CRUXv 36 N 3

129
SKOLIAD No. 124
Lily Yen and Mogens Hansen
Please send your solutions to problems in this Skoliad by 1 July, 2010. A opy
of CRUX with Mayhem will be sent to one pre-university reader who sends
in solutions before the deadline. The de ision of the editors is nal.
Our ontest this month is the sele ted problems from the 10 Annual th
Christopher Newport University Regional Mathemati s Contest, 2009. Our

thanks go to Ron Persky, Christopher Newport University, Newport News,
USA, for providing us with this ontest and for permission to publish it.
th
10 Annual Christopher Newport University
Regional Mathemati s Contest 2009
Sele ted problems
1 . Elves and ogres live in the land of Pixie. The average height of the elves
is 80 m, the average height of the ogres is 200 m, and the average height
of the elves and the ogres together is 140 m. If 36 elves live in Pixie, how
many ogres live there?
2. You are given a two-digit positive integer. If you reverse the digits of
your number, the result is a number whi h is 20% larger than your original
number. What is your original number?
3 . Three squares are pla ed side-by- .......
side inside a right-angled triangle as ...
........... .....
shown in the diagram. The side length
.
...
....... ..
..........................................
of the smallest of the three squares ...
.....
....... .. ..
is 16. The side length of the largest . .
...
.
...... ...
..
..
................... ...................... ..
of the three squares is 36. What is the ..
...... .. ..
........ ... ... ... .
side length of the middle square?
.
..
....................................................................................................
.
4 . Friends Maya and Naya ordered nger food in a restaurant, Maya or-
dering hi ken wings and Naya ordering bite-size ribs. Ea h wing ost the
same amount, and ea h rib ost the same amount, but one wing was more
expensive than one rib. Maya re eived 20% more pie es than Naya did, and
Maya paid 50% more in total than Naya did. The pri e of one wing was what
per entage higher than the pri e of one rib?
5. A 9 12 re tangular pie e of paper is folded on e so that a pair of diag-
onally opposite orners oin ide. What is the length of the rease?
130
6 . In alm weather, an air raft an y from one ity to another 200 kilometres
north of the rst and ba k in exa tly two hours. In a steady north wind, the
round trip takes ve minutes longer. Find the speed (in kilometres per hour)
of the wind.
7. A re tangular oor, 24 feet 40 feet, is overed by squares of sides 1 foot.
A halk line is drawn from one orner to the diagonally opposite orner. How
many tiles have a halk line segment on them?
ieme

10 Con ours mathematique
annuel
de l'Universite
Christopher Newport 2009
problemes
hoisis
1 . Un ertain nombre de fees

et d'ogres demeurent au Pays imaginaire. Les
fees
sont de taille moyenne 80 m, omparativement a 200 m pour les ogres.
et des ogres est de 140 m. Si
Enn la taille moyenne de la totalite des fees
36 fees
demeurent au Pays imaginaire, determiner
le nombre d'ogres qui y
demeurent.
2. On vous donne un entier a deux hires. Si on inverse l'ordre des deux
hires, l'entier qui en resulte
est 20% plus elev
e que l'entier original. Deter-
miner l'entier original.
3. Trois arres sont pla es ote
^ a ote^ ..
a l'interieur
d'un triangle re tangle, tel ................. .....
qu'illustre. Le plus petit des arres a ........ .
...........................................
un ot ^ e de longueur 16, tandis que le
.......
....... .. ..
................................
plus grand a un ot ^ e de longueur 36. .......
. ..
.................... ...
Determiner
la longueur du ot^ e du tri-
.
...... ..
........... ... ... ... .
angle au milieu. .
..........................................................................................................
.
4 . Madeleine et Nadine ont de id

e de manger sur le pou e, Madeleine ayant
ommande des ailes de poulet ontrairement a Nadine qui a hoisi des ote-^
lettes. Toute aile de poulet a le même prix et toute otelette
^ a son propre
prix, une aile de poulet etant
plus outeuse
^ qu'une otelette.
^ Madeleine a
obtenu un nombre d'ailes 20% plus elev e que le nombre de otelettes
^ re ues
par Nadine ; aussi, Madeleine a debours
e 50% de plus au total que Nadine.
Le prix d'une aile de poulet est plus elev
e que le prix d'une otelette
^ par quel
pour entage ?
5. Un mor eau de papier re tangulaire de taille 9 12 est plie une seule fois,
de fa on a e qu'une paire de oins diagonalement opposes se rejoignent.
Determiner
la longueur de la plissure.
6. Dans l'absen e de vent, un avion ee tue un trajet aller-retour d'une
ville a une autre, situee
a 200 kilometres
au nord, dans exa tement deux
heures. Ave un vent onstant du nord, e trajet prend inq minutes de plus.
Determiner
la vitesse du vent, en kilometres
a l'heure.
131
7. Un plan her re tangulaire de taille 24 pieds 40 pieds, est pave de tuiles

arrees
de ot ^ es
1 pied. Une ligne droite est tra ee
d'un oin a son oin dia-
gonalement oppose. Determiner
le nombre de tuiles qui in luent un segment
de ette ligne.
Next follow solutions to the Cariboo College High S hool Mathemati s

Contest, 1990, Junior Final, Part B, whi h appeared in Skoliad 118 in the
Totten Commemorative issue [2009 : 263{265.
1 . A boy on a bi y le oasts down from the top of a hill. He overs 4 metres
in the rst se ond and in ea h su eeding se ond overs 5 metres more than
in the previous se ond. He rea hes the bottom of the hill in 11 se onds.
(a) How long is the hill?
(b) What is the boy's average speed in metres per se ond?
( ) What distan e did he over in the last se ond?
Solution by Gesine Geupel, student, Max Ernst Gymnasium, Bruhl, NRW,
Germany.
The distan e travelled in ea h se ond is easily al ulated:
se ond: 1 2 st nd
3 rd
4 5
th
6
th
7
th
8
th
9
th
10th
11 th th
metres: 4 9 14 19 24 29 34 39 44 49 54
Adding the distan es yields that the hill is 319 m long. The boy's average
speed is therefore 319
11
m/s = 29 m/s. That he overed 54 metres in the last
se ond is in the table above.

Also solved by LENA CHOI, student, E ole Banting Middle S hool, Coquitlam, BC;
ROWENA HO, student, E ole Banting Middle S hool, Coquitlam, BC; MONICA HSIEH,
student, Burnaby North Se ondary S hool, Burnaby, BC; JULIA PENG, student, Campbell
Collegiate, Regina, SK; SZERA PINTER, student, Mos rop Se ondary S hool, Burnaby, BC;
ALISON TAM, student, Burnaby South Se ondary S hool, Burnaby, BC; and KENRICK TSE,
student, Quil hena Elementary S hool, Van ouver, BC.
2 . Two golfers, on their way to the ourse, rea hed a railway rossing just
as a 2.5 km train arrived. Rather than waiting, they de ided to go on to the
next rossing 1 km along in the dire tion the train was going. They travelled
at 50 km/h while the train travelled at 70 km/h.
(a) How long did they have to wait for the train to lear the rossing?
(b) Rather than travelling at 50 km/h, how fast would they have had to tra-
vel to rea h the rossing just as the train was learing the rossing?
Solution by Alison Tam, student, Burnaby South Se ondary S hool, Burnaby,
BC.
The train needs to travel 3.5 km to ompletely pass the se ond rossing.
3.5 km
That takes 70 km/h
= 0.05 h, or 3 minutes. The golfers then just travel 1 km,
132
km
whi h takes them 501km /h
= 0.02 h or 1.2 minutes. The golfers therefore
must wait for 3 1.2 = 1.8 minutes, or 1 minute and 48 se onds.
If the golfers do not want to wait at the se ond rossing, then they
must spend 3 minutes (or 0.05 h) getting there, sin e that is how long the
1 km
train takes to pass. Thus, their speed must be 0.05 h
= 20 km/h.

Also solved by LENA CHOI, student, E ole Banting Middle S hool, Coquitlam, BC; JULIA
PENG, student, Campbell Collegiate, Regina, SK; SZERA PINTER, student, Mos rop Se ondary
S hool, Burnaby, BC; and KENRICK TSE, student, Quil hena Elementary S hool, Van ouver,
BC.
3. A student asks you to hoose a number from 1 to 9 and multiply it by 109,

then asks you to nd the sum of the digits in the produ t. Knowing the sum
of the digits, the student is able to tell you the number with whi h you began.
Explain how this an be done.
Solution by Szera Pinter, student, Mos rop Se ondary S hool, Burnaby, BC.
The table shows the al ulation of the de-
sired digit sum in ea h ase. Sin e all the digit 1 109 = 109 10
sums are dierent, the original number an easily 2 109 = 218 11
be re overed. Indeed, simply subtra t 9 from the 3 109 = 327 12
digit sum. 4 109 = 436 13
You an also predi t the digit sums without 5 109 = 545 14
a tually al ulating them. If the original single- 6 109 = 654 15
digit number is x, then 9x is obviously divisi- 7 109 = 763 16
ble by 9, so its digit sum must be divisible by 9. 8 109 = 872 17
Sin e 9x 81, the digit sum of 9x must therefore 9 109 = 981 18
equal 9. The digit sum of 100x is learly x, so the
digit sum of 109x is x + 9. Hen e, x an be re overed from the digit sum by
subtra ting 9.
Also solved by CINDY CHEN, student, Burnaby North Se ondary S hool, Burnaby, BC;
GESINE GEUPEL, student, Max Ernst Gymnasium, Bruhl, NRW, Germany; ROWENA HO,

student, E ole Banting Middle S hool, Coquitlam, BC; JULIA PENG, student, Campbell
Collegiate, Regina, SK; and KENRICK TSE, student, Quil hena Elementary S hool, Van ouver,
BC.

4 . Suppose you throw 5 darts at a round board with a radius of 25 2 m. If
all 5 darts sti k in the board, show that at least two of them must be within
50 m of ea h other.
Solution by Julia Peng, student, Campbell Collegiate, Regina, SK.

Divide the dart board into four parts using perpendi ular diameters,
like so: . (Let ea h quarter ir le in lude its boundary.) With ve darts,
learly at least one of the four quarter ir les must ontain at least two darts.
133
Consider su h a quarter ir le (see the thi k lined ... ...

... ... ... ... ... ... ... ... ...
part of the gure). By the Pythagorean Theorem, B ... ... ...

... .
.. .
..
...
.
.
............................
.......
.. .....
....
.
..
..
. .. .... ....... ..
|AB|2 |OA|2 + |OB|2
. ....
= . .. ....
....
....
..... ..
..
.. .... .... ..
252 2 + 252 2 = 252 22 , .. 50 ......

.. .. .... ..
=
.. ....
.. ...
.... ..
..
..
.. .. 25 2 ......
. ...
....
.. .
.
..
so |AB| = 25 2 = 50. Then draw the ir le

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

....
.. .. ... .... ..
through points O, A, and B . Sin e AOB = 90,

.. ....
..
..
...
...................25 2
...................................
....
...
.... ..
...
.
.
..
the entre of that ir le is the midpoint of AB .

... .. .
O A
... . ..
.. ... .
.. ... .
... ... .
.. ... ... ... ... ... ...
Thus AB is a diameter for the dotted ir le in the
gure, and the distan e between any two darts
within the thi k quarter ir le is at most |AB|, as required.
Also solved by KENRICK TSE, student, Quil hena Elementary S hool, Van ouver, BC.
The idea that if four parts ontain ve darts then at least one part must ontain at least
two darts is known as the Pigeonhole Prin iple, whi h is useful in many ontest problems.
. The diameter, AB , of a ir le
..............................
...........
5
.....
.......................... ......
is divided into 4 equal parts by the ... . .....
............... ...........
points C , D, and E . Semi ir les are .. . .
.. . ...
. ....
.. ..... ...
............................................
drawn on AC , AD, AE , and AB as
.... ...
................ .... ... ...
shown. Find the ratio of the area of ............. ........ ... ... ..
the shaded parts to the area of the un-
.. .
. ...............................................................................................
. .
.
shaded parts. A C D E B
Solution by Kenri k Tse, student, Quil hena Elementary S hool, Van ouver,
BC, modied by the editor.
Let r be the radius of the smallest semi ir le. Then the other semi ir les
have radii 2r, 3r, and 4r, respe tively. Thus their areas are, respe tively,
1 1 1 9 1
r 2 , (2r)2 = 2r 2 , (3r)2 = r 2 , and (4r)2 = 8r 2 .
2 2 2 2 2
The area of the innermost shaded band is now the dieren e in area
between the two smallest semi ir les, thus 2r2 12 r2 = 32 r2 . Similarly,
the outer shaded band has area 8r2 29 r2 = 72 r2 . Thus the shaded
region has area 32 r2 + 72 r2 = 5r2 .
Now we see that the unshaded region has area 8r2 5r2 = 3r2 .
2
It follows that the desired ratio is 5r
3r 2
5
= , or 5 : 3.
3
Also solved by MONICA HSIEH, student, Burnaby North Se ondary S hool, Burnaby,
BC; JULIA PENG, student, Campbell Collegiate, Regina, SK; and SZERA PINTER, student,
Mos rop Se ondary S hool, Burnaby, BC.
This issue's prize of one opy of CRUX with MAYHEM for the best
solutions goes to Julia Peng, student, Campbell Collegiate, Regina, SK.
We hope that our readers will ontinue to share their joy of mathemati s
by submitting solutions to our problems.
134
MATHEMATICAL MAYHEM
Mathemati al Mayhem began in 1988 as a Mathemati al Journal for and by
High S hool and University Students. It ontinues, with the same emphasis,
as an integral part of Crux Mathemati orum with Mathemati al Mayhem.
The Mayhem Editor is Ian VanderBurgh (University of Waterloo). The
other sta members are Monika Khbeis (Our Lady of Mt. Carmel Se ondary
S hool, Mississauga, ON) and Eri Robert (Leo Hayes High S hool, Freder-
i ton, NB).
Mayhem Problems
Please send your solutions to the problems in this edition by 15 August 2010.
Solutions re eived after this date will only be onsidered if there is time before pub-
li ation of the solutions. The Mayhem Sta ask that ea h solution be submitted on
a separate page and that the solver's name and onta t information be in luded with
ea h solution.
Ea h problem is given in English and Fren h, the o ial languages of Canada.
In issues 1, 3, 5, and 7, English will pre ede Fren h, and in issues 2, 4, 6, and 8,
Fren h will pre ede English.
The editor thanks Jean-Mar Terrier of the University of Montreal for transla-
tions of the problems.
M432 . Proposed by the Mayhem Sta.

Determine the value of d with d > 0 so that the area of the quadrilateral
with verti es A(0, 2), B(4, 6), C(7, 5), and D(d, 0) is 24.
M433 . Proposed by Bru e Shawyer, Memorial University of Newfound-
land, St. John's, NL.
In triangle ABC , AB < BC , L is the midpoint of AC , and M is the
midpoint of AB . Also, P is the point on LM su h that M P = M A. Prove
that P BA = P BC .
M434. Proposed by Heisu Ni olae, P^rjol Se ondary S hool, Ba au,
Romania.
Determine all eight-digit positive integers abcdef gh whi h satisfy the
relations a3 b2 = 2, c3 d2 = 4, 2e f 2 = 7, and g3 h2 = 1.
M435 . Proposed by Mihaly Ben ze, Brasov, Romania.
Prove that

X
n
1 1 n(n + 2)
1+ 2
+ 2
= .
k=1
k (k + 1) n+1
135
M436 . Proposed by Ne ulai Stan iu, George Emil Palade Se ondary

S hool, Buzau, Romania.
Determine the smallest possible value of x + y, if x and y are positive
integers with 2008
2009
x
< <
y
2009
2010
.
M437 . Proposed by Samuel Gomez

Moreno, Universidad de Jaen,
Jaen,

Spain.
Let x denote the greatest integer not ex eeding x. For example,
3.1 = 3 and 1.4 = 2. Let {x} denote the fra tional part of the
real number x, that is, {x} = x x. For example, {3.1} = 0.1 and
{1.4} = 0.6. Determine all rational numbers x su h that x {x} = x.
.................................................................
M432
. Propose par l'Equipe de Mayhem.
Determiner
la valeur de d, d > 0, telle que l'aire du quadrilatere
de
sommets A(0, 2), B(4, 6), C(7, 5) et D(d, 0) soit de 24.
M433 . Propose par Bru e Shawyer, Universite Memorial de Terre-

Neuve, St. John's, NL.
Dans un triangle ABC ave AB < BC , soit L et M les points milieux
respe tifs de AC et AB . De plus, soit P le point sur LM tel que l'on ait
M P = M A. Montrer que P BA = P BC .
M434
. Propose par Heisu Ni olae, E ole se ondaire P^rjol, Ba au, Roumanie.
Trouver tous les entiers positifs a huit hires abcdef gh satisfaisant
les relations a3 b2 = 2, c3 d2 = 4, 2e f 2 = 7 et g3 h2 = 1.
M435 . Propose par Mihaly Ben ze, Brasov, Roumanie.
Montrer que

X
n
1 1 n(n + 2)
1+ + = .
k=1
k2 (k + 1)2 n+1
M436
. Propose par Ne ulai Stan iu, E ole se ondaire George Emil Palade,
Buzau, Roumanie.
Trouver la plus petite valeur possible de x+ y, si x et y sont des entiers
2008 x 2009
positifs tels que 2009
< <
y 2010
.
136
M437 . Propose par Samuel Gomez

Moreno, Universite de Jaen,
Jaen,

Espagne.
On note x le plus grand entier n'ex edant
pas x. Par exemple, on a
3.1 = 3 et 1.4 = 2. On note {x} la partie fra tionnaire du nombre
reel
x, 'est-a-dire
{x} = x x. Par exemple, on a {3.1} = 0.1 et
{1.4} = 0.6. Trouver tous les nombres rationnels x tels que x {x} = x.
Mayhem Solutions
Totten{M1 . Proposed by Shawn Godin, Cairine Wilson Se ondary
S hool, Orleans, ON.
An ient Egyptians wrote all fra tions in terms of distin t unit fra tions
(that is, in terms of distin t fra tions with numerators of 1). For example,
instead of writing 11
12
, they would write 12 + 13 + 12
1
. The unit fra tion 12 an
be written in terms of other unit fra tions as 12 = 13 + 16 . Find an innite
family of unit fra tions ea h of whi h an be written as the sum of two unit
fra tions.
Solution by Katherine Janell Eyre, student, Angelo State University, San
Angelo, TX, USA.
We show that any unit fra tion n1 with n a positive integer and n 2
an be written as the sum of two distin t unit fra tions. To do this, we need
to nd positive integers x and y su h that n1 = x1 + y1 . By multiplying
through by xyn and manipulating, we obtain the equivalent equations
xy = yn + xn ;
xy yn xn + n2 = n2 ;
(x n)(y n) = n2 .
Sin e a possible fa torization of n2 is 1 n2 , then we an let x n = 1 (and

so x = n + 1) and y n = n2 (and so y = n + n2 = n(n + 1)).
Thus,
1 1 1
= +
n n+1 n(n + 1)
for n 2. This represents an innite family of unit fra tions ea h of whi h
an be written as the sum of two distin t unit fra tions.

