CRUXv 34 N 8

449
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 (As ension of Our Lord Se ondary
S hool, Mississauga), Eri Robert (Leo Hayes High S hool, Frederi ton), Larry
Ri e (University of Waterloo), and Ron Lan aster (University of Toronto).
Mayhem Problems
Veuillez nous transmettre vos solutions aux problemes

du present

numero

avant le 15 Mars 2009. Les solutions re ues apres
ette date ne seront prises en
ompte que s'il nous reste du temps avant la publi ation des solutions.
Chaque probleme

sera publie dans les deux langues o ielles du Canada
(anglais et fran ais). Dans les numeros

1, 3, 5 et 7, l'anglais pre
edera

le fran ais,
et dans les numeros
2, 4, 6 et 8, le fran ais pre
edera

l'anglais.
La reda tion

souhaite remer ier Jean-Mar Terrier, de l'Universite de
Montreal,
d'avoir traduit les problemes.

M369. Propose par l'Equipe
de Mayhem.
Soit A(0, 0), B(6, 0), C(6, 4) et D(0, 4) les sommets d'un re tangle.
Par le point P (4, 3), on tra e d'une part une droite horizontale oupant BC
en M et AD en N et d'autre part une droite verti ale oupant AB en Q et
CD en R. Montrer que AP , DM et BR passent toutes par le même point.

de Mayhem.
(a) Montrer que cos(A + B) + cos(A B) = 2 cos A cos B pour tous les
angles A et B .

D
CD
(b) Montrer que cos C + cos D = 2 cos C +
cos
pour tous
2
2
les angles C et D.
( ) Trouver la valeur exa te de cos 20 + cos 60 + cos 100 + cos 140 ,
sans l'aide d'une al ulatri e.

M371. Propose par Panagiote Ligouras, E ole
Se ondaire Leonard

de
Vin i, No i, Italie.
Un segment AB de longueur 3 ontient un point C tel que AC = 2. On
onstruit d'un même ot
^ e de AB deux triangles equilat

eraux

ACF et CBE .
Determiner

l'aire du triangle AKE si K est le point milieu de F C .
450

de Mayhem.
Soit x un nombre reel

satisfaisant x3 = x + 1. Trouver des entiers a, b
7
et c de sorte que x = ax2 + bx + c.
M373. Propose par Kunal Singh, etudiant,

Kendriya Vidyalaya S hool,
Shillong, Inde.
Les ot
^ es
d'un triangle sont mesures
par trois nombres entiers onse u
tifs et le plus grand angle est le double du plus petit. Determiner

la longueur
des ot
^ es
du triangle.
M374. Propose par Mihaly Ben ze, Brasov, Roumanie.
Soit p un nombre premier xe,

ave p 3. Trouver le nombre de solutions de x3 + y3 = x2 y + xy2 + p2009 , ou x et y sont des entiers.

M375. Propose par Ne ulai Stan iu, E ole
Te hnique Superieure

de Saint
Mu eni Sava, Ber a, Roumanie.
Determiner

toutes les solutions reelles

du systeme
d'equations

1
x2
4
y2
9
z2
= 4;
x2 + y 2 + z 2 = 9 ;
xyz =
9
2
.................................................................
M369. Proposed by the Mayhem Sta.
A re tangle has verti es A(0, 0), B(6, 0), C(6, 4), and D(0, 4). A horizontal line is drawn through P (4, 3), meeting BC at M and AD at N . A
verti al line is drawn through P , meeting AB at Q and CD at R. Prove that
AP , DM , and BR all pass through the same point.

(a) Prove that cos(A + B) + cos(A B)
A and B .
(b) Prove that cos C + cos D = 2 cos
C and D .
= 2 cos A cos B
C +D
2
cos
CD
2
for all angles

for all angles
( ) Determine the exa t value of cos 20 + cos 60 + cos 100 + cos 140 ,
without using a al ulator.
M371. Proposed by Panagiote Ligouras, Leonardo da Vin i High S hool,

No i, Italy.
Suppose that the line segment AB has length 3 and C is on AB with
AC = 2. Equilateral triangles ACF and CBE are onstru ted on the same
side of AB . If K is the midpoint of F C , determine the area of AKE .
451

A real number x satises x3
so that x7 = ax2 + bx + c.
= x + 1.
Determine integers a, b, and c
M373. Proposed by Kunal Singh, student, Kendriya Vidyalaya S hool,

Shillong, India.
The side lengths of a triangle are three onse utive positive integers
and the largest angle in the triangle is twi e the smallest one. Determine the
side lengths of the triangle.
M374. Proposed by Mihaly Ben ze, Brasov, Romania.

Suppose that p is a xed prime number with p 3. Determine the
number of solutions to x3 + y3 = x2 y + xy2 + p2009 , where x and y are
integers.
M375. Proposed by Ne ulai Stan iu, Saint Mu eni Sava Te hnologi al

High S hool, Ber a, Romania.
Determine all real solutions to the system of equations
1
x2
4
y2
9
z2
= 4;
x2 + y 2 + z 2 = 9 ;
xyz =
9
2
Mayhem Solutions
M332.
Proposed by Dionne Bailey, Elsie Campbell, and Charles R.

Diminnie, Angelo State University, San Angelo, TX, USA.
A losed right ir ular ylinder has an integer radius and an integer
height. The numeri al value of the volume is four times the numeri al value
of its total surfa e area (in luding its top and bottom). Determine the smallest possible volume for the ylinder.
Solution by Cao Minh Quang, Nguyen Binh Khiem High S hool, Vinh Long,
Vietnam.
Let r and h be the radius and the height of the losed right ir ular
ylinder. The volume of su h a ylinder is V = r2 h and the surfa e area is
A = 2r 2 + 2rh.
From the hypotheses, r2 h = 4(2r2 + 2rh), or rh = 8r + 8h, or
rh 8r 8h + 64 = 64, or (r 8)(h 8) = 64. Note that r 8 > 8 and
h 8 > 8. This gives us the following possibilities:
452
r8
h8
r
h
V
1
64
9
72
5832
2
32
10
40
4000
4
16
12
24
3456
8
84
16
16
4096
16
4
24
12
6912
32
2
40
10
16000
64
1
72
9
46656
Thus, the smallest possible volume for the ylinder is 3456.

Also solved by DENISE CORNWELL, student, Angelo State University, San Angelo,
\Abastos",
TX, USA; RICHARD I. HESS, Ran ho Palos Verdes, CA, USA; RICARD PEIRO, IES
Valen ia, Spain; BILLY SUANDITO, Palembang, Indonesia; and TITU ZVONARU, Comane
sti,
Romania. There was 1 in omplete solution submitted.

Anne and Brenda play a game whi h begins with a pile of n toothpi ks.
They alternate turns with Anne going rst. On ea h player's turn, she must
remove 1, 3, or 6 toothpi ks from the pile. The player who removes the last
toothpi k wins the game. For whi h of the values of n from 36 to 40 in lusive
does Brenda have a winning strategy?
Solution by Ri hard I. Hess, Ran ho Palos Verdes, CA, USA, modied by the
editor.
We an build a table of winning and losing positions for Anne. Her
winning positions start with 1, 3, or 6, sin e she an immediately win by
removing all of the toothpi ks.
Starting with 2 toothpi ks, Anne must remove 1 toothpi k, leaving
Brenda with 1, and so Brenda wins. Starting with 4 toothpi ks, Anne must
remove 1 or 3 toothpi ks, leaving Brenda with 3 or 1 (respe tively), and so
Brenda wins by removing all of the toothpi ks.
Starting with 5 toothpi ks, Anne an remove 3 toothpi ks, thus leaving
Brenda with 2 toothpi ks. Sin e 2 is a losing position for whoever goes rst,
then Brenda loses, so Anne wins.
So far, 1, 3, 5, and 6 are winning positions for Anne, while 2 and 4 are
losing positions for Anne.
Starting with a pile of size n, Anne must redu e the pile to one of size
n 1, n 3, or n 6 and pass to Brenda. If the person who goes rst has
a winning strategy starting with a pile of ea h of these sizes, then Anne will
lose. In other words, if Anne has a winning strategy starting with piles of
size n 1, n 3, and n 6, then Anne will lose starting with a pile of size
n, as Brenda an implement Anne's strategy for the smaller pile and win, no
matter what Anne does. If one or more of these pile sizes are su h that the
rst person does not have a winning strategy, then Anne should redu e to
this size, thus preventing Brenda from being able to win, and so Anne herself
will win.
We an examine the ases from n = 7 to n = 40, obtaining the following lists:
453
Winning positions for Anne: 1, 3, 5, 6, 7, 8, 10, 12, 14, 15, 16, 17, 19, 21,
23, 24, 25, 26, 28, 30, 32, 33, 34, 35, 37, 39.
Losing positions for Anne: 2, 4, 9, 11, 13, 18, 20, 22, 27, 29, 31, 36, 38, 40.
Therefore, Brenda wins for n = 36, 38, 40.
Also solved by JACLYN CHANG, student, Western Canada High S hool, Calgary, AB.
See the Problem of the Month olumn in [2007 : 15{17 for a similar problem with a
more detailed explanation.

(a) Determine all integers x for whi h
x3
3x 2
(b) Determine all integers y for whi h
3y 3 + 3
3y 2 + y 2
is an integer.
is an integer.
I. Solution by Edin Ajanovi , student, First Bosniak High S hool, Sarajevo,
Bosnia and Herzegovina.
Let A be an integer su h that A =
x3
.
3x 2
Then 3A is an integer and
3x 9
3x 2 7
7
3A =
=
=1
.
3x 2
3x 2
3x 2
7
Thus, 3x 2 is an integer; that is, 3x 2
is a divisor of 7, so 3x 2 is
one of 1, 7. Sin e x is an integer, x = 1 or x = 3. This answers part (a).
Now let B be an integer su h that
B
3y 3 + 3
y 2 2y 3
=
y
3y 2 + y 2
3y 2 + y 2
(y 3)(y + 1)
y3
= y
= y
(y + 1)(3y 2)
3y 2
=
y3
is an integer, 3y
is an integer. From the solution to part (a),
2
y = 1 or y = 3, whi h answers part (b).
Sin e
II. Solution by Edward T.H. Wang, Wilfrid Laurier University, Waterloo, ON.
We show that the only integer solutions to part (a) are x = 1 and x = 3.
x3
Let f (x) = 3x
. Then f (0) = 32 , f (1) = 2, f (2) = 14 , and
2
f (3) = 0. Of these, only f (1) and f (3) are integers.
If x > 3, then f (x) is not an integer, sin e 3x 2 > x 3 > 0 for
x3
x > 3 and so 0 <
< 1.
3x 2
s+3
where s 1. Then f (x) = f (s) = 3s
.
+2
Sin e 3s + 2 > s + 3 > 0 for s 1, f (s) is not an integer by a similar
argument so, f (x) is not an integer.
If x
1,
let x
= s
454
Therefore, f (x) is an integer for integer values of x if and only if x = 1
or x = 3.
\Abastos", Valen ia, Spain; CAO MINH QUANG,

Also solved by RICARD PEIRO, IES
Nguyen Binh Khiem High S hool, Vinh Long, Vietnam; and TITU ZVONARU, Comane
sti,
Romania. There was one in orre t and one in omplete solution submitted.
M335. Proposed by the Mayhem sta.

In a sequen e of four numbers, the se ond number is twi e the rst
number. Also, the sum of the rst and fourth numbers is 9, the sum of the
se ond and third is 7, and the sum of the squares of the four numbers is 78.
Determine all su h sequen es.
Solution by Denise Cornwell, student, Angelo State University, San Angelo,
TX, USA.
Let a, b, c, and d represent the rst, se ond, third and fourth number,
respe tively. We an now write the given information as b = 2a, a + d = 9,
b + c = 7 and a2 + b2 + c2 + d2 = 78.
The rst three equations allow us to rewrite b, c, and d in terms of a,
obtaining b = 2a, c = 7 b = 7 2a, and d = 9 a.
Therefore,
a2 + (2a)2 + (7 2a)2 + (9 a)2
a2 + 4a2 + 49 28a + 4a2 + 81 18a + a2 78
5a2 23a + 26
(5a 13)(a 2)
=
=
=
=
78 ,
0,
0,
0,
hen e a = 13
or a = 2.
5
Therefore, the sequen es are a = 13
, b = 26
, c = 95 , d = 32
and
5
5
5
a = 2, b = 4, c = 3, d = 7. Both sequen es satisfy the given requirements.
Also solved by EDIN AJANOVIC, student, First Bosniak High S hool, Sarajevo,
Bosnia and Herzegovina; JACLYN CHANG, student, Western Canada High S hool, Calgary,
AB; RICHARD I. HESS, Ran ho Palos Verdes, CA, USA; CAO MINH QUANG, Nguyen Binh
Khiem High S hool, Vinh Long, Vietnam; KUNAL SINGH, student, Kendriya Vidyalaya S hool,
Shillong, India; BILLY SUANDITO, Palembang, Indonesia; and TITU ZVONARU, Comane
sti,
Romania. There was one in orre t and one in omplete solution submitted.

A latti e point is a point (x, y) in the oordinate plane with ea h of x
and y an integer. Suppose that n is a positive integer. Determine the number
of latti e points inside the region |x| + |y| n.
Solution by Edin Ajanovi , student, First Bosniak High S hool, Sarajevo,
Bosnia and Herzegovina, modied by the editor.
We an rewrite the given inequality as the equations |x| + |y| = 0 and
|x| + |y| = k for 1 k n, where x, y Z.
455
The equation
(x, y) = (0, 0).
|x| + |y| = 0
has one integer solution only, namely
Consider next |x| + |y| = k, for an integer k with 1 k n. We an

remove the absolute values by splitting into four ases:
Case 1. The integers x and y satisfy x + y = k, where x 0 and y 0.
This has solutions (k, 0), (k 1, 1), . . . , (1, k 1), (0, k), for a total of k + 1
solutions.
Case 2. The integers x and y satisfy x y = k, where x 0 and
y < 0.
This has solutions (k 1, 1), (k 2, 2), . . . , 1, (k 1) , (0, k), for
a total of k solutions.
Case 3. The integers x and y satisfy x + y = k, where x < 0 and y 0.
This ase is the same as Case 3, but with the roles of x and y swit hed, so
there are a total of k solutions here as well.
Case 4. The integers x and y satisfy
x < 0 and y < 0.
x y = k, where

This has solutions
1,
(k
1)
,
2,
(k
2)
,
.
.
.
, (k 2), 2 ,

(k 1), 1 , for a total of k 1 solutions.
Thus, for ea h k with 1 k n, the
(k + 1) + k + k + (k 1) = 4k solutions.
equation
|x| + |y| = k
has
Therefore, the number of latti e points inside the region is

1+
n
X
4k
1 + 4
k=1
n
X
k=1
1 + 4
k = 1 + 4 (1 + 2 + + n)
n(n + 1)
= 2n2 + 2n + 1 .
2
\Abastos", Valen ia, Spain. There were one in orre t

Also solved by RICARD PEIRO, IES
and two in omplete solutions submitted.

On sides AB and CD of re tangle ABCD with AD
and E are hosen so that AF CE is a rhombus.
(a) If AB = 16 and BC
(b) If AB = x and BC
= 12,
= y,
< AB ,
points F
determine EF .
determine EF in terms of x and y.
Solution by Kunal Singh, student, Kendriya Vidyalaya S hool, Shillong, India.

We present the solution to (b), whi h is a general version of the spe i
ase in (a).
Suppose that AF = F C = CE = EA = m. Let O be the point of
interse tion of diagonals AC and EF of rhombus AF CE . Note that AC
and EF are perpendi ular and bise t ea h other at O.
456
By the Pythagorean Theorem,
CB 2 ,
y2 ,
y2 ,
x2 + y 2 ,
x2 + y 2
=
.
2x
CF 2 F B 2
m2 (x m)2
2
m x2 + 2mx m2
2mx
=
=
=
=
Now, AF
OA =
x2 + y 2
2x
and AC = AB 2 + BC 2 =
1
AC . Thus, by the Pythagorean Theorem again,
2
= m =
OF 2
= AF 2 OA2
2
2
x + y2
=
2x
Also,
!2
p
x2 + y 2
2
x4 + y 4 + 2x2 y 2
x2 + y 2
x4 + y 4 + 2x2 y 2 x4 x2 y 2
4x2
4x2
p
x2 + y 2 .
2 2
y +x y
4x2
Therefore,
OF =
y 4 + x2 y 2
4x2
and
EF = 2 OF =
In part (a), this yields
2y
p
y 2 (y 2 + x2 )
2x
x2 + y 2
2x
p
x2 + y 2
2x
p
x2 + y 2
x
12 162 + 122
12(20)
EF =
=
= 15 .
16
16
Also solved by EDIN AJANOVIC, student, First Bosniak High S hool, Sarajevo, Bosnia
and Herzegovina; JACLYN CHANG, student, Western Canada High S hool, Calgary, AB (part
\Abastos",
(a) only); RICHARD I. HESS, Ran ho Palos Verdes, CA, USA; RICARD PEIRO, IES
Valen ia, Spain; CAO MINH QUANG, Nguyen Binh Khiem High S hool, Vinh Long, Vietnam;
BILLY SUANDITO, Palembang, Indonesia; LUYAN ZHONG-QIAO, Columbia International
College, Hamilton, ON; and TITU ZVONARU, Comane
sti, Romania.
457
Problem of the Month
Ian VanderBurgh
Here is a problem that might seem to be not very interesting initially,
but turns out to have a whole lot of unexpe ted solutions.
Problem (2005 Canadian Open Mathemati s
3
0
5
6 2
Challenge) In the grid shown, ea h row has 2 5
0
1
y
a value assigned to it and ea h olumn has a
5
2
x
8
0
value assigned to it. The number in ea h ell
0 3
2
3 5
is the sum of its row and olumn values. For 4 7 2 1 9
example, the \8" is the sum of the value assigned to the 3 row and the value assigned to the 4 olumn. Determine
the values of x and y.
It is tempting rst of all to give labels to the values that are assigned
to the rows and olumns in order to be able to dive into some algebra. Let's
label the values assigned to the ve olumns A, B , C , D, E and the values
assigned to the ve rows a, b, c, d, e.
Ea h entry in the table gives us an equation involving two of these
variables. For example, the 3 in row 4, olumn 2 gives us d + B = 3,
and the 9 in row 5, olumn 5, gives us e + E = 9. We ould pro eed and
write down 25 equations, one for ea h entry in the table. These equations
would in lude 12 variables { the 10 that label the rows and olumns together
with x and y. We ould then spend pages and pages wading through algebra
trying to ome up with the answers. At this point, we would hope that there
has to be a better way. Maybe we should have looked before we leapt!
Here are three neat ways to approa h this. (As a point of interest, I
was re ently talking about this problem with a friend while driving and so
neither of us really wanted to do any algebra, and so were for ed to ome up
with better ways to do it.)
Solution 1. If we hoose ve entries from the table whi h in lude one from
ea h row and one from ea h olumn, then the sum of these entries is onstant
no matter how we hoose the entries, as it is always equal to
rd
th
A + B + C + D + E + a + b + c + d + e.
Can you see why? Here are three ways in whi h this an be done (looking at
the underlined numbers in the two grids below and the grid on the following
page):
3
2
5
0
4
0
5
2
3
7
5
0
x
2
2
6
1
8
3
1
2
y
0
5
9
3
2
5
0
4
0
5
2
3
7
5
0
x
2
2
6
1
8
3
1
2
y
0
5
9
458
Therefore,
=
=
3
2
5
0
4
3 + (5) + 2 + 8 + (9)
(4) + (3) + x + 1 + (2)
3 + y + 2 + (2) + 3 ,
0
5
2
3
7
5
0
x
2
2
6
1
8
3
1
2
y
0
5
9
or 1 = x 8 = y + 6. Thus, x = 7 and
y = 7.
Solution 2. Consider the rst two entries in row 1. From the labels above, we
have 3 = A + a and 0 = B + a. Subtra ting these, we obtain the equation
3 = 3 0 = (A + a) (B + a) = A B .
Noti e that whenever we take entries in olumns 1 and 2 from the same
row, their dieren e will always equal A B , whi h is equal to 3. Similarly,
sin e the dieren e between the 0 and the 5 in the rst row is 5, then every
entry in olumn 3 will be 5 greater than the entry in olumn 2 from the same
row. In row 3, we see that x = 2 + 5 = 7.
Also, sin e the dieren e between the 6 and the 2 in the rst row is
8, then every entry in olumn 5 is 8 less than the entry in olumn 4 from the
same row. In row 2, we see that y = 1 8 = 7. Thus, x = 7 and y = 7.
Solution 3. Consider the sub-grid
0
x
1
8
Sin e the 0 is in row 2 and olumn 3, then 0 = b + C . Similarly,

