stewart’s Relations

41:3 (Mar 2015) 110-113
Leibniz’s relation involves the center of mass of n weighted points in d-
dimensional Euclidean space. It deserves to be better known; among other
things, Stewart’s theorem is an easy consequence.

Copyright © Canadian Mathematical Society, 2020

172/ Michel Bataille’s Focus On . . . Index

21 The Product of Two Reflections in the Plane

42:3 (Mar 2016) 109-113
Knowing that the product of two reflections is either a translation or rotation
can help prove geometric theorems and solve geometric problems.
22 Constructions on the Sides
42:5 (May 2016) 211-215
We investigate configurations involving triangles or quadrilaterals constructed
on the sides of triangles or quadrilaterals, and favor proofs using transfor-
mations or complex numbers.

27 Some Relations in the Triangle (I)

43:7 (Sep 2017) 293-297
The goal here is to present a selection of less familiar relations among the
parts of a triangle that are attractive and useful.

28 Some Relations in the Triangle (II)

43:9 (Nov 2017) 389-393
A continuation of Number 27, focusing here on formulas involving lengths
related to the classical cevians.
32 Harmonic Ranges and Pencils
44:7 (Sep 2018) 291-296
Elementary properties of a harmonic conjugate can lead to simple and elegant
solutions to some geometry problems.

36 Geometry with Complex Numbers (I)

45:5 (May 2019) 258-264
We focus here on the use of complex numbers to prove results involving a
triangle and its circumcircle.

37 Geometry with Complex Numbers (II)

45:7 (Sep 2019) 407-412
We continue our discussion from Number 36, here using complex numbers
to deal with regular polygons, similarities, and areas.
39 Introducing SA , SB , SC in Barycentric Coordinates
46:1 (Jan 2020) 26-31
Barycentric coordinates relative to a triangle are appropriate not just for
dealing with affine properties such as collinearity, concurrency, and areas,
but they can often be used also for Euclidean properties such as lengths and
perpendicularity.

Crux Mathematicorum, Vol. 46(4), April 2020

 Michel Bataille’s Focus On . . . Index /173

Calculus

5 Inequalities via Lagrange Multipliers

39:1 (Jan 2013) 24-26
(See under Inequalities.)

10 Some Sequences of Integrals

40:1 (Jan 2014) 21-24
R1
A study of the sequences In = 0 (ax2 + bx + c)n dx with the goal of finding
a sequence (ωn ) such that limn→∞ = ωInn = 1.

11 The Partial Sums of Some Divergent Series

40:3 (Mar 2014) 112-115
an+1
For a sequence (an ) of positive real numbers such that limn→∞ = Pn an =
ω > 1, and the corresponding sequence of its partial sums An = k=1 Ak ,
Ä A establish
we ä a few results about the behavior of theÄ a sequences
ä (An ) and
n+1
An in comparison with the sequences (an ) and n+1
an and offer some
applications.

15 A Formula of Euler
41:1 (Jan 2015) 16-20
(See under Algebra.)

30 Cauchy’s Functional Equation

44:3 (Mar 2018) 106-109
(See under Algebra.)

31 Mean Value and Rolle’s Theorems

44:5 (May 2018) 202-206
The Mean Value Theorem establishes a link between a function and its
derivative. Here we will see it at work in various problems, sometimes rather
unexpectedly.

35 The Asymptotic Behavior of Integrals

45:3 (Mar 2019) 137-143
We keep the same goal as in Number 10 [2014: 21-24] (namely, to determine
the asymptotic behavior of a given sequence of integrals), but here, restrict-
ing ourselves to elementary problems and methods, present simple ways to
obtain such an asymptotic behavior.

Copyright © Canadian Mathematical Society, 2020

174/ Michel Bataille’s Focus On . . . Index

Combinatorics
15 A Formula of Euler
41:1 (Jan 2015) 16-20
(See under Algebra.)

Trigonometry

34 Some Trigonometric Relations

45:1 (Jan 2019) 26-32
We consider a selection of problems involving the values of the circular func-
tions at mπ n for various natural numbers m and n. Solutions depend on
complex numbers and polynomials in addition to the classical trigonometric
identities.

Solutions to Exercises
9 39:9 (Nov 2013) 404-408 From Numbers 2 to 5.
14 40:9 (Nov 2014) 380-385 From Numbers 6 to 11.
19 41:9 (Nov 2015) 386-391 From Numbers 12 to 16.
24 42:9 (Nov 2016) 393-397 From Numbers 17 to 21.
29 44:1 (Jan 2018) 19-24 From Numbers 22 to 26.
33 44:9 (Nov 2018) 377-381 From Numbers 27 to 31.
38 45:9 (Nov 2019) 511-517 From Numbers 32 to 36.

Crux Mathematicorum, Vol. 46(4), April 2020

 Problems /175

PROBLEMS
Click here to submit problems proposals as well as solutions, comments
and generalizations to any problem in this section.

To facilitate their consideration, solutions should be received by June 15, 2020.