Also solved by SAMUEL GOMEZ MORENO, Universidad de Jaen, Jaen,
Spain; RICHARD
I. HESS, Ran ho Palos Verdes, CA, USA; GEOFFREY A. KANDALL, Hamden, CT, USA; HUGO

LUYO SANCHEZ, Ponti ia Universidad Catoli a
IES
del Peru, Lima, Peru; RICARD PEIRO,
\Abastos", Valen ia, Spain; and D.J. SMEENK, Zaltbommel, the Netherlands. There was one
in orre t solution submitted.
137
Totten{M2 . Proposed by Bru e Shawyer, Memorial University of New-

foundland, St. John's, NL.
..............
The boundary of the shadow on the moon is

................ ...............................
......... .......
....... .... ......
.
...... ...
... .....
always a ir ular ar . On a ertain day, the moon

.
.... ... ....
....
. .. ...
...
... ..
. ...
is seen with the shadow passing through diamet-

.. .. ..
... ..
.. ..
..
... ... ...
...
ri ally opposite points. If the entre of the ir ular

.. ..
... ..
... ...
... .
... .. ...
...
ar forming the shadow is on the ir umferen e of
.. ..
... .. .
.
. ..
.. ... ..
.. . ..
the moon, determine the exa t proportion of the

... .. ...
... ..
... .. ...
.... ..
. ....
.
moon that is not in shadow.

. ..
.... ... ....
.....
....... ... .....
......... ... .......
.............. ....... ......................
.......................
Solution by Georey A. Kandall, Hamden, CT, USA.

Let r denote the radius of the moon and C
its entre, and R denote the radius of the ir u- P
lar ar forming the shadow and O the entre of .................................................
.......... ....... ..... ...... .................
this ar . Let P and Q be the points where the

........ ... .... ...... ......
...... .....
..
. .... .. ..... .....
... ....
.. .... ... ....
shadow interse ts the ir umferen e of the moon.

... ... .... .. ...
...
. .. . .
. .
.. ...
..
. . .... .
.
. .... ...
y z
.
Let x denote the area of P OQ, y the area of the x

.
... ...... ..... .. ..
..
... ..... .... ..
.. ..
..
.. ..... ..
.
O
.. ..
region between P Q and the ar through P and Q C

.........
q ..
. ..
........ ...
. ...
. ...
.. ..... ...
. ...
. ..
.
entred at O, and z the area of the region between

... .... .
. .
.. .
.
.. .... .
. .
.. .... .... .. ..
... .... ..
.. ..
.... .... ...
the two ar s through P and Q.

...
... .... .... ... ...
.... .... ..
. ... .....
.... .... . . ..
..... .... ..... ..... ....
Sin e P C = CQ = r, OP = OQ = R, and
....... .... .. ... .....
......... .......
.............. .................. ......................
.......................
P
OQ = 90 (be ause P Q is a diameter), then Q
2R = 2r , and so R = 2r .
Sin e P OQ = 90 , then the sum of areas x and y is one-quarter of

the area of the ir le entred at O, or x + y = 14 R2 = 41 ( 2r)2 = 12 r2 .
Also, the sum of the areas y and z is one-half of the area of the ir le
entred at C , or y + z = 12 r2 .
Thus, x + y = y + z and so x = z . Sin e x equals the area of P OQ

whi h is right-angled at O, then z = x = 12 ( 2r)2 = r2 .
Thus, the area of the region that is not in the shadow equals the area
of the entire ir le entred at C minus z , or r2 r2 = r2 ( 1) and the
r 2 ( 1) 1
exa t proportion of the moon that is not in shadow is = .
r 2

Also solved by RICHARD I. HESS, Ran ho Palos Verdes, CA, USA; and RICARD PEIRO,
IES \Abastos", Valen ia, Spain.
Totten{M3 . Proposed by John Ciriani, Kamloops, BC.

Prove that the quadrati equation ax2 + bx + c = 0 does not have a
rational root if a, b, and c are odd integers.
Solution by Samuel Gomez Moreno, Universidad de Jaen,
Jaen,
Spain.
p
Suppose that the rational number q , where p and q are relatively prime
integers, is a root of the quadrati equation ax2 + bx + c = 0. Then we have
138
2
p p
that a
q
+b
q
+ c = 0, whi h yields ap2 + bpq + cq2 = 0.
We are given that a, b, and c are odd integers. Sin e p and q are rela-
tively prime, they annot both be even. There are three ases to onsider:
If p is even and q is odd, then ap2 and bpq are even and cq 2 is odd, so
ap2 + bpq + cq 2 is odd, and thus annot equal 0.
Similarly, if p is odd and q is even, then ap2 + bpq + cq 2 is odd, and
thus annot equal 0.
If p is odd and q is odd, then ap2 , bpq , and cq 2 are all odd, so again we
see that ap2 + bpq + cq2 is odd, and thus annot equal 0.
Therefore, the quadrati equation ax2 + bx + c = 0 with a, b, and c
all odd integers annot have a rational root.
Also solved by EDIN AJANOVIC, student, First Bosniak High S hool, Sarajevo, Bosnia
and Herzegovina; GEORGE APOSTOLOPOULOS, Messolonghi, Gree e; KATHERINE JANELL
EYRE, student, Angelo State University, San Angelo, TX, USA; RICHARD I. HESS, Ran ho
Palos Verdes, CA, USA; MATT HUBBS, student, Missouri State University, Springeld, MO,
USA; GEOFFREY A. KANDALL, Hamden, CT, USA; KONSTANTINOS AL. NAKOS, Agrinio,
Gree e; RICARD PEIRO, IES \Abastos", Valen ia, Spain; EDWARD T.H. WANG, Wilfrid Laurier
University, Waterloo, ON; and KONSTANTINE ZELATOR, University of Pittsburgh, Pittsburgh,
PA, USA.
Miguel Amengual Covas pointed out that this problem appeared as problem no. 178 in
Eureka, Vol. 2, No. 8 (O tober), 1976, p. 171. [Ed.: Eureka be ame Crux Mathemati orum,
whi h eventually be ame Crux Mathemati orum with Mathemati al Mayhem.
A similar argument to the featured one shows that an even degree polynomial with odd
integer oe ients annot have a rational root.
Totten{M4 . Proposed by Bill Sands, University of Calgary, Calgary,

AB.
In a survey, some students were asked whether they liked the olour
orange. Exa tly 2% of the boys in the survey liked orange, while exa tly 59%
of the girls in the survey liked orange. Altogether, exa tly 17% of the students
in the survey liked orange. Find the smallest possible number of students in
the survey.
Let B be the number of boys in the survey and G the number of girls.
From the given information, 0.02B is the number of boys that liked orange,
0.59G is the number of girls that liked orange, and 0.17(B + G) is the total
number of students that liked orange. Ea h of these three quantities must
be an integer.
Note that 0.02B + 0.59G = 0.17(B + G) whi h gives 2B + 59G =
17B + 17G, or 42G = 15B , or 5B = 14G, or G : B = 5 : 14. Thus, there
is an integer t su h that G = 5t and B = 14t. Sin e we seek the minimum
value of B + G = 19t, we want to nd the minimum value of t.
17 17 19
But 0.17(B + G) = 100 (5t + 14t) =
100
t is also an integer; that
is, 100 | 17 19 t. Sin e gcd(17 19, 100) = 1, then 100 | t. The smallest
su h positive value of t is therefore t = 100.
139
This value of t gives G = 5t = 500 and B = 14t = 1400, whi h are

admissible values of G and B sin e 2% of B is 10 and 59% of G is 295, and
these are integers. Above we saw that 17% of B + G is an integer sin e
100 | t.
Therefore, the smallest possible number of students surveyed is 1900.
Also solved by RICHARD I. HESS, Ran ho Palos Verdes, CA, USA; HUGO LUYO

SANCHEZ, Ponti ia Universidad Catoli a
IES
del Peru, Lima, Peru; and RICARD PEIRO,
\Abastos", Valen ia, Spain.
Totten{M5 . Proposed by Ovidiu Furdui, Campia Turzii, Cluj, Romania.

Let a 6= 1 be a positive real number. Determine all pairs of positive
integers (x, y) su h that loga x loga y = loga (x y).
Suppose that x and
ysatisfy loga x loga y = log
a(x y). Sin e
x
loga x loga y = loga , then loga (x y) = loga xy and so by the
y
one-to-one property of f (u) = loga u, it follows that xy = x y.
From this we obtain x = xy y2 or y2 = x(y 1) whi h gives
y2 y2 1 + 1
x = =
y1 y1
(y 1)(y + 1) + 1 1
= = y+1+ .
y1 y1
1
Sin e we are seeking pairs of positive integers (x, y), then y 1
is an integer.
2
Thus, y 1 = 1, or y = 2, and so x = y y 1 = 4.
Therefore, the only pair (x, y) satisfying the relation is (4, 2).
and Herzegovina; GEORGE APOSTOLOPOULOS, Messolonghi, Gree e; LUIS J. BLANCO (stu-
dent) and ANGEL PLAZA, University of Las Palmas de Gran Canaria, Spain; SAMUEL GOMEZ
MORENO, Universidad de Jaen,
Jaen,
Spain; RICHARD I. HESS, Ran ho Palos Verdes, CA, USA;
KONSTANTINOS AL. NAKOS, Agrinio, Gree e; RICARD PEIRO, IES \Abastos", Valen ia, Spain;
and KONSTANTINE ZELATOR, University of Pittsburgh, Pittsburgh, PA, USA.
Totten{M6 . Proposed by Suzanne Feldberg, Thompson Rivers Univer-

sity, Kamloops, BC.
It is widely known how to draw a 5-pointed star qui kly. To make it
symmetri , one pla es 5 verti es at 72 intervals about a ir le and onne ts
the verti es with line segments of equal length without lifting one's pen. By
starting from a xed point and using the same method, one an draw two
dierent (and symmetri ) 7-pointed stars without lifting one's pen.
140
...r
......
.............................
...... ..r
.............................
.....
....
.
..... ........
..... .....
....
...
... .....
....
.....
......
.
.....
....
...
...
....r......................................................r
... ... ...
...
..
..r.......... .. .. ...........r
... ... ...
...
..
..... .... .... .. .. .. .... ............ ................. ..
.. ..... .. .. .... ................. ... ..
..
... ...... ... ..
..
...
.. .. .... ... .. ..
..
..
..
...
..
..
.. ....
.
........... ..
...
.. ..
..
.. . .. ... .................... ........................... ..
...
.
..
.r........ ...
..
.. .......... ................r .
.
.
. ..r..........................................................................r
...
..
...
.
. .. ..
..
... ............... .......... . ...
. ..
..
... .. ....... .. ..
...
...
...... ...................... ..
.... ....
....
.. ...
...
...
....
......... ............. ....
....
..
.... ....
....
..r.... ......
.........
.......r
...................
......
.... ..r..
....
......
.........
...r
...................
......
....
How many dierent 6-pointed, 8-pointed, or 9-pointed stars an one

draw this way? How many dierent n-pointed stars an one draw this way?
Solution by Ri hard I. Hess, Ran ho Palos Verdes, CA, USA, modied by the
editor.
There are no su h 6-pointed stars, one su h 8-pointed star, and two
su h 9-pointed stars.
Suppose that n 5. We an dene su h an n-pointed star more pre-
isely as joining n regularly spa ed points on a ir le by starting at a xed
point and drawing n ongruent lines onne ting points whi h are su essively
k spa es apart, with 1 < k < 21 n. We ex lude k = 1 sin e this would reate
an n-gon. We ex lude k = 12 n (in the ase where n is even), be ause only two
lines would be drawn. We ex lude 12 n < k < n be ause these would reate
ongruent stars to the stars ounted above by drawing them in the opposite
dire tion.
We note also that k annot be a divisor of n, sin e after nk lines we
would be ba k to the original starting point, and hen e we would not have
drawn n lines.
Similarly, if k and n are not oprime, they share a ommon divisor and
the only points whi h would be onne ted when drawing the star onne ting
points whi h are k spa es apart would be the same points as for the star on-
ne ting points whi h are gcd(k, n) spa es apart. Sin e gcd(k, n) is a divisor
of n, not all of the points would be used.
Thus, in order to omplete an n-pointed star, n and k must be oprime
and 1 < k < 12 n. The number of ways in whi h an n-pointed star an
be drawn equals the number of su h k. There are (n) integers k oprime
to n satisfying 1 k n, and 12 (n) integers k oprime to n satisfying
1 k 12 n (noting that 12 n is either not an integer or is not oprime to n).
Sin e we omit k = 1 whi h is oprime to n, then there are
1
(n) 1
2
stars that an be drawn. This is well-dened, sin e (n) is even for n > 4.
This agrees with the stated results for n = 6, n = 8, and n = 9.
A partially omplete solution was also submitted.
141
Totten{M7 . Proposed by John Grant M Loughlin, University of New

Brunswi k, Frederi ton, NB.
An unmarked ruler is known to be exa tly 1 4
6 m in length. It is possible to exa tly measure ...........................................................................
all integer lengths from 1 m to 6 m using only 2
.. .
.....................................................................
marks, as shown in the diagram, at 1 m and 4 m, 6 m
...
....................................................
...
....................................................
sin e 2 = 6 4, 3 = 4 1, and 5 = 6 1.
... ...
Suppose that an unmarked ruler is known to be exa tly 30 m in length.

(a) Find a way of pla ing 9 or fewer marks on the ruler to be able to exa tly
measure all integer lengths from 1 m to 30 m.
(b) Prove that at least 7 marks are needed to be able to exa tly measure all
integer lengths from 1 m to 30 m.

( ) Determine the smallest number of marks required on the ruler to be
able to exa tly measure all integer lengths from 1 m to 30 m.
Solution to (a) by the Mayhem Sta; solution to (b) by Ri hard I. Hess, Ran-
ho Palos Verdes, CA, USA, modied by the editor.
(a) If marks are pla ed 1, 2, 3, 4, 5, 10, 15, 20, and 25 m from the left end,
then these distan es an be measured from that end of the ruler. Distan es
from 6 m to 9 m in lusive an be measured from the 10 m mark to the
appropriate mark at 1, 2, 3, or 4 m. Distan es from 11 m to 14 m in lusive
an be measured in a similar way using the 15 m mark. Similarly, we an
measure distan es from 16 m to 19 m, 21 m to 24 m, and 25 m to 29 m.
Therefore, all distan es from 1 m to 30 m an be measured.
(b) Suppose there are k marks. Ea h mark an potentially measure a dif-
ferent distan e to ea h end of the ruler. Thus, the marks an measure up
to 2k distin t distan es using the ends of the ruler. Also, ea h pair of marks
an potentially measure a dierent distan e. There are k2 su h pairs. There-

fore, at most 2k + k2 = 12 k(k + 3) distan es an be measured using k marks.
If k = 6, this total is 27. Therefore, at least 7 marks are needed to measure
the 30 distan es.
No omplete solution to part ( ) was re eived, so this part remains open.
Using a omputer, both Hess, and Samuel Gomez Moreno, Universidad de Jaen,
Jaen,

Spain veried that at least 8 marks are needed in part ( ). In addition, Gomez
Moreno provided
some web links to information about \perfe t rulers" and \optimal rulers".
Totten{M8. Proposed by Edward J. Barbeau, University of Toronto,

Toronto, ON.
Let T be the set of all ordered triples (a, b, c) of positive integers su h
that a < b < c. We say that two triples (a, b, c) and (u, v, w) are equivalent
if a : b : c = u : v : w. We use this relation to partition T into equivalen e
lasses. The triple (a, b, c) is geometri if ac = b2 (that is, its terms form a
geometri sequen e) and harmoni if a1 + 1c = 2b (that is, the re ipro als of
its terms form an arithmeti sequen e).
142
(a) Verify that if (a, b, c) is geometri , then all triples equivalent to it are
also geometri .
(b) Verify that if (a, b, c) is harmoni , then all triples equivalent to it are
also harmoni .
( ) Let G be the set of equivalen e lasses of geometri triples and H be the
set of equivalen e lasses of harmoni triples. Determine a one-to-one
orresponden e between G and H .
Solution to (a) and (b) by Ja lyn Chang, student, Western Canada High S hool,
Calgary, AB; solution to ( ) by the proposer, ea h modied by the editor.
Throughout the solution, we use the notation (a, b, c) (u, v, w) to
mean that the triples (a, b, c) and (u, v, w) are equivalent members of T . If
(a, b, c) (u, v, w), then a : b : c = u : v : w by denition, so for some
positive real number k we have (u, v, w) = (ka, kb, kc).
(a) Suppose that (a, b, c) is geometri . Then by denition, ac = b2 . If
(u, v, w) (a, b, c), then (u, v, w) = (ka, kb, kc) for some positive real
number k. Sin e ac = b2 , then
uw = (ka)(kc) = k2 ac = k2 b2 = (kb)2 = v 2 ,
and so (u, v, w) is geometri .
1 1 2
(b) Suppose that (a, b, c) is harmoni . By denition,
a
+
c
= . If
b
(u, v, w) (a, b, c), then (u, v, w) = (ka, kb, kc) for some positive real
1 1 2
number k. Sin e a
+ = ,
c b
then

1 1 1 1 1 1 1 1 2 2 2
+ = + = + = = = ,
u w ka kc k a c k b kb v
so (u, v, w) is harmoni .
( ) Suppose that (a, b, c) is a geometri triple of positive integers with
a < b < c. Then we an write (a, b, c) = (a, ar, ar 2 ) for some positive real
number r. Dene a fun tion f by
f (a, b, c) = f (a, ar, ar 2 ) = (a(r+1), 2ar, ar(r+1)) = (a+b, 2b, b+c) .
Note that a + b < b + b < b + c, sin e a < b < c, so (a + b, 2b, b + c) is a
triple in T . Furthermore,
1 1 r+1 1 2
+ = = = ,
a(r + 1) ar(r + 1) ar(r + 1) ar 2ar
so f (a, b, c) is a harmoni triple. Note also that if (u, v, w) (a, b, c) with
(u, v, w) = (ka, kb, kc), then
f (u, v, w) = f (ka, kb, kc) = (ka + kb, 2kb, kb + kc)
= k(a + b, 2b, b + c) = k f (a, b, c) .
143
This tells us that f maps equivalent triples to equivalent triples, whi h means
that f is a well-dened map from G to H .
Suppose now that (A, B, C) is a harmoni triple of positive integers
with A < B < C . Dene g(A, B, C) = (A2 , AC, C 2 ). Then the ompo-
nents of g(A, B, C) are integers with A2 < AC < C 2 so g(A, B, C) is in
T . Note that A2 C 2 = (AC)2 , so g(A, B, C) is geometri . By dire tly al-
ulating as we did above, we an show that if (U, V, W ) (A, B, C), then
g(U, V, W ) g(A, B, C). This tells us that g maps equivalent triples to
equivalent triples, so g is a well-dened map from H to G.
Finally, we need to show that g(f (a, b, c)) (a, b, c) and also that
f (g(A, B, C)) (A, B, C). This will show that f and g are inverses of ea h
other.
If (a, b, c) is geometri , then (a, b, c) = (a, ar, ar2 ) for some real num-
ber r, so
g(f (a, b, c)) = g(a(r + 1), 2ar, ar(r + 1))
= (a2 (r + 1)2 , a2 r(r + 1)2 , a2 r 2 (r + 1)2 )
= a(r + 1)2 (a, ar, ar 2 )
(a, b, c) .
1 1 2 A+C 2
If (A, B, C) is harmoni , then A
+
C
=
B
and so AC
=
B
, or
2AC
B=
A+C
. Thus,
f (g(A, B, C)) = f (A2 , AC, C 2 )

= (A2 + AC, 2AC, AC + C 2 )

2AC
= (A + C) A, ,C
A+C
= (A + C) (A, B, C)
(A, B, C) .
Therefore, f and g are inverses of ea h other, so they ea h dene a

one-to-one orresponden e between G and H .
Also solved by RICHARD I. HESS, Ran ho Palos Verdes, CA, USA.
Totten{M9 . Proposed by Kirk Evenrude, Kamloops, BC.

