CRUXv 46 N 10

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
Philip McCartney Northern Kentucky University
Guest Editors Yagub Aliyev ADA University, Baku, Azerbaijan

Ana Duff Ontario Tech University
Andrew McEachern York University
Vasile Radu Birchmount Park Collegiate Institute
Aaron Slobodin University of Victoria
Chi Hoi Yip University of British Columbia
Editor-at-Large Bill Sands University of Calgary

Managing Editor Denise Charron Canadian Mathematical Society
IN THIS ISSUE / DANS CE NUMÉRO
482 Editorial Kseniya Garaschuk

483 MathemAttic: No. 20
483 Problems: MA96–MA100
486 Solutions: MA71–MA75
491 Folding Paper Geometry Abbas Galehpour Aghdam
495 Olympiad Corner: No. 388
495 Problems: OC506–OC510
497 Solutions: OC481–OC485
504 Problems: 4591–4600
509 Bonus Problems: B51 – B75
514 Solutions: 4541–4550
482/ Editorial
EDITORIAL
2020 has given us a lot to talk about, to think about, to reconsider, redo and
relearn. Hopefully, we have come out better on the other side, but only the future
will tell.
Year 2020 marked the 46th Volume of Crux. Thanks to the generous support
of our sponsors, Crux continues to flourish as an open access journal, gaining a
wider audience that includes high school students, teachers and numerous other
avid problem solvers of all ages and from all around the globe. The impact of the
journal has grown dramatically since it became freely available online last year.
Over the 12 month period from October 2019 to October 2020, the website has
been visited by over 11,500 unique users accessing the website over 40,000 times.
The number of submissions has also grown drastically to the point where we had
to increase the size of the Editorial Board to moderate the incoming volume of
solutions in a timely fashion. We now routinely receive around 200 submissions
per issue and growing: what a great problem to have!
This year was marked by several losses to the mathematical community. In March
2020, we lost legendary Richard K. Guy whose life and work influenced so many
of us. We dedicated issue 8 of this Volume to the memory of Richard Guy
and received an unprecedented number of submissions. As a result, the memo-
rial issue was the largest Crux issue to date with 103 pages to commemorate
103 years of Richard’s life. The issue was used in the University of Calgary’s
events in honour of Guy held October 1–4, 2020: https://science.ucalgary.
ca/mathematics-statistics/about/richard-guy
With all of our sections going strong, we are looking forward to 2021. The new
year will start with a new cover for Crux and a new regular column Exploring
Indigenous Mathematics.
Stay healthy.
Kseniya Garaschuk
Crux Mathematicorum, Vol. 46(10), December 2020

MathemAttic /483
MATHEMATTIC
No. 20
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 February 15, 2021.
MA96.
a) A circle passes through points with coordinates (0, 1) and (0, 9) and is tangent
to the positive part of the x-axis. Find the radius and coordinates of the centre of
the circle.
b) Let a and b be any real numbers of the same sign (either both positive or both
negative). A circle passes through points with coordinates (0, a) and (0, b) and is
tangent to the positive part of the x-axis. Find the radius and coordinates of the
centre of the circle in terms of a and b.
MA97. In London there are two notorious burglars, A and B, who steal
famous paintings. They hide their stolen paintings in secret warehouses at different
ends of the city. Eventually all the art galleries are shut down, so they start stealing
from each other’s collection. Initially A has 16 more paintings than B. Every week,
A steals a quarter of B’s paintings, and B steals a quarter of A’s paintings. After
3 weeks, Sherlock Holmes catches both thieves. Which thief has more paintings
by this point, and by how much?
MA98. A pair of telephone poles d metres apart is supported by two cables

which run from the top of each pole to the bottom of the other. The poles are 4
m and 6 m tall. Determine the height above the ground of the point T , where the
two cables intersect. What happens to this height as d increases?
Copyright © Canadian Mathematical Society, 2020

484/ MathemAttic
MA99. A flag consists of a white cross on a red field. The white stripes, both
vertical and horizontal, are of the same width. The flag measures 48cm by 24cm.
If the area of the white cross equals the area of the red field, what is the width of
the cross?
MA100. Suppose the equation x3 + 3x2 − x − 1 = 0 has real roots a, b, c.

Find the value of a2 + b2 + c2 .
.................................................................
Les problèmes 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 février 2021.
La rédaction souhaite remercier Rolland Gaudet, professeur titulaire à la retraite à
l’Université de Saint-Boniface, d’avoir traduit les problèmes.
MA96.
a) Un certain cercle passe par (0, 1) et (0, 9) et est tangent à l’axe des x dans sa
partie positive. Déterminer le rayon du cercle et les coordonnées de son centre.
b) Soient a et b deux nombres réels de même signe, les deux étant positifs ou les
deux étant négatifs. Un certain cercle passe par (0, a) et (0, b) et est tangent à l’axe
des x dans sa partie positive. Déterminer le rayon du cercle et les coordonnées de
son centre, en termes de a et b.
MA97. Deux cambrioleurs, A et B, ont la spécialité de voler des œuvres d’art

et de les cacher, chacun dans son propre entrepôt. Éventuellement, toutes les gal-
leries d’art ont été vidées et sont donc fermées. Les deux cabrioleurs commencent
alors à voler l’un de l’autre. À ce moment, A possède 16 œuvres d’art de plus

MathemAttic /485
que B. Par la suite, chaque semaine, A vole le quart des œuvres d’art de B et B
vole le quart de celles de A. Après 3 semaines, on attrappe les deux cambrioleurs.
Lequel cambrioleur a alors le plus d’œuvres d’art, et par combien ?
MA98. Deux poteaux de téléphone, de hauteurs 4 mètres et 6 mètres et à

d mètres de distance, sont stabilisés par deux cables, allant du haut de chaque
poteau jusqu’à la base de l’autre. Déterminer la distance au sol du point T où les
cables se rencontrent. Qu’arrive-t-il à cette hauteur lorsque d augmente ?
MA99. Un drapeau de taille 48 cm par 24 cm consiste d’une croix blanche sur

un fond rouge, les rayures blanches, l’une horizontale et l’autre verticale, étant de
même largeur. Si les surfaces rouge et blanche sont égales, déterminer la largeur
des rayures de la croix.
MA100. Supposer que l’équation x3 + 3x2 − x − 1 = 0 a les racine réelles a,

b, c. Déterminer la valeur de a2 + b2 + c2 .

486/ MathemAttic
MATHEMATTIC
SOLUTIONS
Statements of the problems in this section originally appear in 2020: 46(5), p. 199–210.
MA71. You are given a rectangle OABC from which you remove three
right-angled triangles, leaving a fourth triangle OP Q as shaded in the diagram
below.
How must you position the points P and Q so that the area of each of the three
removed triangles is the same? In other words, what are the ratios P B : P A and
QB : QC?
Originally Problem 2, Vermont State Mathematics Coalition Talent Search, 2009.
We received 12 submissions, all correct. We present the solution by T. Reji and
B. Sneha.
Let OA = a and OC = b be the sides of the given rectangle OABC. Let P and
Q be as given in the figure. Denote the length P B by x and the length BQ by y.
Then length AP = b − x and length QC = a − y.
We need to find the ratio P B : P A and QB : QC subject to the condition that
the area of the three triangles except OP Q are the same. That is, we need the
area of OAP to equal the area of P QB to equal the area of OCQ:
a xy b
(b − x) = = (a − y)
2 2 2
ab − ax = xy = ab − by (1)
The outer equalities of (1) give us ab − ax = ab − by or

ax
y= . (2)
b
Substituting (2) for y into the left-most equalities of (1), ab − ax = xy, gives

MathemAttic /487
x2
b−x= =⇒ x2 + bx − b2 = 0,
b
which has the solution
√
−b + b2 + 4b2 b √
x= = ( 5 − 1).
2 2
Resubstituting this value of x in (2) gives
a √
y= ( 5 − 1).
2
x y
With these values for x and y, the ratios P B : P A =and QB : QC =
b−x a−y
are both √ √
5−1 5+1
P B : P A = QB : QC = √ = ,
3− 5 2
which one recognizes to be the golden section.
MA72. Consider four numbers x, y, z and w. The first three are in arithmetic
progression and the last three are in geometric progression. If x + w = 16 and
y + z = 8, find all possible solutions (x, y, z, w).
Originally (modified) Problem 8, Vermont State Mathematics Coalition Talent
Search, 2009.
We received seven correct and complete and seven incomplete solutions. In each
of the incomplete solutions the case z = 0 was overlooked. We present the solution
by Joel Schlosberg, lightly edited.
Since x, y, z are in arithmetic progression and z = 8 − y,
x = 2y − z = 3y − 8.
Using that y, z, w are in geometric progression and the previous equation,
z 2 = yw = y(16 − x) = y(24 − 3y).
Therefore
0 = z 2 − z 2 = (8 − y)2 − y(24 − 3y) = 4(y − 2)(y − 8),
so either
y = 2 =⇒ (x, y, z, w) = (−2, 2, 6, 18)
or
y = 8 =⇒ (x, y, z, w) = (16, 8, 0, 0).