4531. Proposed by Leonard Giugiuc and Dan Stefan Marinescu.

Let a, b and c be positive real numbers and let x, y and z be real numbers. Suppose
that a + b + c = 2 and xa + yb + zc = 1. Prove that
3
x + y + z − (xy + yz + zx) ≥ .
4

4532. Proposed by Marius Stănean.

Let ABC be a triangle with circumcircle Γ and let M , N , P be points on the sides
BC, CA, AB, respectively. Let M 0 , N 0 , P 0 be the intersections of AM , BN , CP
with Γ different from the vertices of the triangle. Prove that

R2 r
MM0 · NN0 · PP0 ≤ ,
4
where R and r are the circumradius and the inradius of triangle ABC.

4533. Proposed by Leonard Giugiuc and Kadir Altintas.

Let K be the symmedian point of ABC. Let ka , kb and kc be the lengths of the
altitudes from K to the sides BC, AC and AB, respectively. If r is the inradius
and s is the semiperimeter, prove that
Å ã2 Å ã2
1 3 2
+ ≥ 2 .
r s ka + kb2 + kc2

Copyright © Canadian Mathematical Society, 2020

176/ Problems

4534. Proposed by Michel Bataille.

For n ∈ N, evaluate
∞
P 1
k!(n+k+1)
k=0
∞ .
P (−1)k
(n+k+1)!
k=0

4535. Proposed by Mihaela Berindeanu.

Let ABC be an acute triangle with orthocenter H, and let E be the reflection of
H in the midpoint D of side BC. If the perpendicular to DE at H intersects AB
at X and AC at Y , prove that HX · EC + Y C · HE = EX · BE.

4536. Proposed by Leonard Giugiuc and Rovsan Pirkuliev.

Let ABC be a triangle with ∠ABC = 60◦ . Consider a point M on the side AC.
Find the angles of the triangle, given that
√
3BM = AC + max{AM, M C}.

4537. Proposed by Arsalan Wares.

Let A be a regular hexagon with vertices A1 , A2 , A3 , A4 , A5 and A6 . The six
midpoints on the six sides of hexagon A are connected to the six vertices with
12 line segments as shown. The dodecagon formed by these 12 line segments has
been shaded. What part of hexagon A has been shaded?

4538. Proposed by Nguyen Viet Hung.

Let a1 , a2 , . . . , an be non-negative real numbers. Prove that
X p X   X
1 + ai 2 + ai aj ≥ n − 1 + 1 + ( ai )2 .
1≤i≤n 1≤i<j≤n 1≤i≤n

When does equality occur?

Crux Mathematicorum, Vol. 46(4), April 2020

 Problems /177

4539. Proposed by Leonard Giugiuc.

Let ABC be a triangle√with centroid G, incircle ω, circumradius R and semiperime-
ter s. Show that 24R 6 ≥ 25s given that G lies on ω.

4540. Proposed by Prithwijit De.

Given a prime p and an odd natural number k, do there exist infinitely many
natural numbers n such that p divides nk + k n ? Justify your answer.

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

Copyright © Canadian Mathematical Society, 2020

178/ Problems

Cliquez ici afin de proposer de nouveaux problèmes, de même que pour

offrir des solutions, commentaires ou généralisations aux problèmes
proposś dans cette section.

Pour faciliter l’examen des solutions, nous demandons aux lecteurs de les faire parvenir
au plus tard le 15 juin 2020.
La rédaction souhaite remercier Rolland Gaudet, professeur titulaire à la retraite à
l’Université de Saint-Boniface, d’avoir traduit les problèmes.

4531. Proposé par Leonard Giugiuc et Dan Stefan Marinescu.

Soient a, b et c des nombres réels positifs et soient x, y et z des nombres réels. De
plus, supposer que a + b + c = 2 et xa + yb + zc = 1. Démontrer que
3
x + y + z − (xy + yz + zx) ≥ .
4

4532. Proposé par Marius Stănean.

Soit ABC un triangle de cercle circonscrit Γ et soient M, N, P des points situés
sur les côtés BC, CA, AB, respectivement. Soient M 0 , N 0 , P 0 les intersections de
AM, BN, CP avec Γ, mais distincts des sommets du triangle. Démontrer que
R2 r
MM0 · NN0 · PP0 ≤ ,
4
où R et r sont les rayons du cercle circonscrit, puis du cercle inscrit, du triangle
ABC.

4533. Proposé par Leonard Giugiuc et Kadir Altintas.

Soit K un point symédian de ABC. Soient aussi ka , kb et kc les longueurs des
altitudes de K vers les côtés BC, AC et AB, respectivement. Si r est le rayon du
cercle inscrit et s est le demi périmètre, démontrer que
Å ã2 Å ã2
1 3 2
+ ≥ 2 .
r s ka + kb2 + kc2

Crux Mathematicorum, Vol. 46(4), April 2020

 Problems /179

4534. Proposé par Michel Bataille.

Pour n ∈ N, évaluer
∞
P 1
k!(n+k+1)
k=0
∞ .
P (−1)k
(n+k+1)!
k=0

4535. Proposé par Mihaela Berindeanu.

Soit ABC un triangle acutangle d’orthocentre H et soit E la réflexion de H par
rapport au mi point D du côté BC. Si la perpendiculaire vers DE au point H
intersecte AB en X et AC en Y , démontrer que HX · EC + Y C · HE = EX · BE.