A train 900 m long, travelling at 90 km/h, approa hes a 100 m long
bridge.
(a) How many se onds does it take the train to lear the bridge?
(b) Suppose that, just as the train rea hes the bridge, it begins to slow
down at the rate of 0.2 m/s2 . Now how long does it take to lear the
bridge?
144

(a) The train's given speed of 90 km/h is equivalent to 90 000 m/h. Sin e
there are 60 se onds in 1 minute and 60 minutes in 1 hour, then there are
60 60 = 3600 se onds in one hour. Thus, the train's speed is equivalent to
90 000 3600 = 25 m/s.
To lear the bridge, the train must travel 1000 m; that is, the front of
the train travels 900 m from the start of the bridge to the end of the bridge,
then an additional 100 m as the rear of the train nishes rossing the bridge.
To travel 1000 m at 25 m/s, it takes the train 1000 25 = 40 se onds.
(b) The velo ity in metres per se ond of the train t se onds after the train
rea hes the bridge is v(t) = 250.2t, sin e the train de elerates at 0.2 m/s2 .
Note that if t < 125, then v(t) > 0, so the train is moving forwards when
0 t < 125. (Also if the train ontinues \slowing down" for t > 125, then
v(t) < 0, so the train moves ba kwards in that ase.)
The distan e in metres that the front of the train travels t se onds after
the train rea hes the bridge is s(t) = 0 + 25t 12 (0.2)t2 = 25t 0.1t2 . (This
formula is a standard formula in physi s involving the initial displa ement,
initial velo ity, and onstant a eleration.)
To determine the time required to lear the bridge, we solve the equa-
tion s(t) = 1000, or equivalently
25t 0.1t2 = 1000 ;
0 = t2 250t + 10000 ;
0 = (t 50)(t 200) .
Therefore, t = 50 or t = 200. We want the smallest solution, so the time to
lear the bridge is 50 s. (The se ond solution omes from the ase where the
train lears the bridge, ontinues to slow down, omes to a stop, and then
ontinues ba kwards until rea hing the bridge again.)
Also solved by PAUL BRACKEN, University of Texas, Edinburg, TX, USA; JACLYN
CHANG, student, Western Canada High S hool, Calgary, AB; EMILY HENDRYX, student,
Angelo State University, San Angelo, TX, USA; MISSOURI STATE UNIVERSITY PROBLEM
SOLVING GROUP, Springeld, MO, USA; RICHARD I. HESS, Ran ho Palos Verdes, CA, USA;
and RICARD PEIRO, IES \Abastos", Valen ia, Spain.
Totten{M10. Proposed by Ni holas Bu k, College of New Caledonia,

Prin e George, BC.
Show that if p is a prime number, and A and B are positive integers
su h that p divides A, p2 does not divide A, and p does not divide B , then
the Diophantine equation Ax2 + By2 = p2008 does not have any solutions
in positive integers x and y.
For ea h nonnegative integer n, let P (n) be the statement \The equa-
tion Ax2 + By2 = pn has a positive integer solution (x, y)". We will show
that P (n) is false for all su h integers n.
145
Note rst that P (0) and P (1) are false, sin e the fa t that A, B , x, and
y are positive integers and A p (be ause A is divisible by p) implies that
Ax2 + By 2 Ax2 + 1 p + 1, so Ax2 + By 2 6= 1 and Ax2 + By 2 6= p for
all positive integers x and y.
Assume that P (n) is true for some integers n, and that k 2 is the
smallest positive integer for whi h P (k) is true, say with Ar2 + Bs2 = pk .
Sin e p | A and p2 A, then A = pC for some positive integer C with
p C . Thus, pCr 2 + Bs2 = pk , and so Bs2 = pk pCr 2 . Sin e the right
side is divisible by p, then p | Bs2 . Sin e p B , then p | s2 , whi h means
that p | s; say s = pt for some positive integer t.
Then pCr2 + Bp2 t2 = pk , whi h yields Cr2 + Bpt2 = pk1 , or equiv-
alently Cr2 = pk1 Bpt2 . By an argument similar to the above, p | r, and
so r = pu for some positive integer u.
Therefore, Cp2 u2 + Bpt2 = pk1 , or Cpu2 + Bt2 = pk2 , or nally
Au + Bt2 = pk2 . But this means that (x, y) = (u, t) is a positive integer
2
solution to Ax2 + By2 = pk2 , whi h implies that P (k 2) is true, whi h

ontradi ts the minimality of k.
Therefore, P (n) is false for ea h n; in parti ular, Ax2 + By2 = p2008
has no positive integer solutions.
and Herzegovina; and RICHARD I. HESS, Ran ho Palos Verdes, CA, USA.
Problem of the Month

Ian VanderBurgh
Any ommuters in the rowd?

Problem 1 (2009 Sun Life Finan ial Canadian Open Mathemati s Challenge)
Suppose that f and g are fun tions. The real number x is a real xed point
of f if f (c) = c. We say that f and g ommute if f (g(x)) = g(f (x)) for all
real numbers x.
(a) If f (x) = x2 2, determine all real xed points of f .
(b) If f (x) = x2 2, determine all ubi polynomials g that ommute with f .
To solve the rst part we must understand and apply the denition.
. We want to nd all real numbers c for whi h f (c) = c.
Solution to part (a)
Sin e f (x) = x2 2, then we want to solve c2 2 = c, or c2 c 2 = 0.
Fa toring, we obtain (c 2)(c + 1) = 0, and so c = 2 or c = 1. We an
he k that f (2) = 2 and f (1) = 1, so the real xed points of f are x = 2
and x = 1.
We ould solve the se ond part using a lot of algebra.
146
. Here g(x) is a ubi polynomial, so suppose that

Solution 1 to part (b)
g(x) = ax3 + bx2 + dx + e for some real numbers a, b, d, e with a 6= 0.
For f (g(x)) = g(f (x)), we need
f (ax3 + bx2 + dx + e) = g(x2 2) ;
(ax + bx2 + dx + e)2 2
3
= a(x2 2)3 + b(x2 2)2 + d(x2 2) + e .
After some painful algebra, we nd that
(ax3 + bx2 + dx + e)2 2
= a2 x6 + (2ab)x5 + (2ad + b2 )x4 + (2ae + 2bd)x3
+ (2be + d2 )x2 + (2de)x + (e2 2)
and
a(x2 2)3 + b(x2 2)2 + d(x2 2) + e
= ax6 + (6a + b)x4 + (12a 4b + d)x2 + (8a + 4b 2d + e) .
Sin e these two expressions are equal for all values of x, then we an equate
oe ients to obtain the following seven (!) equations:
a2 = a ,
2ab = 0 ,
2ad + b2 = 6a + b ,
2ae + 2bd = 0 ,
2be + d2 = 12a 4b + d ,
2de = 0 ,
e 2 = 8a + 4b 2d + e .
2
Take a deep breath. While there are seven equations, there are only four
variables. In what order should we work through these equations?
Sin e a 6= 0, the rst equation gives a = 1. The se ond equation
be omes 2b = 0, whi h yields b = 0. The third equation be omes 2d = 6,
whi h yields d = 3. The fourth equation be omes 2e = 0, whi h yields
e = 0. We an he k that these values of a, b, d, e satisfy the remaining
three equations.
Therefore, g(x) = x3 3x is the only ubi polynomial that an om-
mute with f (x). The algebra above onrms that it does indeed ommute
with f (x).
That was pretty gruesome. I was pretty surprised when we found that
b = e = 0. Does this tell us anything spe ial about g(x)? It tells us that the
fun tion g(x) is an odd fun tion, sin e
g(x) = (x)3 3(x)
= x3 + 3x
= (x3 3x)
= g(x) .
147
Is there a way of showing that g(x) is odd in advan e, before doing any of
this expansion?
The solution below gives us a way of showing that b = e = 0 and that
the fun tion g(x) is odd. Is this way easier? Likely not. But, it may be more
portable, be ause of one key fa t that it uses:
If p(x) is a polynomial of degree n, then the equation p(x) = 0
has at most n solutions.
We'll a tually use this fa t in a slightly dierent form:
Ifp(x) is a polynomial of degree at most n and the equation
p(x) = 0 has more than n solutions, then p(x) must be the zero
polynomial.
Solution 2 to part (b). Suppose that g(x) is a ubi polynomial for whi h
f (g(x)) = g(f (x)). To show that g(x) is an odd fun tion, we want to show
that g(x) = g(x) for all x.
We note rst that f is an even fun tion, sin e
f (x) = (x)2 2 = x2 2 = f (x)
for all x.
We know that f (g(x)) = g(f (x)) and f (g(x)) = g(f (x)). Sin e f
is even, this se ond equation be omes f (g(x)) = g(f (x)). Thus, we have
that f (g(x)) = f (g(x)) for all x.
Sin e f (x) = x2 2, then for every real number x we have
[g(x)]2 2 = [g(x)]2 2 ;
[g(x)]2 = [g(x)]2 ;
[g(x)]2 [g(x)]2 = 0;
[g(x) g(x)][g(x) + g(x)] = 0.
Therefore, for all x, we have g(x) + g(x) = 0 or g(x) g(x) = 0.

Using the same notation from Solution 1, we note that
g(x) = a(x)3 + b(x)2 + d(x) + e
= ax3 + bx2 dx + e .
Therefore, g(x) + g(x) = 0 is equivalent to

(ax3 + bx2 dx + e) + (ax3 + bx2 + dx + e) = 0 ,
or bx2 + e = 0. Also, g(x) g(x) = 0 is equivalent to

(ax3 + bx2 dx + e) (ax3 + bx2 + dx + e) = 0 ,
or ax3 + dx = 0.
148
Therefore, for every x, at least one of bx2 + e = 0 and ax3 + dx = 0 is

true. Sin e there are innitely many possible x, then one of these equations
is true for innitely many x. Our key fa t tells that this means that one of
the left sides must be the zero polynomial. Therefore, either b = e = 0 or
a = d = 0. Sin e g(x) is a ubi polynomial, then a 6= 0. Thus, b = e = 0
and so g(x) = ax3 + dx.
Finally, we need to determine the values of a and d, and so we must
nd a and d so that
(ax3 + dx)2 2 = a(x2 2)3 + d(x2 2) ;
a x + 2adx4 + d2 x2 2 =
2 6
ax6 6ax4 + (12a + d)x2 + (8a 2d) .
Equating the rst two oe ients, we obtain a2 = a and 2ad = 6a. Sin e
a 6= 0, the rst gives a = 1 from whi h the se ond gives 2d = 6 or d = 3.
Therefore, g(x) = x3 3x as in Solution 1.
A really good question to ask at this point is whether anything was
gained in Solution 2. We did less algebrai work, but we had to do a fair
bit more deep thinking. I would argue that, in fa t, Solution 2 is worth it,
be ause it gives us another tool in our toolbox. What if we wanted to nd
a seventh degree polynomial that ommuted with f (x)? Solution 1 would
be ome a fair bit uglier, whereas the method of Solution 2 would still work;
the initial set-up would be no more ompli ated while the algebrai work at
the end would be signi antly easier than in the Solution 1 approa h.
As a posts ript, I will leave you with part ( ) from this problem as it
appeared:
( ) Suppose that p and q are real-valued fun tions that ommute.
If 2[q(p(x))]4 + 2 = [p(x)]4 + [p(x)]3 for all real numbers x, prove that
q has no real xed points.
Happy ommuting!
149
THE OLYMPIAD CORNER

No. 285
R.E. Woodrow
We begin this number with the problems of the Estonian IMO team
sele tion test 2007. Thanks go to Bill Sands, Canadian Team Leader to the
IMO in Vietnam, for olle ting them for our use.
ESTONIAN IMO TEAM SELECTION CONTEST 2007
1 . A swit hboard has one row of n swit hes, where ea h swit h an be ei-
ther up or down. If a swit h ips down from the up position, then its right
neighbour (if present) automati ally ips. At the start all swit hes are down.
The operator of the board rst ips the leftmost swit h, then ips the se ond
leftmost swit h twi e, and so forth until eventually he ips the rightmost
swit h n times. After all these ips, how many swit hes are up?
2. Let D be the foot of the altitude of triangle ABC drawn from vertex A.
Let E , F be the points symmetri to D with respe t to the lines AB , AC ,
respe tively. Let triangles BDE , CDF have inradii r1 , r2 and ir umradii
R1 , R2 , respe tively. If SK denotes the area of gure K , prove that
|SABD SACD | |r1 R1 r2 R2 | .
bn 1
3 . Let n be a natural number, n 2. Prove that if b1
is a prime power
for some positive integer b, then n is prime.
D...........................................F ................................. C
4. In square ABCD the points E and F
........... ..... ...
... .... . .. ......... ...
are hosen in the interior of sides BC and ... ........ ... .......... . ...
.. ..... .. . . . ...
CD , respe tively. The line drawn from
.. ....... . .
... .. .
... .. ..... ... ... .....
... . . .
F perpendi ular to AE passes through
. ..
.. .... ........ ..... ..... .....
... .. ..... K... .. ..
the interse tion point G of AE and BD. ... .... ..... ......... ... ...
.. ... .........................................
... .. ..................................................................... E
.
A point K is hosen on F G su h that
.. .. .
....
...............................................................................G
.. ....... .............. ... ..
|AK| = |EF |. Find EKF .
...................
A B
5 . Find all ontinuous fun tions f : R R su h that for all x, y R,

f (x + f (y)) = y + f (x + 1) .
6 . Consider a 10 10 grid. A move onsists of olouring 4 unit squares

that lie at the interse tion of some two rows and two olumns. A move is
permitted if at least one of the 4 squares is previously un oloured. What is
the largest possible number of moves that an be taken to olour the whole
grid?
150
Next are the 10 Grade problems of the XXXIII Russian Mathemati al

th
Olympiad, Final Round 2006{2007. Thanks again go to Bill Sands, Canadian

Team Leader to the IMO in Vietnam, for olle ting them for our use.
XXXIII RUSSIAN MATHEMATICAL OLYMPIAD
2006-2007
Final Round
10th Grade
1 . (A. Polyansky) The surfa e of a 9 9 9 ube is divided into unit squares,

and is ompletely overed by 2 1 non-overlapping paper re tangles (ea h
re tangle overs two adja ent squares). Prove that the number of re tangles
whi h are bent over an edge is odd.
2 . (A. Khrabrov) Given a polynomial P (x) = a0 xn + a1 xn1 + + an , let
m = min{a0 , a0 + a1 , . . . , a0 + a1 + + an }. Prove that P (x) mxn
for all x 1.
3. (V. Astakhov) In an a ute triangle ABC , BB1 is a bise tor. Point K is
hosen on the smaller ar BC of the ir um ir le, su h that B1 K and AC
are perpendi ular. Point L is hosen on line AC su h that BL and AK are
also perpendi ular. Line BB1 meets the smaller ar AC at point T . Prove
that points K , L, T are ollinear.
4 . (K. Knop, O. Leontyeva) A magi ian and his assistant want to develop
the following tri k: A spe tator writes an arbitrary sequen e of N de imal
digits on a bla kboard. The assistant then overs two adja ent digits. Then
the magi ian enters the room, looks at the bla kboard, and announ es whi h
digits are overed. Find the smallest value of N for whi h they an ensure
that they an always perform this tri k su essfully.
5 . (N. Agakhanov) A set of n > 2 ve tors is given in the plane. A ve tor in
the set is long if its length is not less than the length of the sum of all the
other ve tors in the set. Prove that if ea h ve tor in the set is long, then the
sum of all the ve tors in the set is the zero ve tor.
6. (S. Berlov) Two ir les 1 and 2 interse t at points A and B . Let P Q
and RS be the segments of ommon tangents to these ir les (points P and
R lie on 1 , while points Q and S lie on 2 ). Ray RB interse ts 2 again at
point W . If RBkP Q, nd the ratio RB/BW .
7 . (D. Karpov) A onvex polyhedron F has a vertex A of degree 5, while
all other verti es are of degree 3. A olouring of the edges of F with three
olours is good, if all three olours are present at ea h vertex B 6= A. Prove
that if the number of good olourings is not divisible by 5, then there exists
a good olouring su h that three edges meeting at A have the same olour.
151
8 . (A. Golovanov) Dima omputed the de imal expansions of the numbers

1
, 1 , . . . , 99!
80! 81!
1
, whi h he wrote on 20 innite strips of paper (for example,
1
the last strip ontains the number 99! = 0. 00 . . 00} 10715 . . .). From one of
| .{z
155 zeros
these strips, Sasha uts a fragment onsisting of N onse utive digits and no
de imal point. Find the largest N for whi h Sasha an ut su h a fragment
without Dima being able to dedu e whi h one of 20 strips was ut.
We ontinue with the problems of the 11 Grade, Final Round of the

th
XXXIII Russian Mathemati al Olympiad 2006-2007. Thanks again go to Bill

Sands, Canadian Team Leader to the IMO in Vietnam, for olle ting these
problems for our use.
XXXIII RUSSIAN MATHEMATICAL OLYMPIAD
2006-2007
Final Round
11th Grade
1 . (N. Agakhanov) The produ t f (x) = cos x cos 2x cos 3x cos 2k x is

written on the bla kboard, where k 10. Prove that it is possible to repla e
one \cos" by a \sin" su h that the new produ t, f1 (x), satises the inequality
|f1 (x)| 3 21k for all real x.
2. (A. Polyansky) The in ir le of a triangle ABC tou hes sides BC , AC ,

and AB at points A1 , B1 , and C1 , respe tively. Segment AA1 interse ts
the in ir le again at point Q. Line is parallel to BC and passes through
A. Lines A1 C1 and A1 B1 interse t at points P and R, respe tively. Prove
that P QR = B1 QC1 .
3 . [Ed.: This is the same as problem 4 of the 10th Grade ontest.
4 . (A. Golovanov) An innite sequen e {xn } n=1 is dened as follows. The
rst term x1 is a rational number greater than 1, and xn+1 = xn + x1 for
n
all positive integers n, where xn is the greatest integer not ex eeding x.
Prove that this sequen e ontains an integer.
5 . (F. Petrov) At ea h vertex of a onvex 100-gon, two distin t numbers are
written. Prove that one an sele t one number at ea h vertex, su h that the
sele ted numbers at two adja ent verti es are always distin t.
6 . (N. Agakhanov, I. Bogdanov) Determine if there exist three nonzero real
numbers a, b, c su h that for every n > 3 there exists a polynomial Pn (x)
of the form Pn (x) = xn + + x3 + ax2 + bx + c whi h has only integer
roots.
152
7. (A. Zaslavsky) Given a tetrahedron T , a pair of its skew edges a and b is

large if T is overed by the spheres with diameters a and b. Does a large pair
ne essarily exist?
8. (I. Bogdanov, G. Chelnokov) Among N ities some pairs of them are
onne ted by nonstop air shuttles. For ea h k with 2 k N and for
any set of k ities, there are at most 2k 2 airlines onne ting these ities.
Prove that all ights an be distributed between two ompanies su h that
every ir ular trip uses the airlines of both ompanies.
Next we give the problems of the 20 Korean Mathemati al Olympiad,

th
Final Round, Mar h 24{25, 2007. Thanks again go to Bill Sands, Canadian
Team Leader to the IMO in Vietnam, for olle ting them for us.
th
20 KOREAN MATHEMATICAL OLYMPIAD
Final Round
March 24, 2007 Time 4.5 hours
1 . Triangle ABC is a ute with ir um ir le and ir um entre O. The ir le

has entre O , is tangent to O at A and to the side BC at D , and interse ts
the lines AB and AC again at E and F , respe tively. The lines OO and EO
interse t again at A and G, respe tively. The lines BO and A G interse t
at H . Prove that DF 2 = AF GH .
2 . Consider the sixteen tiles xed on a .......................................................................................
wall as shown below. How many ways
... .. .. .. ..
..........................................................................
.. .. .. .. ..
are there to write either 0 or 1 on ea h
... ... ... ... ...
.............................................................................
tile so that the produ t of the two num-
.. .. .. .. ..
.. .. .. .. ..
... ... ... ... ...
bers written on every neighbouring pair of .............................................................................
.. .. .. .. ..
tiles (sharing a ommon side) is always 0?
.. .. .. .. ..
.. . . . .
.......................................................................
3 . Find all triples (x, y, z) of positive integers satisfying 1 + 4x + 4y = z2.
March 25, 2007 Time 4.5 hours
4 . Find all pairs (p, q) of primes su h that pp + qq + 1 is divisible by pq.