488/ MathemAttic
MA73. A checkerboard is “almost tileable” if there exists some way of placing

non-overlapping dominoes on the board that leaves exactly one square in each row
and column uncovered. (Note that dominoes are 2 × 1 tiles which may be placed
in either orientation.) Prove that, for n ≥ 3, an n × n checkerboard is almost
tileable if and only if n is congruent to 0 or 1 modulo 4.
We received 3 solutions. We present the solution by Richard Hess, edited.
Possible solutions for n = 4 and n = 5 are shown below.
For any n ≡ 0 or 1 mod 4, we can inductively construct a solution from the

(n − 4) × (n − 4) solution by appending the 4 × 4 solution to the bottom left corner
and packing the rest of the board with dominoes:
..
. Solution for
(n − 4) × (n − 4)
...
We now show that it is impossible to find a1 solution for n ≡ 2 or 3 mod 4. Colour

the squares of the board white and black alternatingly, with the top left corner
white. Place n pawns on squares in any arrangement such that each column and
row has a pawn; the squares with pawns on them will be our candidates for squares
which are not covered by dominoes in an almost tiling. Let w and b be the number
of pawns on white squares and black squares respectively. If all the pawns are on
the main white diagonal then clearly w − b = n. Any arrangement of the pawns
can be transformed into an arrangement with all the pawns on the diagonal by
repeatedly taking the pawns at locations (a, b) and (c, a) and moving them to (a, a)
and (c, b). Note that whenever we swap pawns from two different columns while
keeping the pawns in their respective rows then w − b will either change by 4 or
remain the same. Hence, for any arrangement of pawns we must have w − b ≡ n

1
MathemAttic /489
mod 4. Suppose now that the checkerboard is almost tileable; in any solution,
each domino covers a white square and a black square. We place the pawns on
the uncovered squares. If n is even there are as many white squares on the board
as black squares, so we must have w − b = 0; if n is odd, there is one more white
square than there are black squares, so we must have w − b = 1. This contradicts
the earlier assertion that w − b ≡ n mod 4.
MA74. A set of n distinct positive integers has sum 2015. If every integer in
the set has the same sum of digits (in base 10), find the largest possible value of
n.
We received three submissions, out of which one was correct and complete. We
present the solution by Corneliu Mănescu-Avram, modified by the editor.
The numbers 8, 17, 26, 35, 44, 53, 62, 71, 80, 107, 116, 125, 134, 143, 152, 161,
170, 206, and 305 add up to 2015 and the sum of the digits of each number is 8.
We deduce that n ≥ 19. Since all n integers have to be congruent to each other
modulo 9, their sum has to be at least
n(n − 1) n(9n − 7)
1 + 10 + 19 + · · · + (9n − 8) = 9 +n= .
2 2
From this it follows n < 22.
Let s be the common digit sum. Then ns ≡ 2015 ≡ 8 (mod 9). Thus n cannot
be 21. If n = 20, then 20s ≡ 8 (mod 9), thus s ≡ 4 (mod 9). The smallest 15
integers with digit sum 4 are 4, 13, 22, 31, 40, 103, 112, 121, 130, 202, 211, 220,
301, 310, and 400, which already sum to 2220. The smallest 15 integers with digit
sum 13 are 49, 58, 67, 76, 85, 94, 139, 148, 157, 166, 175, 184, 193, 229, and 238,
which already sum to 2058. If the digit sum is 22 or greater than the smallest
number is at least 499 and the sum of the n numbers must be greater than 2015.
MA75. At the Mathville Tapas restaurant, the dishes come in three types:
small, medium, and large. Each dish costs an integer number of dollars, with
the small dishes being the cheapest and the large dishes being the most expen-
sive. (Tax is already included, different sizes have different prices, and the prices
have stayed constant for years.) This week, Jean, Evan, and Katie order 9 small
dishes, 6 medium dishes, and 8 large dishes. When the bill arrives, the following
conversation occurs:
Jean: “The bill is exactly twice as much as last week.”
Evan: “The bill is exactly three times as much as last month.”
Katie: “If we gave the waiter a 10% tip, the total would still be less than $100.”
Find the price of the group’s meal next week: 2 small dishes, 9 medium dishes,
and 11 large dishes.

490/ MathemAttic

We received 9 submissions of which 4 were correct and complete. We present the
solution by Richard Hess.
Let the prices of dishes be 0 < S < M < L. The language of the problem implies
that S, M and L are distinct and it makes no sense that a small dish would be
free. We are told 9S + 6M + 8L = T < 91, where T is the total bill. We are
further told that T is even, determining that S must be even. Also, T is divisible
by 3, determining that L must be divisible by 3.
The only cases that satisfy the constraints are (S, M, L, T ) = (2, 3, 6, 84) and
(2, 4, 6, 90). The second is not possible since 90/2 = 45 is odd and we cannot
get an odd total when all prices are even. For the remaining case we can get a
total of 42 with 6 large dishes and a total of 28 with 14 small dishes. Many other
combinations achieve totals of 42 or 28. Therefore $97 is the price for 2 small
dishes, 9 medium dishes and 11 large dishes.

Abbas Galehpour Aghdam /491
Folding Paper Geometry

Abbas Galehpour Aghdam
In [1], Ian VanderBurgh discussed a paper folding problem from the UK Interme-
diate Challenge 1999, giving the following problem as a challenge to the readers:
Problem A rectangular sheet of paper ABCD has AB = 8 and BC = 6. The
paper is folded so that corner A coincides with the midpoint, M , of DC. What is
the length of fold?
We will solve this problem by examining four different approaches.
We first need to draw a diagram, but it is better to start with a practical thing.
Get a rectangular sheet of paper with sides 8 and 6 (or with dimensions in the
ratio 4:3), then fold it by following the method given in the problem:
Figure 1: Paper folding
Now, we can draw a suitable diagram as shown in Figure 2.
Figure 2: Basic diagram

492/ Folding Paper Geometry
Let’s redraw the diagram by A F

adding the edges of the full orig- B
inal rectangle (the dotted lines)
and labelling the relevant points as
shown in Figure 3.
We should calculate the length of
EF . How should we start? We
first focus on ∆DEM to deter-
mine the lengths of AE, DE and E
EM , then follow four different ap-
proaches.
Since EM is the folded image of D M C
AE, we have EM = AE. Since the
paper has width 6, then AD = 6 Figure 3: Complete diagram
and DE = 6−AE. Since the paper
has length 8 and ∆DEM is right-
angled at D, we know one of the three side lengths, namely, DM = 4, and we know
the other two side lengths in terms of AE, namely, EM = AE and DE = 6 − AE.
What should we do to determine the length of AE? Let us apply the Pythagorean
Theorem in ∆DEM . We obtain
13
EM 2 = DE 2 + DM 2 , AE 2 = (6 − AE)2 + 42 , AE = .
3
13 13 13 5
Since AE = , we have EM = and DE = 6 − = .
3 3 3 3
Method 1: Using area
Since AF becomes F M after folding, then AF = F M and ∆AEF ∼ = ∆EF M (be-

cause of side-side-side congruency), therefore ∆AEF is equal in area to ∆EF M .
This implies that the area of ABCD is equal to: 2× (the area of the ∆AEF ) +
the area of the ∆DEM + the area of M F BC. Thus
1 1 1
Å ã
8×6=2 AE · AF + DE · DM + BC(F B + M C);
2 2 2
13 10
48 = AF + + 3 (8 − AF ) + 4 ;
3 3
13
AF = .
2
We now have two of the three side lengths of ∆AEF , so we can use the Pythagorean
169 169 2197
Theorem to conclude that EF 2 = AF 2 +AE 2 = + = ; since EF > 0,
√ √ 4 9 36
2197 13 13
then EF = = .
6 6

Abbas Galehpour Aghdam /493
Method 2: Using trigonometry
In Figure 3, let ∠DEM = α and ∠EF M = θ. Since ∆AEF ∼

= ∆M EF , so
∠AF E = θ. This yields
π
∠AEF = − θ. (1)
2
In right-angle triangle ∆M EF , ∠M EF is complementary to ∠EF M , thus
π
∠M EF = − θ. (2)
2
We also have
∠AEF + ∠M EF + ∠DEM = π. (3)
With substitution (1) and (2) into (3), we get
α = 2θ. (4)
DM 12 12
Since ∆DEM , we have tan α = = , so tan 2θ = (because of (4)). This
DE 5 5
yields
2 tan θ 12
2 = , or 6 tan2 θ + 5 tan θ − 6 = 0.
1 − tan θ 5
2
If we solve this equation for tan θ, we get tan θ = . Moreover, in right ∆AEF ,
3
AE AE 2 3 3 13 13
we have tan θ = , therefore = or AF = AE = × = . The
AF AF 3 2 2 3 2
length of EF is calculated similarly.
Method 3: Using trigonometry (again)
In Figure 3, connect AM as shown in the Figure 4.

A F
B
θ
E
α
D M C
Figure 4: Diagram with angles

494/ Folding Paper Geometry
Notice that AD = 6 and√DM = 4. Applying the Pythagorean Theorem to right

∆ADM yields AM = 2 13. Now, let’s do some angle-chasing in ∆AF M . We
know that ∠AF E = ∠EF M = θ, and this yields ∠AF M = 2θ = α. Moreover,
π − ∠AF M π α
since ∆AF M is isosceles, we have ∠F AM = ∠F M A = = − .
2 2 2
Applying the Sine Law on ∆AF M yields:
AF AM
= ;
sin(∠F M A) sin(∠AF M )
AF AM
= ;
sin π2 − α2 sin α
α AM
AF = cos · . (5)
2 sin α
DM 12 DE 5
In ∆DEM , we have sin α = = and cos α = = . Using double-
EM 13 EM 13 √
2 α 9 α α 3 13
angle identity, we have cos = ; since cos > 0, then cos = . Sub-
2 13 2 2 13
13
stituting these values into (5), we get AF = . The length of EF is determined
2
similarly to method 1.
Method 4: Using similar triangles

π α α
We know that ∠F AM = − (Figure 4), so ∠DAM = . This implies that
2 2 2
α
∠DAM = = θ = ∠EF M,
2
AM DM
therefore ∆EF M is similar to ∆ADM , so = . Rearranging yields
EF EM
EM
EF = AM × , so
DM
√
√ 13
3 13 13
EF = 2 13 × = .
4 6
References
[1] I. Vander Burgh, Problem of the month. Crux Mathematicorum with Mathe-
matical Mayhem, Volume 36 (4), April, 2010.

OLYMPIAD CORNER /495
OLYMPIAD CORNER
No. 388
The problems featured in this section have appeared in a regional or national mathematical
Olympiad.
Click here to submit solutions, comments and generalizations to any

problem in this section
OC506. A quadrilateral is called convex if the lines given by its diagonals

intersect inside the quadrilateral. A convex quadrilateral has side lengths 3, 3, 4,
4 not necessarily in this order, and its area is a positive integer. Find the number
of non-congruent convex quadrilaterals having these properties.
OC507. There are 2n consecutive integers written on a blackboard. In each

move, they are divided into pairs and each pair is replaced with their sum and
their difference, which may be taken to be positive or negative. Prove that no 2n
consecutive integers can appear on the board again.
OC508. Let ABC be an isosceles triangle with AC = BC, whose incenter

is I. Let P be a point on the circumcircle of the triangle AIB lying inside the
triangle ABC. The lines through P parallel to CA and CB meet AB at D and
E, respectively. The line through P parallel to AB meets CA and CB at F and
G, respectively. Prove that the lines DF and EG intersect on the circumcircle of
the triangle ABC.
OC509. Prove that for any odd prime p the number of positive integers n
satisfying p|n!+1 is smaller than or equal to cp2/3 where c is a constant independent
of p.
OC510. 2019 points are chosen independently and uniformly in the unit disc
{(x, y) ∈ R2 | x2 + y 2 ≤ 1}. Let C be the convex hull of the chosen points. Which
probability is larger: that C is a polygon with three vertices, or a polygon with
four vertices?
.................................................................