4536. Proposé par Leonard Giugiuc et Rovsan Pirkuliev.

Soit ABC un triangle tel que ∠ABC = 60◦ et soit M un point sur le côté AC.
Déterminer les angles du triangle, étant donné que
√
3BM = AC + max{AM, M C}.

4537. Proposé par Arsalan Wares.

Soit A un hexagone régulier de sommets A1 , A2 , A3 , A4 , A5 et A6 . Les six mi points
des six côtés de l’hexagone sont reliés aux six sommets, à l’aide de 12 segments,
tels qu’indiqués ci-bas, où le dodécagone formé par ces 12 segments est coloré.
Quelle fraction de l’hexagone A est ainsi colorée ?

4538. Proposé par Nguyen Viet Hung.

Soient a1 , a2 , . . . , an des nombres réels non négatifs. Démontrer que
X p X   X
1 + ai 2 + ai aj ≥ n − 1 + 1 + ( ai )2 .
1≤i≤n 1≤i<j≤n 1≤i≤n

Quand est-ce que l’égalité tient ?

Copyright © Canadian Mathematical Society, 2020

180/ Problems

4539. Proposé par Leonard Giugiuc.

Soit ABC un triangle de centroı̈de G, de cercle inscrit ω,√ de rayon de cercle
circonscrit R et de demi périmètre s. Démontrer que 24R 6 ≥ 25s, supposant
que G se situe sur ω.

4540. Proposé par Prithwijit De.

À partir d’un nombre premier p et un entier naturel impair k, existe-t-il un nombre
infini de nombres naturels n tels que p divise nk + k n ? Justifier votre réponse.

Crux Mathematicorum, Vol. 46(4), April 2020

 Bonus Problems /181

BONUS PROBLEMS
These problems appear as a bonus. Their solutions will not be considered for publication.

B1. Proposed by Miguel Ochoa Sanchez and Leonard Giugiuc.

Let ABCD be a cyclic quadrilateral. Let P be a point on the arc BC and let L
and Q be the feet of perpendiculars dropped from P on the sides AD and BC,
respectively. Let M and N be the feet of perpendiculars dropped from P on the
lines AB and DC, respectively. Prove that

AM BQ CN DL
· · · = 1.
M B QC N D LA

B2. Proposed by Leonard Giugiuc.

Find the real numbers x, y, z and t such that
√
xt − yz = −1 and x2 + y 2 + z 2 + t2 − xz − yt = 3.

B3. Proposed by Leonard Giugiuc.

Let a, b and c be positive real numbers such that ab + bc + ca = 3. Prove the
inequality
1 1 1
+ + ≤ 1.
a2 + 2 b2 + 2 c2 + 2

B4. Proposed by Leonard Giugiuc.

Let ABC be a triangle with no angle exceeding 120◦ with BC = a, AC = b and
AB = c. Let T be its Fermat-Torricelli point, that is the point such that the total
distance from the three vertices of ABC to T is minimum possible. Prove that
√
(b + c)|T A| + (c + a)|T B| + (a + b)|T C| ≥ 3(|T A| + |T B| + |T C|)2 − 4Area(ABC).

B5. Proposed by Leonard Giugiuc.

Let n be an integer such that n ≥ 4. Consider real numbers ak , 1 ≤ k ≤ n such
that 2 ≥ a1 ≥ 1 ≥ a2 ≥ · · · ≥ an−1 ≥ an and nk=1 ak = n. Prove that
P
Pn 2
a) k=1 ak ≤ n + 2.
Pn (n−2)(n+1)
b) 1≤i<j≤n ai aj ≥ 2 .

Copyright © Canadian Mathematical Society, 2020

182/ Bonus Problems

B6. Proposed by Leonard Giugiuc.

Let ABC be a triangle such that ∠BAC ≥ 2π3 . Prove that
√
r 2 3−3
≤ ,
R 2
where r is the inradius and R is the circumradius of ABC.

B7. Proposed by Leonard Giugiuc.

Let a, b, c and d be real numbers such that 2 ≥ a ≥ 1 ≥ b ≥ c ≥ d ≥ 0 and
a + b + c + d = 4. Prove that
2 9
+ ≤ 2.
a3 + b3 + c3 + d3 ab + bc + cd + da + ac + bd

B8. Proposed by Leonard Giugiuc and Dan Stefan Marinescu.

Let ABCD be a rectangle with center O. Let M and P be two points in the plane
(not necessarily distinct) such that O lies on the line M P and OM = 3 · OP .
Prove that

M A + M B + M C + M D ≥ P A + P B + P C + P D.

B9. Proposed by Leonard Giugiuc.

Let a, b, c ≥ 1 and 0 ≤ d, e, f ≤ 1 such that a + b + c + d + e + f = 6. Prove that

6 ≤ a2 + b2 + c2 + d2 + e2 + f 2 ≤ 18.

B10. Proposed by Leonard Giugiuc.