1 = b + D , 8 = c + D , and x = c + C .
But then 0 + 8 = (b + C) + (c + D) = (c + C) + (b + D) = x + 1, or
x = 7.
In a similar way, by looking at the sub-grid 18 y0 we an show that
1 + 0 = y + 8, or y = 7. Thus, x = 7 and y = 7.
So there are three dierent but neat solutions to the problem. One
footnote to the nal solution is that in fa t, in any sub-grid of the form
p q
, we must have p + s = q + r. Can you see why?
r s
Another interesting point about this problem is that it might be easier
for those who know less! If we repla ed the x and the y with \?" and gave
it to someone who didn't know a lot of algebra, they might nd the answers
faster than those of us who go immediately to algebra. Sometimes, the extra
ma hinery that we have an get in the way.
As 2008 draws to a lose, the Mayhem Editor has three enormous sets of
thanks to oer. First, to the Mayhem Sta, espe ially to Monika Khbeis and
Eri Robert, for all of their help over the past year. Se ond, to the Editorin-Chief of CRUX with MAYHEM, Va lav

Linek, for all of his help and en ouragement over the past year (as well as for his sharp eyes!). Third, to
the Mayhem readership for their support. Please keep those problems and
solutions oming!
459
THE OLYMPIAD CORNER
No. 274
R.E. Woodrow
With the Winter break oming up, I have de ided to fo us this issue
mainly on providing problems for your puzzling pleasure, and to give some
time for the mails to deliver the solutions to problems from 2008 numbers
of the Corner to restore the readers' solutions le, whi h is parti ularly thin
for the February 2008 number, as you will see later in the olumn.
To start you o we have the problems proposed but not used at the
47 IMO in Slovenia 2006. My thanks go to Robert Morewood, Canadian
Team Leader at the IMO for olle ting them for our use.
th
47th INTERNATIONAL MATHEMATICAL OLYMPIAD

SLOVENIA 2006
Problems Proposed But Not Used
Contributing Countries. Argentina, Australia, Brazil, Bulgaria, Canada,

Colombia, Cze h Republi , Estonia, Finland, Fran e, Georgia, Gree e, Hong
Kong, India, Indonesia, Iran, Ireland, Italy, Japan, Republi of Korea,
Luxembourg, Netherlands, Poland, Peru, Romania, Russia, Serbia and
Montenegro, Singapore, Slovakia, South Afri a, Sweden, Taiwan, Ukraine,
United Kingdom, United States of Ameri a, Venezuela.
Problem Sele tion Committee. Andrej Bauer, Robert Gerets hlager, Geza

Kos,
Mar in Ku zma, Sventoslav Sav hev.
Algebra
A1. Given an arbitrary real number a0 , dene a sequen e of real numbers
a0 , a1 , a2 , . . .
by the re ursion
ai+1 = ai {ai } ,
i 0,
where ai is the greatest integer not ex eeding ai , and {ai }

Prove that ai = ai+2 for su iently large i.
A2. Let a0
= 1
by the re ursion
= ai ai .
and dene the sequen e of real numbers a0 , a1 , a2 , . . .

n
X
ank
k=0
k+1
for n 1. Show that an > 0 for n 1.
= 0
460
A3. Let c0 = 1, c1 = 0 and dene the sequen e c0 , c1 , c2 , . . . by the

re ursion cn+2 = cn+1 + cn for n 0. Let S be the set of ordered pairs
(x, y) su h that
X
X
x =
cj and y =
cj1
jJ
jJ
for some nite set J of positive integers. Prove that there exist real numbers
, , m, and M with the property that an ordered pair of non-negative
integers (x, y) satises the inequality
m < x + y < M
if and only if (x, y) S . (By onvention an empty sum is 0.)
A4. Let a1 , a2 , . . . , an be positive real numbers. Prove that

X
i<j
X
ai aj
n
ai aj .
ai + aj
2(a1 + a2 + + an ) i<j
A5. Let a, b, and c be the lengths of the sides of a triangle. Prove that
b+ca
c+ab
a+bc
3.
b+ c a
c+ a b
a+ b c
Combinatori s
C1. There are n 2 lamps L1 , L2 , . . . , Ln arranged in a row. Ea h of them
is either on or o. Initially the lamp L1 is on and all of the other lamps are
o. Ea h se ond the state of ea h lamp hanges as follows: if the lamp Li
and its neighbours (L1 and Ln ea h have one neighbor, any other lamp has
two neighbours) are in the same state, then Li is swit hed o; otherwise, Li
is swit hed on. Prove that there are
(a) innitely many n for whi h all of the lamps will eventually be o,
(b) innitely many n for whi h the lamps will never be all o.
C2. Let S be a nite set of points in the plane su h that no three of them
are on a line. For ea h onvex polygon P whose verti es are in S , let a(P )
be the number of verti es of P , and let b(P ) be the number of points of S
whi h are outside of P . Prove that for every real number x
X
P
xa(P ) (1 x)b(P ) = 1 ,
where the sum is taken over all onvex polygons with verti es in S . (A line
segment, a point, and the empty set are onvex polygons of 2, 1, and 0
verti es, respe tively.)
461
C3. A ake is shaped as an n n square with n2 unit squares. Strawberries
lie on some of the unit squares so that ea h row or olumn ontains exa tly
one strawberry; all this arrangement A.
Let B be another su h arrangement. Suppose that every grid re tangle
with one vertex at the top left orner of the ake ontains no fewer strawberries of arrangement B than of arrangement A. Prove that the arrangement
B an be obtained from A by performing a sequen e of swaps, where a swap
onsists of sele ting a grid re tangle with only two strawberries, situated at
its top right orner and bottom left orner, and then moving these two strawberries to the other two orners of that re tangle.
C4. An (n, k)-tournament is a ompetition with n players held in k rounds
su h that
(a) Ea h player plays in ea h round, and every two players meet at most
on e.
(b) If player A meets player B in round i, player C meets player D in round
i, and player A meets player C in round j , then player B meets player
D in round j .
Determine all pairs (n, k) for whi h there exists an (n, k)-tournament.
C5. A holey triangle is an upward equilateral triangle of side length n with
upward unit triangular holes ut out. A diamond is a unit rhombus with

angles of 60 and 120 . Prove that a holey triangle T an be tiled with
diamonds if and only if for ea h k = 1, 2, . . . , n every upward equilateral
triangle of side length k in T ontains at most k holes.
n
C6. Let P be a onvex polyhedron with no parallel edges and no edge parallel to a fa e other than the two fa es it borders. A pair of points on P are
antipodal if there exist two parallel planes ea h ontaining one of the points
and su h that P lies between them. Let A be the number of antipodal pairs
of verti es and let B be the number of antipodal pairs of mid-points of edges.
Express A B in terms of the numbers of verti es, edges, and fa es of P .
Geometry
G1. Let ABCD be a trapezoid with
AB||CD and AB > CD . Points

K and L lie on the line segments
AB and CD , respe tively, su h that
AK : KB = DL : LC . Suppose that
there are points P and Q on the line
segment KL satisfying
D ...p..........................L
C
....p........................................................p
..
...p
.......
....... ...
.... ...... ..................... ..............
.......
.
.
.
.
.
.
.
.
.
.
...
.. .......
.
.........
P ..........................................
...
....
....
....
.........
...
... ......
... ................. ......
.....
.......
...
.......
....
....
..
.......
.....
...
..
.... .....................
.......
.
.
.
.
.
.
.
.
.
.
.
.
.
.....
.. ........
. ..........
...
.
.
.
.
.
.
.
p
.
.
....
..
.....
......................
....
.
.
.
.
..... ...
.............
.
.
.
.
.
Q
..... ...
...........
.
.
.
.
.
.
.. ..
..
..p...............................................................p.....................................................................................p
and CQD =
Prove that the points P , Q, B , and C are on y li .
AP B = BCD
K
ABC .
462
G2. Let ABCDE be a onvex pentagon su h that

=
=
CAD = DAE ;
ACD = ADE .
The diagonals BD and CE interse t at

P . Prove that the line AP bise ts the
side CD.
G3. A point D is hosen on the side
of a triangle ABC with
ACB < BAC < 90
in su h a way that BD = BA. The

in ir le of ABC is tangent to AB and
AC at points K and L, respe tively.
Let J be the in entre of triangle BCD.
Prove that the line KL interse ts the
line segment AJ at its mid-point.
....q
.......... ...
............. .....
.
.
.
............................................................
....... .....
.......
.....
.
.....
...
.
.
...
.
....
.
....q K
......... ........
.
...
.
.
.
.
.
... .....
. ...
.
..
.
.
.
.
.
.
.
.. .......
..
.
...
.
.
.
.
.
.
.. ..
.
.
..
.
.
.
.
.
.
q
.
.
.
.
.. .
.
.
.
... .
........................... ..... .....
...
.... .. ..
.. ...
........
..
.
...
... .. ....
..
.
.
..
.
.
.
.
.
.
.. .....
..
...
.
... ..
..
..... .......J ..q........................................................ ...... ..........
...
.
.
.
.
.
.
.
.
.
....q........ ......
...
......
.
... ......
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..q....................................................q.....................q..............................................................q............................................................................................q
G4. In triangle ABC , let J be the
entre of the ex ir le tangent to side

BC at A1 and to the extensions of
sides AC and AB at B1 and C1 , respe tively. Suppose that the lines
A1 B1 and AB are perpendi ular and
interse t at D. Let E be the foot
of the perpendi ular from C1 to line
DJ . Determine the angles BEA1
and AEB1 .
...
.... ..........................................................
........
...............
.......
............
.....
..........
.
.
....
.
1......
....
....
.....
.
.
.
...
... ....
.
.
.
...
.
.
...... ....
..
.
.
..
..... ...
.
.
..
.
.. ... ...
.
..
.
.
...
..
... ...
.
.
.
.
...
.
.
..
.
.
.
.
.
..
.
.
..
.
....
.
.
.
.
.
.
..
.
.
.
.
...
.
.. ....
.
.
.
.
...
.
.
.
.
.
..
..
.. ..... ....
..
.
.
.
.
.
.
.
.
.
.
.... ..
.
.
.
.
...
.
.
.
.
.
.
.
.
.
.
.... .. ..
.
.
..
....... ...
..
....
....
...... ...
....
....
...
....... ...
..
....
....
...... ...
...
...
.......
.
.
.
.
.
.
.
.
.
.
.
.....
.... ..
...
...
....
............
...
...
.....
....
1 ....... ........................... .................
.....
....
.......
.......
.
.. ..... .................
.
....
.
.
.
.
.
.
.
.
.
.
.
....
.
. . .. .....
......
............................................................................................................................................................................................................................................
B q
qJ
Cq
q q
D B
G5. Cir les 1 and 2 with entres O1
and O2 are externally tangent at point

D and internaly tangent to a ir le
at points E and F , respe tively. Line
t is the ommon tangent of 1 and 2
at D. Let AB be the diameter of
perpendi ular to t, so that A, E , and
O1 are on the same side of t. Prove
that the lines AO1 , BO2 , EF , and t
are on urrent.
..........
.........
AC
q
C1
....
...
.....
E
...
.
............q..................................................................... .....
.
.
.
.
.
.
.
.
.
.
.
.
....
...
.... ...
..... ...
......... .....
........ ....
......
....
.... .... ................
.... ..
.......
..
... ...
... ... ............................
......
..... ........
.
.
....
.
.
.
.
.
.
.......
.
..
.....
.
.
.
.
.
.
.
.
.....
.
........
...
.
.
.
.
.
..
...
.
...
.
.
.
... ...
.
...
.....
..
.. ...
.
.
.
.
...
........
.. ....
.
.
..
1
.
.
.
.
.
.
..
.
..
.
...
.. .... ....................2
..
........
.
.
...
.
..
...
... ...
..
..
....
...
.... ...
2
...
.....
.... ....
.
...
......
.
.
.
....
.
....
.........
.
.
.
...
.
.
.
.
.
.
.
.
...
1
.
................................
...
....
..
.
.............................................................................................................................................................................................................................
..
...
.
..
...
..
..
.
.
..
.....
...
..
...
....
....
....
.
.................. .........
qF
q
O
A q
..............
BAC
ABC
.....p
....................................
.........................
.. ...
.
A ...p.......................................................
.
.
... ..
.....
... ..
.........................
... .
... ...
.... ...... .......................
.
........
...
. ....
...
........
...
..
........
...
...
...
...
........
...
.
..
........
.
.
....
.
...
........
..
..
........ ...
..
..
.
...
...
.........
...
...
... ............... ..
.
.
..
.
.
.
.
P .... .. .. .. ..................p
..
...
...
.. .. .. .. ......
.
.. .. .. .. ....... p....
..
...... D
.. .. .. ... .. .. .. ..
..
.
.
...p..................... .. .. .. .. .. .. .. .......
.........
..
.
.
.............
........
.
............. ....... .....................
............. ... ...........
......p...
q
O
qB
463
G6. In a triangle ABC , let Ma , Mb ,
be the respe tive mid-points of

the sides BC , CA, AB and let Ta ,
Tb , Tc be the mid-points of the ar s
BC , CA, AB of the ir um ir le of
ABC not ontaining A, B , C , respe tively. For ea h i {a,b,c}, let i
be the ir le with diameter Mi Ti . Let
pi be the ommon external tangent to
j , k su h that {i,j ,k} = {a,b,c}
and su h that i lies on one side of
pi while j , k lie on the other side.
Prove that the lines pa , pb , pc form a
triangle similar to ABC and nd the
ratio of similitude.
Mc
..................................................................
.........
........
...........................................
........
b ....................
b
.....
.
.
.
.
.
....
..
.. ... .....
.
.
.
.
.
.
...
....
.. ...
..
.
.
.......
.
.
.
.
.
....
.
.
...
.
..
.
.
.
.
.
.
.
.
.
....
......
.
.
..
....
.. ..
..
...
..
....
.. ..
...
...
. .
....
...
.... .....
.. ..
...
...
..
.... ..
..
.. ..
.
.
.
.
.. ....
.... ...
.
.
.
.....................c
.
......
...
..... ..
.
.
......
.....
......
.
..
.
.....
.
.....
..
...
b
.....
...
...
...
..
......
...
..
.....
....
.
.
.
.
.
.
.
.........
.
.
.
...
.
.
.
.................
....
.
.
.
... .......................
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
c
....
...
.
..
.
.
.
.
....
...
..
.
..
.
..
.
.
....
.
.
..
.
.
.
.
.
.
....
...
....
...
...
..
....
..
..
...
....
..
..
...
..
....
...
..
....
... ....
.
.
....
... ...
.... .....
a
... ..
... ..
.....
........................................................................................................................................................................................
.
....
.
....
.
.
.
.
.
...
....
...
...
...
.....
....
..
...
.....
a ..............
.....
.......
..
........
..
....
.
.
..
.
.
.
.........
.
...
.
.
.
......
...........
...........
....
........................
...................................................
A..................................
....
......r
........ ....
Tc r
rT
M
r
r
M
M
r
r
Ta
G7. Let ABCD be a onvex quadrilateral. A ir le passing through A and

D and a ir le passing through B and
C are externally tangent at the point
P in the interior of the quadrilateral.
Prove that if P AB + P DC 90
and P BA + P CD 90 , then
AB + CD BC + AD .
D
r
....................................
...........
........
........
.....
... ...................................................................................
.....
.
..
...................
.
....
.....
.
...
.
.
... ........
.........
.
..
.. ...
..
... ....
... ..
..
.
...
......
.
... ....
...
.
.....
.
... ....
....
.
...
...
.....
.
... ....
...
....
.
.
......
.
...
.. ...
.
.
...
. ....
.
.
.
.
.
.
.
.
.
.....
...
..... .....
.
...
........................ ...........
....
..
....
.
.
.
.
.
.
.
.
.
.
.
.
.
....
.
.
.
.
.
.
.
.
..................
.....
... .........
..
................
........
.............. ............................
............
rC
P r
r
B
..r
A
G8. Points A1 , B1 , C1 are on the sides BC , CA, AB of a triangle ABC , re-
spe tively. The ir um ir les of triangles AB1 C1 , BC1 A1 , CA1 B1 interse t

the ir um ir le of triangle ABC again at points A2 , B2 , C2 , respe tively
(that is, A2 6= A, B2 6= B , C2 6= C ). Points A3 , B3 , C3 are symmetri
to A1 , B1 , C1 with respe t to the mid-points of the sides BC , CA, AB
respe tively. Prove that the triangles A2 B2 C2 and A3 B3 C3 are similar.
..............
.................... ................................................... A
............................
...........
.......
.
......... ..................
...........................
.
.
................................
.
.
.................
.
.
..... .........................................
.
.
............
.
.
.
.
.
... ..... .......
.
A2.............
...
... ..... .......
...
.....
... ... ......
.
.......
.
.
.
.
.
.
... .... .....
.....
.
.
.
.
.
.
.
.
....
... ...
.... ...
....
...
.
.. ..
.
... ...
...
.
.
.
.
.......................... .......
.
.
.
.
.
.
.
.
.
.
.
.
.. ...
.
.
.
.
.
.
.
.
.
.
.
.
............ B1 .....
.
.......
.
.
.
.
.
.
.
.
.
.
.
.
.
.... ...
.
.
.
...
..............
.
......
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
..
...
..
..
...
... .. ....
.....
...
..
..
.. .. ....
....
...
.....
...
.. .... ...... ....
.
....
.
.
.
.
...
...
.
.
.
.
... .... ....
.
...
.. .....
.
.
....
.
... .... ....
..
..
.
.
.. C1 .......................................
... ... ...
.
..........
..
.
..
..........
......
..
... .......
..
.... ................. ....
..... ....
... .......
..
... ........
...
.... ..
...
...
...........
..
.
A
.
.
.
1
.
.
.
.
..................................................................................................................................................................................
.
.
.
....
.
......
B.......
..
.
.
.
.
.
......
.
.
.
.....
......
...
.
.......
.
.
..... .....
.
.....
.
.. ..
........
....
. .
.......
....
.........
.....
........
.
.
.
.
..... ....
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
................
..
.
... ..
....
................
......
... ...
..........................
.
.. ..
.....
...
.. ...
....
... .....
......
.. ......
...
.
.
.
...
.
.
.
....
.... ......
..
....
.....
...
.....
..... ....
......
..... ...
.....
......... C2
.......
..........
........ .....
.........
......
.................
.................. ......................
.............
.
.
.
.
.
.
.
.
.
.
.
.
................................
..............
..
.....
.......
B2 .......................
.......
........
..........
.........
.
.
.
............
.
.
.
.
.
.
.
.....
........................
........................................
rC
464
Number Theory
N1. Given x (0, 1) let y (0, 1) be the number whose n digit after the
th
de imal point is the (2 ) digit after the de imal point of x. Prove that if x
is a rational number, then y is a rational number.
n
th
N2. For ea h positive integer n let

1
f (n) =
n

n
n
n
+
+ +
1
2
n
where x is the greatest integer not ex eeding x.
(a) Prove that f (n + 1) > f (n) for innitely many n.
(b) Prove that f (n + 1) < f (n) for innitely many n.
N3. Find all solutions in integers of the equation

x7 1
= y5 1 .
x1
N4. Let a and b be relatively prime integers with 1 < b < a . Dene the
weight of an integer c, denoted by w(c), to be the minimum possible value
of |x| + |y| taken over all pairs of integers x and y su h that
ax + by = c .
An integer c is alled a lo al hampion if w(c) max{w(c a), w(c b)}.