5 . For the vertex A of ABC , let A be the point of interse tion of the angle
bise tor at A with side BC , and let A be the distan e between the feet of
the perpendi ulars from A to the lines AB and AC , respe tively. Dene B
and C similarly, and let be the perimeter of ABC . Prove that
A B C 1
.
3 64
153
6 . Let N be the set of positive integers, and let f :NN satisfy

kf (n) f (kn) kf (n) + k 1
for all k, n N.
(a) Prove that f (a) + f (b) f (a + b) f (a) + f (b) + 1 for all a, b N.
(b) Show that if f satises f (2007n) 2007f (n) + 2005 for all n N,
then f (2007c) = 2007f (c) for some c N.
As nal sets of problems for this number, we rst give the 2006/7 British
Mathemati al Olympiad, Round 1. Thanks are extended to Bill Sands, Cana-
dian Team Leader to the IMO in Vietnam, for olle ting these problems for
us.
2006/7 BRITISH MATHEMATICAL OLYMPIAD
Round 1
1. Find four prime numbers less than 100 whi h are fa tors of 332 232 .
2. In the onvex quadrilateral ABCD , points M , N lie on the side AB
su h that AM = M N = N B , and points P , Q lie on the side CD su h that
CP = P Q = QD . Prove that
1
Area of AM CP = Area of M N P Q = ( Area of ABCD) .
3
3 . The number 916238457 is an example of a nine-digit number whi h on-

tains ea h of the digits 1 to 9 exa tly on e. It also has the property that the
digits 1 to 5 o ur in their natural order, while the digits 1 to 6 do not. How
many su h numbers are there?
4 . Two tou hing ir les S and T share a ommon tangent whi h meets S at
A and T at B . Let AP be a diameter of S and let the tangent from P to T
tou h it at Q. Show that AP = P Q.
5. For positive real numbers a, b, c prove that
(a2 + b2 )2 (a + b + c)(a + b c)(b + c a)(c + a b) .

6 . Let n be an integer. Show that, if 2 + 2 1 + 12n2 is an integer, then it
is a perfe t square.
154
Lastly, we have the problems of Round 2 of the 2006/7 British Mathe-

mati al Olympiad, ourtesy of Bill Sands.
2006/7 BRITISH MATHEMATICAL OLYMPIAD
Round 2
1. Triangle ABC has integer-length sides, and AC = 2007. The internal

bise tor of BAC meets BC at D. Given that AB = CD, determine AB
and BC .
2. Show that there are innitely many pairs of positive integers (m, n) su h
that
m+1 n+1
+
n m
is a positive integer.
3. Let ABC be an a ute-angled triangle with AB > AC and BAC = 60 .
Denote the ir um entre by O and the ortho entre by H and let OH meet
AB at P and AC at Q. Prove that P O = HQ.
Note: The ir um entre of triangle ABC is the entre of the ir le whi h
passes through the verti es A, B and C . The ortho entre is the point of
interse tion of the perpendi ulars from ea h vertex to the opposite side.
4. In the land of Hexagonia, the six ities are onne ted by a rail network
su h that there is a dire t rail line onne ting ea h pair of ities. On Sundays,
some lines may be losed for repair. The passengers' rail harter stipulates
that any ity must be a essible by rail from any other (not ne essarily di-
re tly) at all times. In how many dierent ways an some of the lines be
losed subje t to this ondition?
Now we return to solutions from our readers to problems given in the

February 2009 number of the Corner and solutions to the Thai Mathemati al
Olympiad Examinations 2005, given at [2009 : 22.
1. Let P (x), Q(x), and R(x) be polynomials satisfying

2xP x3 + Q x x2 = 1 + x + x2 R(x) .
Show that x 1 is a fa tor of P (x) Q(x).

Solved by Matthew Babbitt, home-s hooled student, Fort Edward, NY, USA;
Mi hel Bataille, Rouen, Fran e; and Titu Zvonaru, Comane sti, Romania. We
give Babbitt's write-up.
We need to prove that x 1 is a fa tor of P (x) Q(x). In other words,
we need to show that 1 is a root of P (x) Q(x), or P (1) Q(1) = 0.
155
Sin e x3 1 = (x 1)(x2 + x + 1), the roots of x2 + x

+ 1 are third
1 + i 3 1 i 3
roots of unity, namely w = 2
and w2 = 2
. Note that
x x = (1 x x ) + 1 = 0 + 1 = 1 and x = 1. Thus, we have
2 2 3
2wP (1) + Q(1) = 2w 2 P (1) + Q(1) = 0 .
Hen e
2(w 2 w)P (1) = 0 = P (1) = 0 .
We now have Q(1) = 0 and P (1) Q(1) = 0, hen e (x 1)|(P (x) Q(x)).
2 . Find all fun tions f : R R su h that

f x + y + f (xy) = f f (x + y) + xy
for all x, y R.
Solution by Mi hel Bataille, Rouen, Fran e.
For a R, let fa : R R be dened by fa (x) = x + a. It is
readily he ked that fa is a solution for ea h real number a. We show that
there are no other solutions. To this aim, let f be an arbitrary solution
and let u, v be any real numbers. Choose s R su h that s2 > 4u and
s2 > 4v . Then the quadrati X 2 sX + u = 0 has two solutions x1 ,
y1 satisfying x1 + y1 = s and x1 y1 = u. Taking x = x1 , y = y1 in the
given equation yields f (s + f (u)) = f (f (s)) + u. Similarly, we obtain
f (s + f (v)) = f (f (s)) + v . It follows that if f (u) = f (v), then u = v , that
is, f is inje tive.
Now, let a = f (0). Setting y = 0 in the fun tional equation, we obtain
f (x + a) = f (f (x)) for ea h x, hen e f (x) = x + a and f = fa , as desired.
3 . Let a, b, and c be positive real numbers. Prove that

3 6
1 + .
ab + bc + ca a+b+c
Solved by George Apostolopoulos, Messolonghi, Gree e; Arkady Alt, San

Jose, CA, USA; Mi hel Bataille, Rouen, Fran e; Jose Luis Daz-Barrero, Uni-
versitat Polite ni a
de Catalunya, Bar elona, Spain; Joe Howard, Portales,
NM, USA; and Titu Zvonaru, Comane sti, Romania. We give the write-up by
Howard.
By the well-known inequality (a + b + c)2 3(ab + bc + ca), we have
1 1
.
3(ab + bc + ca) a+b+c
156
By the AM-GM Inequality and the above, we have

3
1 +
ab + bc + ca

3 6 6
2 = ,
ab + bc + ca 3(ab + bc + ca) a+b+c
and the result is proved.

4 . Let n be a positive integer. Prove that n(n + 1)(n + 2) is not a perfe t
square.
Solved by Arkady Alt, San Jose, CA, USA; Miguel Amengual Covas, Cala
Figuera, Mallor a, Spain; Matthew Babbitt, home-s hooled student, Fort
Edward, NY, USA; Edward T.H. Wang and Dexter Wei, Wilfrid Laurier Uni-
versity, Waterloo, ON; and by Konstantine Zelator, University of Pittsburgh,
Pittsburgh, PA, USA. We give the solution of Amengual Covas.
Assume on the ontrary that n(n + 1)(n + 2) is a perfe t square for
some positive integer n.
Sin e gcd(n, n + 1) = 1 and gcd(n + 1, n + 2) = 1, we have that
gcd(n + 1, n(n + 2)) = 1, and therefore ea h of the numbers n + 1 and
n(n + 2) is a perfe t square. Writing n + 1 = u2 and n(n + 2) = v 2 , where
u, v are integers, yields u4 1 = v 2 . This equation an be written in the
form (u2 v)(u2 + v) = 1, when e u2 v = 1 and u2 + v = 1. Solving,
we nd that u2 = 1, v = 0. This implies that n = 0, whi h ontradi ts that
n is a positive integer.
Thus our assumption was wrong, and onsequently n(n + 1)(n + 2),
where n is a positive integer, is not a perfe t square.

5. Find the least positive integer n su h that 2549| n2545 2541 .
Solution by Matthew Babbitt, home-s hooled student, Fort Edward, NY,
USA.
The given statement is equivalent to n2545 8 (mod 2549). Note
that 2549 is prime. Sin e n2545 is not a multiple of 2549, n is not a multi-
ple of 2549. By Fermat's Little Theorem, n2548 1 (mod 2549). Hen e,
8n3 1 (mod 2549), or (2n)3 1 n2548 (mod 2549). Sin e 3 is
not a divisor of 2548, it follows that 2n 1 (mod 2549), whi h means
that 2n 2548 (mod 2549). The least positive integer that satises this
last ongruen e is n = 2548
2
= 1274.
6 . Do there exist positive integers x, y, and z su h that

2548x + (2005)y = (543)z ?
157
and Konstantine Zelator, University of Pittsburgh, Pittsburgh, PA, USA. We
give Babbitt's write-up.
The answer is no.
Assume for the sake of ontradi tion that there are positive integer so-
lutions to the equation. Note that 2548 1 (mod 3), 2005 2 (mod 3),
and 543 0 (mod 3). Hen e the rst term is 1 (mod 3) and the third
term is 0 (mod 3). Therefore, the se ond term must be 0 1 2 (mod 3).
Hen e, y is odd, and so the se ond term is negative. Let y = 2y1 1. We are
now looking for positive integer solutions to 2548x 20052y 1 = (543)z . 1
Note that 2548 0 (mod 4) and 2005 543 1 (mod 4). There-
fore, the rst term is 0 (mod 4) and the se ond and third terms are 1 (mod 4).
Thus, 0 1 3 1 (mod 4), whi h is a ontradi tion.
Therefore, there are no positive integer solutions to the equation.
10 . Assume ABC is an isos eles triangle with AB = AC . Suppose that
P is a point on the extension of side BC . X and Y are points on lines AB
and AC su h that P XkAC and P Y kAB . Let T be the midpoint of ar BC .
Prove that P T XY . (Iran 2004)
Solved by Konstantine Zelator, University of Pittsburgh, Pittsburgh, PA, USA.
We introdu e oordinates. Let a and .. .
....
..
... ....
b be positive reals. Let the verti es of

..
.. ... .. ..
.. .. ..
.. .... .. ..
X(x1 , y1 )
..
triangle ABC be A(a, 0), B(b, 0), and

.. ..
. .. ..
.. .
. ....
.. . ....
... ..... .. ..
.. ....
C(b, 0). The point P (t, 0) is on the x-axis

.. ..
. .
. ..
.. .
.
. ..
..
.. ..... .. ....
.. .. ..
to the right of C(b, 0), so that t > b. We

.. ... .. ..
... .... .. ..
.. . ... ...
.. . .
.. ..... .... ..
also introdu e the points X(x1 , y1 ) and

.. ... .. ..
... .. .. ..
..
.. ... .. ...
Y (x2 , y2 ). The ar in the gure is part of

.. .. .. ..
.. .. .. ..
.. ... .. ...
.. .... ..
.......
the ir um ir le of triangle ABC .

..
A(0, a)
.... ..
.......... ..
...
.......
..
The origin (0, 0) is the midpoint of

... .... .... ...
.. ... ...
. ..
... .... .... ..
..
side AC , and sin e the triangle is isos eles

.... .... .... ..
...
. ... ..
. ..
... .... .... ..
...
with AB = AC , it follows that the en-

... ...
. ...
..
..
... ..
. ... ...
... ...
. .
... ..
..
tre of the ir ums ribed ir le is the mid- B(b, 0)

... ..
. .
... ...
... ...
. ... ..
... ...
. .
...
..
..
point of the line segment AT and AT is

.. ...
C(b, 0) P (t, 0) . . ... ..
... ..
. . ...
...............................................................................................................................................................................................................................
..................................... .
a diameter. The point T has oordinates

. ..
... ....
... .....
..
.. .. ....
.
T (0, k) ... ...
.. .... ..
..
T (0, k), k a positive real number.

... .. ... ....
.... .. .. ..
.. .. ...
.... .. .. ..
.. ..
Let A = 2w; we have

.... ..
.. .. ..
..
... .. .
. ..
.... .. ..
. ...
... .. .
..
..
....
p
.... ...
.... .. .
a2 + b2 ,
.. .. .
....
|AC| =
.. ..
... .. ..
... ..
... .. ...
..... Y (x2 , y2 )
... ....
b
,
....
... ...
sin w =
....
... .... ...
a2
+ b2
.... .
.
.... ...
... ..
a
..
.
.... ..
cos w = 2 ....
.... ..
....
a + b2
..
From the Law of Sines in the triangle ABC , we also have

b
2R sin A = 2b , R = ,
sin A
158
where R is the radius of the ir ums ribed ir le. Furthermore, we have that
sin A = sin 2w = 2 sin w cos w , hen e
2ab b a2 + b2
sin A = sin 2w = , and R = = .
a2 + b 2 sin A 2a
From the gure, |AT | = 2R = a + k, hen e

a2 + b2 b2
k = 2R a = a = .
a a

b2
and T (0, k) = T 0,
a
. Next, we express x1 , x2 , y1 , y2 in terms of a
a
and b. First, the slope of line P X is mP X =
b
; an equation for the line
P X is thus a
y = (x t) . (1)
b
a
The slope of line P Y is mP Y = ; an equation
b
for line P Y is thus
a
y = (x t) . (2)
b
An equation for line AB is
a
y = x + a. (3)
b
An equation for line AC is
a
y = x + a. (4)
b
The point X(x1 , y1 ) lies at the interse tion of the lines AB and P X. By
solving the system of equations (1) and (3), we nd that
tb a(t + b)
x1 = , y2 = . (5)
2 2b
Similarly, Y (x2 , y2 ) is the point of interse tion of the lines AC and P Y . By
solving the system of two equations (2) and (4), we nd that
b+t a(b t)
x2 = , y2 = . (6)
2 2b
Therefore, the slope of XY is
y2 y1 at
mXY = = 2 . (7)
x2 x1 b
The slope of line P T is
0 (k) k b2
mP T = = = . (8)
t0 t at
By (7) and (8), we have mXY mP T = 1, whi h proves that XY PT.
159
Now we turn to solutions to the 46 Ukrainian Mathemati al Olympiad

th
2006, Final Round, given at [2009 : 23{24.

1 . (V.V. Plakhotnyk) Prove that for any rational numbers a and b the graph
of the fun tion
f (x) = x3 6abx 2a3 4b3 , x R
has exa tly one point in ommon with the x-axis.

Mi hel Bataille, Rouen, Fran e; and Konstantine Zelator, University of
Pittsburgh, Pittsburgh, PA, USA. We give Bataille's version.
We will make use of the following result proved at the end: x3 3px+2q
vanishes exa tly on e for x R if and only if q2 > p3 or p = q = 0.
Here, p = 2ab and q = (a3 + 2b3 ), hen e q2 > p3 an be rewritten
as (a + 2b3 )2 > 8a3 b3 , that is, (a3 2b3 )2 > 0. This is ertainly true if
3
a3 6= 2b3 . However, a3 = 2b3 annot o ur if a, b 6= 0, sin e otherwise

the number 2 would be the ube of a nonzero rational number, whi h is
impossible (if m3 = 2n3 for positive integers m and n, then a ontradi tion
arises: the exponent of 2 in the standard fa torization is a multiple of 3 on
the left but not on the right).
Sin e p = q = 0 when a = b = 0, the ondition q2 > p3 or p = q = 0
is satised for all rational numbers a and b, and the result follows.
We now prove the result used above. Let P (x) = x3 3px + 2q. If
p 0, then for a, b R with a 6= b,
P (a) P (b)
= a2 + ab + b2 3p > 0 ,
ab
hen e P is in reasing on R and vanishes only on e.

If p > 0,thenP is in reasing on (, p) and

( p, ) and de-

reasingon ( p, p) . An easy al ulation

gives P (

p) = 2(q + p p)
and P( p) =2(q p p), so that P ( p) > P ( p) and we have that
P ( p) P ( p) = 4(q 2 p3 ). It follows that if q 2 < p3 , then P vanishes
on e in (p, p) as well as in
(, p) and in ( p, ).

If 0 < p3 < q2 , then P ( p) and P ( p) have the same sign and P
vanishes only on e. Lastly if p3 = q2 6= 0, then
2
q 2q
P (x) = x
p
x+
p
,
and P vanishes twi e. The required result follows from these observations.
5 . (O.O. Kur henko) Prove that for any real numbers x and y
| cos x| + | cos y| + | cos(x + y)| 1 .
160
Solved by Arkady Alt, San Jose, CA, USA; and Mi hel Bataille, Rouen, Fran e.
We give the argument of Bataille.
Let v = e2ix and w = e2iy . Then,

|1 + v| = eix (eix + eix ) = eix (2 cos x) = 2| cos x|
and similarly,
|1 + w| = 2| cos y|
and

|v + w| = ei(yx) (ei(x+y) + ei(x+y) ) = 2| cos(x + y)| .
Now, using the Triangle Inequality, we obtain

2 = |(1 + v) + (1 + w) (v + w)|
|1 + v| + |1 + w| + |v + w|

= 2 | cos x| + | cos y| + | cos(x + y)| ,
and the result follows.
6 . (T.M. Mitelman) Find all fun tions f : R R su h that

f x3 + y 3 = x2 f (x) + yf (y 2 )
for all real numbers x and y.

The solutions are the fun tions fm (x) = mx, where m a real number.
It is readily he ked that these fun tions satisfy the identity. We now
show that there are no other solutions. To this aim, let f be any solution.
Taking x = y = 0 in the identity yields f (0) = 0; also, with only y = 0, we
obtain f (x3 ) = x2 f (x), and with only x = 0, we obtain f (y3 ) = yf (y2 ).
Thus, for all real numbers x,
f (x3 ) = x2 f (x) = xf (x2 ) .
From the identity we now obtain f (x3 + y3 ) = f (x3 ) + f (y3 ). Sin e any
real number is the ube of a real number, it follows that
f (a + b) = f (a) + f (b)
for all real numbers a and b.

Consequently, f is odd (take b = a), and f (na) = nf (a) if n Z
and a R. Substituting x + 1 and x 1 for x and y in the identity, we
obtain on the one hand
f ((x+1)3 +(x1)3 ) = f (2x3 +6x) = 2f (x3 )+6f (x) = 2x2 f (x)+6f (x) ,
161
while on the other hand, using the identity and the fa ts established so far,

f (x + 1)3 + (x 1)3

= (x + 1)2 f (x) + f (1) + (x 1)2 f (x) f (1)
= 2x2 f (x) + 2f (x) + 4xf (1) .
By omparison, we see that f (x) = xf (1) for all x, so f (x) = fm (x) with
m = f (1) and we are done.
Next we turn to solutions from our readers to problems of the Cze h-

Polish-Slovak Mathemati al Competition 2006 given at [2009 : 24{25.
1. Five distin t points A, B , C , D , and E lie in this order on a ir le of
radius r and satisfy AC = BD = CE = r. Prove that the ortho entres
of the triangles ACD, BCD, and BCE are the verti es of a right-angled
triangle.
Solved by Mi hel Bataille, Rouen, Fran e; and Titu Zvonaru, Comane sti,
Romania. We give Bataille's solution.
Let O be the entre of the ir le ontaining A, B , C , D, E . Let H1 ,
H2 , H3 and G1 , G2 , G3 be the ortho entres and entroids of the triangles
ACD , BCD , BCE , respe tively. Sin e these triangles have O as a ommon

ir um entre, we have OHk = 3OGk for ea h k. Thus, triangles H1 H2 H3
and G1 G2 G3 are homotheti , and the problem is equivalent to showing that
the triangle G1 G2 G3 is right-angled.
Without loss of generality we take r = 1 and we shall use omplex
axes. Observing that OAC , OCE , OBD are equilateral triangles with the
same orientation, we may suppose that the axes of O, A, C , E , B , D are
0, 1, 2 , , u, u 2 , respe tively, where = e2i/3 and u = ei for
some 0, 3 . The axes g1 , g2 , g3 of G1 , G2 , G3 are then given by
3g1 = 1 2 u 2 , 3g2 = u 2 u 2 , 3g3 = u 2 + .