Let ABC be a nonobtuse triangle with smallest angle A. Prove that

cos (B − C) ≥ cos B + cos C

and determine when equality holds.

B11. Proposed by Michael Rozenberg and Leonard Giugiuc.

Prove that if a, b, c and d are non-negative real numbers such that a + b + c + d = 4,
then »
ab + bc + cd + da + ac + bd ≥ 3 (a2 + b2 + c2 + d2 )abcd.

B12. Proposed by Leonard Giugiuc.

Let x, y and z be positive real numbers such that xyz = 512. Prove that
1 1 1
√ +√ +√ ≥ 1.
1+x 1+y 1+z

Crux Mathematicorum, Vol. 46(4), April 2020

 Bonus Problems /183

B13?. Proposed by Leonard Giugiuc.

Let n be an integer with n ≥ 4. Prove or disprove that for any positive real
numbers ai , i = 1, 2, . . . , n that sum up to 1, we have:
»
n
(1 − an1 )(1 − an2 ) · · · (1 − ann ) ≥ (nn − 1)a1 a2 . . . an .

B14?. Proposed by Leonard Giugiuc.

√
Let k be a real number with k > 7+32 5 . Prove or disprove that for any non-
negative real numbers x, y, z no two of which are zero, we have

3
…
x y z
… …
+ + ≥√ .
ky + z kz + x kx + y k+1

B15. Proposed by Leonard Giugiuc.

Let a, b and c be real numbers such that a ≥ b ≥ 1 ≥ c ≥ 0 and a + b + c = 3.
a) Show that 2 ≤ ab + bc + ca ≤ 3.
45
b) Prove that a3 + b3 + c3 + a2 +b2 +c2 ≤ 18 and study the equality cases.

B16. Proposed by Dao Thanh Oai and Leonard Giugiuc.

Let ABCD be a cyclic quadrilateral. Prove that the following two statements are
equivalent:
a) AC ≥ BD,
b) AB · AD + CB · CD ≥ BA · BC + DA · DC.

B17. Proposed by Dao Thanh Oai and Leonard Giugiuc.

Let ABCD be a cyclic quadrilateral. Prove that
√
AB + AC + AD + BC + BD + CD ≤ 4R( 2 + 1),

where R is the circumradius of ABCD.

B18. Proposed by Leonard Giugiuc and Dorin Marghidanu.

Let n ≥ 2 be a natural number, and ak be real numbers such that 0 < ak < 2 for
all k = 1, 2, . . . , n with nk=1 ak = 1. Prove that
Q

n
X 1 n
√ ≤√ .
k=1
1 + ak 2

Prove further that the condition ak < 2 can be dropped when n = 2 or n = 3.

Copyright © Canadian Mathematical Society, 2020

184/ Bonus Problems

B19. Proposed by Leonard Giugiuc.

Find the maximum value k such that
a2 + b2 + c2 + k(ab + bc + ca) ≥ 3 + k(a + b + c)
for any positive numbers a, b and c such that abc = 1.

B20?. Proposed by Leonard Giugiuc.

Let x, y ∈ (0, 3/2) be real numbers that satisfy (x − 2)(y − 2) = 1. Prove or
disprove that
x3 + y 3 ≥ 2.

B21. Proposed by Marian Cucoanes and Leonard Giugiuc.

Consider an arbitrary triangle ABC with medians ma , mb , mc , circumradius R,
inradius r and exradii ra , rb , rc . Show that
p
ma + mb + mc ≤ 16R2 + 4rR + 9r2 ≤ ra + rb + rc .

B22. Proposed by Leonard Giugiuc.

Let a, b, c, d, e, f be non-negative real numbers such that a + b + c + d + e + f = 4.
If a ≥ b ≥ c ≥ 1 ≥ d ≥ e ≥ f ≥ 0, prove that
a2 + b2 + c2 + d2 + e2 + f 2 + 180abcdef ≤ 10.

B23. Proposed by Miguel Ochoa Sanchez and Leonard Giugiuc.

Given a triangle ABC, let the tangent to its circumcircle at A intersect the line
BC at D, and let the circle through A that is tangent to BC at D intersect the
AB 3
Å ã
EB
circumcircle again at E. Prove that = .
EC AC

B24. Proposed by Ruben Dario Auqui and Leonard Giugiuc.

Let ABCD be a square. Let ω be the circle centered at A with radius AB. A
point M lies inside the square on ω; the line BM intersects the side CD at N .
Prove that CM = 2M N if and only if CM and BN are perpendicular.

B25. Proposed by Ruben Dario Auqui and Leonard Giugiuc.

Let ABC be a triangle with semiperimeter s. The A-excircle of the triangle touches
the side BC at Q and the lines AB and AC at M and N , respectively. Suppose
that AQ intersects M N at P . Prove that
p
s a(s − a)(as + (b − c)2 )
AP = .
b(s − c) + c(s − b)

Crux Mathematicorum, Vol. 46(4), April 2020

 Solutions /185

SOLUTIONS
No problem is ever permanently closed. The editor is always pleased to consider for
publication new solutions or new insights on past problems.
Statements of the problems in this section originally appear in 2019: 45(9), p. 518–523.