Find all lo al hampions and determine their number.
N5. Prove that for every positive integer n, there exists an integer m su h
that 2m + m is divisible by n.
A se ond problem set is the 2005/06 Swedish Mathemati al Contest.

My thanks go to Robert Morewood, Canadian Team Leader at the IMO, for
olle ting them for our use.
SWEDISH MATHEMATICAL CONTEST 2005/2006
Final Round
November 19, 2005 (Time: 5 hours)
1. Find all solutions in integers x and y of the equation

(x + y 2 )(x2 + y) = (x + y)3 .
465
2. A queue in front of a ounter onsists of 12 persons. The ounter is then

losed be ause of a te hni al problem and the 12 people are redire ted to
another one. In how many dierent ways an the new queue be formed if
ea h person maintains the same position as before, or is one step loser to
the front, or is one step farther from the front?
3. In the triangle ABC the angle bise tor from A interse ts the side BC
in the point D and the angle bise tor from C interse ts the side AB in the
point E . The angle at B is greater than 60 . Prove that AE + CD < AC .
4. The polynomial f (x) is of degree four. The zeroes of f are real and form
an arithmeti progression, that is, the zeroes are a, a + d, a + 2d, and a + 3d
where a and d are real numbers. Prove that the three zeroes of f (x) also
form an arithmeti progression.
5. Ea h square on a 2005 2005 hessboard is painted either bla k or white.
This is done in su h a way that ea h 2 2 \sub- hessboard" ontains an odd
number of bla k squares. Show that the number of bla k squares among the
four orner squares is even. In how many dierent ways an the hessboard
be painted so that the above ondition is satised?
6. All the edges of a regular tetrahedron are of length 1. The tetrahedron
is proje ted orthogonally into a plane. Determine the largest possible area
and the least possible area of the image.
Next we look at an alternate solution to problem 2 of the 17 Irish

Mathemati al Olympiad dis ussed at [2007 : 151; 2008 : 88-89.
th
2. Let A and B be distin t points on a ir le T . Let C be a point distin t

from B su h that |AB| = |AC| and su h that BC is tangent to T at B .
Suppose that the bise tor of ABC meets AC at a point D inside T . Show
that ABC > 72 .
C
........
........ ....
........
...
........
.
.
.
.
..
.
.
..
........
..
.....................................................................
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
............. .............
.......
..
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
........
..
.....
..... ......
..
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.......
.....
..
.....
......
...
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.....
.
.....
...
... .............
.
.
.
.
.
.
.
.
.... ....
.....
... ............
...
.
.
.
.
.
.
.
.... ...
.....
. .........
.
.
...
.
.
.
.
.
.
....
.... ...
.. ............
.
..
.
.
.
.
.
.
.
.
....
. ...
. ..
.
..... .........
...
..... .........
.... .....
.. ................
..... .....
... ....................
.
.... ...
.. ...........
.........
.......... ....
......................................................................................................................................................................................................................................................
..............
..
..
..........
.
.
.
.
.............
.
.
.
.
.
.
.
.....
...
.
.
..............
......
...
..............
..............
..
..
..............
.............
.....
.............................
...
..
...
..
...
..
.
.
..
..
......
..
...
Alternate Solution by Luyan ZhongQiao, Columbia International College,

Hamilton, ON.
Let D 6= B be the se ond point of
interse tion of BD with the ir le T .
Let ABD = DBC = . Sin e
AB = AC , we have ACB = 2 .
Also, BC is tangent to the ir le and
AB is a hord, hen e D AB = .
Let ABO = .
466
Sin e BC is tangent to T , we have 2 + = 90 . Adding the angles
in the isos eles triangle ABC yields 4 + CAB = 180 . From these two
equations it follows that CAB = 2 . Sin e D is inside T we have
= D AB > CAB = 2 ,
> . This last inequality together with 2 + = 90 yields

2
5
2 + > 90 , or
> 90 . Hen e, > 36 and ABC = 2 > 72 .
2
2
and therefore
Next we look at a omment on the solution to problem 6 of the 2007

Italian Olympiad [2007: 149{150; 208: 84{85.
6. Let P be a point inside the triangle ABC . Say that the lines AP , BP ,
and CP meet the sides of ABC at A , B , and C , respe tively. Let
x =
AP
P A
y =
Prove that xyz = x + y + z + 2.
BP
P B
z =
CP
P C
Comment by J. Chris Fisher, University of Regina, Regina, SK.

The result is a theorem of Euler featured in the Crux Mathemati orum
arti le \Euler's Triangle Theorem" by G.C. Shephard in [1999 : 148{153.
Euler's proof is very neat, even ni er than any of the solutions found in Crux
Mathemati orum. It an be lo ated in Edward Sandifer's book, How Euler
Did It, and on his webpage: http://www.maa.org/news/howeulerdidit.html
li k on \19 Century Triangle Geometry".
th
Now we turn to solutions from our readers to problems given in the

February 2008 number of the Corner. First a solution to a problem of the
11 Form, Final Round, XXXI Russian Mathemati al Olympiad given in the
Corner at [2008: 20{21.
th
5. (N. Agakhanov) Does there exist a bounded fun tion f

f (1) > 0 su h
that
for all x, y R?
: R R
with
f 2 (x + y) f 2 (x) + 2f (xy) + f 2 (y)
Solved by Li Zhou, Polk Community College, Winter Haven, FL, USA.

Suppose that f is su h a fun tion. Let a0
for n 1. Then
2
f (a1 ) f (a0 ) + 2f (1) + f
=1
1
a0
and an
= an1 +
2f (1) .
1
an1
467
As an indu tion step, assume that f 2 (an ) 2nf (1) for some n 1. Then

1
f 2 (an+1 ) = f 2 an +
an

1
an
2
f (an ) + 2f (1) 2(n + 1)f (1) ,
f 2 (an ) + 2f (1) + f 2
ompleting the indu tion. Hen e f 2 (an ) 2nf (1) for all n
di ting the fa ts that f (1) > 0 and f is bounded.
1,
ontra-
And to omplete our les for the Corner, we look at a problem of the
Taiwan Mathemati al Olympiad, Sele ted Problems 2005, given in [2008:
21{22.
1. A ABC is given with side lengths a, b, and c. A point P lies inside

ABC , and the distan es from P to the three sides are p, q , and r , respe tively. Prove that
a2 + b2 + c2
R
,
18
pqr
where R is the ir umradius of ABC . When does equality hold?
Solved by Arkady Alt, San Jose, CA, USA; Mi hel Bataille, Rouen, Fran e;
and George Tsapakidis, Agrinio, Gree e. We give Bataille's write-up.
Let F denote the area of ABC . We have the well-known relation
abc
2F =
, but also from the denition of p, q, and r we have the equation
2R
2F = ap + bq + cr . Thus, the proposed inequality is equivalent to
abc
2(ap + bq + cr)
or
a2 + b2 + c2
18 3 pqr
(a2 + b2 + c2 )(ap + bq + cr) 9abc
By the AM{GM Inequality,
and
and the inequality (1) now follows from
a2 + b2 + c2 3
a2 b2 c2
pqr .
ap + bq + cr 3
(a2 + b2 + c2 )(ap + bq + cr) 9
(1)
p
3
abcpqr ,
3
3
a2 b2 c2 abc 3 pqr .
That ompletes the Corner for this number, and this Volume. As Joanne
Canape, who has been translating my s ribbles into beautiful LTEX has de ided that twenty-plus years is enough, I want to thank her too for all the
help over the time we've worked together.
A
468
BOOK REVIEW
John Grant M Loughlin

The Symmetries of Things
By John H. Conway, Heidi Burgiel, and Chaim Goodman{Strauss, published
by AK Peters, Wellesley, MA, 2008
ISBN 978-1-56881-220-5, hard over, 423+xxv pages, US$75.00
Reviewed by J. Chris Fisher, University of Regina, Regina, SK
The authors set themselves the ambitious goal of produ ing a book that
appeals to everybody. As far as I an tell from a single reading, they have
su eeded admirably. The rst thing anybody would noti e about the book
is that it is lled with beautiful and fas inating photographs and omputer
drawings. No spe ial knowledge is required for admiring beauty; this book
would be as mu h at home on the living room oee table as on the o e
shelf. Of ourse, it is primarily a mathemati s book.
The ontents have been organized into three parts. The rst of them
des ribes and enumerates the symmetries found in repeating patterns on surfa es; it is written at a level suitable for Crux with Mayhem readers. This part
might well serve as a textbook for a geometry ourse dire ted at university
students spe ializing in mathemati s, edu ation, physi al s ien e, or omputer s ien e. What makes the authors' approa h both novel and elementary
is the introdu tion of what they all the orbifold signature notation. Groups
are not needed here; the on ept an be easily des ribed and qui kly mastered. Here is the idea.
A point in a pattern where two mirrors of symmetry
meet at an angle of m
is alled kaleidos opi and is denoted by m; points
having rotational symmetry of order m (but no kaleidos opi symmetry) are
alled gyrational and are denoted by m (with no asterisk). If a region has
an oppositely oriented image in the pattern that is not explained by mirrors,
then these two regions must be related by a glide re e tion, whi h here is
alled a mira le (short for \mirrorless rossing", they say), denoted by .
To identify the signature of any repeating
plane pattern one writes down the symbols
...
starting from the middle and working out...
...
ward. First lo ate mirror lines and ea h
..
...
kind of kaleidos opi point, if any (where two
3
....................
...........
points are of the same kind if they are related
6 ..................................... ..... ....
.
.
.
.
.
.
..r
.
................................
by a symmetry of the pattern); list them after
..
.. .. ... ... ..
2
the asterisk in de reasing order. Next lo ate
any gyrational points and order them before
any asterisk. Then look for mira les. Typi al
signatures are 632 for a pattern whose symmetry is explained by three kinds
of mirrors
that meet in pairs at angles of 6 , 3 , and 2 ;
632
632 (having no asterisk) for a pattern with
...
...
...
.
.
.
...
...
...
...
...
...
... ....
... ....
... ....
... ...
... ..
......
.
.
.
.
....................................................................................................................................................................................................
. .
..
. ..
... ...
... .....
... ...
..
...
... ....
... .....
...
..
..
...
...
...
.
.
.
.
...
.
...
...
...
...
...
...
...
...
..
...
...
...
..
..
...
.
.
...
.
.
.
.
...
...
.
..
...
.
.
.
.
.
...
.
...
.
.
.
.
.
.
.
... .
... ..
... ...
... ...
.
.
.
.
.
.
.
...........................................................................................................................................................................................
.
.
.
.
.
.
. ..
. ..
... ....
... ....
...
...
..
...
..
...
..
...
...
...
...
...
...
.
..
...
.
.
...
.
...
..
.
.
..
.
.
.
.
.
.
...
...
.
...
...
...
...
...
...
..
..
...
..
...
... .....
... ....
... .....
... ..
... ...
... ..
......
.
.
.
.
..........................................................................................................................................................................................................
.
... ...
... ...
... ...
... .....
... .....
... .....
..
..
..
...
...
...
469
6{fold, 3{fold, and 2{fold gyrational points but no re e tions or
222 for a pattern with two kinds of kaleidos opi points where
mira les;
a pair of
mirrors interse t at right angles, and one point where there is a half-turn
symmetry but no mirror.
.......................................................................................................................................................................................................................................
...
...
...
...
....
....
....
....
...
...
...
...
....
....
....
....
..
....
....
....
..........................................................................................................................................................................................................................................
.....
.....
.....
.....
.....
....
....
....
....
....
...
...
...
...
...
..
.
....
....
....
.
.
.
.
.
.
.
.
.
.
.
.............................................................................................................................................................................................................................................
..
..
.
.
.
.
...
...
....
....
.
.
.
....
.....
....................................................................
...
...
..
...
...
..
..
......................................................................................................................................................................................................................................................
..
..
.
....
....
.
....
....
.... ......................................................... .....
....
....
....
....
....
....
.
....
....
.
....
.
.
..
.
.
.
.
.
.
.
.
.......................................................................................................................................................................................................................................
...
...
.
.
.
.
....
....
....
....
.
.
.
...
...
....
....
....
....
....
....
....
..
..
..
..
..
.............................................................................................................................................................................................................................................
...
.....
.....
.....
.....
.
..
.
....
....
.
....
...
...
...
....
...
....
....
....
....
....
.
.
.
.
.
.
.
.
..............................................................................................................................................................................................................................
2
2
632
222
Unlike most other notation systems that have been devised for des ribing plane symmetry, these orbifold signatures an also be used to des ribe
frieze patterns and spheri al patterns. (We learn in Part III that they work
equally well for des ribing hyperboli patterns.) But how does one know that
the resulting lists of 17 signatures for plane patterns, 7 signatures for frieze
patterns, and 14 signatures for spheri al patterns are orre t and omplete?
There is a \Magi Theorem" that assigns a ost to every symbol in the signature in su h a way that plane patterns and frieze patterns ost exa tly $2
while spheri al patterns ost a bit less. That theorem tells us immediately
whi h signatures are feasible. To establish the Magi Theorem, a pattern
on the surfa e is asso iated with a folded surfa e they all an orbifold. The
orbifold is obtained by identifying points related by a symmetry of the pattern (whereby points of an orbit are folded atop of one another so that a
single representative point of every orbit lives on the orbifold). This sounds
a bit s ary, but the authors manage to explain the details in a gentle way
using suitable pi tures and simple examples. They then state the Classi ation Theorem for Surfa es and provide Conway's elementary and intuitive Zip
proof. They also prove that these surfa es an be distinguished using Euler's
formula (involving the numbers of verti es, edges, and fa es of a suitable map
on the surfa e), whi h they also prove. Sin e the orbifolds are easily lassied
using Euler's formula, the orresponding patterns are thereby lassied.
Remarkably, all the proofs should satisfy the professional mathemati ian even though they are dire ted at an elementary audien e. The authors
a hieve this feat by repeatedly redu ing te hni al di ulties down to problems that are postponed to the following hapter. This way they present
one on ept at a time, as ompared to the typi al textbook's initial barrage
of poorly motivated denitions and lemmas. Their proofs are every bit as
brilliant as their notation. The illustrations are not just beautiful, but they
470
have been arefully hosen to larify the exposition. I really appre iated
the authors' de ision to repeat pi tures that they require for illustrating new
ideas | instead of making the reader turn ba k to a pi ture on an earlier
page, they reprodu ed a smaller version of it whenever needed. The authors
learly have fun oining whimsi al new words; their terminology will not appeal to everybody, but the informal nature of their dis ussions makes for
enjoyable reading. I rather liked the word mira le in pla e of the standard,
but awkward and misleading term glide re e tion; however I saw little need
for gyrational in pla e of rotational or wandering in pla e of translation. We
will have to wait to see whi h words at h on.
What I have des ribed so far is the ontent of the 116 pages of the
rst nine hapters. Originally, a ording to the prefa e, this was all that
the authors had intended to write. But they de ided it was worthwhile to
extend the signature to olour symmetry, and the book grew from there.
For a areful reading of Part II the reader needs some group theory and
a bit of mathemati al maturity. The authors' main goal for this part is to
present their analysis and notation for olour symmetry. They enumerate the
p{fold olour types for plane, spheri al, and frieze patterns (for all primes
p). The omplete lassi ations appear in a book for the rst time. Along
the way the authors show how their orbifold notation orresponds to previous lassi ation systems, whi h gives them the opportunity to dis uss the
short omings of those systems. Also in this part, they enumerate the isohedral tilings of the sphere and plane, and they extend to n = 2009 the
Bes he-Ei k-O'Brien table of the number of abstra t groups of ea h order n.
The informative lists of Part II an probably be understood by readers
who might not take an interest in the a ompanying te hni al arguments.
Similar omments apply to Part III, whi h the authors expe t to be ompletely
understood only by a few professional mathemati ians. Still, as they point
out, mu h of Part III an protably be explored by other readers, while many
more will enjoy inspe ting the pretty pi tures. Here, among other things, the
authors dis uss hyperboli groups and Ar himedean polyhedra and tilings;
they list the 219 rystallographi spa e groups (and explain why hemists
distinguish 230 groups), and they provide a omplete list for the rst time in
print of the 4{dimensional Ar himedean polytopes. Apparently they ould
have kept writing, but they de ided to leave something for the rest of us to
do. Their nal words are, \A universe awaits | Go forth!"
I thank Bru e Shawyer for inviting me to serve as Book Review Editor. I am
grateful to the late Jim Totten for guiding me during his tenure; Bru e Crofoot
for insightful ommentary; Va lav