496/ 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.
OC506. Un quadrilatère est dit convexe si les lignes associées aux diagonale
intersectent à l’intérieur du quadrilatère. Or un certain quadrilatère convexe a des
côtés de longueurs 3, 3, 4 et 4, pas nécessairement dans cet ordre ; de plus, sa
surface est donnée par un entier positif. Déterminer le nombre de quadrilatères
covexes non congrus ayant ces proriétés.
OC507. Sur un tableau à craie sont écrits 2n entiers consécutifs. Par la suite,
on regroupe ces entiers en paires, puis chaque paire est effacée et remplacée par
deux nouveaux entiers, la somme et la différence des deux entiers de la paire, où
la différence peut être prise en positif ou en négatif. Démontrer que quel que soit
le nombre de fois qu’on répète ce processus, on ne verra jamais réapparaı̂tre 2n
entiers consécutifs.
OC508. Soit ABC un triangle isocèle tel que AC = BC et soit I le centre

de son cercle inscrit. Soit P un point sur le cercle circonscrit du triangle AIB,
se situant à l’intérieur du triangle ABC. Les lignes passant par P et parallèles à
CA et CB rencontrent AB en D et E, respectivement. La ligne passant par P et
parallèle à AB rencontre CA et CB en F et G, respectivement. Démontrer que
les lignes DF et EG intersectent en un point qui se trouve sur le cercle circonscrit
du triangle ABC.
OC509. Démontrer que pour tout nombre premier impair p, le nombre

d’entiers positifs n tels que p|n! + 1 est plus petit ou égal à cp2/3 , où c est une
constante indépendante de p.
OC510. 2019 points sont choisis de façon aléatoire et indépendante les

uns des autres, selon une distribution uniforme dans le disque unitaire {(x, y) ∈
R2 | x2 + y 2 ≤ 1}. Soit C l’enveloppe convexe de ces points. Laquelle probabilité
est la plus élevée : C est un polygone à 3 sommets, ou C est un polygone à 4
sommets?

OLYMPIAD CORNER
SOLUTIONS
OC481. In the plane, there are circles k and l intersecting at points E and
F . The tangent to the circle l drawn from E intersects the circle k at point H
(H 6= E). On the arc EH of the circle k, which does not contain the point F ,
choose a point C (E 6= C 6= H) and let D be the intersection of the line CE with
the circle l (D 6= E). Prove that triangles DEF and CHF are similar.
Originally Czech-Slovakia Math Olympiad, 3rd Problem, Category B, Regional
Round 2017.
We received 12 submissions. We present the solution by the UCLan Cyprus Prob-
lem Solving Group.
Since C, E, F, H are concyclic, with C on the opposite side of EF than H, then
∠CHF = 180◦ − ∠CEF = ∠F ED .
We also have
∠F CH = ∠F EH = ∠EDF .
Here, the first equality follows since C, E, F, H concyclic, and the second by the
Chord-Tangent Theorem as HE is tangent to ` at E. So, triangles DEF and
CHF have equal angles, which means that they are similar.

OC482. Let a1 , a2 , . . . , a2017 be real numbers such that
a1 + a2 + · · · + a2017 = 2017.
Find the largest number of pairs (i, j) for which 1 ≤ i < j ≤ 2017 and ai + aj < 2.
Originally Bulgaria Math Olympiad, 4th Problem, Grade 11, Second Round 2017.
We received 8 submissions of which 7 were correct and complete. We present 2
solutions.
Solution 1, by UCLan Cyprus Problem Solving Group.
Ç å
2016
We claim that we can have pairs but no more.
2
This number
Ç of åpairs is achieved by taking a1 = · · · = a2016 = 0 and a2017 = 2017.
2016
Then all pairs not including a2017 has sum less than 2.
2
Let us now show that we cannot have more pairs. Equivalently, we will show that
at least 2016 pairs do not have the property.
We may assume that a1 6 · · · 6 a2017 . For 1 6 i 6 1008 we define
bi = ai + a2018−i .
We also define b1009 = 2a1009 . Then

b1009
b1 + · · · + b1008 + = a1 + · · · + a2017 = 2017 = 2 · 1008 + 1 .
2
So there is at least one 1 6 i 6 1009 such that bi > 2.
Note that if bi > 2, then for each j > i and k > 2018 − i we have
aj + ak > ai + a2018−i = bi > 2 .
There are Ç å
i (4035 − 3i)i
(2018 − i)i − i − =
2 2
such pairs (j, k) with j < k. (Since there are a total of (2018 − i)i pairs
Ç å (j, k) with
i
j > i and k > 2018 − i, out of which i of them satisfy j = k and of them
2
satisfy j > k.)
We have
(4035 − 3i)i (i − 1)(4032 − 3i)
− 2016 = >0
2 2
with equality if and only if i = 1. So at least 2016 pairs do not have this property
as required.

Solution 2, by Roy Barbara.

More generally, let n ≥ 6 and a1 , a2 , . . . , an ∈ R with ni=1 ai = n. Then, the
P
largest number of pairs (i, j), 1 ≤ i < j ≤ n such that ai + aj < 2 is exactly
Ç å Ç å
n n−1
− (n − 1) = (1)
2 2
For 1 ≤ i < j ≤ n, call (i, j) a green pair (a red pair) if ai + aj < 2 (if ai + aj ≥ 2).
Note that the number of green (red) pairs is invariant by any rearrangement of
the ai ’s, so we may assume, when needed, that the ai ’s are ordered as a1 ≤ a2 ≤
. . . ≤ an .
The number n − 1 of red pairs (i, j) is reached with a1 = a2 = . . . = an−1 = 0 and
an = n.
Next we show that in any configuration a1 , a2 , . . . , an the number of red pairs is
at least n − 1 and (1) will follow.
Lemma. Let n ≥ 6 and a1 , a2 , . . . , an ∈ R such that ni=1 ai = n. Then, either
P
there are at least n − 1 red pairs or there is a red pair (i, j) with ai ≤ 1.
Proof. If n = 2k, k ≥ 3, we take a1 ≤ a2 ≤ . . . ≤ a2k such that 2k a = 2k and
P
P2k i=1 i
if n = 2k + 1, k ≥ 3, we take a0 ≤ a1 ≤ . . . ≤ a2k such that i=0 ai = 2k + 1. We
have two cases.
(i) ak > 1. Then, 1 < ak ≤ ak+1 ≤ . . . ≤ a2k clearly shows that the number of
k+1 k(k + 1)
red pairs is at least k+1

2 . As k ≥ 3, then 2
≥ 2, so
2
≥ 2k, that
k+1

is 2 ≥ 2k, whence 2k ≥ n − 1.
(ii) ak ≤ 1. We claim that ak + a2k ≥ 2 (and we are done with the red pair
(k, 2k)). Otherwise, since a1 ≤ aP 2 ≤ . . . ≤ a2k , we would get ai P
+ak+i < 2 for
i = 1, 2, . . . , k. Adding, we get ki=1 (ai + ak+i ) < 2k, that is 2k i=1 ai < 2k.
If n = 2k, this is a contradiction. If n = 2k+1, adding a0 ≤ 1 (that obviously
holds) to this inequality would imply 2k i=0 ai < 2k + 1, contradiction.
P
We are now ready to prove by induction on n that in any configuration a1 , a2 , . . . , an

the number of red pairs is at least n − 1.
P6
For n = 6, let a1 ≤ a2 ≤ . . . ≤ a6 be real numbers
P6 with i=1 ai = 6. If we had
a1 + a6 < 2, a2 + a5 < 2 and a3 + a4 < 2, then i=1 ai < 6, contradiction. Hence,
we must have either a1 + a6 ≥ 2 or a2 + a5 ≥ 2 or a3 + a4 ≥ 2.
If a1 + a6 ≥ 2, the following pairs are clearly red: (1, 6), (2, 6), (3, 6), (4, 6), (5, 6).
If a2 + a5 ≥ 2, the following pairs are clearly red: (2, 5), (2, 6), (3, 5), (3, 6), (4, 5),
(4, 6), (5, 6).
If a3 + a4 ≥ 2, the following pairs are clearly red: (3, 4), (3, 5), (3, 6), (4, 5), (4, 6),
(5, 6).

In any case, there are at least 5 red pairs, and the number 5 is reached with
a1 = a2 = a3 = a4 = a5 = 0, a6 = 6.
Now, assume that the thesis holds for some n ≥ 6. Let a0 , a1 , . . . , an ∈ R with
n
X
ai = n + 1.
i=0
By the lemma, either there are at least n red pairs (and we are done) or there
is a red pair (i, j) with ai ≤ 1. In this case, without loss of generality, we may
assume that this red pair is (0, n). Thus, we have a0 ≤ 1 and a0 + an ≥ 2 for
i = 0, 1, . . . , n − 1 and bn = an + a0 − 1 (bn ≤ an ). Then,
n
X
bi = n.
i=1
By the inductive hypothesis, there are at least n − 1 pairs (i, j), 1 ≤ i < j ≤ n
with bi + bj ≥ 2. Since ai ≥ bi for i = 1, 2, . . . , n, then the same n − 1 pairs (i, j)
satisfy ai + aj ≥ 2. If we add the red pair (0, n), we obtain (at least) n red pairs
for the sequence a0 , a1 , . . . , an .
OC483. Prove that for each prime number p > 2, there is exactly one positive
integer n such that the number n2 + np is a perfect square.
Originally Poland Math Olympiad, 1st Problem, Second Round 2017.
We received 23 submissions, of which 22 were correct and complete. We present 3
solutions.
Solution 1, by Fernando Ballesta Yagüe.
Let n be a positive integer, and p a given prime, p > 2. Then, n2 + np is a perfect
square if and only if n2 + np = k 2 for some positive integer k, i.e. if and only if
4n2 + 4np = 4k 2 for some positive integer k. Then,
(2n + p)2 − p2 = 4k 2 ⇐⇒ (2n + p)2 − 4k 2 = p2 ,
i.e.
(2n + p − 2k)(2n + p + 2k) = p2 .
Since p is a prime number, the only decompositions of p2 as a product of two
natural numbers are p · p and 1 · p2 . As k > 0,
2n + p + 2k > 2n + p − 2k,
so it cannot be
2n + p + 2k = 2n + p − 2k = p.
Therefore, it must be 2n + p + 2k = p2 , 2n + p − 2k = 1. Adding them up:
p2 − 2p + 1 p−1 2
Å ã
4n + 2p = p2 + 1 ⇐⇒ n = =
4 2

p−1 2
Å ã
So n = is the only value that makes n2 + np a perfect square. Notice
2
that p > 2 is prime, so it is an odd number, so p − 1 is even, which implies that n
is a positive integer.
Solution 2, by Vincent Chan.