4481. Proposed by Warut Suksompong.

Find all functions f : R → R such that

f (x2 + y 2 ) = f (x + y)f (x − y) + 2f (y)y

for all x, y ∈ R.
We received 16 submissions, of which 7 were correct and complete. We present
the solution by Michel Bataille.
There are three functions that satisfy the equation:
• f0 (x) = 0 for all x ∈ R,
• f1 (x) = x,
• f2 (x) = 0 for x 6= 0 and f2 (0) = 1.
The functions f0 , f1 are obvious solutions. As for the function f2 , consider first
the case when x2 + y 2 = 0. Then x = y = 0 and

f2 (x2 + y 2 ) = f2 (0) = 1 = f2 (0)f2 (0) + 2f2 (0) · 0 = f2 (x + y)f2 (x − y) + 2f2 (y)y.

If x2 +y 2 6= 0, then x+y and x−y cannot equal 0 both, hence f2 (x+y)f2 (x−y) = 0
and 2f2 (y)y = 0 (since f2 (y) = 0 if y 6= 0) so that the required functional equation
still holds and f2 is also a solution.
Conversely, let f be an arbitrary solution. For convenience, we let E(x, y) denote
the equality f (x2 + y 2 ) = f (x + y)f (x − y) + 2f (y)y. From E(0, 0) we deduce that
f (0) = (f (0))2 , hence f (0) = 0 or f (0) = 1.
First, suppose that f (0) = 0. From E(x, x) and E(x, −x) we obtain

2xf (x) = −2xf (−x) = f (2x2 ),

so that f (−x) = −f (x) if x 6= 0 and f is an odd function. With the help of this
result and comparing E(x, y) and E(y, x), we get that

f (x + y)f (x − y) = xf (x) − yf (y)

for all x, y. Taking y = 0, this yields (f (x))2 = xf (x) and so f (x) = 0 or f (x) = x
for all x. If f 6= f0 , we have f (a) = a 6= 0 for some a. Now, we consider x 6= 0
and assume that f (x) = 0. Then, f (x + a)f (x − a) = xf (x) − a2 = −a2 6= 0

Copyright © Canadian Mathematical Society, 2020

186/ Solutions

and we must have f (x + a) = x + a and f (x − a) = x − a. But it follows that

(x + a)(x − a) = −a2 , contradicting x 6= 0. In consequence f (x) = x and we can
conclude that f = f0 or f = f1 in the case when f (0) = 0.
Second, suppose f (0) = 1. Again E(x, x) and E(x, −x) give xf (x) = −xf (−x) so
that f (−x) = −f (x) whenever x 6= 0. Comparing E(x, y) and E(y, x) now gives
f (x + y)f (x − y) = xf (x) − yf (y) when x 6= y. To conclude, consider x 6= 0. Then
y = − x2 6= x2 , hence
x x x x x x x  x
f + f − = f + f − ,
2 2 2 2 2 2 2 2

that is, f (x) = 0. Thus, f = f2 when f (0) = 1 and the proof is complete.
Editor’s comments. With respect to incorrect solutions, there were two common
errors: to note that if f (0) = 0 then for x 6= 0 it must be true that either f (x) = x
or f (x) = 0, but to omit the explanation of why only one of these statements must
hold for all x 6= 0, and to miss the solution which has f (0) = 1.

4482. Proposed by Miguel Ochoa Sanchez and Leonard Giugiuc.

Let ABC be a triangle with incenter I. The line AI intersects BC at D. A line
l passes through I and intersects the sides AB and AC at P and Q, respectively.
Show that

[BDIP ] [CDIQ] BC 2
AC · + AB · = 2 · BC + ,
[AP I] [AQI] AB + AC

where square brackets denote area.

All 18 of the submissions we received provided complete solutions; we feature the

solution by Marie-Nicole Gras.
We put
a = BC, b = CA, c = AB, u = AP, v = AQ,

Crux Mathematicorum, Vol. 46(4), April 2020

 Solutions /187

and use the label A to also denote the angles ∠BAC = ∠P AQ. The desired result
is a consequence of the following known facts:
2bc A 2bc A AD a+b+c
AD = cos , AI = cos , so that = .
b+c 2 a+b+c 2 AI b+c

Moreover, since P Q passes through I, AI is the internal bisector of the angle at

A in ∆P AQ, so that
2uv A 2bc A u+v a+b+c
AI = cos = cos , which implies = ·
u+v 2 a+b+c 2 uv bc
Putting these facts together, we have
[BDIP ] [CDIQ] [ABD] − [AP I] [ACD] − [AQI]
b +c = b +c
[AP I] [AQI] [AP I] [AQI]
cAD sin A2 − uAI sin A2 bAD sin A2 − vAI sin A2
= b + c
uAI sin A2 vAI sin A2
bc AD bc AD
= + −b−c
u AI v AI
AD  u + v 
= bc −b−c
AI uv
a+b+c a+b+c
= bc −b−c
b+c bc
(a + b + c)2
= − (b + c)
b+c
(a + b + c)2 − (b + c)2
=
b+c
a(a + 2b + 2c) a2 BC 2
= = 2a + = 2BC + ·
b+c b+c AB + AC