Linek for re ent support; Shawn Godin for
leadership with Mayhem. Thanks go to all the reviewers, but espe ially this
trio: Chris Fisher, a dependable sour e of thought provoking reviews usually
on erning geometry; Ed Barbeau, an e le ti mathemati ian who is eager to
help; and my su essor, Amar Sodhi. Amar's passion for mathemati s will
shine as he assumes this role. Wel ome Amar! Thanks to the CRUX with
MAYHEM ommunity for an enjoyable journey. | John Grant M Loughlin
471
Old Idaho Usual Here
Robert Israel, Stephen Morris, and Stan Wagon
N. Kildonan [1 raised the following problem. Take an arbitrary word

using at most 10 distin t letters, su h as DON ALD COXET ER. Can one
subsititute distin t digits for the 10 letters so as to make the resulting base
10 number divisible by d? The answer depends on d. If d has 100 digits then
the answer is learly NO. If d = 100 then the answer is again NO, sin e ER
annot be 00. If d is 2, the answer is learly YES: just let the units digit be
even; d = 5 or d = 10 are just as easy.
Kildonan proved that divisibility by d = 3 an always be a hieved and
R. Israel and R. I. Hess extended this to d = 9; the ase of d = 7 was
left unresolved. In this note we settle all ases. The reader interested in an
immediate hallenge should try to prove that divisibility by d = 45 is always
possible. This appears to be the hardest ase.
To phrase things pre isely, a word is a string made from 10 or fewer
distin t letters; for ea h word and ea h possible substitution of distin t digits
for the letters, there is an asso iated value: the base 10 number one gets
after making the substitution. If all substitutions yield a value for the word
w that is not divisible by d, then w is alled a blo ker for d. If any word
ending (on the right) with w fails to be divisible by d, then w is alled a
strong blo ker for d. An integer d is alled attainable if the value of every
su iently long word an be made divisible by d by some substitution of
distin t digits for letters. Thus d is not attainable if there exist arbitrarily
long blo kers. The use of arbitrarily long strings is important be ause, for
example, AB is a blo ker for 101, but only be ause it is too short. An integer
d is strongly attainable if the value of every word an be made divisible by d
by an appropriate substitution.
In this paper we will nd all attainable integers; moreover, they are
all strongly attainable. Note that any divisor of an attainable number is
attainable.
Some ases, su h as d = 2, d = 5, or d = 10 are extremely easy to
attain, and it is just about as easy to attain d = 4 or d = 8. It takes a little
work to show that d = 3 and d = 9 are attainable (proofs given below). Our
main theorem resolves the attainability status of all integers.
Theorem 1 An integer is attainable if and only if it divides one of the integers
18, 24, 45, 50, 60,
or 80.
To start, we dis uss the ases of d = 3 and d = 9 for ompleteness and

to introdu e the ideas needed later. We use the well known fa t that when
d is 3 or 9, then d divides a number if and only if d divides the sum of its
digits.
Copyright
c 2008

Canadian Mathemati al So iety
Crux Mathemati orum with Mathemati al Mayhem, Volume 34, Issue 8
472
The number d = 3 is attainable (Kildonan [1). Given a word, let
the 10 letters be grouped as Ai , Bi , and Ci , where ea h Ai has a multipli ity
(perhaps 0) that is divisible by 3, ea h Bi has a multipli ity of the form 3k+1,
and ea h Ci has a multipli ity of the form 3k + 2. Look for one, two, or three
pairs among the Bi and repla e them with digits 1 and 2, and 4 and 5 for the
se ond pair, and 7 and 8 for the third pair. Then look for pairs of the Ci
and repla e them with digits in any of the still-available pairs among (1, 2),
(4, 5), and (7, 8). These substitutions take are of Bi Ci ex ept possibly
four letters (sin e we used three pairs) and we an substitute 0, 3, 6, and 9
for them. The letters Ai an be assigned the remaining digits in any order.
Thus, the nal number has a digit sum divisible by d = 3.
The number d = 9 is attainable (Solution II by Israel and Hess [1).
Suppose a word has length n. Suppose some letter o urs k times, where
n k is not divisible by 3. Assign 9 to this letter and assign 0 to 8 arbitrarily
to the other letters. Let the value of the resulting number be v (mod 9).
Now repla e ea h digit from 0 to 8 by the next higher digit, wrapping ba k
to 0 in the ase of 8. This adds n k to the value modulo 9. But n k is
relatively prime to 9, so we an do this v/(n k) times, where the division
uses the inverse of n k modulo 9, in order to obtain the value 0 modulo 9.
The other ase is that every letter has a multipli ity k n (mod 3).
If in fa t every multipli ity is ongruent to n (mod 9), then any assignment
will yield a value ongruent to n(0 + 1 + + 9) = 45n 0 (mod 9).
Otherwise there is a multipli ity k n (mod 3) but k 6 n (mod 9), and
then we pro eed as in the rst half of the proof: assign 9 to this letter, 0 to
8 to the other letters, and then y li ally permute the values 0 to 8. Ea h
permutation adds n k modulo 9 and this will eventually transform the
value v, whi h is divisible by 3, to a value divisible by 9, be ause 3 divides
n k but 9 does not.
Now to the proof of Theorem 1, whi h follows from these four lemmas.
Lemma 1 Any integer divisible by a prime greater than 5 is not attainable.
Lemma 2 The largest attainable powers of 2, 3, and 5 are 16, 9, and
respe tively.
Lemma 3 The numbers 36, 48, 75, 90, 100, and 120 are not attainable.
Lemma 4 The numbers 18, 24, 45, 50, 60, and 80 are attainable.
25,
The ordering of these lemmas indi ates how Theorem 1 was found. First
the ases of d = 7 and d = 11 were settled and that led to the general result
of Lemma 1. It followed that the only andidates for attainability had the
form 2a 3b 5c . On e the powers of 2, 3, and 5 were resolved (Lemma 2), the
andidate list was redu ed to the 45 divisors of 3600 = 16 9 25. Resolving
the situation for those divisors, with some omputer help, led to Lemmas 3
and 4. Finally, the omputer sear hes were eliminated and the whole thing
was redone by hand. Theorem 1 follows from the lemmas be ause Lemmas
3 and 4 settle the status of all 45 divisors of 3600.
473
A key idea is that the ten digits sum
blo ker
to 45. So we begin with Lemma 3, d
AAB
whi h shows how unattainability is 27
ABBAB
proved. We use the fa t that an in- 32
(ABCDEF GHJ )5 J 4 K
teger is ongruent modulo 9 (hen e 36
modulo 3) to the sum of its digits. 48 (ABCDEF GH)2 KKJ K
AABA
Let A, B , C , D, E , F , G, H , J , 75
90
A6 (BCDEF GHJ )7 J K
K be the ten letters and let w g be
AB
the on atenation of g opies of word 100
120
ABCDEF GHJ J J K
w . The table at right lists the blo kBBA
ers needed for Lemmas 2 and 3; most 125
were found by a omputer sear h.
We show that the words in the table are blo kers. The easiest ase is
d = 100, sin e the value of any word ending in AB is not divisible by 100.
Case 1. The number d = 36 an be blo ked. We have
(ABCDEF GHJ )5 J 4 K K + 4J + 5(45 K) 4J 4K (mod 9)
The only way 4(J K) is divisible by 9 is if J K is either 90 or 09, and neither

is divisible by 4. Extension on the left by A9i preserves the value modulo 36,
be ause 111111111 is divisible by 9.
Case 2. The number d = 48 an be blo ked. The rightmost 4 digits of the
word (ABCDEF GH)2 KKJ K must be one of 0080, 2272, 4464, 6656, or
8848, as these are the only words of the form KKJ K that are divisible by
16. However, now the value of the word modulo 3 is one of the entries
below, where we work with ve tors and ignore K whi h o urs three times:

2 (45, 45, 45, 45, 45) (8, 7, 6, 5, 4) (0, 2, 4, 6, 8) + (8, 7, 6, 5, 4)
= (82, 79, 76, 73, 70) ,
and no entry is divisible by 3. Left extension by A3i preserves the value

modulo 48.
Case 3. The number d = 75 an be blo ked. Here we have the ongruen e
AABA 51A + 10B (mod 75). Multiplying by 53 transforms the ongruen e to 3A + 5B 0 (mod 75). However, 3 3A + 5B 69, so
the ongruen e is never satised. Left extension by A3i preserves the value
modulo 75.
A6 (BCDEF GHJ )7 J K 6A + 7(45 A) + J (mod 9)
be ause K must be 0. The expression simplies to J A modulo 9, whi h

annot be divisible by 9 be ause 0 is already assigned to K . Left extension
by A9i preserves the value modulo 9.
474
ABCDEF GHJ J J K 45 + 2J 2J (mod 3)
However, J J K must be either 440 or 880 to obtain divisibility by 40, therefore, 2J is either 8 or 16, and so is not divisible by 3. Left extension by A3i
preserves the value modulo 120.
Case 6. The number d = 32 an be blo ked. Any word ending in ABBAB
has a value satisfying 10010A + 1101B 26A + 13B (mod 32). If this is
ongruent to 0 modulo 32, then we may an el 13, leaving 2A+B . However,
this sum is between 1 and 18 + 8 = 26, so it is not divisible by 32.
Case 7. The number d = 125 an be blo ked. A number is divisible by 125 if
and only if it ends in 125, 250, 375, 500, 625, 750, 875, or 000. Thus, BBA
is a strong blo ker for 125.
Case 8. The number d = 27 an be blo ked. The value of AAB satises the
ongruen e 110A + B 2A + B (mod 27). However, 1 2A + B 26,
whi h is not divisible by 27. This shows nonattainability, be ause we an add
the prex A27i , whi h leaves the value modulo 27 un hanged.
Next we prove Lemma 1. Our rst proof of this was a little ompli ated
(see the Proposition that follows), but when we fo used on words involving two letters only we dis overed Theorem 2, whi h yields Lemma 1 in all
ases ex ept d = 7. Re all Euler's theorem, that a(d) 1 (mod d) when
gcd(a, d) = 1. It follows that if d is oprime to 10, then there is a smallest
positive integer, denoted by ordd (10), su h that 10ord (10) 1 (mod d).
d
Theorem 2 Let d be oprime to 10 and greater than 10 with e = ordd (10).
Then w = Ake1 B is a blo ker for d for any positive integer k.
Proof: Assume rst that k = 1 so that w is just Ae1 B . If 3 does not divide
d then the value of w satises the ongruen e
B+A
e1
X
i=1
Sin e
10i = B A + A
10e 1
B A (mod d)
9
d > 10, d annot divide B A. Now suppose that 3 divides d and

d > 81. Suppose the value of w , in the formula just given, is a multiple of d.
Then multiplying by 9 yields 9(B A) + A (10e 1) = 9Kd, and hen e d
divides 9(B A). However, A 6= B and 81 9(B A) 81, so d > 81
annot divide 9(B A), a ontradi tion.
There remain the ases where 3 divides d and 11 d 81, namely
d {21, 27, 33, 39, 51, 57, 63, 69, 81}. Suppose that d is one of these
but d 6= 21, 27, 81; then ordd (10) = ord3d (10). This means that from
B A + A (10e 1) /9 = Kd, we have 9(B A) + A (10e 1) = 3K(3d),
when e 3d divides 9(B A). Thus, d divides 3(B A), whi h means that
d 27, a ontradi tion.
475
For d = 21 the value of w modulo 21 is B A, whi h is not divisible by
For d = 27 the value of w modulo 27 is 2A + B and 1 2A + B 26,
so the value is not divisible by d. For d = 81 the value of w modulo 81 is
8A + B and 1 8A + B 80, so the value is not divisible by d.
The extension to the ase of general k is straightforward.
21.
The pre eding result blo ks all primes greater than 10. We need to deal
also with d = 7. One an give an alternate onstru tion in the general ase
that in ludes d = 7, and we give the following without proof.
Proposition Suppose that d is oprime to 10 and d does not divide 9. Let
w = KJ K e HK e GK e F K e EK e DK e CK e BK e AK e ,
where e = ordd (10)

ongruent to 9(d 2+ 1)
1.
Then after any substitution the value of

(mod d), and so is not divisible by d.
is
For d = 7 the word w of the Proposition has length 55. A dierent

approa h led to the mu h shorter example OLD IDAHO USUAL HERE ,
its value is always 3 45 (mod 7). This 17- hara ter word is thus a blo ker
for d = 7, and it an be made arbitrarily long by prepending E 6 .
On to Lemma 2. The positive results for d = 16 and d = 25 are not
di ult, but they are omitted as they follow from the ases of d = 80 and
d = 50 (proved below); the ase of d = 9 was dis ussed earlier, as were the
negative results for d = 32, 27, and 125. It remains only to prove Lemma 4.
Case 1. The number d = 50 is strongly attainable. Just use 00 or 50 for the
rightmost two digits.
Case 2. The number d = 80 is strongly attainable. Assign 0 to the rightmost
letter; 8 to the next new letter that o urs reading from the right, 4 to the
next one, and 2 to the next one after that. The value is then divisible by 16
and also by 5; divisibility by 80 is only ae ted by the four rightmost digits.
Case 3. The number d = 18 is strongly attainable. Given a word, nd an
assignment that makes it divisible by 9. If the rightmost digit is even, we are
done. Otherwise, repla e this digit y with 9 y. This preserves divisibility
by 9 and makes the rightmost digit even.
Case 4. The number d = 60 is strongly attainable. The word ends in either
AA or BA. In either ase, assign 0 to A and 6 to B . Let the eight remaining
letters be grouped as Ai , Bi , and Ci as in the proof for d = 3. Use the pairs
(1, 2), (4, 5), and (7, 8) on whatever pairs of letters an be found within Bi
or within Ci . This leaves at most two single letters in the B and C groups.
Use 3 and 9 for the two singletons, and any remaining digits for the A group.
The nal value is then divisible by 3, 4, and 5.
Case 5. The number d = 24 is strongly attainable. The idea is to modify the
proof for d = 3 so as to guarantee divisibility by 8. Re alling the proof that
d = 3 is strongly attainable, all two letters mat hed if they are repla ed in
476
that proof by 1 and 2, or by 4 and 5, or by 7 and 8. If the word ends in AAA
just make sure A is either 0 or 8. The remaining ases are that the word ends
in one of the patterns ABC , ABB , BBA, or ABA.
If the ending is ABC with A and C mat hed, then use 152 or 192,
depending on whether B is part of a mat hed pair or not. If A and C are
unmat hed use 320 or 360 a ording as B is part of a mat hed pair or not.
If the ending is ABB with A and B mat hed, then use 488, sin e the
mat hing an use 4 and 8 as well as 1 and 2. If both A and B are unmat hed,
then use 600. If A is mat hed and B is not, use 800. If B is mat hed and A
is not, use 088.
If the ending is BBA, then pro eed as if the ending was ABB , but use
instead 448, 336, 008, and 880 for the four sub ases.
If the ending is ABA, then pro eed similarly, using 848, 696, 808, and
080 for the four sub ases.
Case 6. The number d = 45 is strongly attainable. Let m(X) denote the
multipli ity of the letter X in the given word redu ed modulo 9; let X denote
the digit assigned to X . Let A be the rightmost letter and assign 0 to it, thus
ensuring divisibility by 5.
Assume rst that the multipli ities of at least eight of the nine remaining digits are all mutually ongruent modulo 3, and assign 9 to the other
letter. Let the m-values of the eight letters be 3ai + c, where ea h ai is a
non-negative integerPand c {0, 1, 2}. Let Li be the digits assigned to these
eight letters.
Li = 36P
, whi h 9 divides, the value modulo 9 of the
P Sin e
word is 3 ai Li . So we want ai Li to be divisible by 3. Assign the pairs
(1, 2), (4, 5), and (7, 8) to pairs of letters with equal ai . Assign 3 and 6 to
the remaining two. The total is then divisible by 9 and therefore by 45.
In the other ase we an hoose a letter, K say, with m(K) not ongruent modulo 3 to the length of the word; therefore S , the sum of the mul = 9.
tipli ities of the nine letters other than K , is not divisible by 3. Let K
= 1 and
Case 6a. There is a letter, say B , with m(B) = m(A). Then let B
assign the remaining digits arbitrarily.
Case 6b. There is no letter as in Case 6a. Then we an nd two letters among
B , C , D , E , F , G, H , J , say C and D , with m(C) 6 m(D) (mod 3).
Consider B ; we know m(B) 6= m(A). Set B to be the non-negative residue
modulo 9 of m(D) m(C) and note that B A m(D) m(C) (mod 9)
whi h is not divisible by 3. Assign unused digits to the letters C and D so that
D
m(B) m(A) (mod 9); there are enough digits left for this to be
C
possible. Assign the remaining digits arbitrarily.
Now we an treat both ases to get the result. The assignment produ es
some total value, redu ed modulo 9 to v. If v 6= 0 then repla e ea h digit
between 0 and 8 by the next higher digit, wrapping ba k to 0 in the ase of 8.
modulo
This adds S to the value modulo 9 and does not alter B A or C D
9. But S is relatively prime to 9, so we an do this v/S times, where the
division uses the inverse modulo 9 of S , in order to a hieve divisibility by 9.
If A is 0 or 5, then we are done.
477
If A is 3 or 6, then swit h digits of A and B , where we know that B is
not divisible by 3; this is be ause the value modulo 3 of B A, whi h starts
out nonzero, does not hange in the translational step. If we are in Case 6b,
; the net hange is
then also swit h C and D

A
m(A) m(B) + C
D
m(D) m(C)
B

A
D
C
+ C
D
B
A
0 (mod 9) ,
B
so divisibility by 9 is preserved.
As A is now not divisible by 3, we an multiply ea h digit less than 9
by 5/A (mod 9). This preserves divisibility by 9 and makes A = 5. The
total is now divisible by 45. The proof that d = 45 is strongly attainable
is omplete, as is the proof Lemma 4, and thus the proof of Theorem 1 is
omplete.
There are several variations to this problem that one might onsider,
su h as using bases other than 10. Another variant is to restri t the alphabet
to the two letters A and B . We use the terms 2-attainable and 2-blo ker in
this ontext. Using te hniques similar to those presented, we obtained the
following result.
Theorem 3 A number is 2-attainable if and only if it divides one of 24, 50,
60, 70, 80,
or 90.
The negative part of the proof required nding a 2-blo ker for ea h
reader might enjoy nding them; they are
d {28, 36, 48, 120, 175}. The

all short, of length at most 7.
A knowledgment
We thank Bill Sands for bringing this problem to our attention.

Referen es
[1 N. Kildonan, Problem 1859 (Solution), Crux Mathemati orum, 20:6,

June 1994, pp. 168{170.
Robert Israel
University of British Columbia
Van ouver, BC, Canada
israel@math.ubc.ca
Stan Wagon
Ma alester College
St. Paul, MN, USA
wagon@macalester.edu
Stephen Morris
Newbury, Berkshire, England
stephenmor@gmail.com
478
A Limit of an Improper Integral Depending
on One Parameter
Iesus C. Diniz
In this arti le we will al ulate the following limit of an improper integral that depends on one parameter R+
lim
0+

exp K(x + l)n Knxn1 dx ,
where K and l are positive real numbers and n is a positive integer.