If n ≡ 0 (mod p), say n = pk for an integer k, then
n2 + np = p2 k 2 + p2 k = p2 (k 2 + k).
But this cannot be a perfect square, since
k 2 < k 2 + k < k 2 + 2k + 1 = (k + 1)2 .
Therefore, gcd(n, p) = 1. In this case, gcd(n, p + n) = 1, so n2 + np = n(n + p) is

a perfect square if and only if both n and n + p are perfect squares, say
n = a2 ,
n + p = b2 .
Subtracting, this gives p = b2 − a2 = (b + a)(b − a). But p is prime, hence b − a = 1

and b + a = p, yielding
p−1
a= .
2
Therefore, the unique value of n for which n2 + np is a perfect square is
(p − 1)2
n= .
4
Solution 3, by UCLan Cyprus Problem Solving Group.

Assume that n2 + np = m2 . Then np = (m − n)(m + n). Since m + n > n, then
p > m − n. Since p is prime, then p|(m − n) or p|(m + n). But p > m − n, so
p|(m + n). So we can write m = kp − n for some positive integer k.
Then np = (kp − 2n)kp which gives k 2 p = n(2k + 1). Since k 2 and 2k + 1 are
coprime (if a prime q divides k 2 and 2k + 1, then it also divides k and therefore 1
as well, a contradiction) then k 2 |n. So n = k 2 r for some positive integer r. But
then k 2 p = k 2 r(2k + 1), so p = r(2k + 1). Since p is prime and 2k + 1 > 1, then
r = 1. This gives p = 2k + 1 and n = k 2 .
p−1 2
Å ã
So if n2 + np is a perfect square, then we must have n = . Since p is odd
2
Å 2 ã2
2 p −1
then this is an integer and it is easy to check that in this case n +np =
4
is a perfect square.

OC484. Let x be a real number with 0 < x < 1 and let 0.c1 c2 c3 . . .
be the decimal expansion of x. Denote by B(x) the set of all subsequences
of c1 , c2 , c3 . . . that consist of six consecutive digits. For instance, B(1/22) =
{045454, 454545, 545454}.
Find the minimum number of elements of B(x) as x varies among all irrational
numbers with 0 < x < 1.
Originally Italy Math Olympiad, 5th Problem, Final Round 2018.
We received 5 submissions. We present the solution by UCLan Cyprus Problem
Solving Group.
Given n ∈ N and x ∈ (0, 1) we write Bn (x) for the set of all subsequences of n
consecutive digits in the decimal expansion of x. This is uniquely defined for an
irrational x. We will show that |Bn (x)| > n + 1. This is best possible as
x = 0. 0 · · 0} 1 |0 ·{z
| ·{z · · 0} 1 |0 ·{z
· · 0} 1 · · ·
n n+1 n+2
is irrational with |Bn (x)| = n + 1. To see the irrationality of x note that for
any k, the decimal expansion of x cannot have period of k as there are infinitely
many 1’s in the decimal expansion which are followed by k or more 0’s. To see
the second claim about the size of Bn (x) we just note that Bn (x) consists of all
possible sequences with at most one digit equal to 1 and all the other digits equal
to 0).
We proceed to prove our claim by induction on n. The case n = 1 is trivial
as x 6= 0. Assume that |Bk (x)| > k + 1 for each irrational x ∈ (0, 1). For
the inductive step, we want to show that |Bk+1 (x)| > k + 2 for each irrational
x ∈ (0, 1). So assume for contradiction that there is an irrational y ∈ (0, 1) such
that |Bk+1 (y)| 6 k + 1. Given a = a1 a2 · · · ak ∈ Bk (y) we write Bk+1 (y; a) for all
elements of Bk+1 (y) which begin with a. Then |Bk+1 (y; a)| > 1 for each a ∈ Bk (y)
and Bk+1 (y; a) ∩ Bk+1 (y; b) = ∅ for each a 6= b. Since |Bk (y)| > k + 1, we get that
|Bk+1 (y)| > k + 1. Therefore |Bk+1 (y)| = k + 1 and the equality occurs if and only
if |Bk+1 (y; a)| = 1 for each a ∈ Bk (y). But in this case, every k consecutive digits
in the decimal expansion of y completely determine the next digit. We will show
that in this case y is rational. This contradiction completes the induction step
and therefore our claim that |Bn (x)| > n + 1 for each n ∈ N and each irrational
x ∈ (0, 1).
We want to show that given the decimal expansion of y ∈ (0, 1), if every k con-
secutive digits of y completely determine the next digit, then y is rational. Let
y = 0.c1 c2 c3 . . .. Since there are only finitely many sequences of k consecutive
digits that can occur, then there are r < s such that cr+1 = cs+1 , . . . , cr+k = cs+k .
Then cr+k+1 = cs+k+1 as every k consecutive digits completely determine the next
one. Inductively, we can now easily get cr+n = cs+n for every n ∈ N showing that
y is rational as required.
Therefore the minimum possible number of elements of B(x) = B6 (x) is 7.

OC485. Prove that a continuous function f : R → R is increasing if and only

if Z b Z c
(c − b) f (x) dx ≤ (b − a) f (x) dx,
a b
for all real numbers a < b < c.
Originally Romania Math Olympiad, 3rd Problem, Grade 12, District Round 2018.
We received 12 submissions. We present the solution by Corneliu Avram Manescu.
If f is increasing and a < b < c, then
Z b Z c
(c − b) f (x) dx ≤ (c − b)(b − a)f (b) = (b − a)(c − b)f (b) ≤ (b − a) f (x) dx.
a b
Conversely, if a and b are real numbers, a < b, let F : R → R be the primitive of

f . If x and y are real numbers such that a < x < y < b, then the given relation
implies
F (x) − F (a) F (b) − F (y)

f (a) = F 0 (a) = lim+ ≤ lim− = F 0 (b) = f (b).
x→a x−a y→b b−y
Remark from the solver. The condition that the function f is continuous is not
necessary: for the direct implication the monotony of f implies the integrability
of f on every compact interval and for the converse implication it suffices that f
have primitives on R.

504/ Problems
PROBLEMS
Click here to submit problems proposals as well as solutions, comments
and generalizations to any problem in this section.
4591. Proposed by Pericles Papadopoulos.

Let point P be inside triangle ABC and let A0 , B 0 and C 0 be the points where the
internal bisectors of ∠BP C, ∠CP A and ∠AP B intersect sides BC, CA and AB,
respectively.
Show that lines AA0 , BB 0 and CC 0 concur at a point K satisfying

1 1
Å ã
AK
= PA + .
KA0 PB PC
4592. Proposed by Michel Bataille.

Let ABC be a triangle with ∠BAC 6= 90◦ and let O be its circumcentre. Let γ
be the circumcircle of ∆BOC. The perpendicular to OA at O intersects γ again
at M and the line AM intersects γ again at N . Prove that
NO 2OA2
= .
NA AB · AC
4593. Proposed by Diaconu Radu.

Solve the system of equations in real numbers:
 2

 a + bc = 7,
ab + bd = 3,


 ac + dc = 2,
bc + d2 = 6.


Problems /505
4594. Proposed by Nguyen Viet Hung.

Prove that for any point M on the incircle of triangle ABC,
M A2 M B2 M C2
+ + = 2R + r,
ha hb hc
where ha , hb and hc are the lengths of the altitudes from A, B and C respectively,
while R and r denote circumradius and inradius, respectively.
4595. Proposed by Nguyen Viet Hung.

»
Let n > 2 be an integer and let Sn = nk=2 1 + 2
Determine bSn c.
P
k2 .
4596. Proposed by Boris C̆olaković.

Let a, b, c be the lengths of the sides of triangle ABC with inradius r and circum-
radius R. Show that
a b c R 1
+ + ≤ −
b+c a+c a+b r 2
4597. Proposed by George Apostolopoulos.

Let a, b, c be positive real numbers with a + b + c = 1. Prove that
3
a2 + b2 + c2 + 1 1 1 ≥ 2(ab + bc + ca).
a + b + c
4598. Proposed by George Stoica.

Let P (z) be a polynomial of degree n with complex coefficients and with no zeroes
z satisfying |z| < 1. Prove that |P (z)| ≤ 2n |P (rz)| for all |z| ≤ 1 and 0 < r < 1.
4599. Proposed by Albert Natian.

The sum of squares of the sides of a triangle ABC is 133. By enlarging two sides
of ABC by a factor of 27, and a third side by a factor of 8, ABC is deformed into
a larger but similar triangle RST whose area is 324 times that of ABC. Find the
side lengths of ABC.
4600. Proposed by Semen Slobodianiuk, modified by the Editorial Board.

It is known (for example, by a formula of Euler, often attributed to Nicolas Fuss,
giving the distance between the centers in terms of the two radii) that given a
bicentric quadrilateral inscribed in one circle and circumscribed about a second,
then every point A of the circumcircle is the vertex of a bicentric quadrilateral
ABCD that is inscribed in the first circle, and circumscribed about the second.
Determine the locus of the centroid of the vertex set {A, B, C, D} as the bicentric
quadrilateral ABCD travels around the first circle while its sides stay tangent to
the second.

506/ 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.

4591. Proposée par Pericles Papadopoulos.
Soit P un point à l’intérieur du triangle ABC et soient A0 , B 0 et C 0 les points

où les bissectrices internes de ∠BP C, ∠CP A et ∠AP T intersectent les côtés BC,
CA et AB, respectivement.
Démontrer que les lignes AA0 , BB 0 et CC 0 sont concourantes en un point K tel

que
1 1
Å ã
AK
= P A + .
KA0 PB PC
4592. Proposée par Michel Bataille.
Soit ABC un triangle tel que ∠BAC 6= 90◦ et soit O le centre de son cercle
circonscrit. Aussi, soit γ le cercle circonscrit de ∆BOC. Enfin, la perpendiculaire
vers OA en O intersecte γ de nouveau en M et la ligne AM intersecte γ de nouveau
en N. Démontrer que
NO 2OA2
= .
NA AB · AC

Problems /507
4593. Proposée par Diaconu Radu.

Résoudre le système d’équations suivant pour les nombres réels:
 2

 a + bc = 7,
ab + bd = 3,


 ac + dc = 2,
bc + d2 = 6.

4594. Proposée par Nguyen Viet Hung.

Démontrer que pour tout point M sur le cercle inscrit de ABC, la suivante tient :
M A2 M B2 M C2
+ + = 2R + r
ha hb hc
où ha , hb et hc sont les longueurs des altitudes émanant de A, B et C respective-
ment, et R et r sont les rayon du cercle circonscrit et du cercle inscrit, respective-
ment.
4595. Proposée par Nguyen Viet Hung.

n …
X 2
Soit n > 2 un entier et soit Sn = 1 + 2 . Déterminer bSn c.
k
k=2
4596. Proposée par Boris C̆olaković.