4483. Proposed by Paul Bracken.

For non-negative integers m and n, evaluate the following sum in closed form
m Ç å
2 j+n
X
j .
j=0
j

We received 15 solutions, all correct. We present the solution by Marie-Nicole

Gras.
We will use the hockey-stick identity
m Ç å Ç å
X n+k m+n+1
= .
n n+1
k=0

Copyright © Canadian Mathematical Society, 2020

188/ Solutions

m Ç å
X
2 j+n
For all m, n ≥ 0 let Sm,n = j . We have
j=0
j

m Ç å
X n+j
Sm,n = [j + j(j − 1)]
j=0
n

m Ç å m Ç å
X n+j X n+j
= j + j(j − 1)
j=1
n j=2
n

m Ç å m Ç å
X n+j X n+j
= (n + 1) + (n + 1)(n + 2)
j=1
n+1 j=2
n+2

m−1 Ç å m−2 Ç å
X n+1+k X n+2+k
= (n + 1) + (n + 1)(n + 2) .
n+1 n+2
k=0 k=0

Apply the hockey-stick identity to each sum to get

Ç å Ç å
m+n+1 m+n+1
Sm,n = (n + 1) + (n + 1)(n + 2) .
n+2 n+3

4484. Proposed by Leonard Giugiuc and Michael Rozenberg.

Let a, b, c ∈ [0, 2] such that a + b + c = 3. Prove that

4(ab + bc + ac) ≤ 12 − ((a − b)(b − c)(c − a))2

and find when the equality holds.

We received nine submissions, eight of which are correct and the other is incom-
plete. We present the solution by Digby Smith, modified slightly by the editor.
Without loss of generality we may assume that a ≥ b ≥ c.
Let x = a − b, y = b − c, and z = a − c. Then x, y, z ≥ 0 and x + y = z ≤ 2. Hence,
4xy ≤ (x + y)2 ≤ 4 so xy ≤ 1, with equality if and only if x = y = 1.
Next, note that

xy − (x + y)2 = xy − z(x + y) = xy − yz − zx
= (a − b)(b − c) + (b − c)(c − a) + (c − a)(a − b)
= (ab + bc + ca) − (a2 + b2 + c2 )
= (ab + bc + ca) − ((a + b + c)2 − 2(ab + bc + ca))
= 3((ab + bc + ca) − 3). (1)

Crux Mathematicorum, Vol. 46(4), April 2020

 Solutions /189

By (1), the given inequality is equivalent in succession to

12 ((ab + bc + ca) − 3) + 3(a − b)2 (b − c)2 (c − a)2 ≤ 0,

4(xy − (x + y)2 ) + 3x2 y 2 (x + y)2 ≤ 0,
(x + y)2 (4 − 3x2 y 2 ) − 4xy ≥ 0. (2)

Finally, since xy ≤ 1, we have

(x + y)2 (4 − 3x2 y 2 ) − 4xy ≥ 4xy(4 − 3x2 y 2 ) − 4xy

= 4xy(3 − 3x2 y 2 )
= 12xy(1 − xy)(1 + xy) ≥ 0,

so (2) holds and the proof of the inequality is complete. Clearly, equality holds if
and only if x = y = 0 or x = y = 1 which implies that (a, b, c) = (1, 1, 1) or any
permutation of (2, 1, 0).

4485. Proposed by Jonathan Parker and Eugen J. Ionascu.

For every square matrix with real entries A = [ai,j ]i=1..n,j=1,2...n , we define the
value
GM (A) = max {min{a1π(1) , a2π(2) , ..., anπ(n) }}
π∈Sn

where Sn is the set of all permutations of the set [n] := {1, 2, 3, ..., n}.
Given the 6 × 6 matrix
 
20 9 7 26 27 13

 19 18 17 6 12 25 

 22 24 21 11 20 11 
A :=  

 20 8 9 23 5 14 
 22 17 4 10 36 33 
21 16 23 35 15 34

find the value GM (A).

There were 11 correct solutions, most straightforward.
Let π = (15)(263). Then

min{a1,π(1) , a2,π(2) , . . . , a6,π(6) } = min{a15 , a26 , a32 , a44 , a51 , a63 }

= min{27, 25, 24, 23, 22, 23}
= 22.

Therefore GM (A) ≥ 22.

On the other hand, since for each permutation π, the set being minimized contains
a number aπ−1 (1),1 from the first column, the minimum cannot exceed 22. Thus
GM (A) ≤ 22. Therefore GM (A) = 22.