This is an interesting example where one annot inter hange the order
of the limit and the integral. Through a ni e appli ation of elementary tools
su h as hange of variables in integrals and the Binomial Theorem, one is
able to obtain this limit.
This problem arose in the al ulation of a lower bound for a probability of theoreti al interest in the study of multidimensional Poisson point
pro esses, namely
lim
0+

exp vn (1) (r + l)n r n vn (1)nr n1 exp vn (1)r n dr ,
where vn (1) is the volume of the n{dimensional unit ball, is the Poisson
intensity, l is the distan e between two distinguished points in Rn , and r
is the distan e from the rst point to the losest o urren e in the Poisson
point pro ess. We shall show that this limit is equal to 1.
Proposition For all positive real numbers K and l, and for ea h positive in-
teger n we have
lim
0+

exp K (x + l)n Knxn1 dx = 1 .
Proof: The ase n = 1 is a straightforward al ulation:

lim
0+
=
=

exp K(x + l) K dx

lim exp K(x + l)
0+
lim exp (Kl) = 1 .
0+
Hen eforth, we take n > 1.

Copyright
c 2008

479
Making rst the hange of variable
v = Kxn ,
dv = Knxn1 dx ,
in the integral we obtain

Z
h
1
n i
1
exp v n + (ln K) n
dv .
Changing variables again, this time a ording to

1
u n = v n + (ln K) n
dv =
we obtain
lim
0+
lim
0+
lim
0+
du ,
h
1
n i
1
exp v n + (ln K) n
dv
exp (u)
ln K
( n1 1)
u
( n1 1)
u
v
du

n n1 n1
l K
exp (u) 1
du ,
ln K
whi h by the Binomial Theorem be omes
n1
X

n k
n1
k l K n
(1)
du
u
0+ ln K
k
k=0
n nk
n1
X n 1 Z
l K
1 + lim
exp (u) (1)k
du .
0+
u
k
ln K
k=1
lim
exp (u)
After yet another hange of variable,

1
w = un
dw =
1 ( n1 1)
u
du ,
n
the pre eding expression be omes

1 + lim
0+
n1
X
k=1

k Z

n1
1
n
n lK
exp w n w n1k dw .
1
k
lK n
It remains to show that
lim
0+
exp (w n ) w n1k dw
lK n
is zero for ea h k {1 , 2, . . . ,
n 1}.
(1)
480
To do this, we make a nal hange of variables
z = n w ,
dz = n dw ,
whi h yields
Z
k
n
exp (w n ) w n1k dw
1
lK n
1
(ln K) n
< n

exp z n z n1k dz

exp z n z n1k dz .
(2)
It now su es for us to show that the last integral in (2) is nite for
ea h k {1, 2, . . . , n 1}.
Setting
Z
1
C =
exp (z n ) z n1k dz
and noting that if z 1, then z n1k

Z

exp z n z n1k dz
z n1 , we nally obtain
Z

C +
exp z n z n1 dz
1
= C +
1
ne
whi h is a nite number. This ompletes the proof.
Acknowledgments
This work was supported by FAPESP grant 0415864-1.
The author also thanks J.C.S. de Miranda for many dis ussions about
this problem.
Iesus C Diniz
Institute of Mathemati s and Statisti s
University of S~ao Paulo
S~ao Paulo, Brazil
iesus@usp.br
481
Sliding Down In lines with Fixed Des ent
Time: a Converse to Galileo's Law of Chords
Je Babb
Suppose that a ve tor is an hored at the origin and lies along the positive x-axis. Consider rotating the ve tor ounter lo kwise about the origin
through an angle A, with 0 < A < . Consider a parti le of mass m whi h
is initially at rest and then slides, under gravity and without fri tion, from a
starting point on the in lined ve tor down towards the origin. Let DA denote
the distan e of the starting point from the origin. For ea h value of A, suppose that DA is hosen to ensure that the parti le requires exa tly T se onds
to rea h the origin. Determine the urve hara terized by the starting points
of the parti les.
For the parti le under onsideration, let dA (t) be the distan e travelled
along the ve tor at time t, vA (t) = dA (t) be the velo ity along the ve tor at
time t, and let aA (t) = vA

(t) be the a eleration along the ve tor at time t.
Note that by denition, DA = dA (T ).
If aA (t) is some onstant K , then
dA (t) =
K 2
t .
2
(1)
This may be onrmed by integrating aA (t) twi e with respe t to time and
applying the initial onditions dA (0) = 0 and vA (0) = 0 to obtain zero for
both onstants of integration.
Sin e the parti le is sliding down a fri tionless in line in the Earth's
gravitational eld, the omponent of a eleration along the in line is
K = g sin A, where g is the a eleration due to gravity at the Earth's surfa e.
If the des ent time is xed at T se onds, then
DA = dA (T ) =
g
2
T 2 sin A .
(2)
A point (r, ) in polar oordinates may be expressed in Cartesian oordinates

as (x, y), where x = r cos and y = r sin . Consider the following equation, whi h is expressed in polar oordinates as
r = 2c sin ,
where r
yields
Copyright
> 0
and 0
< < .
Multiplying both sides of equation (3) by r
r 2 = 2cr sin
c 2008

(3)
(4)
482
By setting x = r cos and y
Cartesian oordinates as
= r sin ,
equation (4) may be re-expressed in
x2 + y 2 = 2cy .
Subtra ting 2cy from ea h side and ompleting the square on y2 2cy yields
x2 + (y c)2 = c2
(5)
whi h is the equation of a ir le of radius c entred at the point (0, c).
Thus, as depi ted in the gure at right,
the lo us of points dened by equation (2) is
a ir le
resting
upon the origin (0, 0), entred

g 2
at 0, 4 T on the verti al axis, and of radius
g 2
T .
4
On an histori al note, the motivation for

r 0, g T 2
this problem arose while onsidering the Law
4
of Chords, whi h was stated and proven by
Galileo Galilei in his 1638 masterpie e Dialogues Con erning Two New S ien es. Galileo
onsidered rates of des ent along a verti al irA
le and, with his Proposition VI, established
the Law of Chords (see [1, p. 212):
If from the highest or lowest point in a verti al ir le there be
drawn any in lined planes meeting the ir umferen e, the times
of des ent along these hords are ea h equal to the other.
Galileo's proof of the Law of Chords is presented via a series of geometri propositions, whi h require familiarity with many of Eu lid's theorems.
The question the author wished to address was whether the verti al
ir le is the only urve with the property that des ent time to the lowest
point on the urve is onstant for all hords. This paper demonstrates that
the verti al ir le, or one of its omponent ar s interse ting at the lowest
point of the ir le, is indeed the only su h urve.
....
...
....
.
.
........................................................
.
.
.
.
.
.
.
.
.
.
.........
......
.
.
.
.
.......
.
.
.
....
.......
.....
.
.
.
.
...
.....
.....
....
....
....
....
.
.
.
.
.
....
..
...
.
.
.
.
...
...
...
...
.
.
...
..
...
.
.
..
.....
....
..
..
...
....
..
...
...
...
...
....
...
..
....
....
..
...
....
..
.
...
.
.
...
.......
.
.
.
.
.
.. ..
..
.
.
.
.
.
.
.
.
..
.
... ...
..
.
.
.
.
.
.
.
..
....
...
..
.....
....
...
...
....
...
..
....
...
....
....
..
...
.
.
.
.
.
.
.
.
.
....
.
.
....
....
....
.....
...
......
.....
....
.....
.... ...
.....
..... .... ...........
...
.......
........
.. .......
........
.
..........
.....
... ......
.....
........... ...
..............................................................................................................................................................................................
Referen es
[1 Galilei, Galileo. 1638/1952 Dialogues Con erning Two New S ien es,
translated by H. Crew and A. de Salvio, En y lopaedia Britanni a,
Chi ago, 1952.
Je Babb
Department of Mathemati s and Statisti s
University of Winnipeg
Winnipeg, MB, R3B 2E9
Canada
j.babb@uwinnipeg.ca
483
PROBLEMS
Toutes solutions aux problemes

dans e numero
doivent nous parvenir au plus
tard le 1er juin 2009. Une etoile

() apres
le numero
indique que le probleme

a et
e
soumis sans solution.
Chaque probleme

sera publie dans les deux langues o ielles du Canada
(anglais et fran ais). Dans les numeros

1, 3, 5 et 7, l'anglais pre
edera

le fran ais,
et dans les numeros

2, 4, 6 et 8, le fran ais pre
edera

l'anglais. Dans la se tion des
solutions, le probleme
sera publie dans la langue de la prin ipale solution present

ee.

La reda tion

souhaite remer ier Rolland Gaudet, de College
universitaire de
Saint-Bonifa e, Winnipeg, MB et Jean-Mar Terrier, de l'Universite de Montreal,

d'avoir traduit les problemes.

3371. Corre tion. Propose par George Tsintsifas, Thessalonique, Gre e.

^ es
respe tifs a, b et c, et soit M un de
Soit ABC un triangle de ot
ses points interieur.

Les droites AM , BM et CM oupent respe tivement
les ot
^ es
opposes
aux points A1 , B1 et C1 . Les droites passant par M et
perpendi ulaires aux ot
^ es
oupent respe tivement BC , CA et AB en A2 ,
B2 , et C2 . Soit p1 , p2 et p3 les distan es respe tives de M aux ot
^ es
BC ,
CA et AB . Montrer que
[A2 B2 C2 ]
[A1 B1 C1 ]
(ap1 + bp2 )(bp2 + cp3 )(cp3 + ap1 )

2a2 b2 c2
ou [KLM ] designe

l'aire du triangle KLM .
a
p1
b
p2
c
p3
3389. Propose par Mihaly Ben ze, Brasov, Roumanie.
Pour a R, la suite (xn ) est denie

par x0 = a et xn+1 = 4xn x2n
pour tout n 0. Montrer qu'il existe un nombre inni de valeurs a R
telles que la suite (xn ) est periodique.

3390. Propose par Mihaly Ben ze, Brasov, Roumanie.

Montrer que si A, B , C et D sont les solutions de
X
3 5
5
8
alors A2007 + B 2007 + C 2007 + D2007 = O, ou O est la matri e nulle de taille
2 2.
3391. Propose par Mi hel Bataille, Rouen, Fran e.
onvexe tel que AC et BD se oupent a

Soit ABCD un quadrilatere
angle droit en P , et soit respe tivement I , J , K et L les milieux de AB , BC ,
CD et DA. Montrer que les er les (P IJ ), (P J K), (P KL) et (P LI) sont
ongruents si et seulement si ABCD est y lique.
484
3392. Propose par Mi hel Bataille, Rouen, Fran e.

Soit A, B , C , D et E on y liques ave respe tivement V et W sur les
a CB par V
droites AB et AD. Montrer que si la droite CE , la parallele
et la parallele
a CD par W sont on ourantes, alors les triangles EV B et
EW D sont semblables. La re iproque

est-elle vraie ?
3393. Propose par Dragoljub Milosevi et G. Milanova , Serbie.

Soit le triangle ABC , ou
demi perim
etre.

Montrer que
a = BC , b = AC , c = AB
et ou s est le
y+z
A
z+x
B
x+y
C
9
2
x
a(s a)
y
b(s b)
z
c(c a)
s
ou les angles A, B et C sont mesures

en radians et x, y et z sont des nombres
reels
positifs quel onques.
3394. Propose par Dragoljub Milosevi et G. Milanova , Serbie.

edre
; designons

par hA et mA les longueurs resSoit ABCD un tetra
pe tives de la hauteur et de la mediane

issues du sommet A sur la fa e
opposee
BCD. Si V est le volume du tetra
edre,

montrer que
128
(hA + hB + hC + hD )(m2A + m2B + m2C + m2D ) V
3
3395. Propose par Tai hi Maekawa, Takatsuki, Prefe ture

d'Osaka, Japon.
Soit le triangle ABC ; denotons

par H son ortho entre et par R le rayon
de son er le ir ons rit. Montrer que 4R3 (l2 +m2 +n2 )R lmn = 0,
ou AH = l, BH = m et CH = n.
3396. Propose par Neven Juri , Zagreb, Croatie.

Soit n un entier positif et, pour i, j et k dans {1, 2, . . . , n}, posons
aijk
1 + mod(k i + j 1, n) + nmod(i j + k 1, n)
+ n2 mod(i + j + k 2, n) ,
ou mod(a, n) est le residu

de a modulo n, hoisi parmi 0, 1, . . . , n 1.
Pour quels n le ube portant les valeurs aijk est-il un ube magique ? (I i,
le mot \magique" se ref
ere
au fait que la somme des aijk est onstante si
deux des indi es restent xes et l'autre varie, et de plus que les sommes des
diagonales prin ipales du ube sont aussi egales

a ette même onstante.)
485
3397. Propose par Jose Luis Daz-Barrero, Universite Polyte hnique de

Catalogne, Bar elone, Espagne.

Evaluer
1
lim
n n2
n2 x2
dx .
2 + xx
3398. Propose par Bru e Shawyer, Universite Memorial de Terre-Neuve,

St. John's, NL.
Soit l'equation

j
n
10

j k
n
2n
n 10
10log10 n =
10
(a) montrer que n = 5294117647058823 est une solution de ette equation,

(b) determiner

toute autre solution entiere
et positive de ette equation.

3399. Proposed by Vin entiu Radules u, Universite de Craiova, Craiova,

Roumanie.
Montrer qu'il n'existe au une fon tion positive et deux fois dierentiable

f : [0, ) R telle que f (x)f (x) + 1 0 pour tout x 0.
3400. Propose par Yakub N. Aliyev, Universite d'Etat de Bakou, Bakou,
Azerbadjan.
Pour des entiers positifs m et k, posons (m)k = m(1 + 10 + 102 +
+ 10k1 ) ; par exemple, (1)2 = 11 et (3)4 = 3333. Determiner

tous les
nombres reels
tels que
j
10n
k
j
k
p
5 9
(1)2n + = (3)2n
6
est valide pour tout entier positif n, ou x denote

le plus grand entier ne
depassant

pas x.
.................................................................
3371. Corre tion. Proposed by George Tsintsifas, Thessaloniki, Gree e.
Let ABC be a triangle with a, b, and c the lengths of the sides opposite
the verti es A, B , and C , respe tively, and let M be an interior point of
ABC . The lines AM , BM , and CM interse t the opposite sides at the
points A1 , B1 , and C1 , respe tively. Lines through M perpendi ular to the
sides of ABC interse t BC , CA, and AB at A2 , B2 , and C2 , respe tively.
Let p1 , p2 , and p3 be the distan es from M to the sides BC , CA, and AB ,
respe tively. Prove that
[A2 B2 C2 ]
[A1 B1 C1 ]
(ap1 + bp2 )(bp2 + cp3 )(cp3 + ap1 )

2a2 b2 c2
where [KLM ] denotes the area of triangle KLM .
a
p1
b
p2
c
p3
486
3389. Proposed by Mihaly Ben ze, Brasov, Romania.
For a R dene a sequen e (xn ) by x0 = a and xn+1 = 4xn x2n

for all n 0. Prove that there exist innitely many a R su h that the
sequen e (xn ) is periodi .
3390. Proposed by Mihaly Ben ze, Brasov, Romania.

Prove that if A, B , C , and D are the solutions of
X
3 5
5
8
then A2007 + B 2007 + C 2007 + D2007 = O, where O is the 2 2 zero matrix.
3391. Proposed by Mi hel Bataille, Rouen, Fran e.
Let ABCD be a onvex quadrilateral su h that AC and BD interse t

in right angles at P , and let I , J , K , and L be the mid-points of AB , BC ,
CD , and DA, respe tively. Show that the ir les (P IJ ), (P J K), (P KL),
and (P LI) are ongruent if and only if ABCD is y li .
3392. Proposed by Mi hel Bataille, Rouen, Fran e.
Let A, B , C , D, and E be on y li with V and W on the lines AB and

respe tively. Show that if the line CE , the parallel to CB through V ,
and the parallel to CD through W are on urrent, then triangles EV B and
EW D are similar. Does the onverse hold?
AD ,
3393. Proposed by Dragoljub Milosevi and G. Milanova , Serbia.
Let ABC be a triangle with a = BC , b = AC , c = AB , and semiperimeter s. Prove that

y+z
A
z+x
B
x+y
C
9
2,
x
a(s a)
y
b(s b)
z
c(c a)
s
where the angles A, B , and C are measured in radians and x, y, and z are
any positive real numbers.
3394. Proposed by Dragoljub Milosevi and G. Milanova , Serbia.
Let ABCD be a tetrahedron with hA and mA the lengths of the altitude and the median from vertex A to the opposite fa e BCD, respe tively.
If V is the volume of the tetrahedron, prove that
128
3
(hA + hB + hC + hD )(m2A + m2B + m2C + m2D ) V
3395. Proposed by Tai hi Maekawa, Takatsuki City, Osaka, Japan.
Let triangle ABC have ortho entre H and ir umradius R. Prove that
4R3 (l2 + m2 + n2 )R lmn = 0, where AH = l, BH = m, and
CH = n.
487
3396. Proposed by Neven Juri , Zagreb, Croatia.
Let n be a positive integer, and for i, j , and k in {1, 2, . . . , n} let
aijk
1 + mod(k i + j 1, n) + nmod(i j + k 1, n)
+ n2 mod(i + j + k 2, n) ,
where mod(a, n) is the residue of a modulo n in the range 0, 1, . . . , n 1.