Now, the ve tors 3G2 G1 and 3G2 G3 have axes
3(g1 g2 ) = 1 u , 3(g3 g2 ) = + u 2 .
From 1 u = 1 ei = ei/2 (e(i)/2 ei/2 ) = 2iei/2 sin(/2) and

2 2
+ u 2 = e2i/3 + e( 3 )i = 2ei/2 cos
3 2

we dedu e that gg3
g
g2
= ki for some real number k. This means that G2 G1
1 2

and G2 G3 are orthogonal; that is, triangle G1 G2 G3 is right-angled.
162
2. There are n hildren sitting at a round table. Erika is the oldest among
them and she has n andies. No other hild has any andy. Erika distributes
the andies as follows. In every round, all the hildren with at least two
andies show their hands. Erika hooses one of them and he/she gives one
andy to ea h of the hildren sitting next to him/her. (So in the rst round
Erika must hoose herself to begin the distribution.) For whi h n 3 is it
possible to redistribute the andies so that ea h hild has exa tly one andy?
Solution by Oliver Geupel, Bruhl,
NRW, Germany.
The redistribution is possible if and only if n is odd.
To prove this, let N be the set of positive integers, and let 7 denote
the binary relation on Nn su h that (a1 , a2 , . . . , an ) 7 (a1 , a2 , . . . an ) if and
only if there is an index k su h that ak1 = ak1 + 1, ak = ak 2, and
ak+1 = ak+1 + 1, where indi es are taken modulo n, that is, ak+n = ak .
First, assume that
(0) (0) (1) (1)
(n, 0, 0, . . . , 0) = (a1 , a2 , . . . , a(0) (1)
n ) 7 (a1 , a2 , . . . , an )
n ) = (1, 1, . . . , 1) .
(m) (m)
7 7 (a1 , a2 , . . . , a(m)
For 1 k n, let bk denote the number of indi es j {1, 2, . . . m}

su h that a(j)k = ak
(j1)
2. Then, for 2 k n, the number of indi es
j {1, . . . , m} su h that ak = ak + 1 is ak ak + 2bk = 2bk + 1,
(j) (j1) (m) (0)
and on the other hand is bk1 + bk+1 , sin e an in rease in the k omponent th
results from a de rease of a neighbour. Hen e, bk+1 = 2bk bk1 + 1 for

2 k n. By indu tion one has bk+1 = kb2 (k1)b1 +k(k1)/2. Hen e,
b1 = bn+1 = nb2 (n 1)b1 + n(n 1)/2, and thus b1 = b2 + (n 1)/2.
Consequently, n is odd.
For onvenien e, let ak denote the sequen e a, a, . . . , a onsisting of
k o urren es of the symbol a, and let 7+ denote the transitive losure of
the relation 7.
We will show by mathemati al indu tion that for ea h k > 0 there is
a path (0k , 2k + 1, 0k ) 7+ (12k+1 ) su h that both the rst and the last
omponent of our n-tuple have only one in reasing step ea h.
The statement is obvious for k = 1. Assume it holds for k. We obtain
(0k+1 , 2k + 3, 0k+1 ) 7+ (0, 1k , 3, 1k , 0) (by hypothesis)
+
7 (1, 0, 1 k1
, 3, 1 k1
, 0, 1) (shift from the entre outwards)
+ 2
7 (1 , 0, 1 k2
, 3, 1 k2 2
, 0, 1 ) (shift from the entre outwards)
+ +
7 7 (1 k+3
) (iterate shifting from the entre outwards) .
This ompletes the proof.
3 . The sum of four real numbers is 9 and the sum of their squares is 21.
Prove that these four numbers an be labelled as a, b, c, and d so that the
inequality ab cd 2 holds.
163
Solution by Jose Luis Daz-Barrero, Universitat Polite ni a

de Catalunya,
Bar elona, Spain.
Without loss of generality we an assume that a b c d. We will
onsider two ases: a + b 5 and a + b < 5.
Case 1. If a + b 5, then taking into a ount that c + d 2cd, we have
2 2
a2 + b2 + 2ab = (a + b)2 25
= 4 + a2 + b2 + c2 + d2 4 + a2 + b2 + 2cd ,
from whi h follows ab cd 2.

Case 2. Now we will see that a + b < 5 is not possible. In fa t, if a + b < 5,
then c + d > 4 and therefore 4 < c + d a + b < 5. From (a d)(b c) 0
and (a b)(c d) 0 it follows that ab + cd ac + bd ad + bc. Sin e
(ab + cd) + (ac + bd) + (ad + bc)
(a + b + c + d)2 (a2 + b2 + c2 + d2 )
= = 30 ,
2
we have ab+cd 10. From 4 < c+d a+b < 5 and (a+b)+(c+d) = 9,
it follows that (a + b)(c + d) 20, and hen e from
(a + b)2 + (c + d)2 + 2(a + b)(c + d) = 81
we obtain (a + b)2 + (c + d)2 < 41. We now have

41 = 21 + 2 10
(a2 + b2 + c2 + d2 ) + 2(ab + cd)
= (a + b)2 + (c + d)2
< 41 ,
a ontradi tion, and we are done.
Next we look at readers' solutions to problems of the XXI Olimpiadi

Italiano della Matemati a, Cesenati o 2006 given at [2009 : 25{26.
1 . Rose and Savino play a game with a de k of traditional Neapolitan playing
ards whi h onsists of 40 ards of four dierent suits, numbered 1 to 10. At
the start ea h player has 20 ards. Taking turns, one shows a ard on the
table. Whenever some ards on the table add to exa tly 15, these are then
removed from the game (if the sum 15 an be obtained in more than one way,
the player who last moved de ides whi h ards adding to 15 to remove). At
the end of the game only one ard, a 9, is left on the table. Savino holds two
ards numbered 3 and 5, and Rose holds one ard. What is the number of
Rose's ard?
164
John Grant M Loughlin, University of New Brunswi k, Frederi ton, NB and
Titu Zvonaru, Comane
sti, Romania. We give the write-up of Babbitt.
We shall prove that Rose is holding an 8.
Let Rose's ard have a value of R. Let the number of sweeps o urring
in the game be d. The sum of all the ards is 4(1+2+ +10) = 455 = 220.
Therefore, 220 = 3 + 5 + 9 + R + 15d, whi h implies 203 = R + 15d. Sin e
R is less than 15, R is the remainder when 203 is divided by 15. Now,
203 8
15
= 13 +
15
, hen e R = 8.
2. Find all values of m, n, and p su h that

pn + 144 = m2 ,
where m and n are positive integers and p is a prime number.

Solved by Matthew Babbitt, home-s hooled student, Fort Edward, NY,
USA; Mi hel Bataille, Rouen, Fran e; Jose Luis Daz-Barrero, Universitat
Polite ni a
de Catalunya, Bar elona, Spain; Edward T.H. Wang, Wilfrid
Laurier University, Waterloo, ON; Konstantine Zelator, University of Pitts-
burgh, Pittsburgh, PA, USA; and Titu Zvonaru, Comane sti, Romania. We
give Babbitt's solution.
The only su h triples are (m, n, p) {(13, 2, 5), (20, 8, 2), (15, 4, 3)}.
Write pn = m2 144 = (m 12)(m + 12). Sin e p is a prime, m 12
and m+12 must be integral powers of p. Either p divides ea h fa tor m12,
or m 12 = 1 and m = 13. The se ond ase yields (m, n, p) = (13, 2, 5).
The rst ase yields p | [(m + 12) (m 12)] = 24, or p {2, 3}.
If p = 2, then we need to nd two powers of 2 whose dieren e is 24.
Note that m 12 < 32, sin e 2n+1 2n > 24 for n 5. So m 12 is
one of 21 , 22 , 23 , 24 . Che king ea h ase yields only m 12 = 8 = 23 , or
m = 20, and thus (m, n, p) = (20, 8, 2).
If p = 3, then we need to nd two powers of 3 whose dieren e is 24.
Now m 12 < 27, sin e 3n+1 3n > 24 for n 3. So m 12 is either 3
or 9. Only m 12 = 3 works, yielding (m, n, p) = (15, 4, 3).
Therefore, these are the only triples.
3 . Let A and B be two points on a ir le su h that AB is not a diameter.
Let P be a point on dierent from A and B , and let H be the ortho entre
of the triangle ABP . Find the lo us of H as P varies over all points of
dierent from A and B .
Solved by Mi hel Bataille, Rouen, Fran e; Oliver Geupel, Bruhl,
NRW, Ger-
many; and Titu Zvonaru, Comane
sti, Romania. We give Bataille's write-up.
Let O be the entre of and M be the midpoint of AB . When P
traverses {A, B}, the xed point O remains the ir um entre of P AB .
165

It is well known that P H = 2OM . It follows that H is the image of P

under T , where T is the translation by the ve tor 2OM . Let = T ().
The ir le has the same radius as and its entre is the symmetri point
of O about M . The lo us of H is learly {A , B }, where A = T (A)
and B = T (B).
5. Consider the inequality
(x1 + + xn )2 4(x1 x2 + x2 x3 + + xn x1 ) .
(a) Determine for whi h n 3 the inequality holds true for all possible
hoi es of positive real numbers x1 , x2 , . . . , xn .
(b) Determine for whi h n 3 the inequality holds true for all possible
hoi es of any real numbers x1 , x2 , . . . , xn .
Solved by Edward T.H. Wang and Dexter Wei, Wilfrid Laurier University,
Waterloo, ON; and Titu Zvonaru, Comane sti, Romania. We give the write-
up of Wang and Wei.
For part (a) we will show that the inequality holds for all n 4 but fails
for n = 3. When n = 3, (x1 , x2 , x3 ) = (2, 1, 1) provides a ounterexample.
Next, we onsider the ase n = 4. We have
(x1 + x2 + x3 + x4 )2 4(x1 x2 + x2 x3 + x3 x4 + x4 x1 )
= x21 + x22 + x23 + x24 + 2(x1 x3 + x2 x4 )
2(x1 x2 + x2 x3 + x3 x4 + x4 x1 )
= (x1 x2 + x3 x4 ) 0 ,
2
so the inequality in fa t holds for all real numbers x1 , x2 , x3 , and x4 .

Now we use indu tion to show that the inequality holds for n 5 and
for all nonnegative real numbers x1 , x2 , . . . , xn . Suppose that n 4 and
(x1 + x2 + + xn )2 4(x1 x2 + x2 x3 + + xn x1 ), where xi 0 for
ea h i. We are to show that
(x1 + x2 + + xn+1 )2 4(x1 x2 + x2 x3 + + xn+1 x1 ) . (1)
Sin e both sides of (1) are invariant under y li permutations of the
indi es, we may assume, without loss of generality, that x1 xn+1 .
Then, by the indu tion hypothesis, we have
(x1 + x2 + + xn+1 )2 = (x1 + x2 + + (xn + xn+1 ))2
4(x1 x2 + x2 x3 + + xn1 (xn + xn+1 ) + (xn + xn+1 )x1 )
= 4(x1 x2 + x2 x3 + + xn xn+1 + xn+1 x1 )
+ 4xn1 xn+1 + 4xn (x1 xn+1 )
4(x1 x2 + x2 x3 + + xn+1 x1 ) .
This ompletes the indu tion and hen e, the proof of our laim.
166
For part (b) we laim that the inequality holds only for n = 4.
That it fails for n = 3 and holds for n = 4 was shown in part (a).
Hen e, it remains to show that it fails for all n 5. For onvenien e, we let
A and B denote the left side and right side of the inequality, respe tively.
If n = 2k where k 2, then we let x1 = 1 for i = 1, 2, . . . , k, and
xi = 1 for i = k + 1, k + 2, . . . , 2k = n. Then A = 0, while on the other
hand B = 4[(n 2) 2] = 4(n 4) > 0.
If n = 2k + 1 where k 2, then we let x1 = 1 for i = 1, 2, . . . , k + 1,
and xi = 1 for i = k + 2, k + 3, . . . , 2k + 1 = n. Then A = 1, while on
the other hand B = 4[(2k 1) 2] = 4(2k 3) = 4(n 4) 4.
This ompletes the proof.
Next we open our le of solutions to problems posed in the Mar h

2009 number of the Corner, beginning with a solution to a problem of the
19 Korean Mathemati al Olympiad 2006 given at [2009 : 80{81.
th
4. Given three distin t real numbers a1 , a2 , and a3 , dene three real num-

bers b1 , b2 , and b3 as follows

aj ai aj ak
bj = 1+ 1+ , where {i, j, k} = {1, 2, 3} .
aj ai aj ak
Prove that

1 + |a1 b1 + a2 b2 + a3 b3 | 1 + |a1 | 1 + |a2 | 1 + |a3 | .
When does equality hold?
Solved by Mi hel Bataille, Rouen, Fran e; and Oliver Geupel, Bruhl,
NRW,
Germany. We give Bataille's write-up.
For onvenien e, let
a2 a3 a3 a1 a1 a2
c1 = , c2 = , c3 = ;
a2 a3 a3 a1 a1 a2
so b1 = (1 + c3 )(1 c2 ), b2 = (1 c3 )(1 + c1 ), and b3 = (1 + c2 )(1 c1 ).
Simple al ulations then yield
a1 b1 + a2 b2 + a3 b3
= a1 + a2 + a3 + a1 a2 + a2 a3 + a3 a1 a1 c2 c3 a2 c3 c1 a3 c1 c2
and
a1 c2 c3 a2 c3 c1 a3 c1 c2
a1 a2 a3 [a21 (a3 a2 ) + a22 (a1 a3 ) + a23 (a2 a1 )]
=
(a2 a3 )(a3 a1 )(a1 a2 )
= a1 a2 a3 .
167
Thus, using the Triangle Inequality,

1 + |a1 b1 + a2 b2 + a3 b3 |
= 1 + |a1 + a2 + a3 + a1 a2 + a2 a3 + a3 a1 + a1 a2 a3 |
1 + |a1 | + |a2 | + |a3 | + |a1 a2 | + |a2 a3 | + |a3 a1 | + |a1 a2 a3 |
= 1 + |a1 | + |a2 | + |a3 | + |a1 ||a2 | + |a2 ||a3 | + |a3 ||a1 | + |a1 ||a2 ||a3 |
= (1 + |a1 |)(1 + |a2 |)(1 + |a3 |) .
and the desired inequality follows.
Equality holds if and only if it holds where we applied the Triangle
Inequality, that is, if and only if a1 , a2 , a3 , a1 a2 , a2 a3 , a3 a1 , a1 a2 a3 all
have the same sign. Due to the presen e of the produ ts a1 a2 , a2 a3 , a3 a1 ,
this sign is the positive one and equality o urs pre isely when a1 , a2 , a3 are
nonnegative.
Now we turn to solutions from readers to problems of the Olympiade

Suisse de mathematiques
2005, tour nal, given at [2009 : 82{83.
3. Pour tout a1 , . . . , an > 0, prouver l'inegalit
e suivante et determiner
tous
les as d'egalit
e
X
n
n X
n
kak + akk .
k=1
2 k=1
Solutions and omments by Mi hel Bataille, Rouen, Fran e; Oliver Geupel,

Bruhl,
NRW, Germany; Edward T.H. Wang, Wilfrid Laurier University,
Waterloo, ON; and Titu Zvonaru, Comane sti, Romania. We give Bataille's
omment.
Ce probleme
a dej a et
e propose lors de la 52 olympiade polonaise
ieme

en 2001. Son enon

e et sa solution se trouvent dans CRUX with MAYHEM:
[2004 : 19; 2005 : 445.
6 . Soient a, b, c des nombres reels
positifs ave abc = 1. Determiner
toutes
les valeurs que peut prendre la somme
1+a 1+b 1+c
+ + .
1 + a + ab 1 + b + bc 1 + c + ca
Solved by George Apostolopoulos, Messolonghi, Gree e; Mi hel Bataille,

Rouen, Fran e; and Titu Zvonaru, Comane
sti, Romania. We give the write-
up of Apostolopoulos.
Sin e abc = 1, we have
1+b 1+b a(1 + b)
= 1
= ,
1 + b + bc 1+b+ 1 + a + ab
a
168
so
1+a 1+b 1 + a + a(1 + b) a
+ = = 1 +
1 + a + ab 1 + b + bc 1 + a + ab 1 + a + ab
and
1
1+c 1+ ab + 1
= 1
ab
1
= .
1 + c + ca 1+ + 1 + a + ab
ab b
Finally,
1+a 1+b 1+c
+ +
1 + a + ab 1 + b + bc 1 + c + ca

a ab + 1
= 1+ +
1+a+b 1 + a + ab

1 + a + ab
= 1+ = 1+1 = 2.
1 + a + ab
7. Soit n 1 un nombre naturel. Determiner

toutes les solutions entieres

positives de l'equation

7 4n = a2 + b2 + c2 + d2 .

The following quadruples and their permutations yield all solutions:
(2n , 2n , 2n , 2n+1 ) ,
(2n1 , 2n1 , 2n1 , 5 2n1 ) ,
(2n1 , 3 2n1 , 3 2n1 , 3 2n1 ) .
It is readily he ked that ea h quadruple yields a solution. Conversely,

let (a, b, c, d) be any solution. Then, a2 + b2 + c2 + d2 0 (mod 4) and,
sin e x2 0 or 1 (mod 4) a ording as the integer x is even or odd, we must
have a2 b2 c2 d2 0 (mod 4) or a2 b2 c2 d2 1 (mod 4).
In the latter ase, a, b, c, d are odd, hen e a2 b2 c2 d2 1 (mod 8),
and we an set
a2 = 1 + 8a1 , b2 = 1 + 8b1 , c2 = 1 + 8c1 , d2 = 1 + 8d1 ,
where a1 , b1 , c1 , d1 are nonnegative integers. From the equation

1 + 2(a1 + b1 + c1 + d1 ) = 7 4n1
we see that n = 1 and a1 + b1 + c1 + d1 = 3. Up to permutations,

(a1 , b1 , c1 , d1 ) is one of the quadruples (0, 0, 0, 3), (0, 0, 1, 2), (0, 1, 1, 1),
when e (a, b, c, d) = (1, 1, 1, 5) or (1, 3, 3, 3) (up to permutations).
169
Now, suppose that a, b, c, d are all even positive integers. We write

them as
a = 2 t , b = 2 u , c = 2 v , d = 2 w ,
for some odd positive integers t, u, v, w and positive integers , , , for
whi h we assume without loss of generality. Then,
22 (t2 + 22() u2 + 22() v 2 + 22() w 2 ) = 7 22n .
If > , then = n and t2 + 22() u2 + 22() v2 + 22() w2 = 7,

whi h is learly impossible.
Thus, = , and
22 (t2 + u2 + 22() v 2 + 22() w 2 ) = 7 22n .
Sin e t2 +u2 is twi e an odd integer, we must have = (otherwise the total
exponent of 2 would be odd on the left and even on the right) and nally,
22 (t2 + u2 + v 2 + 22() w 2 ) = 7 22n .
If = , then 22 (t2 + u2 + v 2 + w 2 ) = 7 22n , and hen e < n and

t + u2 + v 2 + w 2 = 7 22(n) . By the previous results, = n 1
2
and (t, u, v, w) = (1, 1, 1, 5) or (1, 3, 3, 3) (up to permutations), and thus

(a, b, c, d) = (2n1 , 2n1 , 2n1 , 52n1 ) or (2n1 , 32n1 , 32n1 , 32n1 )
(up to permutations).
If > , then t2 + u2 + v2 + 22() w2 is odd, hen e
22 (t2 + u2 + v 2 + 22() w 2 ) = 7 22n .
In this last ase, the only possibility is t = u = v = 1 and 22() w2 = 4, so

that = + 1 and w = 1. It follows that (a, b, c, d) = (2n , 2n , 2n , 2n+1 ).
9. Trouver toutes les fon tions f : R+ R+ , veriant
la ondition suivante
pour tout x,y > 0:

f yf (x) (x + y) = x2 f (x) + f (y) .