Soit a, b et c les longueurs des côtés d’un triangle ABC dont le cercle inscrit et le
cercle circonscrit ont les rayons r et R respectivement. Démontrer que
a b c R 1
+ + ≤ −
b+c a+c a+b r 2
4597. Proposée par George Apostolopoulos.

Soient a, b, c des nombres réels positifs tels que a + b + c = 1. Démontrer que
3
a2 + b2 + c2 + 1 1 1 ≥ 2(ab + bc + ca).
a + b + c
4598. Proposée par George Stoica.

Soit P (z) un polynôme de degré n à coefficients complexes avec aucune racine z
telle que |z| < 1. Démontrer que |P (z)| ≤ 2n |P (rz)| pour tout |z| ≤ 1 et 0 < r < 1.

508/ Problems
4599. Proposée par Albert Natian.

La somme des carrés des côtés d’un triangle ABC est 133. En allongeant deux
côtés de ABC par un facteur de 27 et le troisième par un facteur de 8, ABC est
transformé en un triangle similaire RST dont la surface est 324 celle de ABC.
Déterminer les longueurs des côtés de ABC.
4600. Proposée par Semen Slobodianiuk, modifé par le comité éditorial.

C’est un fait connu (par exemple, par une formule d’Euler, souvent attribuée à
Nicolas Fuss, donnant la distance entre centres en termes de rayons) : pour tout
quadrilatère bicentrique inscrit dans un cercle et circonscrit autours d’un second,
chaque point A sur le cercle circonscrit est le sommet d’un quadrilatère bicentrique
ABCD inscrit dans le premier cercle et circonscrit autours du second. Déterminer
le lieu géométrique de l’ensemble de sommets {A, B, C, D} lorsque le quadrilatère
bicentrique ABCD se déplace autours du premier cercle, tout en faisant en sorte
que ses côtés restent tangents au second.

Bonus Problems /509
BONUS PROBLEMS
These problems appear as a bonus. Their solutions will not be considered for publication.
B51. Proposed by Leonard Giugiuc.

Let a ∈ (1, e) be a fixed real number. Find
1
Ä ä
sup a−x + a− x .
x>0
B52. Proposed by Dao Thanh Oai and Leonard Giugiuc.

Let ABCD be a convex quadrilateral and let E, H, G, F be the midpoints of AB,
BC, CD, DA. Show that M N OP is a parallelogram and area(ABCD)/area(KJIL)=9.

Let S be a point in the interior of triangle ABC. Prove that for the existence
and uniqueness of a triangle XY Z for which (i) its sides Y Z, ZX, XY contain the
points A, B, C, respectively, and (ii) the lines XA, Y B, and ZC are concurrent in
S, it is sufficient that the largest of the three areas [SAB], [SBC], [SCA] is less
than the sum of the other two.

Let ABCD be a convex quadrilateral with AB p = a, BC = b, CD = c, DA = d
and s = a+b+c+d
2 . Prove that if ab + cd = 2 (s − a)(s − b)(s − c)(s − d), then
p p
there √exist positive numbers
√ x, y, z such that a = x2 + y 2 , b = z x2 + y 2 ,
c = x 1 + z 2 and d = y 1 + z 2 .
B55. Proposed by Mihaela Berindeanu.

In M ABC, ] (BAC) = 60◦ , O = circumcenter, I = incenter and H = orthocenter.
If BO k HI, find ] (HAO).

510/ Bonus Problems
B56. Proposed by Arsalan Wares

In figure below, two circles touch each other internally at point P . Line segments
P C and P D are chords to the larger circle that intersect the smaller circle at
points A and B, respectively. Point M and N are on the larger circle so that
points M, A, B and N are collinear. Suppose BD : AC = 3 : 4, BN = 10, and
AM = 20. Find the exact length of AB.

Given the acute triangle ABC and its circumcircle Γ, define E to be the point
where the bisector of the angle at C intersects the tangent to Γ at B. Furthermore,
for the midpoint F of AB define X = EF ∩ AC, and Y to be the second point
where BX intersects Γ. Prove that Y B = Y A.

Let A0 and B 0 be the points where the tangents to a given circle at B and A,
respectively, meet the tangent that touches the circle at C. If O is the center of
the circle, define
X = AB ∩ OB 0 , Y = AB ∩ OA0 , and Z = A0 X ∩ B 0 Y.
Prove that the circle determined by X, Y , and the midpoint of A0 B 0 also contains
the midpoint of OZ.

Let ai with 1 ≤ i ≤ 6 be real numbers such that
6 6
X 15 X 45
ai = and a2i = .
i=1
2 i=1
4
Find the extrema for the expression

X
ai aj ak al .
1≤i<j<k<l≤6

Bonus Problems /511

√
Let k be a real number with k ≥ 3. Consider non-negative real numbers a, b, c
such that
ab + bc + ca + (k 2 − 3)abc = k 2 .
Prove that
a + b + c + (2k − 3)abc ≥ 2k.

Let a, b and c be positive real numbers. Prove that
p p p
a2 c2 − abc + b2 + b2 c2 − abc + a2 ≥ a2 + ab + b2 .
When does the equality hold?
B62. Proposed by Leonard Giugiuc and Daniel Dan.

Find all real numbers k such that for all triangles ABC we have
3(1 + k)
cos2 A + cos2 B + cos2 C + k(cos A cos B + cos A cos C + cos B cos C) ≥ .
4
B63. Proposed by Leonard Giugiuc and George Apostolopoulos.

Prove that in any triangle ABC with the customary notations, we have
m2a m2b m2c 9r

+ + ≥ .
b2 + c2 c2 + a2 a2 + b2 4R

Let 0 ≤ a < b < c ≤ √1 and a + b + c = 1. Find the best upper bound for
3
arctan a − arctan b arctan b − arctan c arctan c − arctan a

+ + .
a−b b−c c−a
B65? . Proposed by Leonard Giugiuc.

Let n ≥ 4 be an integer and let ai for i = 1, 2, . . . , n be positive real numbers.
Prove or disprove the following:
Å 2 √ √
a + a22 + · · · + a2n n( n − n − 1) a1 + a2 + · · · + an
ã
(n − 1) 1 + 1 1 1 ≥
√ .
n a1 + a2 + · · · + an
n

512/ Bonus Problems
B66. Proposed by Leonard Giugiuc and Nguyen Viet Hung.

Let ABC be a non-obtuse angles triangle, none of whose angles are less that π/4.
Prove that
cos 2A cos 2B cos 2C
+ + ≤ −2.
sin B sin C sin C sin A sin A sin B
B67. Proposed by Lorian Saceanu.

a) For any triangle ABC, prove that
aA + bB + cC 3r π
≥ arccot ≥ ,
a+b+c s 3
where s is the semiperimeter of ABC and r is the inradius.
b) For an acute angled triangle ABC, prove that
aA + bB + cC r
≤ arccos ,
a+b+c R
where r is the inradius and R is the circumradius.
B68. Proposed by Daniel Sitaru.

√
3
Prove that for 3 ≤ a, b, c ≤ 1, we have
! …
√ √
…
−1 ab + bc + ca ab + bc + ca
· tan−1 ( abc).
3 3
abc · tan ≤
3 3
When does equality occur?
B69. Proposed by George Apostolopoulos.

Let ABC be a triangle with circumradius R and inradius r. Prove that
√ Å R ã2
csc A + csc B + csc C − cot A − cot B − cot C ≤ 3 .
2r
B70. Proposed by Leonard Giugiuc, Diana Trailescu and Dan Stefan Mari-
nescu.
Let f : [0, ∞) → R be a differentiable function, whose derivative is concave for
all positive real numbers and f (0) = 0. Prove that for any integer n ≥ 3 and any
non-negative numbers xk , k = 1, 2, . . . , n, we have
n
X n
X
f (x1 + · · · + xn ) + (n − 2) f (xk ) ≤ f (xi + xj ).
k=1 1≤i<j≤n

Bonus Problems /513
B71. Proposed by Hoang Le Nhat Tung.

Let a, b, c be positive real numbers such that a + b + c = 3. Prove that
a b c 1
p +p +p ≥
2(b4 + c4 ) + 7bc 2(a4 + c4 ) + 7ac 2(a4 + b4 ) + 7ab 3
B72. Proposed by Daniel Sitaru.

Prove that in triangle ABC, the following relationship holds:
sin A sin B sin C 2s
B C
+ C A
+ A B
≥ .
sin 2 sin 2 sin 2 sin 2 sin 2 sin 2 r
B73. Proposed by Leonard Giugiuc and Diana Trailescu.

Let a, b and c be real numbers that satisfy a ≥ 1 ≥ b ≥ c, a + b + c < 0 and
ab + bc + ca = 3. Determine the maximum value of a + b + c and the values of a,
b and c for which this maximum is attained

Let a, b, c ≥ 0 be real numbers with the property a2 + b2 + c2 = 1. Show that:
a2 ln (1 + bc) b2 ln (1 + ac) c2 ln (1 + ab) 64

+ + ≥ ln .
bc ac ab 27
B75. Proposed by George Apostolopoulos.

If a, b, c are positive real numbers such that a + b + c = 1, prove that
4(ab + bc + ca)2 ≤ ab + bc + ca + 3abc.

514/ Solutions
SOLUTIONS
No problem is ever permanently closed. The editor is always pleased to consider for
publication new solutions or new insights on past problems.
4541. Proposed by Michel Bataille.

Let a, b, c be positive real numbers such that abc = 1 and let Sk = ak + bk + ck .
Prove that
S2 + S4 p
≥ 1 + 1 + S3 .
2
We received 24 submissions, of which 22 were correct and complete. We present
the solution by Corneliu Manescu-Avram.
From the AM-GM inequality we have S3 ≥ 3. On the other hand,
S2 + S4 (a2 − a)2 + (b2 − b)2 + (c3 − c)2

− S3 = ≥ 0,
2 2
√
therefore it suffices to prove that S3 ≥ 1 + 1 + S3 . Separating the radical,
squaring and reducing similar terms, we get S3 (S3 − 3) ≥ 0, which is true.
Editor’s comment. Roy Barbara, Walther Janous, and Borche Joshevski submitted
generalizations.
4542. Proposed by Leonard Giugiuc and Alexander Bogomolny.