For whi h n is the ube with entries aijk a magi ube? (Here "magi " means
that the sum of aijk is onstant if two indi es are xed and the third index
varies, and also the sums along the great diagonals of the ube are equal to
this onstant.)
3397.
Proposed by Jose Luis Daz-Barrero, Universitat Polite ni a

de
Catalunya, Bar elona, Spain.
Evaluate
Z n 2
1
n x2
lim 2
dx .
x
n
n
2+x
3398. Proposed by Bru e Shawyer, Memorial University of Newfoundland, St. John's, NL.
Given the equation
j
n
10

j k
n
2n
n 10
10log10 n =
10
(a) show that n = 5294117647058823 is a solution,
(b) nd all other positive integer solutions of the equation.
3399. Proposed by Vin entiu Radules u, University of Craiova, Craiova,

Romania.
Prove that there does not exist a positive, twi e dierentiable fun tion
f : [0, ) R su h that f (x)f (x) + 1 0 for all x 0.
3400. Proposed by Yakub N. Aliyev, Baku State University, Baku, Azer-
baijan.
For positive integers m and k let (m)k = m(1+10+102 + +10k1 ),
for example, (1)2 = 11 and (3)4 = 3333. Find all real numbers su h that
j
10n
k
j
k
p
5 9
(1)2n + = (3)2n
6
holds for ea h positive integer n, where x is the greatest integer not ex eeding x.
488
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.
3229. [2007 : 170, 172; 2007 : 179{181 Proposed by Mihaly Ben ze,
Brasov, Romania.
(a) Let x and y be positive real numbers, and let n be a positive integer.
Prove that
n
(x + y)
n
X
k=0
1
n nk k
x
y
k
n+1+2
n ni
X
X
i=1 k=0
n
k
n
k+i
(n + 1)2 .
(b) Let x1 , x2 , . . . , xk be positive real numbers, and let n be a positive

integer. Determine the minimum value of
(x1 + x2 + + xk )n
i1 ++ik =n
i1 , . . . , ik 0
i1 !i2 ! ik !
n!xi11 xi22 xikk
Solution to (b) by Sergey Sadov, Memorial University of Newfoundland,

St. John's, NL.
We may assume that x1 + x2 + + xk = 1 without loss of generality.
The minimum o urs when x1 = x2 = = xk = k1 . We will obtain
this as a onsequen e of a more general proposition. For onvenien e, let
X = (x1 , x2 , . . . , xk ) be a ve tor in Rk+ (that is, xi > 0 for all i) and let
Q = (q1 , q2 , . . . , qk ) be a multi-index with non-negative integer entries qi .
Let
|Q|
|X|
Q!
XQ
Finally, let = {X
dimension k 1.
n
X
i=1
n
X
qi ;
xi ;
i=1
= q1 !q2 ! qk ! ;
= xq11 xq22 xqkk .
Rk+ : |X| = 1}
be the positive unit simplex in Rk of
489
Theorem 1 With the above notation, let
X
P (X) =
|Q|=n
cQ
XQ
be a homogeneous rational fun tion of degree n in the variables xi . Suppose that cQ 0 for ea h Q and that P (X) is a symmetri fun tion, that
is, inter hanging any xi and xj does not hange the value of P (X). Then
the minimum value of P (X) over the simplex exists and is attained when
1
x1 = x2 = = xk = .
k
Proof. If P (X) is identi ally zero, then there is nothing to prove, so we assume at least one oe ient cQ is not zero. Note that P (x) has a minimum
value, sin e P (X) is ontinuous on and tends to innity as any of the
xi approa hes 0. Suppose that the minimum o urs at a point with at least
two unequal oordinates. Without loss of generality (due to the symmetry of
P (X)) we may assume that x1 > x2 . We will show that by slightly de reasing x1 and slightly in reasing x2 (while keeping all other variables and the
sum x1 + x2 un hanged) the value of P (X) will be ome smaller, ontrary to
our assumption. Fixing x3 , x4 , . . . , xn auses P (X) to be ome a symmetri
fun tion of two variables
X
F (u, v) = P (u, v, x3 , . . . , xn ) =
j+sn
Aj,s
uj v s
where the oe ients Aj,s depend on x3 , x4 , . . . , xn . Note that ea h Aj,s

is non-negative and at least one of these is positive, and that Aj,s = As,j .
Thus, F (u, v) is a linear ombination with non-negative oe ients, not all
zero, of fun tions of the form
Fj,s (u, v) =
us v j
where j and s are non-negative integers with j+s n.
and v(t)
= x2 + t,
so that
the \time derivative"

d Fj,s
uj v s
du
= 1
dt
and

d
Fj,s u(t), v(t)
dt

u(t), v(t)
dt
=
=
dv
= 1.
dt
Now let u(t) = x1 t

It su es to prove that
is negative at t = 0. We have

Fj,s u, v
Fj,s u, v
u
v

j
s
1
s
j
1
u
v uj v s
u
v us v j
Dierentiating on e again, we obtain

d2 Fj,s u(t), v(t)
=
dt2

!
j
s
j
s 2
1
+ 2 +
+
2
j
u
v
u
v
u vs
j
s
+ 2 +
2
v
u
j
s
v
u
2 !
1
us v j
490
d2 F
dF
uv
> 0. Sin e
= 0 when u = v (this o urs when t =
),
dt2
dt
2
dF
follows that dt < 0 when t = 0. We have obtained a ontradi tion by
Hen e,
it
assuming x1 > x2 at a point a hieving the minimum. This proves that the
minimum o urs when all the xi are equal.
It follows that the minimum sought in part (b) is
kn
i1 !i2 ! . . . ik !
n!
i1 ++ik =n
i1 ,...,ik 0
No other solutions to part (b) were re eived.

Regarding part (a) Sadov remarks that
min
x+y=1
n
X
k=0
n
k
xnk yk
= 2
n
X
j =0
n = n + 1 + 2
j
n1
X nj
X
j =0 i=0
n
j
n
i+j
where the rst equality follows Theorem 1 and the last expression is the minimum obtained
by Bataille [2007 : 179{181. He notes that the rst equality yields a minimum of at least
2n+1 , onsiderably
improving the lower bound of (n + 1)2 if n > 4. He observes that if

bi = ni xni yi , then by the AM{HM Inequality

(x + y)n
1 X 1 1
1 X
,
bi =
n+1
bi
n+1
n+1
hen e (x + y)2 b1
(n + 1)2 , whi h yields a qui k proof of part (a).
i
Sadov omments that the fun tion P (X) in Theorem 1 is S hur- onvex, referring to [1
for the denition of this term and appli ations. He indi ates that the AM-GM Inequality and
Pthe
AM{HM Inequality an be obtained by taking P (X) = (x1 x2 xn )1 and P (X) = xi
in Theorem 1, respe tively.
He mentions that parti ular ases (and other theorems of a more general nature) of
Theorem 1 an be found in [2, Chapter 3, Se tion G, Examples G.1.k and G.1.m; though he
believes that Theorem 1 is present somewhere in the existing literature.
Finally, he refers to [3, Se tion 2.18, for a treatment of (the related) Muirhead's
Inequality.
P
Referen es
[1 M.L. Clevenson and W. Watkins, Majorization and the Birthday Inequality, Math.
Magazine, vol. 64, No. 3 (1991), pp. 183{188.
[2 A. Marshall and I. Olkin, Inequalities: Theory of Majorization and Its Appli ations,
A ademi Press, 1979.
[3 G.H. Hardy, J.E. Littlewood, and G. Polya, Inequalities, Cambridge University Press,
1952.
3289. [2007 : 485, 487 Proposed by Virgil Ni ula, Bu harest, Romania.

Let ABC be a triangle for whi h there exists a point D in its interior
su h that DAB = DCA and DBA = DAC . Let E and F be points
on the lines AB and CA, respe tively, su h that AB = BE and CA = AF .
Prove that the points A, E , D, and F are on y li .
491
A omposite of similar solutions by John G. Heuver, Grande Prairie, AB and
George Tsapakidis, Agrinio, Gree e.
Triangles ADC and BDA are similar (be ause their angles are assumed
to be equal), when e
CD
CA
2CA
CF
=
=
=
AD
AB
2AB
AE
It follows that DCF DAE (DCF = DCA = DAB = DAE ,

while the adja ent sides are proportional from the previous step), so that
AF D (= CF D) = AED . Noting that E and F both lie on the same
side of AD, we on lude that A, E , F , and D lie on the same ir le, as
desired.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Gree e; SEFKET
University of Sarajevo, Sarajevo, Bosnia and Herzegovina; RICARDO
ARSLANAGIC,
BARROSO CAMPOS, University of Seville, Seville, Spain; MICHEL BATAILLE, Rouen,
Fran e; CHIP CURTIS, Missouri Southern State University, Joplin, MO, USA; APOSTOLIS
K. DEMIS, Varvakeio High S hool, Athens, Gree e; OLIVER GEUPEL, Bruhl,
NRW, Germany;
GEOFFREY A. KANDALL, Hamden, CT, USA; ANDREA MUNARO, student, University of
Trento, Trento, Italy; PETER Y. WOO, Biola University, La Mirada, CA, USA; TITU ZVONARU,
Comane
sti, Romania; and the proposer.
Bataille omments that D an be onstru ted as the point inside ABC where the ir le
F AE interse ts the ir le through B, C, and the ir um entre O. It lies on the rst ir le by
the result of this problem. It lies on the se ond ir le be ause BOC = BDC = 2 BAC
as follows: on the one hand the angle at A is ins ribed in the ir le BAC that is entred at O
and is therefore 12 BOC; on the other hand the similar triangles ADC and BDA t together
at A and at D in su h a way that the two exterior angles at D (that form BDC) sum to twi e
the sum of the two interior angles at A (that form BAC).
Let ABCD be a trapezoid with AD k BC . Denote the lengths of

AD and BC by a and b, respe tively. Let M be the mid-point of CD , and
let P and Q be the mid-points of AM and BM , respe tively. If N is the
interse tion of DP and CQ, prove that N belongs to the interior of ABM
if and only if 1 < a < 3.
3
Solution by Joel S hlosberg, Bayside, NY, USA.

There is a misleading subtlety in the statement of the problem: we shall
see that the on lusion fails should AD = BC ; in other words, our trapezoid
ABCD must not be a parallelogram.
Sin e all of the onditions of the problem are invariant under an ane
transformation, we an assume without loss of generality that
AB AD
and
AB = 4 .
We therefore introdu e a Cartesian oordinate system with the origin at A,

and with B on the positive x{axis and D on the positive y{axis; then A, B ,
492
C , and D
have oordinates
A(0, 0) , B(4, 0) , C(4, b) , D(0, a) ,
where a and b are positive. It follows that the oordinates of M , P , and Q

are

a+b
a+b
a+b
M 2,
, P 1,
, Q 3,
,
2
so that the lines DP and CQ satisfy

y =
when e N
b 3a
4
x + a
and
y =
= DP CQ has oordinates

4b (a b)2
,
N
a+b a+b
3b a
4
x + a 2b ,
A point in the plane is on the same side of AB as M if and only if its

is positive. The y{ oordinate of N satises
y { oordinate
(a b)2
0,
a+b
with equality if and only if a = b. Sin e P = DP AM has x{ oordinate

1, a point on line DP is on the same side of AM as B if and only if its
x{ oordinate ex eeds 1. Lastly, sin e Q = CQ BM has x{ oordinate 3,
a point on line CQ is on the same side of BM as A if and only if its
x{ oordinate is less than 3. Therefore, N is within the interior of ABM
if and only if a 6= b and 1 <
4b
< 3.
a+b
This inequality is equivalent to
1
a
<
< 3,
3
b
whi h is the desired inequality; it is equivalent to N belonging to the interior

of ABM as long as a 6= b.
Also solved by ROY BARBARA, Lebanese University, Fanar, Lebanon; MICHEL
BATAILLE, Rouen, Fran e; CHIP CURTIS, Missouri Southern State University, Joplin, MO, USA;
IES Alvarez

FRANCISCO JAVIER GARC I A CAPITAN,
Cubero, Priego de Cordoba,

Spain; OLIVER
GEUPEL, Bruhl,
NRW, Germany; PETER Y. WOO, Biola University, La Mirada, CA, USA; TITU
ZVONARU, Comane
Let ABC be an isos eles triangle with AB = AC . Find all points P

su h that the sum of the squares of the distan es of the points A, B , and C
from any line through P is onstant.
493
Solution by Mi hel Bataille, Rouen, Fran e.
We will introdu e a Cartesian oordinate system. Label the points as
A(0, a), B(b, 0), C(b, 0), and P (u, v), where a and b are positive and u and
v are variables. The equation of a line through P is x cos + y sin = p,
where p = u cos + v sin and is an arbitrary real number. Let d(Q, )
denote the distan e between a point Q and the line , and let
S = d(A, )2 + d(B, )2 + d(C, )2 .
We then have
d(A, )2 = (a sin p)2 , d(B, )2 = (p + b cos )2 ,

d(C, ) = (p b cos )2 . After a simple al ulation, we obtain

3(u2 v 2 )
a2
2
S = (cos 2)
+ av
+b
2
2
3(u2 + v 2 )
a2
+ (sin 2)(3uv au) +
av +
+ b2 .
2
2
2
and
The number S is independent of if and only if

and
3uv au = 0
whi h gives
u = 0
or
and
3(u2 v 2 ) + 2av a2 + 2b2 = 0 ,

3v 2 + 2av a2 + 2b2 = 0 ,
(1)

a2
2a2
2
v =
and 3 u
(2)
+
a2 + 2b2 = 0 .
3
9
3
Now, (1) is satised if and only if 0 < a b 3 (that is, B = C 60 )

and the oordinates of P are
!
!
p
p
a + 2(3b2 a2 )
a 2(3b2 a2 )
or
.
0,
0,
3
3
Similarly, (2) is satised if and only if a b 3 (that is, B = C 60 )

and the oordinates of P are
!
!
p
p
2(a2 3b2 ) a
2(a2 3b2 ) a
,
or
,
.
3
3
3
3
a
In on lusion, if ABC is equilateral, only the entre of ABC is a solution

for P . Otherwise, there are two solutions for P :
P
!
p
2(3b2 a2 )
3
(if A > 60 ) ,
!
p
2(a2 3b2 ) a
,
3
3
(if A < 60 ) .
0,
494
In both ases, the two solutions are symmetri about the entroid of the
triangle; in the rst ase, the two points lie on the median through A and
in the se ond ase the two points lie on a line parallel to BC and passing
through the entroid of ABC .
Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Gree e; ROY BARBARA,

Lebanese University, Fanar, Lebanon; CHIP CURTIS, Missouri Southern State University,

Joplin, MO, USA; FRANCISCO JAVIER GARC I A CAPITAN,
IES Alvarez
Cubero, Priego de

Cordoba,

Spain; WALTHER JANOUS, Ursulinengymnasium, Innsbru k, Austria; V ACLAV
Y,
Big Rapids, MI, USA (2 solutions); SKIDMORE COLLEGE PROBLEM SOLVING
KONECN
GROUP, Skidmore College, Saratoga Springs, NY, USA; GEORGE TSAPAKIDIS, Agrinio, Gree e;
PETER Y. WOO, Biola University, La Mirada, CA, USA; TITU ZVONARU, Comane
sti, Romania;
and the proposer.
3292. [2007 : 485, 488 Proposed by Mihaly Ben ze, Brasov, Romania.
Let a, b, c, and d be arbitrary real numbers. Show that
11a2 + 11b2 + 221c2 + 131d2 + 22ab + 202cd + 48c + 6
98ac + 98bc + 38ad + 38bd + 12a + 12b + 12d .
Solution by Oliver Geupel, Bruhl,

NRW, Germany.
The proof is by ontradi tion. Assume that
11a2 + 11b2 + 221c2 + 131d2 + 22ab + 202cd + 48c + 6
< 98ac + 98bc + 38ad + 38bd + 12a + 12b + 12d .
We onsider the quadrati f (x) = 11x2 + px + q with

p
q
=
=
22b 98c 38d 12 ,

11b2 + 221c2 + 131d2 + 202cd + 48c + 6 98bc 38bd
12b 12d .
Then f (a) < 0, and the leading oe ient of f is positive, hen e f has two
distin t real roots; that is, the dis riminant of f is positive. By omputing
the dis riminant, we nd p2 44q = 120(c+6d1)2 0, a ontradi tion.
Also solved by APOSTOLIS K. DEMIS, Varvakeio High S hool, Athens, Gree e; MICHEL
BATAILLE, Rouen, Fran e; CHIP CURTIS, Missouri Southern State University, Joplin, MO, USA;

RICHARD I. HESS, Ran ho Palos Verdes, CA, E-U;
WALTHER JANOUS, Ursulinengymnasium,
Innsbru k, Austria; KEE-WAI LAU, Hong Kong, China; JOEL SCHLOSBERG, Bayside, NY, USA;
STAN WAGON, Ma alester College, St. Paul, MN, USA; PETER Y. WOO, Biola University, La
Mirada, CA, USA; and the proposer.
495
Prove that
n arcsin
Y
k=1
9k + 2
3
27k + 54k 2 + 36k + 8
arctan
3k + 1
= 3n .
Composite of similar solutions by Mi hel Bataille, Rouen, Fran e and

Douglass L. Grant, Cape Breton University, Sydney, NS, modied by the
editor.
For ea h k = 1, 2, . . . , n let Pk denote the orresponding fa tor under
the produ t sign. We prove that in fa t, Pk = 3 for ea h k.
3
Note rst that 27k
+ 54k2 + 36k + 8 = (3k + 2)3 . For xed k, let

1
= tan1
. Then tan = 1
implies sin = 1 .
3k + 1
3k + 1
3k + 2
Sin e tan1 is an in reasing fun tion, we have

0 < tan

1
2
< tan
hen e, 0 < 3 < 2 .

From
sin(3)
we obtain
= 3 sin 4 sin3
4
9k + 2
3
p
= p
=
3
3k + 2
(3k + 2)
(3k + 2)3
sin1
sin1
and it follows that Pk =

9k + 2
27k3 + 54k2 + 36k + 8

!
9k + 2
p
= sin1 (sin 3) = 3 ,
(3k + 2)3
3
= 3.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Gree e; SEFKET
University of Sarajevo, Sarajevo, Bosnia and Herzegovina; DIONNE BAILEY,
ARSLANAGIC,
ELSIE CAMPBELL, CHARLES DIMINNIE, and KARL HAVLAK, Angelo State University, San
Angelo, TX, USA; ROY BARBARA, Lebanese University, Fanar, Lebanon; CHIP CURTIS,
Missouri Southern State University, Joplin, MO, USA; OLEH FAYNSHTEYN, Leipzig,
student, Sarajevo
Germany; OLIVER GEUPEL, Bruhl,

NRW, Germany; SALEM MALIKIC,
College, Sarajevo, Bosnia and Herzegovina; HENRY RICARDO, Medgar Evers College
(CUNY ), Brooklyn, NY, USA; JOEL SCHLOSBERG, Bayside, NY, USA; GEORGE TSAPAKIDIS,
Agrinio, Gree e; PETER Y. WOO, Biola University, La Mirada, CA, USA; TITU ZVONARU,
Comane
496
For all positive integers m and n, show that
m(m + 1)n2 (n + 1)2 (2n2 + 2n 1) n(n + 1)m2 (m + 1)2 (2m2 + 2m 1)
is divisible by 720.
Solution by Chip Curtis, Missouri Southern State University, Joplin, MO,
USA.

Let f (k) = k(k + 1) 2k2 + 2k 1 . Write
A (m, n)

= m (m + 1) n2 (n + 1)2 2n2 + 2n 1

n (n + 1) m2 (m + 1)2 2m2 + 2m 1
= mn (m + 1) (n + 1) [f (n) f (m)] .
Let C(m, n)
= mn(m + 1)(n + 1) and D(m, n) = f (n) f (m), so

A(m, n) = C(m, n) D(m, n). Sin e 720 = 24 32 5, it su es to show
A(m, n) is divisible by 16, 9, and 5.
that
that
(a) Divisibility by 16. The residues of f (n) modulo 4 are given in the following table.
n (mod 4) 0
f (n) (mod 4) 0
1 2
2 2
3
0
(i) If n 0 or 3 (mod 4), or m 0 or 3 (mod 4), then C(m, n)

is divisible by 8 and D(m, n) is divisible by 2, so that A(m, n) is
divisible by 16.
(ii) Otherwise, C(m, n) is divisible by 4 and D(m, n) is divisible by
4, and therefore, A(m, n) is divisible by 16.
(b) Divisibility by 9. The residues of f (n) modulo 9 are given in the next
table.
n (mod 9)
f (n) (mod 9)
0 1
0 6
2 3
3 6
4 5
6 6
6 7
3 6
8
0
(i) If n 0 or 8 (mod 9), or m 0 or 8 (mod 9), then C(m, n) 0

(mod 9), so that A(m, n) is divisible by 9.
(ii) If n 2 or 6 (mod 9), or m 2 or 6 (mod 9), then C(m, n)
and D(m, n) are ea h divisible by 3, and therefore, A(m, n) is
divisible by 9.
(iii) Otherwise, f (n) 6 (mod 9) and f (m) 6 (mod 9), so that
D(m, n) is divisible by 9. Thus, A(m, n) is divisible by 9.
497
( ) Divisibility by 5. The residues of f (n) modulo 5 are given in the table
below.
n (mod 9)
f (n) (mod 9)
0 1
0 1
2 3
1 1
4
0
(i) If n 0 or 4 (mod 5), or m 0 or 4 (mod 5), then C(m, n) is

divisible by 5, so that A(m, n) is divisible by 5.
(ii) Otherwise, D(m, n) is divisible by 5, and therefore, A(m, n) is
divisible by 5.