The fun tion f (x) = x1 is the unique solution. To this aim, let f be a
solution and let (E) denote the identity. Taking y = x = 1 in (E) yields
f (a) = a where a = f (1). Taking x = a, y = 1 in (E) yields 2a2 a1 = 0,
and sin e a > 0 we have a = 1 and f (1) = 1. Finally, setting x = 1 in (E)
with arbitrary y, we obtain f (y)(1 + y) = 1 + f (y), when e f (y) = y1 .
That ompletes the Corner for this issue. Send me your ni e solutions
and generalizations.
170
BOOK REVIEWS
Amar Sodhi
Mythemati s: Solving the Twelve Labors of Her ules

By Mi hael Huber, Prin eton University Press, Prin eton and Oxford, 2009
ISBN 978-0-691-13575-5, hard over, 183+xix pages, US$24.95
Reviewed by Edward Barbeau, University of Toronto
The twelve labours of the lassi al Greek hero, Her ules, were per-
formed at the behest of Eurystheus in atonement for Her ules' murder of
his hildren and those of his brother in a t of madness. As they stand, these
labours do not appear to oer mu h potential for mathemati al elaboration,
but the author uses them as a pretext to reate a number of problems that
an be given to mathemati al undergraduates of the rst two years.
Ea h of the twelve hapters opens with the des ription by the Greek
author Apollodorus of one of the labours, followed by the statement and
nally the solutions of two or three problems. Appendi es relate problems
and their mathemati al areas, ll in the tale of Her ules up to the time of the
labours, dis uss various versions of the Her ules legend, and provide a brief
primer on the Lapla e transform.
For example, the twelfth labour is to bring to Eurystheus Cerberus,
the three-headed dog that guards Hades. In the rst of the two problems,
Her ules is lo ated at the point (2, 1, 7) on the surfa e z = 2x+ y2 2xy
and has to make his des ent to Hades. The reader is asked what the rate of
des ent is above the line y = x and also for the dire tion of steepest des ent;
the solution of this uses a ve tor ow diagram. To apture Cerberus, Her ules
must render the beast un ons ious by su iently impeding the blood ow
to all of the three heads by hoking ea h of the three ne ks in turn. This is a
problem of exponential de ay in whi h the blood ow has to be redu ed from
6 to below 2.5 ml per se ond; he has to ontend with the slow restoration of
the ow when ea h ne k is released. The solution uses a spreadsheet. As
a bonus, the reader is hallenged by one of the three Sudoku puzzles in the
book; this one has the novel feature of requiring that ea h row, olumn and
blo k is lled by one ea h of the rst six digits and three sevens.
Many of the problems in the book are familiar ones provided with a
new and sometimes arti ial setting: the shooting towards a target of an
arrow subje t to gravity, the Monty Hall ( ar and goats) problem, minimizing
the time to ross a river to a point along the opposite bank, maximizing the
angle subtended at the eye by a pi ture hanging on a wall, resonan e of an
undamped harmoni os illator with a for ing fun tion, logisti growth. Even
though the book does not have a real spark, the treatment of the problems
is solid and the sour ing of the myth is meti ulous.
171
PROBLEMS
Solutions to problems in this issue should arrive no later than 1 O tober 2010.
An asterisk () after a number indi ates that a problem was proposed without a
solution.
Ea h problem is given in English and Fren h, the o ial languages of Canada.
In issues 1, 3, 5, and 7, English will pre ede Fren h, and in issues 2, 4, 6, and 8,
Fren h will pre ede English. In the solutions' se tion, the problem will be stated in
the language of the primary featured solution.
The editor thanks Jean-Mar Terrier of the University of Montreal for transla-
tions of the problems.
3527 . Proposed by Hung Pham Kim, student, Stanford University, Palo

Alto, CA, USA.
Let a, b, and c be nonnegative real numbers su h that a + b + c = 3.
Prove that
X 3

3

75
a2 b + b2 c + .
y li
2 2 4
3528 . Proposed by Hiroshi Kinoshita and Katsuhiro Yokota, Tokyo, Japan.

The in ir le of triangle ABC tou hes the sides BC , AC , AB at the
points A , B , C , respe tively. Let , ra , rb , rc denote the inradii of the ir-
les A B C , AB C , BC A , CA B , respe tively, and let r be the inradius
of the triangle ABC . Prove that
1
r = ( + ra + rb + rc ) .
2
3529 . Proposed by Mi hel Bataille, Rouen, Fran e.

Let A be a point on a ir le with entre O and t be the tangent to
at A. Triangle P OQ is su h that P is on , Q is on t, and P OQ = 90 .
Find the envelope of the perpendi ular to AP through Q as P OQ varies.
3530 . Proposed by Ovidiu Furdui, Campia Turzii, Cluj, Romania.
Let f : [0, 1] R be an integrable fun tion whi h is ontinuous at 1.
Let k be a xed positive integer, and let
Z 1 f (x)
an = dx .
0 (1 + xn )(1 + xn+k )
Find L = n
lim an and lim n(L an ).
n
172
3531 . Proposed by K.S Bhanu, Institute of S ien e, Nagpur, India, and

M.N. Deshpande, Nagpur, India.
Let a, b be positive integers. On the real line, A stands at a and B
stands at b. A fair oin is tossed, and if it shows heads then A moves one
unit to the right, while if it shows tails then B moves one unit to the left.
The pro ess stops when A or B rea hes the origin.
Let PA (a, b) be the probability that A rea hes the origin before B , and
dene PB (a, b) similarly. Prove that
E(a, b) = 2aPA (a + 1, b) + 2bPB (a, b + 1) ,
where E(a, b) is the expe ted number of tosses before the pro ess termi-
nates.
Let triangle ABC have ir umradius R, inradius r, and let a , b , c
be the distan es from the entroid to the sides BC , CA, AB , respe tively.
Prove that r
a + b + c R
r .
3 2
3533 . Proposed by Cao Minh Quang, Nguyen Binh Khiem High S hool,
Vinh Long, Vietnam.
Let a, b, c be positive real numbers su h that a + b + c = 1. Let m and
n be positive real numbers satisfying 6m 5n. Prove that
ma + nbc mb + nca mc + nab 3m + n
+ + .
b+c c+a a+b 2
3534 . Proposed by Mihaly Ben ze, Brasov, Romania.

Let x1 , x2 , . . . , xn be positive real numbers, where n 2, and let
1. Prove that
! !
X
n X
n
1
(n 1) x
k xk
k=1 k=1
! !1
X X
n X
n
2
xi xk + (n 1) 1
x+1
k xk .
1i<jn k=1 k=1
3535 . Proposed by Walther Janous, Ursulinengymnasium, Innsbru k,

Austria.
Let a, b, and c be positive real numbers and let 0. Prove that

a2 + bc b2 + ca c2 + ab
+ + 31 (a + b + c) .
b+c c+a a+b
173
3536 . Proposed by Samuel Gomez Moreno, Universidad de Jaen, Jaen,

Spain.
Find all positive integers n and k su h that the equation {x2n } = {x}
has 2010 roots inside the interval [k, k + 1), where x is the greatest integer
not ex eeding x and {x} = x x.
3537 . Proposed by Marian Marines u, Monbonnot, Fran e.
Let f : [0, 1] R be ontinuous, and let g : [0, 1] R be monotoni
and dierentiable with g(0) = 0. Prove that there is a number 0 < a < 1
su h that
Z a Z 1 Z a
f (x)g(x) dx = f (x) dx g(x) dx .
0 0 0
3538 . Proposed by Vi tor Oxman, Western Galilee College, Israel.

In the plane you are given a triangle ABC with its internal angle bi-
se tor BD, a point E on the side BC su h that ED is the bise tor of angle
AEC , and the ir um ir le of the triangle ABC (but not its entre). Con-
stru t the entre of that ir le using only a straightedge.
[Ed.: The Pon elet{Steiner Theorem says that given a ir le with its
entre, we an arry out all the ruler- ompass onstru tions in the plane of
that ir le by straightedge only. See Crux problems 2694, 2695, and 2696
[2002 : 553-557.
.................................................................
3527. Propose par Pham Kim Hung, etudiant,

Universite de Stanford, Palo

Alto, CA, E-U.
Soit a, b, et c trois nombres reel
non-negatifs tel que a + b + c = 3.
Montrer que
X 3

3

75
a2 b + b2 c + .
y li
2 2 4
3528 . Propose par Hiroshi Kinoshita et Katsuhiro Yokota, Tokyo, Japon.

Le er le ins rit d'un triangle ABC tou he respe tivement les ot ^ es

BC , AC et AB aux points A , B et C . On denote
respe tivement par , ra ,
rb et rc les rayons des er les ins rits des triangles A B C , AB C , BC A
et CA B , et soit r le rayon du er le ins rit du triangle ABC . Montrer que
1
r = ( + ra + rb + rc ) .
2
174
3529 . Propose par Mi hel Bataille, Rouen, Fran e.

Soit A un point sur un er le de entre O et t la tangente a en A.
Soit P OQ le triangle tel que P soit sur , Q sur t et P OQ = 90 . Trou-
ver l'enveloppe des perpendi ulaires a AP passant par Q lorsque le triangle
P OQ varie.
3530 . Propose par Ovidiu Furdui, Campia Turzii, Cluj, Roumanie.

Soit f : [0, 1] R une fon tion integrable, ontinue en 1. Soit k un
entier positif xe et soit
Z 1 f (x)
an = dx .
0 (1 + xn )(1+ xn+k )
Trouver L = n
lim an et lim n(L an ).
n
3531 . Propose par K.S. Bhanu, Institut des S ien es, Nagpur, Inde, et
M.N. Deshpande, Nagpur, Inde.
Soit a et b deux entiers positifs. On pla e deux points sur la droite reelle,

A en a et B en b. On lan e une pie e de monnaie non pipee ; si elle tombe
sur pile, on depla e
B d'une unite vers la gau he, si elle tombe sur fa e, on
depla e
A d'une unite vers la droite. Le jeu s'arrête des que l'un des points
atteint l'origine.
Soit PA (a, b) la probabilite que A atteigne l'origine avant B et PB (a, b)
elle du as ontraire. Montrer que
E(a, b) = 2aPA (a + 1, b) + 2bPB (a, b + 1) ,
ou E(a, b) est le nombre de lan ers attendu avant la n du jeu.

3532 . Propose par Mi hel Bataille, Rouen, Fran e.
Dans un triangle ABC , on note R le rayon du er le ir ons rit, r elui
du er le ins rit, et soit a , b et c les distan es respe tives de son entre de
gravite aux ot
^ es
BC , CA et AB . Montrer que
r
a + b + c R
r .
3 2
3533 . Propose par Cao Minh Quang, Nguyen Binh Khiem High S hool,
Vinh Long, Vietnam.
Soit a, b et c trois nombres reels
positifs tels que a + b + c = 1. Soit
m et n deux nombres reels positifs tels que 6m 5n. Montrer que
ma + nbc mb + nca mc + nab 3m + n
+ + .
b+c c+a a+b 2
175
3534 . Propose par Mihaly Ben ze, Brasov, Roumanie.

Soit x1 , x2 , . . . , xn , n 2, des nombres reels
positifs et 1.
Montrer que
! !
X
n X
n
1
(n 1) x
k xk
k=1 k=1
! !1
X X
n X
n
2
xi xk + (n 1) 1
x+1
k xk .
1i<jn k=1 k=1
3535 . Propose par Walther Janous, Ursulinengymnasium, Innsbru k,

Autri he.
Soit a, b et c trois nombres reels
positifs et 0. Montrer que

a2 + bc b2 + ca c2 + ab
+ + 31 (a + b + c) .
b+c c+a a+b
3536 . Propose par Samuel Gomez Moreno, Universite de Jaen, Jaen,

Espagne.
Trouver tous les entiers positifs n et k tels que l'equation
{x2n } = {x}
ait 2010 ra ines a l'interieur
de l'intervalle [k, k + 1), ou x est le plus grand
entier n'ex edant
pas x et {x} = x x.
3537 . Propose par Marian Marines u, Monbonnot, Fran e.
Soit f : [0, 1] R ontinue, et g : [0, 1] R monotone et dierentiable

ave g(0) = 0. Montrer qu'il existe un nombre 0 < a < 1 tel que
Z a Z 1 Z a
f (x)g(x) dx = f (x) dx g(x) dx .
0 0 0
3538 . Propose par Vi tor Oxman, Western Galilee College, Israel.

Dans le plan, on onsidere un triangle ABC dont on donne la bisse -
tri e interieure
BD , un point E sur le ot
^ e BC tel que ED soit la bisse tri e
de l'angle AEC , et enn le er le ir ons rit de ABC , mais pas son entre.
Ave la regle
seulement, trouver le entre de e er le.
[Ed. : Le theor
eme
de Pon elet-Steiner dit qu'etant
donne un er le
ave son entre, on peut exe uter
toutes les onstru tions ave la regle
et
le ompas dans le plan de e er le a l'aide de la regle
seulement. Voir les
problemes
de Crux 2694, 2695 et 2696 [2002 : 553-557.
176
SOLUTIONS
No problem is ever permanently losed. The editor is always pleased
to onsider for publi ation new solutions or new insights on past problems.
3425 . Corrigenda. The following orre tions should repla e the orre-
sponding text in CRUX with MAYHEM Vol. 36, No. 2 at the indi ated pla e.
Our apologies for these errors.
1 eu ev e0.5
p. 125, line 26: u

u
=
v

u
1
p. 125, line 31: 1
(v 1)ev + + (v 2)e 4 2
v
v
1 ln(t2 + 0.5t + 1)
p. 126, line 10: 1t2 1t 2 t + 0.5t + 1
(t)
p. 126, line 14: (12t3 + 2t2 + 12t 11) in the numerator of t
, with
orresponding hanges on lines 18, 19, and 23.
p. 126, line 16: 11 12t3 2t2 12t 0
1
p. 126, line 20: 3
(18t2 13t + 3) > 0
p. 126, line 24: 5(2t 1)(2t2 + t + 2)

3.32 2.496 > 0 , if 0.6 t 0.7 ,
p. 126, line 25: 7.36 6.024 > 0 , if 0.7 t 0.8 .
p. 127, line 6: (0) = 0
3426. [2009 : 172, 174 Proposed by Salvatore Tringali, student, Mediter-

ranea University, Reggio Calabria, Italy.
Find all prime numbers p, q, and r su h that p + q = (p q)r .
Solution by Harry Sedinger, St. Bonaventure University, St. Bonaventure,
New York, USA.
Clearly, p 6= q. Assume p > q and let p q = n. The equation then
be omes n + 2q = nr , or
2q = n(nr1 1) .
If r = 2, then 2q = n(n 1), whi h has a solution with q prime only

if n = 3; and then q = 3.
177
If r = 3, then 2q = n(n2 1) = (n 1)n(n + 1), so that n = 2,

q = 3, p = 5 and we have a solution, (p, q, r) = (5, 3, 3).
If r > 3, then
2q = n(nr1 1) = n(n 1)(nr2 + + 1) ,
whi h implies n = 2, and then q = 2r2 + + 1 = 2r1 1 7. Now,

q = 2r1 1, 2r1 , and p = q + 2 = 2r1 + 1 are three onse utive integers,
so that one of them must be divisible by 3. Sin e 2r1 is not divisible by 3,
then either q or p is divisible by 3, and therefore not a prime. Thus there are
no solutions with r > 3.
We have shown that, under the assumption p > q, the given equation
has a solution only if r = 3. Now, if we assume p < q, then r must be even
and we an rewrite the given equation as q + p = (q p)r . But we have
already shown that this equation has a solution with q > p only if r = 3.
This shows that the solution found above is unique.
Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Gree e; SEFKET
ARSLANAGIC, University of Sarajevo, Sarajevo, Bosnia and Herzegovina; ROY BARBARA,
Lebanese University, Fanar, Lebanon; RICARDO BARROSO CAMPOS, University of Seville,
Seville, Spain; MICHEL BATAILLE, Rouen, Fran e; BRIAN D. BEASLEY, Presbyterian
College, Clinton, SC, USA; CHIP CURTIS, Missouri Southern State University, Joplin, MO,
USA; CHARLES R. DIMINNIE, Angelo State University, San Angelo, TX, USA; OLIVER
GEUPEL, Bruhl,
NRW, Germany; JOHN HAWKINS and DAVID R. STONE, Georgia Southern
University, Statesboro, GA, USA; RICHARD I. HESS, Ran ho Palos Verdes, CA, USA; JOSE

HERNANDEZ SANTIAGO, student, Universidad Te nologi a
de la Mixte a, Oaxa a, Mexi o;
WALTHER JANOUS, Ursulinengymnasium, Innsbru k, Austria; KEE-WAI LAU, Hong Kong,
China; THANOS MAGKOS, 3rd High S hool of Kozani, Kozani, Gree e; MISSOURI STATE
UNIVERSITY PROBLEM SOLVING GROUP, Springeld, MO, USA; CRISTINEL MORTICI,
Valahia University of Târgoviste, Romania; TROY MULHOLLAND, student, St. Bonaventure
University, New York, USA (2 solutions); MICHAEL PARMENTER, Memorial University of
Newfoundland, St. John's, NL; JOEL SCHLOSBERG, Bayside, NY, USA; DIGBY SMITH, Mount
Royal College, Calgary, AB; ALBERT STADLER, Herrliberg, Switzerland; EDMUND SWYLAN,
Riga, Latvia; KONSTANTINE ZELATOR, University of Pittsburgh, Pittsburgh, PA, USA; TITU
ZVONARU, Comane sti, Romania; and the proposer.
Note that the above argument does not rely on r being a prime. Barbara, Morti i, and
the Missouri State University Problem Solving Group also submitted solutions that relaxed the
restri tion on r.
3427 . [2009 : 172, 174 Proposed by Jose Luis Daz-Barrero, Universitat

Polite ni a
de Catalunya, Bar elona, Spain.
The numbers a, b, c, and d all lie in the interval (1, ) and are su h
that a + b + c + d = 16. Prove that
X 11
loga
4
bcd + a .
y li
2

NRW, Germany.
Repeated appli ation of Arithmeti Mean { Geometri Mean Inequality
178
yields the following hain of inequalities:

X
4
X 1 1

loga bcd + a loga 2(bcd) 8 a 2
y li y li
X ln 2 1

ln b ln c ln d

1

= + + + +
y li
ln a 8 ln a ln a ln a 2

1 ln a ln b ln a ln c ln a ln d ln b ln c
= + + + + + + +
8 ln b ln a ln c ln a ln d ln a ln c ln b
X 1
ln b ln d ln c ln d
+ + + + + 2 + ln 2
ln d ln b ln d ln c ln a y li
1 X 1
(2 + 2 + 2 + 2 + 2 + 2) + 2 + ln 2
8 ln a y li
7 X 1
= + ln 2 .
2 ln a y li
By the Arithmeti Mean { Harmoni Mean Inequality we have