Let ABC be a triangle with centroid G. Denote by D, E and F the midpoints of
the sides BC, CA and AB respectively. Find the points M on the plane of ABC
such that
M A + M B + M C + 3M G = 2(M D + M E + M F ).
We received 12 solutions, 5 of which were incomplete. All of the complete solu-

tions used Hlawka’s inequality. As several solvers (Walther Janous, Ioannis Sfikas,
UCLan Cyprus Problem Solving Group) pointed out, closely related problems have
occurred both in this journal (problems 3052 and 2482) and The American Math-
ematical Monthly (problem 12015). The problems in this journal prove that in-
equality holds, while the notes to problem 2482 provide a reference to the case of
equality. For completeness, we include a solution by Theo Koupelis here.
By considering the coordinates of the points in the complex plane we have
2D = B + C, 2E = C + A, 2F = A + B, and 3G = A + B + C.
Letting
x = M − A, y = M − B, and z = M − C,

Solutions /515
we get
2(M −D) = y +z, 2(M −E) = z +x, 2(M −F ) = x+y, and 3(M −G) = x+y +z.
Therefore, the given equality is equivalent to
|x| + |y| + |z| + |x + y + z| = |x + y| + |y + z| + |z + x|. (1)
However, Hlawka’s inequality for complex x, y, z is
|x| + |y| + |z| + |x + y + z| ≥ |x + y| + |y + z| + |z + x|. (2)
We can prove this inequality by multiplying both sides by |x| + |y| + |z| + |x + y + z|
and by rearranging the terms to find the equivalent inequality
(|x|+|y|−|x+y|)(|z|+|x+y+z|−|x+y|)+(|y|+|z|−|y+z|)(|x|+|x+y+z|−|y+z|)
+(|z| + |x| − |z + x|)(|y| + |x + y + z| − |x + y|) ≥ 0. (3)

This inequality is obvious by the triangle inequality; in each of the three terms,
both factors are non-negative. (In rearranging the terms to prove this inequality
we used the identity
|x|2 + |y|2 + |z|2 + |x + y + z|2 = |x + y|2 + |y + z|2 + |z + x|2 .)
From (3) we see that equality is reached (and therefore we obtain (1)) when each
of the three terms in (3) vanishes. Considering the first term of (3) as an example,
we see that it vanishes when M is on the line AB (excluding the interior points
of segment AB) or on the segment CG. Similarly for the other two terms in (2).
Therefore, equality occurs when M is a vertex of the triangle or its centroid.
4543. Proposed by Cherng-tiao Perng.

Let n = 4k + 2 (k ≥ 1) be an integer and A1 A2 · · · An be a polygon with parallel
opposite sides, i.e.
Ai Ai+1 k An/2+i An/2+i+1 , i = 1, 2, · · · , n/2,
where one sets An+1 = A1 . Starting with a point B1 and a circle C through
B1 , define B2 , B3 , · · · , Bn+1 inductively by requiring that the circle (Ai Ai+1 Bi )
intersects C again at Bi+1 , for i = 1, 2, · · · , n. Prove that Bn+1 = B1 .
We received 2 solutions; we present them both.
Solution 1, by the UCLan Cyprus Problem Solving Group, annotated and slightly
edited.
Let An+1 = A1 . For 1 ≤ i ≤ n, denote by Γi the circumcircle of 4Ai Bi Ai+1 .
We show in Figure 1 an example of the construction with n = 6. We will use this
specific example to illustrate the main steps of the proof.

516/ Solutions
Figure 1
Note that, in order to show that Bn+1 and B1 coincide, it is sufficient to show
that B1 is on Γn or equivalently, that B1 , An , A1 and Bn are concyclic. We use
directed angles, for which we use the symbol ].
2k
Claim 1: 0 = ]A2r+1 A2r+2 A2r+3 . In Figure 2a, these are the angles drawn in
P
r=0
green.
Figure 2a Figure 2b
Consider the closed polygon A1 A2 · · · A2k+1 A2k+2 A1 (shaded in Figure 2b). The
sum of the directed angles in this polygon is 0; that is,
]A1 A2 A3 + ]A2 A3 A4 + · · · + ]A2k+1 A2k+2 A1 + ]A2k+2 A1 A2 = 0 . (1)

Solutions /517
A2 A1 k A2k+2 A2k+3 gives us that ]A2k+2 A1 A2 = ]A1 A2k+2 A2k+3 , so
]A2k+1 A2k+2 A1 + ]A2k+2 A1 A2 = ]A2k+1 A2k+2 A1 + ]A1 A2k+2 A2k+3

= ]A2k+1 A2k+2 A2k+3 .
Substitute in (1) to get

2k+1
X
Aj Aj+1 Aj+2 = 0. (2)
j=1
Finally, for each 1 6 i 6 k, we have
A2i A2i+1 k A2i+2k+1 A2i+2k+2 and A2i+1 A2i+2 k A2i+2k+2 A2i+2k+3 ,
so
]A2i A2i+1 A2i+2 = ]A2i+2k+1 A2i+2k+2 A2i+2k+3 .
In our example, the equal angles are those decorated with three lines in Figure 2b;
that is, ]A2 A3 A4 = ]A5 A6 A1 . Substitute in (2) to obtain equality in Claim 1.
Claim 2: A1 , B1 , Bn , Bn+1 are concyclic.
Figure 3
For 0 ≤ r ≤ 2k we have
]A2r+1 A2r+2 A2r+3 = ]A2r+1 A2r+2 B2r+2 + ]B2r+2 A2r+2 A2r+3

= ]A2r+1 B2r+1 B2r+2 + ]B2r+2 B2r+3 A2r+3 ,
where in the last line we used the fact that by construction A2r+1 , A2r+2 , B2r+2
and B2r+1 are concyclic (they are all on Γ2r+1 ), as are A2r+2 , B2r+2 , A2r+3 and
B2r+3 (which are all on Γ2r+2 ). In Figure 3, we show the relevant angles in our
example for the case r = 0.

518/ Solutions
Substituting into the equality proven in Claim 1 we get

2k
X
0= ]A2r+1 A2r+2 A2r+3
r=0
2k
X
= (]A2r+1 B2r+1 B2r+2 + ]B2r+2 B2r+3 A2r+3 )
r=0
2k
X
= ]A1 B1 B2 + ]Bn Bn+1 A1 + (]B2r B2r+1 A2r+1 + ]A2r+1 B2r+1 B2r+2 )
r=1
2k
X
= ]A1 B1 B2 + ]B4k+2 B4k+3 A1 + ]B2r B2r+1 B2r+2 , (3)
r=1
where the third line in the above follows from regrouping the terms. By construc-
tion, all the B points are concyclic (they are all on the circle C), so
]B2r B2r+1 B2r+2 = ]B2r B1 B2r+2 .
Angle addition gives us that

2k
X
]B2r B1 B2r+2 = B2 B1 B4k+2 ,
r=1
so from (3) and n = 4k + 2 we conclude that
0 = ]A1 B1 B2 + ]Bn Bn+1 A1 + ]B2 B1 Bn

= ]A1 B1 Bn + ]Bn Bn+1 A1 .
From this we deduce that A1 , B1 , Bn and Bn+1 are concyclic, as claimed.

However, by the definition of the point Bn+1 , the circle through A1 = An+1 , Bn ,
Bn+1 is Γn , so B1 is on Γn . Hence Γn and the circle C intersect at points B1 , Bn
and Bn+1 , so we must have that B1 = Bn+1 as required.
Solution 2, by the proposer, slightly edited.

For j = 1, . . . , n, associate the points Aj and Bj with complex numbers zj and
wj . It suffices to show that the four points An , Bn , B1 , A1 lie on the same circle;
that is, that the cross ratio
(zn − wn )(w1 − z1 )
R := (z1 , wn ; w1 , zn ) =
(zn − z1 )(w1 − wn )
is a real number.
Note that by construction, for j = 1, . . . , n − 1,
(zj − wj )(wj+1 − zj+1 )
Rj := (zj+1 , wj ; wj+1 , zj ) =
(zj − zj+1 )(wj+1 − wj )

Solutions /519
is a real number since Aj , Bj , Aj+1 and Bj+1 are concyclic. Consider the product
P = R1 R2−1 R3 · · · Rn−2
−1
Rn−1 ,
which must also be a real number. Since in the given polygon opposite sides are
parallel, we have for j = 1, . . . , 2k + 1 that
zj+1 − zj
z2k+2+j − z2k+1+j
is real. Thus up to sign or a real factor, by rearranging terms and cancellation,

we get that P is equivalent to
(z1 − w1 )(zn − wn ) (w2 − w3 )(w4 − w5 ) · · · (wn − w1 )

· .
(z2k+1 − z2k+2 )(w1 − wn ) (w1 − w2 )(w3 − w4 ) · · · (wn−1 − wn )
The first factor in the above formula is a real multiple of R since A2k+1 A2k+2 k
An A1 . To prove that R is real, it thus remains to prove that
(w2 − w3 )(w4 − w5 ) · · · (wn − w1 )

(w1 − w2 )(w3 − w4 ) · · · (wn−1 − wn )
is real. We note that this expression is equal to

2k
Y
− (w2j , w2j+2 ; w2j+1 , w1 ).
j=1
Each of the cross ratios (w2j , w2j+2 ; w2j+1 , w1 ) is real since the points B1 , B2j ,
B2j+1 and B2j+2 are on the same circle C, concluding the proof.
Editor’s comments. In the original question it was not addressed how to handle
the case when Ai , Ai+1 and Bi are collinear for some i. However, the proof
should work the same way if we consider the line through Ai , Ai+1 (and Bi ) as a
generalized circle, and define Bi+1 as the second intersection of the line with C.
4544. Proposed by Burghelea Zaharia.