University of Sarajevo, Sarajevo, Bosnia and
Also solved by SEFKET
ARSLANAGIC,
Herzegovina; ROY BARBARA, Lebanese University, Fanar, Lebanon; MICHEL BATAILLE,
Rouen, Fran e; BRIAN D. BEASLEY, Presbyterian College, Clinton, SC, USA; OLIVER GEUPEL,
Bruhl,
NRW, Germany; RICHARD I. HESS, Ran ho Palos Verdes, CA, USA; WALTHER JANOUS,
Ursulinengymnasium, Innsbru k, Austria; ANDREA MUNARO, student, University of Trento,
Trento, Italy; MICHAEL PARMENTER, Memorial University of Newfoundland, St. John's,
NL; SKIDMORE COLLEGE PROBLEM SOLVING GROUP, Skidmore College, Saratoga Springs,
NY, USA; PETER Y. WOO, Biola University, La Mirada, CA, USA; KONSTANTINE ZELATOR,
University of Toledo, Toledo, OH, USA; TITU ZVONARU, Comane
sti, Romania; and the
proposer.
3295. [2007 : 486, 488 Proposed by Mi hel Bataille, Rouen, Fran e.

Let u : R R be a bounded fun tion. For x > 0, let
and
Prove that
f (x)
g(x)
=
=
sup {u(t) : t > ln(1/x)}

sup {u(t) xet : t R} .
lim f (x) = lim g(x).
x0+
Solution by Peter Y. Woo, Biola University, La Mirada, CA, USA, modied

by the editor.
The proof onsists of showing that both limits are equal to the limit
L = lim sup u(t), whi h exists sin e the fun tion u(t) is bounded. Let K be
t
su h that |u(t)| < K for all t R. Given any > 0 and any t0 > 0, we have
(i) there exists some t > t0 su h that L < u(t), and
(ii) there exists some t1 > t0 su h that for all t > t1 , u(t) < L + .
Let > 0 be given. For a xed x > 0 we have t
lim xet = 0, so there exists
t0 > 0 su h that xet < for any t > t0 . By part (i), there exists t > t0
su h that L < u(t), hen e
L 2 < u(t) xet g(x) .
Thus, g(x) > L 2 for ea h x > 0.
498
On the other hand, there exists t1 R su h that u(t) < L + for all
Sin e et > 0, let M > 0 be su h that K M et < L + . We
laim that if x > M , then g(x) L + . For this it su es to show that
u(t) xet < L + whenever x > M and t R.
Indeed, if x > M and t t1 , then u(t) xet < K M et < L + ,
while if x > M and t > t1 , then u(t) xet < u(t) < L + .
Therefore, for x > M we have L 2 < g(x) < L + 2, hen e
lim g(x) = L.
x
Finally, writing S(v) = sup {u(t) : t > v} and making two hanges of
variable in the limit yields
t > t1 .
lim f (x) =
x0+

lim S ln(1/x) = lim S(ln y) = lim S(z) = L .
x0+
Also solved by OLIVER GEUPEL, Bruhl,

NRW, Germany; and the proposer.
3296. [2007 : 486, 488 Proposed by Mi hel Bataille, Rouen, Fran e.

Find the greatest onstant K su h that
b2 c2
c2 a2
a2 b2
+
+
> K
a2 (a b)(a c)
b2 (b c)(b a)
c2 (c a)(c b)
for all distin t positive real numbers a, b, and c.

Solution by Peter Y. Woo, Biola University, La Mirada, CA, USA, expanded
by the editor.
We prove that K = 10.
Let L denote the left side of the given inequality. Sin e L is ompletely
symmetri in a, b, and c, we may assume without loss of generality that
a < b < c.
Note rst that L = P
where Q = a2 b2 c2 (b a)(c a)(c b) and
Q
P = b4 c4 (c b) c4 a4 (c a) + a4 b4 (b a).
Observing that P = 0 when c = b or b = a or c = a, we nd by
straightforward but tedious omputations that
P

= b4 c4 (c b) a4 c5 b5 + a5 c4 b4

= (c b) b4 c4 a4 c4 + c3 b + c2 b2 + cb3 + b4

+ a5 c3 + c2 b + cb2 + b3
499
=
=

(c b) c4 b4 a4 + a4 (a b) c3 + c2 b + cb2 + b3

(c b)(b a) c4 b3 + b2 a + ba2 + a3

a4 c3 + c2 b + cb2 + b3

(c b)(b a) b3 c4 a4 + cb2 a c3 a2

+ c2 ba2 c2 a2 + c3 a3 (c a)

(c b)(b a)(c a) b3 c3 + c2 a + ca2 + a3

+ cb2 a c2 + ca + a2 + c2 ba2 (c + a) + c3 a3 .
Hen e,
W
P
W
= 2 2 2,
Q
a b c
where
b3 c3 + c2 a + ca2 + a3
+ cb2 a c2 + ca + a2
+ c2 ba2 (c + a) + c3 a3 .
By writing W as a sum of 10 terms and using the AM{GM Inequality, we

readily see that W 10 a20 b20 c20 1/10 = 10a2 b2 c2 , from whi h it follows
that L 10. Sin e a, b, and c are distin t, equality annot hold. Thus,
L > 10.
Finally, if we set a = 1, b = 1 + , and c = 1 + 2 and let 0+ ,
then the value of L an be made arbitrarily lose to 10 from the right. Hen e,
K = 10.
Also solved by CHIP CURTIS, Missouri Southern State University, Joplin, MO, USA;
OLIVER GEUPEL, Bruhl,
NRW, Germany; RICHARD I. HESS, Ran ho Palos Verdes, CA, USA;
and the proposer. There were one in orre t and three in omplete solutions (whi h only showed
that L > 10 and then on luded immediately that K = 10).
3297. [2007 : 486, 488 Proposed by Stanley Rabinowitz, MathPro Press,

Chelmsford, MA, USA.
If A, B , and C are the angles of a triangle, prove that
1+ 5
sin A + sin B sin C
2
When does equality hold?
500
Solution by D.J. Smeenk, Zaltbommel, the Netherlands.
The following are equivalent
1+ 5
sin A + sin B sin C
;
2
2 sin A + cos(B C) cos(B + C) 1 + 5 ;
2 sin A + cos A + cos(B C) 1 + 5 .
Let be the rst quadrant angle with cos = 2 and sin =

5
(the angle is approximately 26.6 ). The last inequality then be omes
5 sin(A + ) + cos(B C) 1 + 5 .
However, sin(A + ) 1 and cos(B C) 1, so the last inequality is true.

Equality holds when A = arcsin 2 63.4 and B = C 58.3 .
5

University of Sarajevo, Sarajevo, Bosnia
ARSLANAGIC,
and Herzegovina; ROY BARBARA, Lebanese University, Fanar, Lebanon; MICHEL
BATAILLE, Rouen, Fran e; BRIAN D. BEASLEY, Presbyterian College, Clinton, SC,
USA; CHIP CURTIS, Missouri Southern State University, Joplin, MO, USA; APOSTOLIS
K. DEMIS, Varvakeio High S hool, Athens, Gree e; OLEH FAYNSHTEYN, Leipzig,
NRW, Germany; RICHARD I. HESS, Ran ho Palos Verdes,

Y,

CA, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbru k, Austria; V ACLAV
KONECN
Big Rapids, MI, USA; KEE-WAI LAU, Hong Kong, China; THANOS MAGKOS, 3rd High S hool
student, Sarajevo College, Sarajevo, Bosnia and
of Kozani, Kozani, Gree e; SALEM MALIKIC,

Herzegovina; JUAN-BOSCO ROMERO MARQUEZ,
Universidad de Valladolid, Valladolid, Spain;
JOEL SCHLOSBERG, Bayside, NY, USA; SKIDMORE COLLEGE PROBLEM SOLVING GROUP,
Skidmore College, Saratoga Springs, NY, USA; PANOS E. TSAOUSSOGLOU, Athens, Gree e;
GEORGE TSAPAKIDIS, Agrinio, Gree e; PETER Y. WOO, Biola University, La Mirada, CA, USA;
KONSTANTINE ZELATOR, University of Toledo, Toledo, OH, USA; TITU ZVONARU, Comane
sti,
Romania; and the proposer. There was one in orre t solution submitted.
Janous proved a more general result, namely, for > 0
sin A + sin B sin C

with equality if and only if A = arccos
1
4 2 +1
1+
42 + 1
2
and B = C.
3298. [2007 : 486, 489 Proposed by Stanley Rabinowitz, MathPro Press,

Chelmsford, MA, USA.
Let ABC be a triangle of area
opposite vertex A. Prove that
1
2
in whi h
a2 + csc A
is the length of the side
5.
[Ed.: The proposer's only proof of this is by omputer. He is hoping

that some CRUX with MAYHEM reader will nd a simpler solution.
501
Solution by Kee-Wai Lau, Hong Kong, China.
Let b = AC , c = AB and let S denote the area of triangle ABC .
Sin e S = 12 bc sin A = 12 , we obtain bc = csc A 1.
By the Law of Cosines we have (regardless of the sign of cos A) that
a2 + csc A
= a2 + bc = b2 + c2 2bc cos A + bc
q
2
2
b + c 2bc 1 sin2 A + bc
p
= b2 + c2 2 b2 c2 (bc sin A)2 + bc
p

= b2 + bc + c2 2 b2 c2 1
p
3bc 2 b2 c2 1 .
For x 1, let y = 3x 2 x2 1. Then y is positive and from

(y 3x)2 = 4(x2 1) we get 5x2 6xy + y 2 + 4 = 0. Sin e x is real, the
dis riminant of the quadrati polynomial above must be non-negative.

(6y)2 20(y 2 + 4) 0, or 16y 2 80 0, from whi h we
Thus,
obtain y 5. The result now follows by setting x = bc.
Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Gree e (two solutions);

University of Sarajevo, Sarajevo, Bosnia and Herzegovina; MICHEL
SEFKET
ARSLANAGIC,
BATAILLE, Rouen, Fran e; ROY BARBARA, Lebanese University, Fanar, Lebanon; CHIP
CURTIS, Missouri Southern State University, Joplin, MO, USA; APOSTOLIS K. DEMIS,
Varvakeio High S hool, Athens, Gree e (two solutions); OLEH FAYNSHTEYN, Leipzig,
NRW, Germany; RICHARD I. HESS, Ran ho Palos Verdes,

Y,
Big Rapids, MI, USA; SOTIRIS LOURIDAS, Aegaleo, Gree e;
CA, USA; V ACLAV
KONECN
PANOS E. TSAOUSSOGLOU, Athens, Gree e; PETER Y. WOO, Biola University, La Mirada,
CA, USA; KONSTANTINE ZELATOR, University of Toledo, Toledo, OH, USA; TITU ZVONARU,
Comane
sti, Romania; and the proposer. There was one in orre t solution submitted.
From
dedu e that equality holds if and only if b = c
q the proof given above, it is easy to

and a = 4 45 , in whi h ase A = cos1 23 48.19 . This was pointed out by Barbara,
Geupel, Hess, Kone ny,
Tsapakidis, and Zvonaru.
Both Barbara
and Demis generalized to
an arbitrary triangle of area k > 0, proving that
a2 + csc A 8k + 1 and the lower bound 8k + 1 is the best possible.
3299. [2007 : 487, 489 Proposed by Vi tor Oxman, Western Galilee College, Israel.
Given positive real numbers a, b, and wb , show that
(a) if a triangle ABC exists with BC = a, CA = b, and the length of
the interior bise tor of angle B equal to wb , then it is unique up to
isomorphism;
(b) for the existen e of su h a triangle in (a), it is ne essary and su ient
that
2a |a wb |
b >
0;
2a wb
( ) if ha is the length of the altitude to side BC in su h a triangle in (a),

we have b > |a wb | + 12 ha .
502
Solution by Mi hel Bataille, Rouen, Fran e.
(a) For onvenien e we write w = wb . Let the interior bise tor of B
meet AC at W . We assume that ABC exists with BC = a, CA = b, and
BW = w , and shall produ e a (impli itly dened) formula for the third side
2ac cos B
2
c = AB . We re all that w =
, so that
a+c
B
aw
w =
2
c
2a cos
By the Law of Cosines we have

standard formula
ab
WC =
,
a+c
a2 b2
(a + c)2
W C 2 = a2 + w 2 2aw cos
we therefore have
B
;
2

B
aw 2
= a2 w 2a cos
w = a2
2
c
using the
It follows that c must be the unique positive solution of f (x) = a, where f

is the de reasing fun tion on (0, ) given by
f (x) =
ab2
w2
+
(a + x)2
x
Thus, if a triangle ABC with the given parameters does exist, then its side
lengths are uniquely determined, and (a) is proved.
(b) First, it is easy to see that the two given inequalities are equivalent
to the onjun tion of the following three inequalities
w < 2a ;
(2a + b)w < 2a(a + b) ;
2a(a b) < (2a b)w .
(1)
Se ond, the existen e of a suitable triangle ABC is equivalent to the fa t

that the solution c of f (x) = a satises |a b| < c < a + b; that is,
f |a b|
> a > f (a + b) .
(2)
To show that (1) and (2) are equivalent, we rst suppose that the inequalities
in (1) hold. The inequality f (a + b) < a redu es to the equivalent inequality
(2a + b)w < 2a(a + b), whi h holds by (1). As for f |a b| > a, it is
equivalent to
ab2 |a b| + w 2 a + |a b|
2
> a |a b| a + |a b|
2
This redu es to w2 b2 > 0 when a b, so it holds then; but it also holds if

b < a sin e then it be omes w(2a b) > 2a(a b), whi h holds by (1). We
have proved that (1) implies (2).
Conversely, we assume that (2) holds (that is, that a triangle ABC
w
c
B
with the given parameters exists). Then 2a
=
cos ; hen e w < 2a.
a+c
2
503
Moreover, from f (a + b) < a we obtain (2a + b)w < 2a(a +
b). As for the
ondition 2a(a b) < (2a b)w, it follows from f |a b| > a if a > b,
and from (2a w)(a b) < aw if a b (be ause 2a > w, the left side is
negative). The desired equivalen e follows.
( ) Note that b > |a w| + 21 ha is equivalent to
ab > a |a w| + Area(ABC) ;
c
B
that is, to ab > a |a w| + a +
w sin . Sin e a |a w| <
2
2
(from part (b)), the latter will ertainly hold if
b (a + c) sin
This inequality is equivalent to
sin B (sin A + sin C) sin
or to
2 cos
or nally to
B
2 sin
2
A+C
2
1 cos
cos
AC
2
whi h is ertainly true. The result follows.
(2a w)b
2
B
2
AC
2
Also solved by CHIP CURTIS, Missouri Southern State University, Joplin, MO, USA;
OLIVER GEUPEL, Bruhl,

NRW, Germany; PETER Y. WOO, Biola University, La Mirada, CA,
USA (part ( ) only); and the proposer.
Parts (a) and (b) of our problem appear on page 11 of D.S. Mitrinovi et al., Re ent
Advan es in Geometri Inequalities, Kluwer A ademi Publishers, 1989 as the rst of 40 existen e
results from a 1952 paper (in Cze h) by G. Petrov.
In addition to his solution, Oxman also addressed the question of onstru tibility. Exer ise 4 on page 142 of Gunter

Ewald's Geometry: An Introdu tion (Wadsworth Publ., 1971) says
that in general a triangle annot be onstru ted by ruler and ompass given the lengths a, b,
and wb , even when that triangle exists. The author suggests that the proof of his laim an be
simplied by taking both the given side lengths equal to 1. The formula f(x) = 1 from part
(a) of the featured solution (with a = b = 1, and w2 hosen to be rational) is a ubi equation
with rational oe ients. One simply has to hoose a value of w for whi h the resulting ubi equation has no rational root. The theory of Eu lidean onstru tions then tells us that the
positive root, namely c, annot be onstru ted by using ruler and ompass.
3300. [2007 : 487, 489 Proposed by Arkady Alt, San Jose, CA, USA.
Let a, b, and
dene
be positive real numbers. For any positive integer
Fn =
X bn + cn
3(an + bn + cn )
a+b+c
b+c
y li
(a) Prove that Fn 0 for n 5.

(b) Prove or disprove that Fn 0 for n 6.
504
Solution by Cao Minh Quang, Nguyen Binh Khiem High S hool, Vinh Long,
Vietnam.
Sin e F1 = 0, we take n > 1. We note that xn1 yn1 (x y) 0
for all positive x and y, with equality if and only if x = y. We have
(a + b + c)Fn
3 (an + bn + cn ) (a + b + c)
(an + bn + cn )
=
=
X bn + cn
b+c
y li
X a (bn + cn )
b+c
y li

X
a (bn + cn )
an
b+c
y li
"

#
X ab an1 bn1
ac an1 cn1
+
(b + c)
(b + c)
y li

X ab an1 bn1 (a b)
0.
(b + c)(c + a)
y li
Equality holds if and only if a = b = c.

University of Sarajevo, Sarajevo, Bosnia and
ARSLANAGIC,
^
Herzegovina; ROY BARBARA, Lebanese University, Fanar, Lebanon; VASILE CIRTOAJE,
University of Ploiesti, Romania; CHIP CURTIS, Missouri Southern State University, Joplin, MO,
USA; NIKOLAOS DERGIADES, Thessaloniki, Gree e; OLIVER GEUPEL, Bruhl,
NRW, Germany;
WALTHER JANOUS, Ursulinengymnasium, Innsbru k, Austria; PANOS E. TSAOUSSOGLOU,
Athens, Gree e (part (a) only); STAN WAGON, Ma alester College, St. Paul, MN, USA (part (a)
only); TITU ZVONARU, Comane
C^rtoaje mentioned that this problem was posted (together with a solution similar to
the one featured above) by Wolfgang Berndt (Spanferkel) on the Mathlinks Forum website
http://www.mathlinks.ro/Forum/viewtopic.php?p=607167 in August 2006. Barbara, C^rtoaje,
and Dergiades proved the following generalization: If a1 , a2 , . . . , am are positive real numbers, m 2, and
Fn =
n
n
X an + + an
m(an
m
1 + a2 + + am )
2
a1 + a2 + + am
a2 + + am
y li
then Fn 0 for all n 1. Alt ultimately proved that if a, b, c, p, and q are positive real
numbers and
F (p, q) =
X ap + bp
3(ap + bp + cp )
aq + bq + cq
aq + bq
y li
then (p q)F (p, q) 0.
Mathematical Association of America Presents
Problems from Murray Klamkin: The Canadian Collection

Andy Liu & Bruce Shawyer, Editors
Murray Klamkin was a dedicated problem solver and problem proposer,

who left indelible marks on the problemist community. After working in industry and academe in the United States, he spent the last thirty of his eighty
four years in Canada. He was famous for his Quickies, problems that have
quick and neat solutions. In this book you will find all of the problems that he
proposed for CRUX with MAYHEM, including all of his Quickies. His problems covered a very wide range of topics, and show a great deal of insight
into what is possible in these areas. The problems are arranged into sets
according to topic,and the lightly edited solutions are as published in CRUX
with MAYHEM.
Problems from Murray Klamkin: The Canadian Collection is published
by the Mathematical Association of America (MAA) in collaboration with the
Canadian Mathematical Society (CMS). It is the first volume in the Canadian
Collection.