X 1 16 16
P =
y li
ln a ln a ln(abcd)
y li
16 16 2
4 = = .
a+b+c+d 4 ln 4 ln 2
ln
4
Thus,
X 7 2 11
loga
4
bcd + a + ln 2 = .
y li
2 ln 2 2
By the ondition for equality in the AM{GM{HM Inequality, equality holds

if and only if a = b = c = d = 4.
Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Gree e; ROY BARBARA,
Lebanese University, Fanar, Lebanon; MICHEL BATAILLE, Rouen, Fran e; WALTHER JANOUS,
Ursulinengymnasium, Innsbru k, Austria; and the proposer.
Apostolopoulos and Barbara gave similar solutions, but using Jensen's inequality for
1/ ln(x) instead of the AM{HM Inequality.
3428 . [2009 : 172, 174 Proposed by J. Walter Lyn h, Athens, GA, USA.
Fix an integer n > 2 and let I be the interval of all positive ratios r
su h that there exists an n-gon whose sides onsist of n terms of a geometri
sequen e with ommon ratio r. Prove that the endpoints of I are re ipro als
of ea h other.
[Ed.: The proposer refers to Crux M67 [2003 : 430-431 and 3082 [2006 : 477
for the spe ial ases n = 3 and n = 4.
179
Composite of solutions by Roy Barbara, Lebanese University, Fanar, Lebanon

and Oliver Geupel, Bruhl,
NRW, Germany.
We will prove that there exists a real number between 2 n2 and 2
su h that the rangeI of possible ratios r onsists of those numbers in the
open interval 1 , .
By onsidering a regular n-gon, we see that 1 I . Next, let r I ;
that is, we assume that there is an n-gon whose sides are a, ar, . . . , arn1
in some order. That same polygon has sides b, rb , . . . , r n1
b
for b = arn1 .
This observation shows that r I if and only if 1r I . It remains to show
that for r > 1, the values of r in I form a nite open interval. A ne essary
and su ient ondition that a, ar, . . . , arn1 be the sides of an n-gon (when
a > 0 and r > 1) is that
ar n1 < a + ar + + ar n2 .
This is equivalent to
r n1 1
r n1 < ,
r1
and, therefore, to f (r) = rn 2rn1 + 1 < 0. Be ause the derivative of
2
f (r) is zero when r = 2 , negative to the left, and positive to the right,
n h i
2
r n 2r n1 + 1 stri tly de reases on the interval 1, 2 , and stri tly
n
in reases when r > 2 n2 . Be ause f (1) = 0, we dedu e that f (2 n2 ) < 0.
Be ause f (2) = 1 > 0, there must be a unique value of r, all it , between
2
2 and 2 where f () = 0. It follows immediately that I must be the open
n
1
interval
, , whose endpoints are re ipro als of one another, as laimed.
Also solved by CHIP CURTIS, Missouri Southern State University, Joplin, MO, USA;
JOHN HAWKINS and DAVID R. STONE, Georgia Southern University, Statesboro, GA, USA;
MISSOURI STATE UNIVERSITY PROBLEM SOLVING GROUP, Springeld, MO, USA; and
ALBERT STADLER, Herrliberg, Switzerland.
3429 . [2009 : 172, 175 Proposed by Va lav Kone ny, Big Rapids, MI,
USA.
The line passes through the point A and makes an a ute angle with
the segment AB . The line m passes through B and is perpendi ular to AB .
Constru t a point C on the line and a point P on the line m su h that the
triangle BP C is isos eles with BP = P C and
(a) the line CP trise ts BCA,
(b) the line CP bise ts BCA.
180
Composite of solutions by Mi hel Bataille, Rouen, Fran e and by Oliver

Geupel, Bruhl,
NRW, Germany.
(a) Fix a point L on ................
.................................................................... ...
....
su h that = BAL
............
............ ......... ...
................
. ......... ....
.
.......... ........
.. ....
is a ute; for ea h position

.
..... ........
.
.... .
....... ....
.. ... .
. ..... ....... ....
. ........
of C on AL dene P to
.. . .
.. . . ...
.... ...........
.... .... .....
...
be the interse tion of m
... ... .....
.... ... ...
m
.. ... ...
with the perpendi ular bi-

.
.. .... ...
L
. ..
... ..
. ..
. ..... ..
...
se tor of BC (to satisfy the

..
........ q ... ...
... ..... ..... .... ... .........
... ..... .... .. ... ................
requirement that P lies

.
q . .......
C
... . ..... .
..... ... .
. ............
..
... .. ..... .... .............................
. ....
... ..... .... .
... .. ..
on m and BP = P C ),
..........
...
... 2 . .....
..... ...
.. .....
...........
......... .....
... .. ...
... .. ..
........ ... .... .....
....
..... ....
. ..... .........
.... .
. . .
. . .
and set = 13 BCA.

...
... ........................ .... ..... ... ....
..... .. ..... ... .. .. .. .
.. ......... .. ..... ......... .... ....
........
... .. . ...
... q .
........... .... . ..... ........ ....
.
.
We shall rst establish that

q ........ .....
P
.. . .......
.. ......... ... ... ..
..
.. ....
............ .... .... ....
......
when < 45 , there are

... .
...... .
. . .
..
... ............. .. ... ....
.
.
.
.. .
... ......... .... .... ....
........
three positions of C on
... ......... .... ... ...
... ......... .... ... .....
.... ........
.... ..... .
.. ....... ....
AL for whi h CP trise ts

.... ................ .. ..
q q ................................................................................................................................................................................................................................
BCA. Clearly A annot A B

lie between L and su h a
point C ; nor an CBA be
too large (that is, it an-
not ex eed 135 2 ). In q P3 ..
.......
.......
both these ases the en- . .. . .....
... .. ....
... . ...
tire line CP , ex ept for ..
... . ...
.. .. ...
.
the point C , would lie out-

.. . ...
... .. ....
... ... .....
..
side ABC , hen e ould

. ...
... ... ....
... .. ....
..
not trise t an interior an-

. ....
... .
. ...
... ....
.. ..
gle. Consider the ase

. ....
... .
. ...
... ....
.. .
when A and C are on op-

. . ....
... .. ...
.. . ....
.
.. ..
posite sides of m, and C

. . ....
... . ....
.. . ...
L
.
..
satises CBA = 90 + ;
. .
. ....
.. ....
m 2
.
..
. .. ....
q .. .......
that is, P CB = . Us-
. ...
..
. . ... ........
. ... ........
.. .. ...
... .........
.
.. ... ........
ing the angles of ABC ,

q . . ..
. ... ...
.....
..
.......
. ..... ..............................
C2 ...
... .... .... ... .... .. .. .
............... ..
namely , 90 + , and
....
.... .. ... ........................... ... ...
... .
.
. .... .
..
.. ..
. .
. .
... ... .
.......
... .. . .... . .......... .....
... ................ ..... ..........
3 , we determine that =
... ... ....
... ... .
........... .... ................
.
... ...
q ... . .................
. ..... ... ...
..... .... .... ....
22.5 . The se ond
... ......
P
...
.........
.. ...
2 q . .. .
...
.. .
. . .
. . ..
... ....
.. ...... .
. .
.
. ... ...
...
......
4 .. .
.... . .
.
. ..
. .. . . .
position of C , also sepa-

.. ......
. . .
. .
22
.
C3 q
. .
.... .
. .
. ..
... ................ . .
.
. ...
............... .. ....
. ... .
. . . . . . . . . .....
......
rated from A by m, satis-

. ..........
.................. ..
.
..... ...............................
........... ..................................
3 23 q ........
........ ..
......... ....
.................. ...
..................
... ...
.... ...
es CBA = 90 + 2 ;
. ..................
q ...... .
. .
. .
. ..
........ . ............... ...
..................................................................................................................................................................................................................
q
that is, P CB
= 2 , and A B
= 18 . Finally, C
5
an be on the same side of

m as A: here CBA = 2 90 , and = 54 ; it follows that we must
5
have < 45 in order that CBA = 2 90 = 18 2 5
> 0.
181
We now turn to the onstru tions. For the rst position of C we on-
stru t on the same side of AB as L the ir ular ar that onsists of all points
3
X for whi h AXB = (90 ). This ar meets the ray AL at the
4
desired point C . To see why, note that in triangle BCP , BP = P C as
required; moreover, the angles at B and C are equal. By onstru tion,
3 5
ACB = (90 ), when e CBA = 90 > 90 , and the a ute
4 4 4
angle between BC and m, namely CBP , satises
5 1
CBP = 90 90 = 22.5 = BCA .
4 4 4 3
Sin e CBP = BCP , it follows that CP trise ts BCA as laimed.
A similar onstru tion works for the other two positions of C in the
sense that we draw the ar of an appropriate ir le subtended by the hord
3
AB | in the se ond ase AB subtends an angle of 54 , while in the
5
3
third ase the angle subtended is 162 5 | and we dene C to be the
point where the ir le meets AL. Note, however, that the onstru tion for
the rst position of C requires dividing the angle 90 into four equal
pie es, whi h an easily be a hieved by ruler and ompass. By ontrast, the
onstru tion for the remaining two positions of C requires dividing 90
and 270 into ve equal pie es, hen e in general it annot be a hieved
using the traditional Eu lidean tools.
(b) There is exa tly one position of C on AL for whi h there exists a point
P on m su h that BP = P C and CP bise ts BCA; the point C must
satisfy CBA = 120 3 . For the onstru tion, on the same side of AB
as L draw the ir ular ar that onsists of all points X for whi h AXB =
2
(90 ). This ar meets the ray AL at the desired point C , and again we
3
dene P to be the point where the perpendi ular bise tor of the hord BC
meets m so that BP = P C and the angles at B and C are equal. Moreover,
sin e CBA = 120 3 , we have
1 1
P CB = CBP = 120 90 = 30 = AXB = ACB .
3 3 2 2
In other words, CP bise ts BCA, as desired. Here again, the onstru tion
annot generally be a omplished with only ruler and ompass (be ause of
the ne essity of trise ting the angle 90 ). It is amusing to observe that
when the tools are restri ted to ruler and ompass, the bise tion (in part (b))
is not possible while a trise tion (in part (a)) is possible.
Also solved by RICARDO BARROSO CAMPOS, University of Seville, Seville, Spain; CHIP
CURTIS, Missouri Southern State University, Joplin, MO, USA; EDMUND SWYLAN, Riga,
Latvia; and the proposer.
182
3430 . [2009 : 173, 175 Proposed by Mi hel Bataille, Rouen, Fran e.

Let n be a positive integer. Determine the oe ients of the unique
polynomial Pn (x) for whi h the relation

cos2n + sin2n = Pn sin2 (2)
holds for all real numbers .

Similar solutions by Joel S hlosberg, Bayside, NY, USA and Paul Deiermann,
Southeast Missouri State University, Cape Girardeau, MO, USA.
We shall show that
n/2
X nl n
Pn (x) = 22l (1)l xl .
l=0
l nl
Let x = sin2 2, so that 1 x = cos2 2 and

0 x 1 for all . The
double-angle formulae along with cos 2 = 1 x yield

1 1x 1 1x
2
cos = and sin = 2
.
2 2
Using the Binomial Theorem, we see that
Pn (x()) = cos2n + sin2n

= 2n [1 1 x]n + [1 1 x]n

X
n
n
= 2n (1 x)k/2 [(1)k + (1)k ]
k=0
k
2n/2
X n
= 2(n1) (1 x)k/2
k=0
k
k even
n/2
X n
= 2(n1) (1 x)q
q=0
2q
n/2 q
X n X q
= 2(n1) (1)l xl
q=0
2q l=0 l
n/2 n/2
X X n q
= 2 (n1)
(1)l xl .
l=0 q=l
2q l
Next, we use the known \Moriarty" formula (see formula (1.2) of [1)
n/2
X n k nj n
n2j1
= 2
k=j
2k j j nj
183
to omplete the proof:

n/2 n/2
X X n q
Pn (x) = 2(n1) (1)l xl
l=0 q=l
2q l
n/2
X nl n
= 22l (1)l xl .
l=0
l nl
Also solved by ROY BARBARA, Lebanese University, Fanar, Lebanon; CHARLES

R. DIMINNIE, Angelo State University, San Angelo, TX, USA; OLIVER GEUPEL, Bruhl, NRW,
Germany; WALTHER JANOUS, Ursulinengymnasium, Innsbru k, Austria; ALBERT STADLER,
Herrliberg, Switzerland; and the proposer
Deiermann also gave the following re ursion for the polynomial in the problem:
X n
n/2
Pn (x) = 1 22k xk Pn2k (x) .

k
k=1
Referen es
[1 M. Shattu k, Combinatorial Proofs of Some Moriarty-Type Binomial Coe ient Identi-
ties, Integers: Ele troni Journal of Combinatorial Number Theory 6 (2006), No. A35, p. 1
http://www.emis.de/journals/INTEGERS/papers/g35/g35.pdf

Let f : R R be a ontinuous fun tion that satises

f (x + y) = f f (x) f (y)
for all real numbers x and y. Prove that f is onstant.

Solution by Charles R. Diminnie, Angelo State University, San Angelo, TX,
USA.
If f (0) = 0, then for all real x we have
f (x) = f (x + 0) = f (f (x) f (0)) = f (0) = 0 .
and f is onstant.
If f (0) 6= 0, then a = f (0)2 > 0, and
f (a) = f (f (0) f (0)) = f (0 + 0) = f (0) .
Sin e f is ontinuous on [0, a], f attains a maximum and a minimum

on [0, a]. Furthermore, f (0) = f (a) implies that at least one of these values
is attained at a point b (0, a). Without loss of generality, we an assume
that f has a maximum at b.
Let c = min{a b, b}.
Let x [0, c]. Dene g : [b x, b] R by
g(t) = f (t + x) f (t) .
184
By the denition of c we have 0 b x b + x a, thus

g(b x) = f (b) f (b x) 0 ,
and
g(b) = f (b + x) f (b) 0 .
Sin e g is ontinuous on [b x, b], by the Intermediate Value Theorem
there exists some y [b x, b] su h that g(y) = 0, or f (y) = f (x + y).
It follows that
f (x) = f ((x + y) y) = f (f (x + y) f (y))
= f (f (y) f (y)) = f (y y) = f (0) ,
hen e, f (x) = f (0) for all x [0, c].

Finally, for all real numbers x we have
f (x) = f (x + 0) = f (f (x) f (0))

= f (f (x) f (c)) = f (x + c) .
Thus f is periodi with period c. Sin e f is onstant on [0, c], and

periodi with period c, it follows that f is onstant on all of R.
CHARLES R. DIMINNIE, Angelo State University, San Angelo, TX, USA (se ond solu-
tion); OLIVER GEUPEL, Bruhl, NRW, Germany; MISSOURI STATE UNIVERSITY PROBLEM
SOLVING GROUP, Springeld, MO, USA; JOEL SCHLOSBERG, Bayside, NY, USA; ALBERT
STADLER, Herrliberg, Switzerland; and the proposer. One in orre t solution was submitted.
The
ontinuity of f is needed, as the following example by Ni olae Strungaru shows.
Let {1, 2} B bePa Hamel basis for R over Q. Ea h x R then has aunique expression
x = rx + sx 2 + xb b, for suitable rx , sx , xb in Q. If f(x) = 1 + sx 2, then f satises
bB
the identity of our problem but f is not onstant.

Let a, b, and c be real numbers satisfying a < 2(b + c), b < 2(c + a),
and c < 2(a + b). Prove that

4 a3 + b3 + c3 + 15abc
3 < 6.
(a + b + c)(ab + bc + ca)

Solution by Sefket Arslanagi , University of Sarajevo, Sarajevo, Bosnia and
Herzegovina.
First we prove that a + b + c and ab + bc + ca are positive quantities.
Adding the three onstraint inequalities yields a + b + c < 4(a + b + c),
hen e 0 < 3(a + b + c) and a + b + c is positive.
185
To prove that ab + bc + ca > 0, we may assume a = max{a, b, c}.

Sin e a + b + c > 0, we have a > 0. Now we obtain
ab + bc + ca = a(b + c) + bc
> a(a b c) + bc
= (a b)(a c) 0 ,
and hen e ab + bc + ca is positive.

By multiplying by (a + b + c)(ab + bc + ca) and olle ting terms, we
see that the given inequality is equivalent to the following two inequalities:

4 a3 + b3 + c3 + 6(a + b + c)(ab + bc + ca) 15abc > 0, (1)

3
4 a +b +c 3 3
3(a + b + c)(ab + bc + ca) + 15abc 0. (2)
After some algebrai work, the left side of (1) an be rewritten to obtain
(2b + 2c a)(2c + 2a b)(2a + 2b c) > 0 ,
whi h is learly true due to the hypotheses.

Further algebrai work shows that the left side of (2) an be rewritten
to obtain
(a b)2 (2a + 2b c) + (b c)2 (2b + 2c a) + (c a)2 (2c + 2a b) 0 ,
whi h is also true be ause of the hypotheses.

The inequality is proved.
Equality holds if and only if a = b = c, with a positive ommon value.
Also solved by ROY BARBARA, Lebanese University, Fanar, Lebanon; OLIVER GEUPEL,
Bruhl,
NRW, Germany; WALTHER JANOUS, Ursulinengymnasium, Innsbru k, Austria; KEE-
WAI LAU, Hong Kong, China; SALEM MALIKIC, student, Sarajevo College, Sarajevo, Bosnia
and Herzegovina; and the proposer. There were two in omplete solutions submitted.
The proposer remarked that some variables an be negative, for instan e a = 1, b = 4,
and c = 4 satisfy the onstraints.
3433 . [2009 : 173, 175 Proposed by an unknown proposer.

For ea h positive integer n prove that
1
X
n
1 2k 2n 2k 24n 2n
= .
k=0
2k + 1 k nk 2n + 1 n

NRW, Germany.
Our solution employs generating fun tions.
186
The Ma laurin series below are well known (see for example 1.641, #2
and 1.515, #4 in Table of Integrals, Series, and Produ ts by I.S. Gradshteyn
and I.M. Ryzhik, A ademi Press, 5 edition, 1996):
th

p X

(1)n 2n
sinh1 x = ln x + 1 + x2 = x2n+1 ;
n=0
4n (2n + 1) n

sinh1 x X

(1)n 4n 2n 1
= x2n+1 ;
1 + x2 n=0
2n + 1 n
where both series expansions are valid for x2 1.

Let f (x) = sinh1 x for x2 1. Then we have
!
X

(1)n X
n
1 2k 2n 2k
x2n+1
n=0
4n k=0
2k + 1 k nk
! !
X

(1)n 2n X

(1)n 2n
2n+1 2n
= x x
n=0
4n (2n + 1) n n=0
4n n

sinh1 x X
(1)n 4n 2n 1 2n+1
= f (x)f (x) = = x .
1 + x2 n=0
2n + 1 n
Comparing the oe ients of the terms x2n+1 yields the identity. The
al ulations are justied by the onvergen e of the power series for x2 1.
Also solved by MICHEL BATAILLE, Rouen, Fran e (two solutions); PAUL BRACKEN,
University of Texas, Edinburg, TX, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbru k,
Austria; JOEL SCHLOSBERG, Bayside, NY, USA; ALBERT STADLER, Herrliberg, Switzerland;
and the proposer.
Bataille pointed out that this problem is equivalent to problem 11356 of the Ameri an
Mathemati al Monthly (Vol. 115, No. 4, April, 2008, solution to appear ). [Ed: If we multiply
the identity in problem 11356 by 2n n
, then we obtain our problem. It is not lear whether
these two problems were submitted by the same proposer. Janous lo ated the identity as item
4.2.6.17 on p. 623 of Integrals and Series (Elementary Fun tions), Nauka, Mos ow, 1981 (in
Russian).
Stan Wagon remarked that by applying Mathematica and entering the summation in the
proposed identity, the answer on the right side omes out immediately.
3434. [2009 : 173, 175 Proposed by Bru e Shawyer, Memorial University

of Newfoundland, St. John's, NL.
Given the line segment LM N with LM : M N = 1 : and > 0, and
given the triangle ABC with ABC = x + y and tan x
tan y
1
= , onstru t the

angle x using only a straight edge and ompass.
187
I. Solution by Va lav Kone ny, Big Rapids, MI, USA.