Calculate
2
x4 + 4
Z Å ã
dx
ln .
1 x2 + 4 x
We received 17 correct, 1 incomplete, and 1 incorrect solutions. The majority

of the solutions used transformations using the substitution method for definite
integrals and virtuoso elimination of non-elementary integrals. There were also
some solutions which used power series and their definite integrals where Euler’s
answer for the Basel Problem was helpful. Some of the mistakes in the solutions
could have been avoided if the solvers checked their answer (using, for example,
online integral calculators. This integral is too difficult for modern mathematical

520/ Solutions
software, at least for now, but the approximate answer 0.24, which these programs
can easily calculate for this definite integral, could potentially help when solvers did
their double check.) W. Janous and S. Jason generalized the problem by proving
that Z a Å 4
x + a2 dx 1
ã
2
ln 2 + a2
= (ln a) .
1 x x 2
Here we present the solution by the Missouri State University Problem Solving
Group.
In the integral, let x = 2y, then
2
x4 + 4
Z Å ã
dx
I= ln (1)
1 x x2 + 4
Z 1 Å 4
4y + 1 dy
ã
= ln
1
2
y2 + 1 y
Z 1 Z 1
1 1
= ln(1 + 4y 4 ) dy − ln(1 + y 2 ) dy.
1
2
y 1
2
y
√ √
In the first integral, let z = 2y 2 , or y = z/ 2, then
2 1
1 1
Z Z
2
I= ln(1 + z ) dz − ln(1 + y 2 ) dy
1
2
2z 1
2
y
Z 2 Z 1
1 1 1 1
= ln(1 + z 2 ) dz − ln(1 + y 2 ) dy
2 1 z 2 12 y
1 2 −2 1 1 1 1
Z Z
2
= (ln z + ln(1 + z )) dz − ln(1 + z 2 ) dz
2 1 z 2 12 z
Z 2
ln z 1 2 1 1 1 1
Z Z
= dz + ln(1 + z −2 ) dz − ln(1 + z 2 ) dz
1 z 2 1 z 2 1
2
z
Z 2 Z 1
1 2 1 1 1 1
= (ln 2) + ln(1 + z −2 ) dz − ln(1 + z 2 ) dz. (2)
2 2 1 z 2 21 z
However, let y = 1/z, we will have

1
2 1
1 1
Z Z 2
Z
ln(1 + z −2 ) dz = ln(1 + y 2 )y(−y −2 )dy = ln(1 + y 2 ) dy.
1 z 1 1
2
y
Hence, the second and the third terms in Eq. (1) add to zero.
As a result, we have
2
x4 + 4 1
Z Å ã
dx 2
I= ln = (ln 2) .
1 x2 + 4 x 2

Solutions /521
4545. Proposed by Mihaela Berindeanu.

Solve the following equation over N:
î p ó
6n − 19 = 5 n2 + 4n .
We received 36 submissions of which 33 were correct and complete. We present

the solution of Jiahao Chen and Madhav Modak (independently), slightly modified.
The only solution to the equation is n = 2.
It is easy to check that n = 1 is not a solution.
î √ ó
When n = 2, we have 62 − 19 = 17 = 5 12 , so n = 2 is a solution. Finally, if
n ≥ 3, by binomial theorem, the left-hand side of the equation is
Ç å
n n n n
6 − 19 = (5 + 1) − 19 ≥ 5 + · 5 + 1 − 19 ≥ 5n + 5n − 18 ≥ 5n + 107,
1
while the right-hand side of the equation is

î p ó î p ó
5 n2 + 4n < 5 n2 + 4n + 4 = 5n + 10,
thus the equation has no solution when n ≥ 3.
4546. Proposed by Thanos Kalogerakis, Leonard Giugiuc and Kadir Altintas.

Let D be a point on the side BC of triangle ABC and consider the following
tri-tangent circles:
(K1 , k1 ) is the incircle and (L1 , l1 ) is the A-excircle of ABC
(L2 , l2 ) is the incircle and (K2 , k2 ) is the A-excircle of ABD
(L3 , l3 ) is the incircle and (K3 , k3 ) is the A-excircle of ACD
Prove that Area(4K1 K2 K3 ) = Area(4L1 L2 L3 ).

522/ Solutions
We received 2 solutions, one of which was incomplete. We present the solution by

the proposer.
Let a = BC, b = CA, c = AB, x = AD, y = CD, and z = BD. For a point X,
denote by ~rX the position vector of X. In ∆ABC, we have
a~rA + b~rB + c~rC −a~rA + b~rB + c~rC
~rK1 = and ~rL1 = ,
a+b+c −a + b + c
and in ∆ABD, we have
−z~rA + x~rB + c~rD z~rA + x~rB + c~rD
~rK2 = and ~rL2 = .
−z + x + c −z + x + c
But
y~rB + z~rC
~rD = ,
y+z
so that
Ä ä
y~
rB +z~rC
−z~rA + x~rB + c y+z −z(y + z)~rA + [x(y + z) + cy]~rB + cz~rC
~rK2 = = .
−z + x + c (y + z)(−z + x + c)
Analogously,
z(y + z)~rA + [x(y + z) + cy]~rB + cz~rC
~rL2 = .
(y + z)(z + x + c)
Similarly in ∆ACD we have
−y(y + z)~rA + by~rB + [x(y + z) + bz]~rC
~rK3 =
(y + z)(−y + x + b)
and
y(y + z)~rA + by~rB + [x(y + z) + bz]~rC
~rL3 = .
(y + z)(y + x + b)
Thus we can express the barycentric coordinates with respect to ∆ABC of the six
points as
Å ã
a b c
K1 = , ,
a+b+c a+b+c a+b+c
−z(y + z) x(y + z) + cy
Å ã
cz
K2 = , ,
(y + z)(−z + x + c) (y + z)(−z + x + c) (y + z)(−z + x + c)
−y(y + z) x(y + z) + bz
Å ã
by
K3 = , ,
(y + z)(−y + x + b) (y + z)(−y + x + b) (y + z)(−y + x + b)
−a
Å ã
b c
L1 = , ,
−a + b + c −a + b + c −a + b + c
z(y + z) x(y + z) + cy
Å ã
cz
L2 = , ,
(y + z)(z + x + c) (y + z)(z + x + c) (y + z)(z + x + c)
y(y + z) x(y + z) + bz
Å ã
by
L3 = , ,
(y + z)(y + x + b) (y + z)(y + x + b) (y + z)(y + x + b)

Solutions /523
We deduce that |K1 K2 K3 | = |L1 L2 L3 | if and only if

|∆1 | |∆2 |
= , (1)
(a + b + c)(x + c − z)(x + b − y) (−a + b + c)(x + c + z)(x + b + y
where

a b c

∆1 = −z(y + z) x(y + z) + cy cz ,

−y(y + z) by x(y + z) + bz
and
−a b c

∆2 = z(y + z) x(y + z) + cy cz .

y(y + z) by x(y + z) + bz
Since ∆1 = −∆2 , |∆1 | = |∆2 |, so it suffices to show that the denominators in (1)
are equal. This is equivalent to
ax2 + ayz = b2 z + c2 y,
which is true by Stewart’s Theorem.
4547. Proposed by George Stoica, modified by the Editorial Board.

Consider the complex numbers a, b, c such that |a| = |b| = |c| = 1. Prove that if
|a + b − c|2 + |b + c − a|2 + |c + a − b|2 = 12,
then a, b, c represent the vertices of an equilateral triangle inscribed in the unit
circle.
We received four submissions, all correct, and feature the solution by Jiahao Chen.
Because |z| = 1 implies that z̄ = z1 , we see that the remaining assumption has the
following sequence of implications:
1 1 1
X Å ã
(a + b − c) + − = 12
cyc
a b c
X a Å ã
b b c c a
+ − − − − = 3
cyc
b a c b a c
X a Å ã
b
+ = −3
cyc
b a
X
a2 b + ab2

= −3abc
cyc
(a + b + c)(ab + bc + ca) = 0.
It follows, therefore, that a + b + c = 0 or ab + bc + ca = 0. Should a + b + c = 0,
which (because a 6= 0) is equivalent to 1 + ab + ac = 0, implies
Å ã Å ã
b c b c
Im = −Im and Re + Re = −1.
a a a a

524/ Solutions
The left equality together with ab = ac = 1 implies that also

Å ã
b c
Re = ±Re .
a a
We conclude that
√
1 3
Å ã Å ã
b c b c
Re = Re =− and Im = −Im =± .
a a 2 a a 2
Thus, 1, ab , ac represent the vertices of an equilateral triangle inscribed in the unit
circle and, therefore, so do a, b, c.
It remains to show that ab + bc + ca = 0 likewise implies that a, b, c represent the
vertices of an equilateral triangle. But this equation is equivalent to
ā + b̄ + c̄ = a + b + c = 0,
which immediately implies that a + b + c = 0, and, as we have already seen, this
implies that a, b, c represent the vertices of an equilateral triangle.
Editor’s comments. Mihaela Berindeanu informed us that she proposed a problem
that was identical to our 4547 for use on a Romanian mathematics contest back
in March of 2019. The problem was subsequently published in a supplement to
the Gazeta Matematică (dealing with the National Olympiad that was written in
Deva, Romania, April 22-26, 2019). By coincidence, the journal appeared in 2019
at almost the same time as Stoica’s problem appeared in Crux.
4548. Proposed by Lazea Darius.

Find the maximum k for which
ab + bc + ca + k(a − b)2 (b − c)2 (c − a)2 ≤ 3
for all non-negative real numbers a,b,c such that a + b + c = 3.
We received 12 submissions of which 5 were correct and complete. We present the
solution by the UCLan Cyprus Problem Solving Group.
Since
(a − b)2 + (b − c)2 + (c − a)2 = 2(a2 + b2 + c2 ) − 2(ab + bc + ca)
= 2(a + b + c)2 − 6(ab + bc + ca)
= 18 − 6(ab + bc + ca)
the problem becomes to maximize k such that
(a − b)2 + (b − c)2 + (c − a)2 > 6k(a − b)2 (b − c)2 (c − a)2
under the stated conditions. Without loss of generality we assume that a > b > c.
Let x = a − b and y = b − c. Then x, y > 0, x + 2y = 3 − 2c, and we are to
maximize k so that
x2 + y 2 + (x + y)2 > 6k(xy(x + y))2

Solutions /525
holds for all x, y > 0 for which x + 2y 6 3. (Given any such x, y it is easy to
determine a, b, c and check that they satisfy the given conditions.)
We may assume that x, y > 0 since otherwise the inequality is satisfied for any k.
We will maximize
(xy(x + y))2
x2 + y 2 + (x + y)2
under the conditions x, y > 0 and x + 2y 6 3.
If x + 2y < 3, letting x0 = rx and y 0 = ry where r = 3
x+2y , we have x0 , y 0 > 0,
r > 1, and x0 + 2y 0 = 3. Thus,
(x0 y 0 (x0 + y 0 ))2 (xy(x + y))2 (xy(x + y))2