Copyright © Canadian Mathematical Society, 2020

190/ Solutions

4486. Proposed by Marian Cucoaneş and Marius Drăgan.

Let a, b > 0, c > 1 such that a2 ≥ b2 c. Compute
√ √ √
lim (a − b c)(a − b 3 c) · · · (a − b n c).
n→∞

We received 12 solutions, one of which was incomplete and one of which was
incorrect. We present the solution by Michel Bataille.
n √ √
Let Pn = (a − b k c). From the hypotheses, we have a ≥ b c. We show that if
Q
√ k=2 √
a = b c, then lim Pn = 0 while if a > b c, then lim Pn = ∞ if a > b + 1 and
n→∞ n→∞
lim Pn = 0 if a ≤ b + 1.
n→∞
√
If a = b c, then Pn = 0 for all n ≥ 2 and so lim Pn = 0.
n→∞
√ √
From√now on, we suppose that a > b c. Note that a > b and that a − b k c >
a − b c > 0 for all k ≥ 2 (so that Pn > 0 for all n ≥ 2).
If a > b + 1, we choose m such that a − b > m > 1. Since

Pn+1 √ √
= a − b n+1 c and lim n+1
c = 1,
Pn n→∞

we have lim PPn+1n

= a − b. We deduce that for some N ∈ N, we have Pn+1
Pn >m
n→∞
for all n ≥ N . Then, for any r ∈ N, we obtain

PN +r PN +1 PN +2 PN +r
= · · ··· · > mr
PN PN PN +1 PN +r−1

and therefore PN +r > mr · PN for all r ≥ 1. Since lim mr = ∞, we must have

r→∞
lim Pn = ∞.
n→∞

If a < b + 1, similarly, we choose q such that a − b < q < 1 and then for some
N ∈ N, we have PPn+1n
< q for all n ≥ N . As above we deduce that PN +r < q r PN
and so lim Pn = 0 (since lim q r = 0).
n→∞ r→∞
√ √
Lastly, suppose that a = b + 1. Then a − b n c = 1 − b( n c − 1) and when n → ∞,
we have
√ √ b ln(c)
ln(a − b n c) ∼ −b( n c − 1) ∼ −
n
√
(since ln(1 − x) ∼ −x as x → 0 and n c − 1 = exp((ln(c))/n) − 1 ∼ ln(c) n as
n → ∞).
√
The series n≥2 ln(a − b n c) (whose terms are all negative for n large enough) is
P

divergent (since n≥2 n1 is divergent). This means that lim ln(Pn ) = −∞ and
P
n→∞
so lim Pn = 0.
n→∞

Crux Mathematicorum, Vol. 46(4), April 2020

 Solutions /191

4487. Proposed by Martin Lukarevski.

Let a, b, c be the sides of a triangle ABC, ma , mb , mc the corresponding medians
and R, r its circumradius and inradius respectively. Prove that
a2 b2 c2 4r
+ 2 + 2 ≥ .
m2b 2
+ mc 2
mc + ma ma + m2b R

We received 20 correct solutions. We present the solution by Ioannis Sfikas.

By inequality on page 52 of Geometric Inequalities by O. Bottema and R. Z.
Djordjevic, we have
1 2 1
9R2 ≥ a2 + b2 + c2 a + b2 + c 2 ≥


or
3 27R2
and
2s2 ≥ 27Rr or 2(a + b + c)2 ≥ 108Rr,
where s is the semiperimeter of the triangle ABC. Also, we will use the well-known
relation
3 2
m2a + m2b + m2c = a + b2 + c2 .


4
By Titu’s lemma (a special case of the Cauchy-Schwarz inequality),
a2 b2 c2 (a + b + c)2
+ 2 + 2 ≥
m2b 2
+ mc 2
mc + ma 2
ma + mb 2 (m2a + m2b + m2c )

2(a + b + c)2
=
3 (a2 + b2 + c2 )

108Rr 4r
≥ = .
27R2 R

4488. Proposed by George Apostolopoulos.

Let ABC be an acute-angled triangle. Prove that
√ √ √
…
A B C
cot A + cot B + cot C ≤ cot · cot · cot .
2 2 2

We received 14 submissions, all correct, and we present the solution by Leonard

Giugiuc.
We set x = tan A2 , y = tan B2 , and z = tan C2 . Then 0 < x, y, z < 1 and it is well
known that xy + yz + zx = 1.
Since
1 − x2 1 − tan2 A/2 1 1 − y2 1 − z2
= = = cot A, = cot B, and = cot C,
2x 2 tan A/2 tan A 2y 2z

Copyright © Canadian Mathematical Society, 2020

192/ Solutions

the given inequality is equivalent to

…  
X 1 − x2 1 X yz(1 − x2 )
≤√ or ≤ 1. (1)
cyc
2x xyz cyc
2

By the AM-GM inequality, we have

 
yz(1 − x2 ) 1 1 − x2 1
Å ã
≤ yz + = (2yz + 1 − x2 ).
2 2 2 4
Similarly,  
zx(1 − y 2 ) 1
≤ (2zx + 1 − y 2 ),
2 4
and  
xy(1 − z 2 ) 1
≤ (2xy + 1 − z 2 ).
2 4
Adding up the three inequalities above then yields
 
X yz(1 − x2 ) 1
2(xy + yz + zx) + 3 − (x2 + y 2 + z 2 )


≤
cyc
2 4
1
= (5 − (x2 + y 2 + z 2 ))
4
1
≤ (5 − (xy + yz + zx)) = 1,
4
so (1) holds and the proof is complete.