Let L be a onvenient point on
the hal ine BA that denes ABC , C .q
and let N be some point within that
..
...
angle for whi h L N = LN , and M

..
D..... N
.
..................................................
be the point on the segment L N for

........... ............................................
..q q ........
. ....... .......... ..
..
. ...... .
.. ........ ..
whi h L M = LM . Let the parallel

... ....
.
.
.. ......
.. ..
..
. ... ..
...
. ... ..
..
to AB through N meet the hal ine

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

. .. .. ..
. ... .. ..
..
. ... ..
BC at D . Constru t the ir le with

..
.. .. ..
..
... ..
... .. ...
..
..
.. ....
... .. . ..
.. ......................
diameter BD and all it . Let the
... .. .. .
..
. .. ... .......................
.
E q qM
...
.. .. ............... ..
... .
parallel to BA through M meet for

.. .
... . .. .
. .......
....
....
....
...
....
.
.
. ... ............... ..
..
........... .... ..
... y
.. ... ..
...........
1
..
... ........... ... ..
the rst time at the point E . Claim:

..
.... ............ ... ..
A
.... .....
.. . .
.q.......................x
.. .....
...... ............. ...................
.
........... .
.
..
..
..
ABE is the desired angle x. To see

.............
........................................................q.......................q..................q .. .
. . .
.......... ............
B
...

.................
...............................
F L
why, let F be at the interse tion of the
lines DE and BA. Be ause L F , M E , and N D are parallel segments,
we have F E : ED = L M : M N = 1 : . Dene x = ABE and
y = EBC . Be ause BD is a diameter of , BE is perpendi ular to DE .
In the right triangles BEF and BED we have
tan x . FE 1
FE ED
=
BE BE
= = ,
tan y ED
as desired. By denition, ABC = x + y. Note that the triangle was not
needed for the onstru tion|the verti es A and C were used only to name
the given angle at B .
II. Solution by Edmund Swylan, Riga, Latvia.
Eq
Pla e the segment LN so that B ..q ............
...............................................................
N oin ides with C and L lies within

......... ... .........
......... ... .... ........
....... ....... .... ... ... ... .......
... ... ... .. .....
.. .... ..... .. ... ..... .... .....
ABC but not on AC . Denote by

.. .
... ........ ... ... .. ... .. ....
... ... . .
. ... .. . ....
...
.. y x........
. . . ... . . ....
. ...
. .. . .. ..
D the point where the parallel to AL

.. .. ...
. .
. . ...
...
.. .. ... ... ..
... .... .
..
... ... . .
...
...
.
...
...
..... .... . . . ... ..
through M meets AC . Then we have

. .. ... ..
.... ... .. ...
..... .
..
..
.. ..
. .... ... ...... .
... .
.. ..
...
. .... ... .
. . .
AD : DC = 1 : . Draw the ir um-

.. ... .. .. ...
.... .... ..
....
. ..
.
.
. .. .
...
.
. ..
. ...
. .... .. ... .. .. ...
. ... .. . .
. ... .
ir le of ABC , and draw the hord

.... .. .. ..
.... ...
.... ... ..
.
..
. ...
.
.
...
.
...
.
...
..
. .... .. ... .. ... ... .
.
. .. .. ... ... . ..
EP through D perpendi ular to AC

.... .. . .. .
... ....... .. ... ... .
.. ... ..
.
.
.. ...
.... ... ..
.. .... ..
.. ..
. .. ... .. .... .. ..
with E on the same side of AC as B .

.... ....
... ... .. .. ...
D ... .....
.... .. ... .. ...
... .... .
. .
.
. .
....q.... q .......................................................................................................................................................................
....q
In the right triangles DEA and CED C = N ........................ M
.......... .... .. .... ..
.. .. ..
A
.... ..... ... .....
....
..... .... ..
. . .............
we have
...... . .. ...
. ........
...............q.
. .
P
....... . . . .. .
... ... .... ...... ..
.........
........... ... .... ................ .....
................ ... .. ......
...
.......... q .............................................
... ............. .........
tan DEA . AD 1 1 .. ........
q
AD DC
..
.
..
= = =
..
. L
tan CED DE DE DC
But, P BA = P EA = DEA, while CBP = CED, when e
BP is the desired hal ine that partitions ABC into the angles x = ABP
and y = P BC .
Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Gree e; ROY BARBARA,
Lebanese University, Fanar, Lebanon; MICHEL BATAILLE, Rouen, Fran e; OLIVER GEUPEL,
Bruhl,
NRW, Germany; ALBERT STADLER, Herrliberg, Switzerland; and the proposer.
188
3435 . [2009 : 173, 176 Proposed by Dragoljub Milosevi , Gornji

Milanova , Serbia.
Let a, b, c, and d be positive integers. Prove that
1 1 5
.
a+b+c+d+2 (a + 1)(b + 1)(c + 1)(d + 1) 48
Solution by Albert Stadler, Herrliberg, Switzerland.

Let s = a + b + c + d. For s > 7 the inequality follows from
1 1 1 5
< .
a+b+c+d+2 (a + 1)(b + 1)(c + 1)(d + 1) 10 48
By assumption ea h of a, b, c, d is at least 1, and we now onsider the ase

where s 7. By the AM{GM Inequality,
4
a+1+b+1+c+1+d+1
(a + 1)(b + 1)(c + 1)(d + 1)
4

a+b+c+d 4
= 1+
4
4
s
= 1+
4
.
Then for s = 5, 6, 7 we have

1 1

a+b+c+d+2 (a + 1)(b + 1)(c + 1)(d + 1)
1 1
4
s+2 s
1+
4
5
< .
48
If a = b = c = d = 1, then s = 4 and equality holds.
Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Gree e; SEFKET
Lebanese University, Fanar, Lebanon; MICHEL BATAILLE, Rouen, Fran e; CHIP CURTIS,
Missouri Southern State University, Joplin, MO, USA; OLIVER GEUPEL, Bruhl, NRW, Germany;
RICHARD I. HESS, Ran ho Palos Verdes, CA, USA; WALTHER JANOUS, Ursulinengymnasium,
Innsbru k, Austria; CRISTINEL MORTICI, Valahia University of Târgoviste, Romania; JOEL
SCHLOSBERG, Bayside, NY, USA; TITU ZVONARU, Comane sti, Romania; and the proposer.
Both Geupel and Zvonaru proved that the inequality holds for nonnegative integers. Hess
observed that the inequality fails for a = b = c = d = 1.1, hen e the ondition that all
variables be integers annot be relaxed.
189
3436 . [2009 : 174, 176 Proposed by Dragoljub Milosevi , Gornji

Milanova , Serbia.
Let ABC be a right-angled triangle with hypotenuse AB . Let ma and
mb be the lengths of the medians to the sides BC and AC , respe tively.
Prove that
5 ma + mb 3
< .
2 a+b 2
Solution by Missouri State University Problem Solving Group, Springeld,

MO, USA.
Introdu e a oordinate system with
vertex A at the origin and AC along the ....
x-axis. The verti es of the triangle are

...
..
... B
then
A(0, 0), B(b, a), and C(b, 0); with
...
.
...
.
. ......
. .. .
a ... ..
the midpoint of side BC . Intro-
..
... ....
D b, ... ..
2 ... ... ...
du e the points E(a, b) and F (a, 0); ... ............. D
..
... .
.
... .... ......... ..... .
then EAF is ongruent to ABC . F
.. .. .....
.. ... ..... ... a/2
b a A
Furthermore, G a, 2 is the mid-
... ........ ..
. ... .
..................................................................................... ...........................................................
. ...
... .
...
. . .
point of EF , so ma = AD and mb = AG. b/2 ............................................................... ...... b C
. ..... ... . .
.
. ........ ..... .
By the triangle inequality in ACD a

, we G ...... ...................
. . ..
..
have AD < AC + CD = b + 2 , and ... ........ ...
.
.......... ...
..
likewise the triangle inequality in AF G
..
E ...
yields AG < AF + F G = a + 2b . Hen e,

3
ma + mb < (a + b) .
2
Similarly, the triangle inequality applied to ADG yields

2
a+b
AD + AG DG = (a + b)2 + ,
2
and therefore,
5
ma + mb (a + b) ,
2
whi h ompletes the proof.
Also solved by ARKADY ALT, San Jose, CA, USA; MIGUEL AMENGUAL COVAS,
Cala Figuera, Mallor a, Spain; GEORGE APOSTOLOPOULOS, Messolonghi, Gree e; SEFKET
Lebanese University, Fanar, Lebanon; RICARDO BARROSO CAMPOS, University of Seville,
Seville, Spain; MICHEL BATAILLE, Rouen, Fran e; CAO MINH QUANG, Nguyen Binh Khiem
High S hool, Vinh Long, Vietnam; CHIP CURTIS, Missouri Southern State University, Joplin,
MO, USA; OLIVER GEUPEL, Bruhl, NRW, Germany; JOHN HAWKINS and DAVID R. STONE,
Georgia Southern University, Statesboro, GA, USA; RICHARD I. HESS, Ran ho Palos Verdes,
CA, USA; JOHN G. HEUVER, Grande Prairie, AB; WALTHER JANOUS, Ursulinengymnasium,

Innsbru k, Austria; V ACLAV KONECN Y,
Big Rapids, MI, USA (2 solutions); KEE-WAI LAU,
Hong Kong, China; THANOS MAGKOS, 3rd High S hool of Kozani, Kozani, Gree e; SALEM
190
MALIKIC, student, Sarajevo College, Sarajevo, Bosnia and Herzegovina; CRISTINEL MORTICI,
Valahia University of Târgoviste, Romania; JOEL SCHLOSBERG, Bayside, NY, USA; ALBERT
STADLER, Herrliberg, Switzerland; KONSTANTINE ZELATOR, University of Pittsburgh, Pitts-
burgh, PA, USA; TITU ZVONARU, Comane sti, Romania; and the proposer.
3437 . [2009 : 174, 176 Proposed by Pham Huu Du , Ballajura, Australia

and Vo Quo Ba Can, Can Tho University of Medi ine and Pharma y, Can
Tho, Vietnam.
Let a, b, and c be positive real numbers. Prove that

a2 + bc b2 + ca c2 + ab
+ + 3(a + b + c) .
b+c c+a a+b
Solution by Titu Zvonaru, Comane

sti, Romania.
P P
[Ed: Throughout we shall use for .
y li
By the Generalized Holder

Inequality, we easily dedu e that
r r
X a2 + bc X a2 + bc X
(a2 + bc)2 (b + c)
b+c b+c
X 3
(a2 + bc) . (1)
Hen e, it su es to prove that

X 3 X
(a2 + bc) 3(a + b + c) (a2 + bc)2 (b + c) . (2)
By tedious but straightforward omputations, we nd that (2) is equiv-

alent, in su ession, to
X 3 X
(a2 + bc) 3 (a2 + bc)2 (b2 + c2 + ab + ac + 2bc) ;
X X
(a2 + bc)3 + 3 (a2 + bc)2 (b2 + ca)
X
+3 (a2 + bc)(b2 + ca)2 + 6(a2 + bc)(b2 + ca)(c2 + ab)
X X
3 (a2 + bc)2 (b2 + ca + c2 + ab) + 6 (a2 + bc)2 bc ;
X
[(a2 + bc)3 (a2 + bc)2 bc] + 6(a2 + bc)(b2 + ca)(c2 + ab)
X
5 (a2 + bc)2 bc ;
191
X X X
a2 (a2 +bc)2 + 12a2 b2 c2 + 6 a3 b3 + 6 a4 bc
X
5 (a2 + bc)2 bc ;
X X X
15a2 b2 c2 + a6 + 8 a4 bc + 6 a3 b3
X X X
5 a4 bc + 10 a2 b2 c2 + 5 b3 c3 ;
X X X
a6 + 3 a4 bc + a3 b3 15a2 b2 c2 .
The last
P inequalityPabove is true, sin e by Pthe AM{GM Inequality we
have 6
a 3a2 b2 c2 , a4 bc 3a2 b2 c2 , and a3 b3 3a2 b2 c2 .
y li y li y li
The required inequality now follows from (1) and (2).

Note that equality holds if and only if a = b = c.
Also solved by OLIVER GEUPEL, Bruhl,
NRW, Germany; WALTHER JANOUS, Ursulinen-
gymnasium, Innsbru k, Austria; ALBERT STADLER, Herrliberg, Switzerland; and the proposer.
3438
. Proposed by Vo Quo Ba Can, Can Tho University of Medi ine
and Pharma y, Can Tho, Vietnam.
Let a, b, and c be nonnegative real numbers. Prove the inequality
below for all 0, or give a ounterexample:

X a2 + bc
2+ .
y li
b2 + c2 2
Solution by Albert Stadler, Herrliberg, Switzerland.

We prove that the statement is false for > 32.
Let a = 0 and x = bc , where b 6= c. Then the inequality reads
r
x 1
+x+ 2+ ,
x2 + 1 x 2
whi h is equivalent to
2
1
x+ 2
x
q x ,
1

2 x2 + 1
whi h in turn is equivalent to
2
4(x2 + 1)2 1 x 2(x2 + 1) x2 + 1 2

2
+
x2 + 1
=
x
+ 2 .
x2 x
192
2
2 2
The fun tion f (t) = 2(t t+ 1) t +1
t
+ 2 attains its minimum at t = 1,
and that minimum value is 32.
So, if > 32, then a triple (a, b, c) exists (with x su iently lose to 1)
that violates the inequality.
JOHN HAWKINS and DAVID R. STONE, Georgia Southern University, Statesboro, GA, USA;
and RICHARD I. HESS, Ran ho Palos Verdes, CA, USA.
Stadler also proved that the inequality holds for = 0. He ommented that numeri al
eviden e suggests that the inequality holds true for 0 32.
Crux Mathemati orum

with Mathemati al Mayhem
Former Editors / An iens Reda teurs:
Bru e L.R. Shawyer, James E. Totten
Crux Mathemati orum

Founding Editors / Reda teurs-fondateurs:
Leopold
Sauve & Frederi k G.B. Maskell
G.W. Sands, R.E. Woodrow, Bru e L.R. Shawyer
Mathemati al Mayhem
Founding Editors / Reda teurs-fondateurs:
Patri k Surry & Ravi Vakil
Philip Jong, Je Higham, J.P. Grossman,
Andre Chang, Naoki Sato, Cyrus Hsia, Shawn Godin, Je Hooper

CRUXv 36 N 3

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 36 N 3

Uploaded by

Copyright:

Available Formats

129

SKOLIAD No. 124

Lily Yen and Mogens Hansen

Christopher Newport University Regional Mathemati s Contest, 2009. Our

1 . Un ertain nombre de fees

4 . Madeleine et Nadine ont de id

7. Un plan her re tangulaire de taille 24 pieds 40 pieds, est pave de tuiles

Next follow solutions to the Cariboo College High S hool Mathemati s

3. A student asks you to hoose a number from 1 to 9 and multiply it by 109,

Solution by Julia Peng, student, Campbell Collegiate, Regina, SK.

Consider su h a quarter ir le (see the thi k lined ... ...

part of the gure). By the Pythagorean Theorem, B ... ... ...

252 2 + 252 2 = 252 22 , .. 50 ......

so |AB| = 25 2 = 50. Then draw the ir le

through points O, A, and B . Sin e AOB = 90,

the entre of that ir le is the midpoint of AB .

M432 . Proposed by the Mayhem Sta .

M436 . Proposed by Ne ulai Stan iu, George Emil Palade Se ondary

M437 . Proposed by Samuel Gomez

M433 . Propose par Bru e Shawyer, Universite Memorial de Terre-

M437 . Propose par Samuel Gomez

Sin e a possible fa torization of n2 is 1 n2 , then we an let x n = 1 (and

Totten{M2 . Proposed by Bru e Shawyer, Memorial University of New-

The boundary of the shadow on the moon is

always a ir ular ar . On a ertain day, the moon

is seen with the shadow passing through diamet-

ri ally opposite points. If the entre of the ir ular

the moon, determine the exa t proportion of the

moon that is not in shadow.

Solution by Geo rey A. Kandall, Hamden, CT, USA.

this ar . Let P and Q be the points where the

shadow interse ts the ir umferen e of the moon.

Let x denote the area of P OQ, y the area of the x

region between P Q and the ar through P and Q C

entred at O, and z the area of the region between

the two ar s through P and Q.

Totten{M3 . Proposed by John Ciriani, Kamloops, BC.

Totten{M4 . Proposed by Bill Sands, University of Calgary, Calgary,

This value of t gives G = 5t = 500 and B = 14t = 1400, whi h are

Totten{M5 . Proposed by Ovidiu Furdui, Campia Turzii, Cluj, Romania.

Totten{M6 . Proposed by Suzanne Feldberg, Thompson Rivers Univer-

How many di erent 6-pointed, 8-pointed, or 9-pointed stars an one

Totten{M7 . Proposed by John Grant M Loughlin, University of New

Suppose that an unmarked ruler is known to be exa tly 30 m in length.

Totten{M8. Proposed by Edward J. Barbeau, University of Toronto,

f (g(A, B, C)) = f (A2 , AC, C 2 )

Therefore, f and g are inverses of ea h other, so they ea h de ne a

Totten{M9 . Proposed by Kirk Evenrude, Kamloops, BC.

Solution by Geo rey A. Kandall, Hamden, CT, USA.

Totten{M10. Proposed by Ni holas Bu k, College of New Caledonia,

solution to Ax2 + By2 = pk2 , whi h implies that P (k 2) is true, whi h

Problem of the Month

Any ommuters in the rowd?

. Here g(x) is a ubi polynomial, so suppose that

Therefore, for all x, we have g(x) + g(x) = 0 or g(x) g(x) = 0.

Therefore, g(x) + g(x) = 0 is equivalent to

or bx2 + e = 0. Also, g(x) g(x) = 0 is equivalent to

Therefore, for every x, at least one of bx2 + e = 0 and ax3 + dx = 0 is

THE OLYMPIAD CORNER

5 . Find all ontinuous fun tions f : R R su h that for all x, y R,

6 . Consider a 10 10 grid. A move onsists of olouring 4 unit squares

Next are the 10 Grade problems of the XXXIII Russian Mathemati al

Olympiad, Final Round 2006{2007. Thanks again go to Bill Sands, Canadian

7. Un plan her re tangulaire de taille 24 pieds 40 pieds, est pave de tuiles

M432 . Proposed by the Mayhem Sta.

M433 . Propose par Bru e Shawyer, Universite Memorial de Terre-

M437 . Propose par Samuel Gomez

Solution by Georey A. Kandall, Hamden, CT, USA.

How many dierent 6-pointed, 8-pointed, or 9-pointed stars an one

Therefore, f and g are inverses of ea h other, so they ea h dene a

Solution by Georey A. Kandall, Hamden, CT, USA.

Solution by Jose Luis Daz-Barrero, Universitat Polite ni a

4. Given three distin t real numbers a1 , a2 , and a3 , dene three real num-