= r4 2 > 2
(x0 )2 0 2
+ (y ) + (x + y )0 0 2 2
x + y + (x + y) 2 x + y 2 + (x + y)2
and we may assume that x + 2y = 3. Letting y = t and x = 3 − 2t, we will
maximize
((3 − 2t)t(3 − t))2
f (t) =
(3 − 2t)2 + t2 + (3 − t)2
for t ∈ (0, 3/2). Motivated by the symmetry of f about t = 3/2, we make the
substitution t = s + 3/2 and for s ∈ (−3/2, 0) let
2
3 (−2s(s + 32 )( 32 − s))2 4s2 s2 − 49 s2 (4s2 − 9)2
Å ã
g(s) = f s + = 3 3 = 9 = .
2 (−2s)2 + (s + 2 )2 + ( 2 − s)2 6s2 + 2 6(4s2 + 3)
Next, we let u = s2 and for u ∈ (0, 9/4) we will maximize
√ u(4u − 9)2
h(u) = 6g( u) = .
4u + 3
We have
(4u + 3)((4u − 9)2 + 8u(4u − 9)) − 4u(4u − 9)2
h0 (u) =
(4u + 3)2
(4u − 9)((4u + 3)(12u − 9) − 4u(4u − 9))
=
(4u + 3)2
2
(4u − 9)(32u + 36u − 27)
= .
(4u + 3)2
3
√ 3
√
The zeroes of 32u2 + 36u − 27 are u+ = − 16 (3 − 33) and u− = − 16 (3 + 33)
with
3(6 − 3) 9
u− < 0 < u+ < < .
16 4
Furthermore,
32(4u − 9)(u − u+ )(u − u− )
h0 (u) =
(4u + 3)2

526/ Solutions
and so, h0 (u) > 0 in (0, u+ ) and h0 (u) < 0 in (u+ , 9/4). So, h is maximized at u+
with maximum
16u+ (16u+ − 36)2
h(u+ ) =
64(16u+ + 12)
√ √
3( 33 − 3)(3 33 − 45)2
= √
64(3 33 + 3)
√
27(69 − 11 33)
= .
32
The maximum of g is therefore
√
9(69 − 11 33)
64
p √
and is achieved at s = − 14 3 33 − 9. The maximum of f is the same as that of
p √
g and is achieved at t = 2 − 14 3 33 − 9. Hence,
3
√ √
64 32(69 + 11 33) 69 + 11 33
k6 √ = = .
6 · 9(69 − 11 33) 27 · 768 648
Equality is achieved when
» √ » √
3 1 3 1
c = 0, b = y = − 3 33 − 9 and a = 3 − b = + 3 33 − 9.
2 4 2 4
4549. Proposed by Lorian Saceanu, Leonard Giugiuc and Kadir Altintas.

Let ak and bk be real numbers for k = 1, 2, . . . , n. Prove that
Ã Ã Ã Ã
Xn n
X Xn n
X
(2ak − bk )2 + (2bk − ak )2 ≥ a2k + b2k .
k=1 k=1 k=1 k=1
There were 13 correct solutions. We present three of them here.

Solution 1, by Anil Kumar, Prithwijit De, C.R. Pranesachar, and Zoltan Retkes,
all done independently.
Let a = (a1 , a2 , . . . , an ) and b = (b1 , b2 , . . . , bn ). Then

2kak = k2a − b + bk ≤ k2a − bk + kbk
and
2kbk = k2b − a + ak ≤ k2b − ak + kak,
where kxk is the Euclidean norm of the n−vector x. Adding these inequalities and
rearranging terms yields the required
k2a − bk + k2b − ak ≥ kak + kbk.

Solutions /527
Equality occurs when 2a − b = λb and 2b − a = µa where λ and µ are nonnegative

and satisfy 4 = (1 + λ)(1 + µ). Thus, both must lie in the interval [0, 3]. Hence,
equality holds if and only if a = νb where
1 1+λ 2
≤ν= = ≤ 2.
2 2 1+µ
Solution 2, by Oliver Geupel.

Let a and b be the vectors defined in Solution 1. Suppose that ABC is the triangle
whose respective vertices are at 2a, 2b and 0 in n−space. Let V be the midpoint
of AC, U be the midpoint of BC and G be the intersection of the medians AU
and BV . Then
k2a − bk + k2b − ak = |AU | + |BV |

= |AG| + |GU | + |BG| + |GV |
= |AG| + |GV | + |BG| + |GU |
≥ |AV | + |BU |
= |V C| + |CU |
= kak + kbk.
Solution 3, by Marie-Nicole Gras.

ÄX ä1/2 ÄX ä1/2
Let x = a2i , y= b2i , z=
P
ai bi .
By the Cauchy-Schwarz Inequality, x2 y 2 ≥ z 2 . Now

X
(2ai − bi )2 = 4x2 + y 2 − 4z ≥ 4x2 + y 2 − 4xy = (2x − y)2
and X
(2bi − ai )2 ≥ (2y − x)2 .
Hence the left side of the desired inequality is greater than or equal to
|2x − y| + |2y − x| ≥ |(2x − y) + (2y − x)| = x + y.
Comment by the editor. Walther Janous generalized the inequality to

m
X m
X
kλxk − xk+1 − · · · − xk+m−1 k ≥ (λ − m + 1) kxk k
k=1 k=1
and Borche Joshevski to
ksa − tbk + ksb − tak ≥ |s − t|(kak + kbk),
where s and t are nonnegative.

528/ Solutions
4550. Proposed by Leonard Giugiuc and Kunihiko Chikaya.

Let α be a real number greater than 2 and let x, y and z be positive real numbers
such that x ≥ y ≥ z. Prove that
(xα − y α )(y α − z α )(z α − xα ) α3

≥ ((x − y)3 + (y − z)3 + (z − x)3 ).
(xα−1 + y α−1 )(y α−1 + z α−1 )(z α−1 + xα−1 ) 24
When does equality hold?

We received 11 correct solutions and 1 incomplete submission. We present the
solution by Michel Bataille.
Since (x − y) + (y − z) + (z − x) = 0, the identity
X 3 + Y 3 + Z 3 − 3XY Z = (X + Y + Z)(X 2 + Y 2 + Z 2 − XY − Y Z − ZX)
yields
(x − y)3 + (y − z)3 + (z − x)3 = 3(x − y)(y − z)(z − x).
We deduce that equality holds if at least two of the numbers x, y, z are equal
(then both sides are 0). Otherwise, that is, if x > y > z, we show that the strict
inequality holds, or equivalently, that
(xα − y α )(y α − z α )(xα − z α )

α3
< (x − y)(y − z)(x − z)(xα−1 + y α−1 )(y α−1 + z α−1 )(z α−1 + xα−1 )
8
or y Å ã
z z
f ·f ·f < 1,
x y x
where
2(1 − tα )
f (t) = .
α(1 − t)(1 + tα−1 )
y z z
Since , , are in the interval (0, 1) and f (t) > 0 if t ∈ (0, 1), it suffices to prove
x y x
that f (t) < 1 or equivalently that g(t) = α − 2 − αt + αtα−1 − (α − 2)tα is positive
whenever t ∈ (0, 1).
Now g 0 (t) = −αh(t) with h(t) = 1 − (α − 1)tα−2 + (α − 2)tα−1 .
For t ∈ (0, 1) we have h0 (t) = −(α−1)(α−2)(1−t)tα−3 < 0, hence h(t) > h(1) = 0
and so g 0 (t) < 0 and therefore g(t) > g(1) = 0. This completes the proof.
Editor’s comments: All of the solvers used the methods of Differential and Integral
Calculus. For the majority of the solvers, the convexity of some functions was the
main tool for the proof of the inequality.

CRUXv 46 N 10

Uploaded by

Copyright:

Available Formats

CRUXv 46 N 10

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

CRUXv 46 N 10

Uploaded by

Copyright:

Available Formats

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

c SOCIÉTÉ MATHÉMATIQUE DU CANADA 2020 TOUS DROITS RÉSERVÉS.

Supported by / Soutenu par :

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

MathemAttic Editors John McLoughlin University of New Brunswick

Olympiad Corner Editors Alessandro Ventullo University of Milan

Articles Editor Robert Dawson Saint Mary’s University

Associate Editors Edward Barbeau University of Toronto

Guest Editors Yagub Aliyev ADA University, Baku, Azerbaijan

Editor-at-Large Bill Sands University of Calgary

482 Editorial Kseniya Garaschuk

Crux Mathematicorum, Vol. 46(10), December 2020

Click here to submit solutions, comments and generalizations to any

To facilitate their consideration, solutions should be received by February 15, 2021.

MA98. A pair of telephone poles d metres apart is supported by two cables

Copyright © Canadian Mathematical Society, 2020

MA100. Suppose the equation x3 + 3x2 − x − 1 = 0 has real roots a, b, c.

Cliquez ici afin de soumettre vos solutions, commentaires ou

MA97. Deux cambrioleurs, A et B, ont la spécialité de voler des œuvres d’art

Crux Mathematicorum, Vol. 46(10), December 2020

MA98. Deux poteaux de téléphone, de hauteurs 4 mètres et 6 mètres et à

MA99. Un drapeau de taille 48 cm par 24 cm consiste d’une croix blanche sur

MA100. Supposer que l’équation x3 + 3x2 − x − 1 = 0 a les racine réelles a,

Copyright © Canadian Mathematical Society, 2020

The outer equalities of (1) give us ab − ax = ab − by or

Crux Mathematicorum, Vol. 46(10), December 2020

which has the solution

Resubstituting this value of x in (2) gives

Using that y, z, w are in geometric progression and the previous equation,

z 2 = yw = y(16 − x) = y(24 − 3y).

0 = z 2 − z 2 = (8 − y)2 − y(24 − 3y) = 4(y − 2)(y − 8),

Copyright © Canadian Mathematical Society, 2020

MA73. A checkerboard is “almost tileable” if there exists some way of placing

For any n ≡ 0 or 1 mod 4, we can inductively construct a solution from the

We now show that it is impossible to find a1 solution for n ≡ 2 or 3 mod 4. Colour

Crux Mathematicorum, Vol. 46(10), December 2020

Copyright © Canadian Mathematical Society, 2020

Originally Problem 1, Vermont State Mathematics Coalition Talent Search, 2015.

Crux Mathematicorum, Vol. 46(10), December 2020

Folding Paper Geometry

Figure 1: Paper folding

Now, we can draw a suitable diagram as shown in Figure 2.

Figure 2: Basic diagram

Copyright © Canadian Mathematical Society, 2020

Let’s redraw the diagram by A F

Method 1: Using area

Since AF becomes F M after folding, then AF = F M and ∆AEF ∼ = ∆EF M (be-

Crux Mathematicorum, Vol. 46(10), December 2020

Method 2: Using trigonometry

In Figure 3, let ∠DEM = α and ∠EF M = θ. Since ∆AEF ∼

∠AEF + ∠M EF + ∠DEM = π. (3)

With substitution (1) and (2) into (3), we get

Method 3: Using trigonometry (again)

In Figure 3, connect AM as shown in the Figure 4.

Figure 4: Diagram with angles

Copyright © Canadian Mathematical Society, 2020

Notice that AD = 6 and√DM = 4. Applying the Pythagorean Theorem to right

Method 4: Using similar triangles

Crux Mathematicorum, Vol. 46(10), December 2020