4489. Proposed by Arsalan Wares.

Regular hexagon A has its vertices at points A1 , A2 , A3 , A4 , A5 and A6 . Six
circular congruent arcs are drawn inside hexagon A and all six pass through the
center of A. The terminal points of each of the six arcs divide the sides of A in the
ratio 3 : 7. The six regions within A that are bounded only by circular arcs have
been shaded. Find the ratio of the area of A to the area of the shaded region.

Crux Mathematicorum, Vol. 46(4), April 2020

 Solutions /193

We received 17 correct solutions. We present the solution by Jason Smith.

Let the side length of the hexagon be 10. The area of the entire hexagon is
1 π √
Ahex = 6 · · 102 sin = 150 3.
2 3
Let C denote the center of the hexagon and B1 , B2 , ..., B6 the tips of the flower
petals. For the squared distance r2 between consecutive tips, the law of cosines
applied to triangle B1 A2 B2 gives
2π
r2 = 32 + 72 − 2 · 3 · 7 · cos = 79.
3
We can use this value of r2 to find the area of the of the two circular segments
with vertices B1 and C. Since the area of one of these circular segments is the
area of a sector of the circle minus the area of a triangle, we have
Ç √ å
1 2 1 2 1 π π  79 π 3
Aseg = r θ − r sin θ = · 79 − sin = − .
2 2 2 3 3 2 3 2

There are twelve such identical circular segments comprising the shaded region, so
√ å
√
Ç
79 π 3
Ash = 12 · − = 79(2π − 3 3).
2 3 2

Therefore, the ratio we seek is equal to

√
Ahex 150 3
= √ ≈ 3.025.
Ash 79(2π − 3 3)

4490. Proposed by Borislav Mirchev and Leonard Giugiuc.

A line ` through the orthocenter H of the acute triangle ABC meets the cir-
cumcircle at points K on the smaller arc AC and L on the smaller arc BC. If
M, N , and P are the feet of the perpendiculars to ` from the vertices A, B, and
C, respectively, prove that P H = |KM − LN |.
All 7 of the solutions we received were complete and correct, although four of them
failed to express clearly the result that they were proving. We feature a composite
of the independent solutions by Michel Bataille and by J. Chris Fisher.
As stated, the problem is not correct: see the accompanying figure where P H =
KM + LN . Moreover, the statement is unnecessarily restrictive. We shall show
the following result, which remains true independent of the figure:
A line ` through the orthocenter H of an arbitrary triangle ABC meets
the circumcircle at points K and L. If M, N, and P are the feet of the
perpendiculars to ` from the vertices A, B, and C, respectively, then
−−→ −−→
P H = kKM + LN k.

Copyright © Canadian Mathematical Society, 2020

194/ Solutions

−−→ −−→
Therefore, we have that P H = KM + LN if KM and LN have the same sense,
and P H = |KM − LN | otherwise.

For the proof, we can assume a unit circumradius, place the circumcenter at the
origin, and rotate the figure about O so that ` is parallel to the x-axis. If the
coordinates of vertices are

A(a, a0 ), B(b, b0 ), and C(c, c0 ),

where a2 + a02 = b2 + b02 = c2 + c02 = 1, then the orthocenter has coordinates

H(a + b + c, a0 + b0 + c0 ).

By assumption, ` is the line y = a0 + b0 + c0 . The x-coordinates of M, N , and P

will therefore be a, b, and c, respectively, while the x-coordinates of K and L have
values k and −k, which we have no need to calculate. Thus,

P H = |(a + b + c) − c| = |a + b|.
−−→
When M and L are both inside or both outside the circumcircle, the vectors KM
−−→
and LN point in opposite directions, whence the quantities k − a and −k − b have
opposite signs. It follows that

|KM − LN | = |k − a + (−k − b)| = |a + b| = P H.

When one of M or N is inside the circumcircle and the other outside, then k − a
and −k − b have the same sign, and

KM + LN = |k − a + (−k − b)| = |a + b| = P H,

as claimed.

Crux Mathematicorum, Vol. 46(4), April 2020

 Solutions /195

Editor’s comments. When the editors revised the statement of the proposers’
problem, their assumption that both M and N lie inside the circumcircle was
unfortunately omitted. There seems to be no obvious criteria for predicting the
relative behaviour of M and N . It is easy to check that the assumption of acute
angles is not sufficient. Consider, for example, an isosceles triangle ABC with
CA = CB. When ∠C = 60◦ (and H = O) then for all choices of the point K, `
is a diameter, and both M and N are on or inside the circumcircle. But should
∠C > 60◦ (and H be between C and O as in the above figure) then K can be
chosen sufficiently close to A on the arc AC that omits B so that M is outside the
circumcircle while N is inside. (This serves as a counterexample to the problem
as revised by the editors.) Note that the value of the y-coordinate plays no role
in the argument so that an analogous result holds for lines parallel to `. Note,
finally, that when ∠C = 90◦ , we have a rather nice, easily proved result, namely
KM = CN (which is a consequence of our result for right triangles since P H = 0
and C = L).

Copyright © Canadian Mathematical Society, 2020