Order your copy today!

Call (301) 617-7800
Series: Problem Books,
Catalog Code: CAN-01
280 pp., Hardbound, 2009,
ISBN 9780-88385-828-8,
List: $61.50 Member: $49.95
(CMS members get member price for this book. Be sure to mention you are a CMS member when you call.)
506
YEAR END FINALE

This marks my rst year as Editor/Co-Editor of CRUX with MAYHEM, and
my rst year end nale. As I write in this spa e, whi h Jim Totten wrote in just
one year ago, I am reminded of him and the non-permanen e of things. Just as
the surf swirls sand and pebbles on a bea h, so does time alter words and ourselves.
Islands disappear and are built up somewhere else. When I remember Jim's in redible
enthusiasm and passion for mathemati s and his outrea h work, I am moved. An
o ean is there for us to feel its power and its ool refreshing spray, to play in it and
be leansed by its soothing sound.
Whi h reminds me that issues of CRUX with MAYHEM are now delayed by
several weeks! My apologies to all the readers for this onsiderable delay, and in
2009 we will work hard to lose the gap.
Many people have served this past year (I will borrow Jim's habit of apitalizing
their names). First, I shall forever be grateful to the late JIM TOTTEN for all the
support he gave me during the time I Co-Edited with him (until the end of February,
2008). After Jim's sudden passing on Mar h 9, 2008, three individuals in parti ular
ame to my res ue. I thank CHRIS FISHER for his humour and good spirits in those
tough early days. I thank BRUCE CROFOOT for his solid support as Mission Control in
the CRUX o e at Thompson Rivers University and for listening to me when I needed
someone to talk to. I thank BRUCE SHAWYER for his ontinuing help and advi e on
CRUX with MAYHEM, indeed for ee tively stepping into the role of Editor-at-Large
(whi h Jim was going to take up after stepping down as Editor-in-Chief). I also thank
Bru e S. for his hospitality in Newfoundland and for providing me with a mu h needed
sense of ontinuity.
I thank the members of the Editorial Board for their hard work and for bringing their unique talents to CRUX: JEFF HOOPER, the Asso iate Editor, for his areful
reading and sound judgment; IAN VANDERBURGH, the MAYHEM Editor, whose
Problem of the Month olumn is a joy to read and who is ahead of s hedule (!);
ROBERT BILINSKI, for serving as Skoliad Editor for the past four years; ROBERT
WOODROW, for handling the Olympiad Corner on top of his immense administrative
duties; JOHN GRANT M LOUGHLIN, for his solid support as Book Reviews Editor
and for his fantasti proof reading skills. This is John's last issue as Book Reviews Editor, but he will ontinue as the Guest Editor for the Jim Totten spe ial issue of May,
2009. I wel ome AMAR SODHI to the Board who is taking over as Book Reviews
Editor from John. I thank my olleague here in Winnipeg and Arti les Editor JAMES
CURRIE, for taking on the position and then turning a ba klog of arti les into a ornu opia. I thank the Problems Editors for their ontinuing support and hard work:
ILIYA BLUSKOV, CHRIS FISHER, MARIA TORRES, and EDWARD WANG. The job of
being a Problems Editor is perhaps one of the most di ult on the Board, with tight
deadlines and a deluge of problem proposals and solutions in all areas of mathemati s oming from all over the world et hed, s rawled, or digitally re orded on every
on eivable medium known to man. A big thank you to the four of you.
I thank GRAHAM WRIGHT, the Managing Editor, for his solid support and also
for providing that vital sense of ontinuity.
Others who are not on the Editorial Board but whose work is just as ru ial are
MONIKA KHBEIS and ERIC ROBERT who serve on the Mayhem sta; a big thank
you Monika and Eri . I thank JEAN-MARC TERRIER for translating the problems
that appear in CRUX with MAYHEM. The Canadian Mathemati al So iety re ently
honoured Jean-Mar at the 2008 Winter Meeting for his many years of servi e. It is
507
well deserved! I look forward to working with Jean-Mar in the New Year. As well,
I wel ome ROLLAND GAUDET of College
Universitaire Saint-Bonifa e, Winnipeg.
Rolland is also helping with translations, espe ially in times of risis! I thank our
past CRUX editor BILL SANDS for his proof reading and sound advi e.
My olleagues in the Dept. of Mathemati s and Statisti s have lent their support. Those who have taken pity on this Editor-in-Chief are ANNA STOKKE, ROSS
STOKKE, TERRY VISENTIN, and JEFF BABB. Our se retary, JULIE BEAVER, has also
helped out in a pin h as little emergen ies have arisen. I thank the previous Dean
of S ien e, GABOR KUNSTATTER, for his understanding and foresight in supporting
CRUX with MAYHEM at the University of Winnipeg.
The LATEX expertise of JOANNE CANAPE at the University of Calgary, and TAO
GONG and JUNE ALEONG at Wilfrid Laurier University goes a long way to produ ing
high quality opy. Joanne is nishing up her work in this area, so after my one year
of servi e here I thank her for giving twenty! I wel ome JILL AINSWORTH on board
who is taking over from Joanne to prepare the Olympiads.
Thanks go to the University of Toronto Press and to Thistle Printing, in parti ular TAMI EHRLICH. The quality nished opy and the purple overs are just lovely.
I thank you MICHAEL DOOB and CRAIG PLATT for te hni al support, and JUDI
BORWEIN for putting CRUX on the net.
Someone spe ial who has helped me through my rst year with her areful proof
reading and knowledge of geometry is CHARLENE PAWLUCK. Thank you for sharing
your opy of Eu lid with me, and so mu h more.
Finally, I thank you, the readers, for all that you have done for me and for the
journal. CRUX with MAYHEM is built from your ontributions and all the time, are
and reativity that you put into your submissions is re e ted in these pages. I wish I
ould mention all the truly marvellous people that I have gotten to know in the last
year, but this margin is too small to hold all the names and praises.
I lose with a all for a new Skoliad Editor. Please refer anyone to us you may
think is suitable. Skoliad is missing from this issue, but will be ba k next year; from
now on please send all Skoliad materials dire tly to the Editor-in-Chief (or resend
your past submissions if you did not re eive a reply from us).
I wish ea h of you the very best in 2009 in all areas of life,
Va lav

(Vazz) Linek
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
508
INDEX TO VOLUME 34, 2008

Contributor Proles
February
Peter Y. Woo ................................................... 1
Skoliad Robert Bilinski
February
No. 107 ......................................................... 2
Mar h
No. 108 ....................................................... 65
April
No. 109 ...................................................... 129
May
No. 110 ...................................................... 195
September No. 111 ...................................................... 257
O tober
No. 112 ...................................................... 321
November No. 113 ...................................................... 385
Mathemati al Mayhem Ian VanderBurgh
February
.................................................................. 8
Mar h
................................................................. 69
April
............................................................... 136
May
............................................................... 200
September ............................................................... 266
O tober
............................................................... 327
November ................................................................ 396
De ember
............................................................... 449
Mayhem Problems
February
M326{M331 .................................................... 8
Mar h
M332{M337 ................................................... 69
April
M338{M343 ................................................. 136
May
M344{M349 ................................................. 200
September M350{M356 ................................................. 266
O tober
M357{M362 ................................................. 327
November M363{M368 ................................................. 396
De ember M369{M375 ................................................. 449
Mayhem Solutions
February
M276{M281 ................................................... 10
Mar h
M282{M287 ................................................... 71
April
M288{M293 ................................................. 138
May
M294{M300 ................................................. 202
September M301{M312 ................................................. 268
O tober
M313{M324 ................................................. 329
November M325{M331 ................................................. 398
De ember M332{M337 ................................................. 451
Problem of the Month Ian VanderBurgh
February
................................................................. 15
Mar h
................................................................. 76
April
............................................................... 144
May
............................................................... 208
September ............................................................... 279
O tober
............................................................... 338
November ................................................................ 404
De ember
............................................................... 457
509
Mayhem Arti les
Adding Up
Bru e Shawyer ............................................................ 406
The Olympiad Corner R.E. Woodrow
February
No. 267 ....................................................... 18
Mar h
No. 268 ....................................................... 79
April
No. 269 ...................................................... 147
May
No. 270 ...................................................... 211
September No. 271 ...................................................... 282
O tober
No. 272 ...................................................... 341
November No. 273 ...................................................... 408
De ember No. 274 ...................................................... 459
Book Reviews John Grant M Loughlin
How Euler Did It
by C. Edward Sandifer
Reviewed by J. Chris Fisher ............................................... 34
Nonplussed! Mathemati al Proof of Implausible Ideas
by Julian Havil
Reviewed by Robert D. Poodia k .......................................... 36
Math Through the Ages: A Gentle History for Tea hers and Others
by William P. Berlingho and Fernando Q. Gouvêa
Reviewed by John Grant M Loughlin....................................... 95
Minnesota Math League XXV 1980{2005
by A. Wayne Roberts
Reviewed by Robert L. Crane .............................................. 96
The Magi Numbers of the Professor
by Owen O'Shea and Underwood Dudley
Reviewed by Je Hooper .................................................. 161
Geometri Puzzle Design
by Stewart Con
Reviewed by Jim Totten ................................................... 163
The Liar Paradox and the Towers of Hanoi:
the Ten Greatest Math Puzzles of All Time
by Mar el Danesi
Reviewed by Amar Sodhi .................................................. 164
Problems of the Week
by Jim Totten
Reviewed by John Grant M Loughlin ..................................... 229
Digital Di e
by Paul J. Nahin
Reviewed by Amar Sodhi .................................................. 297
Cal ulus Gems:
Brief Lives and Memorable Mathemati s
by George F. Simmons
Reviewed by Robert D. Poodia k ......................................... 356
From Zero to Innity:
What Makes Numbers Interesting
by Constan e Reid
Reviewed by John Grant M Loughlin ..................................... 357
510
Impossible? Surprising Solutions and Counterintuitive Conundrums
by Julian Havil
Reviewed by Ed Barbeau .................................................. 422
The Symmetries of Things
by John H. Conway, Heidi Burgiel, and Chaim Goodman-Strauss
Reviewed by J. Chris Fisher ............................................... 468
Crux Arti les James Currie
A Useful Inequality
Roy Barbara ............................................................... 38
Sharpening the Hadwiger-Finsler Inequality
Cezar Lupu and Cosmin Pohoata .......................................... 97
Industrial Grade Primes with a Money-Ba k Guarantee
Mi hael P. Abramson .................................................... 165
The Proof of Three Open Inequalities
Vasile C^rtoaje ............................................................ 231
The Sum of a Cube and a Fourth Power
Thomas Mauts h and Gerhard J. Woeginger ............................. 358
Twin Problems on Non-Periodi Fun tions
Eugen J. Ionas u .......................................................... 424
Old Idaho Usual Here
Robert Israel, Stephen Morris, and Stan Wagon ......................... 471
A Limit of an Improper Integral Depending on One Parameter
Iesus C. Diniz ............................................................. 478
Sliding Down In lines with Fixed Des ent Time:
a Converse to Galileo's Law of Chords
Je Babb .................................................................. 481
Problems
February
3301{3312 ..................................................... 44
Mar h
3313{3325 .................................................... 102
April
3326{3337 .................................................... 170
May
3282, 3338{3350 ............................................. 239
September 3351{3362 .................................................... 298
O tober
3363{3375 .................................................... 362
November 3376{3388 .................................................... 430
De ember 3371, 3389{3400 .............................................. 483
Solutions
February
3201{3213 ..................................................... 49
Mar h
3214{3225, 3227 ............................................. 107
April
3226, 3228{3238, 3242 ....................................... 176
May
Klamkin{05, 3239{3241, 3243{3250 ......................... 244
September 3251{3362 .................................................... 303
O tober
3263{3275, 3282 ............................................. 367
November 3276{3281, 3283{3288 ....................................... 435
De ember 3229, 3289{3300 ............................................. 488
Mis ellaneous
Editorial .................................................................... 193
In Memoriam: James Edward Totten ...................................... 194
MAA Book Promotion .......................................................505
Year End Finale ............................................................. 506
511
Proposers and solvers appearing in the SOLUTIONS se tion in 2008:
Proposers
Anonymous Proposer 3211
Gregory Akulov 3285
Arkady Alt 3300
Mi hel Bataille 3217, 3237, 3238, 3251, 3252, 3266, 3267, 3295, 3296
Mihaly Ben ze 3204, 3205, 3206, 3213, 3214, 3216, 3228, 3229, 3230,
3239, 3240, 3253, 3292, 3293, 3294
K.S. Bhanu 3284

Alper Cay 3258
Vasile C^rtoaje 3209, 3210, 3274, 3275, Klamkin{05,
M.N. Deshpande 3283, 3284
Jose Luis Daz-Barrero 3212, 3247, 3263, 3269, 3281, 3282
Marian Din a 3205
J. Chris Fisher 3224
Ovidiu Furdui 3218, 3226, 3227, 3261, 3262, 3277, 3288
G.P. Henderson 3201
Ignotus 3231
Neven Juri 3259, 3276, 3286
Georey A. Kandall 3235
Va lav Kone ny 3256
Tidor Mitev 3236
Virgil Ni ula 3242, 3241, 3260, 3264, 3265, 3270, 3271, 3273, 3278, 3279,
Vi tor Oxman 3299

Fran is o Pala ios Qui~nonero 3212
Pantelimon George Popes u 3269, 3281, 3282
A hilleas Pavlos Poryfyriadis 3223
D.E. Prithwijit 3272
Stanley Rabinowitz 3297, 3298
Juan-Bos o Romero Marquez 3221
Bill Sands 3257, 3268
D.J. Smeenk 3202, 3203, 3250
Marian Tetiva 3220, 3246
George Tsapakidis 3225
G. Tsintsifas 3232, 3233, 3234, 3243, 3244, 3254, 3255
Pham Van Thuan 3222
Dan Vetter 3219
Harley Weston 3224
Shaun White 3208, 3215
John Wiest 3268
Peter Y. Woo 3207
Paul Yiu 3245
Titu Zvonaru 3248, 3249
3280, 3287, 3289, 3290, 3291
Featured Solvers | Individuals

Mohammed Aassila 3248, 3249
Arkady Alt 3241, 3263, 3276

Sefket
Arslanagi 3249, 3253, 3257, 3264
Dionne Bailey 3228, 3237, 3247, 3269, 3271
Roy Barbara 3214, 3232, 3235
Ri ardo Barroso Campos 3267
Mi hel Bataille 3204, 3205, 3209, 3213, 3217,
3222, 3227, 3229, 3232,
3245, 3249, 3256, 3257, 3264, 3269, 3272, 3273, 3276, 3278, 3281, 3283,
3285, 3286, 3288, 3291, 3293, 3299
Brian D. Beasley 3220

Mihaly Ben ze 3230, 3253
Manuel Benito 3253, 3261, 3262
Elsie Campbell 3228, 3237, 3247, 3269, 3271
Cao Minh Quang 3230, 3253, 3281, 3300
s ar Ciaurri Ramrez 3253, 3261, 3262
O
Vasile C^rtoaje Klamkin{05, 3275
Chip Curtis 3204, 3221, 3226, 3234, 3261, 3274, 3276, 3294
Apostolis K. Demis 3211, 3219, 3224, 3255, 3273
Charles R. Diminnie 3221, 3228, 3237, 3247, 3269, 3271
Emilio Fernandez Moral 3253, 3261, 3262
J. Chris Fisher 3238
Fran is o Javier Gar a Capita n 3221
Oliver Geupel 3263, 3273, 3282, 3284, 3292
Douglass L. Grant 3293
P.C. Hammer 3254
Karl Havlak 3271
John Hawkins 3240
Ri hard I. Hess 3221, 3232, 3283, 3286
John G. Heuver 3245, 3289
Joe Howard 3287
Walther Janous 3245
Georey A. Kandall 3221
Paula Ko a 3271
Va lav Kone ny 3202, 3232, 3244
Kee-Wai Lau 3242, 3239, 3246, 3298
Tom Leong 3253, 3260
Nguyen Thanh Liem 3249
Tai hi Maekawa 3250, 3279

Thanos Magkos 3276
Salem Maliki 3214, 3220, 3234, 3245, 3249, 3280
Dragoljub Milosevi 3214
Andrea Munaro 3253, 3265, 3280
Vedula N. Murty 3221
Jose H. Nieto 3201, 3202, 3257, 3259
Mi hael Parmenter 3252, 3284
Alex Remorov 3249
Henry Ri ardo 3204,
Juan-Bos o Romero Marquez 3221
Luz Ron al 3261, 3262
Xavier Ros 3212, 3221, 3236, 3250, 3257
Sergey Sadov 3229
Bill Sands 3268
Joel S hlosberg 3215, 3233, 3290
D.J. Smeenk 3202, 3221, 3235, 3264, 3279, 3297
T. Jeerson Smith 3254
David Stone 3240
Edmund Swylan 3202, 3206, 3207, 3238
Marian Tetiva 3220
Phi Thai Thuan 3249
Daniel Tsai 3251
Panos E. Tsaoussoglou 3249
George Tsapakidis 3289
G. Tsintsifas 3232
Apostolis Vergos 3208
Vo Quo Ba Can 3210, 3216, 3223
Edward T.H. Wang 3253
Shaun White 3208
Peter Y. Woo 3203, 3205, 3218, 3221, 3225, 3244, 3245, 3295, 3296
Paul Yiu 3245
Roger Zarnowski 3237
Titu Zvonaru 3243, 3270
Featured Solvers | Groups

Missouri State University Problem Solving Group 3259, 3266
Skidmore College Problem Solving Group 3245
512
Other Solvers | Individuals
Mohammed Aassila 3223, 3237, 3241, 3281, 3283
Gregory Akulov 3285
Arkady Alt 3221, 3222, 3223, 3226, 3227, 3230, 3236, 3246, 3247, 3248,
3249, 3281, 3287, 3300
Miguel Amengual Covas 3203, 3223, 3250

Christos Anastassiades 3250
George Apostolopoulos 3264, 3265, 3270, 3276, 3278, 3284,
3285, 3286,
3287, 3289, 3291, 3293, 3298

Sefket
Arslanagi 3203, 3206, 3221, 3222, 3223, 3248, 3249, 3250, 3251,
3256, 3260, 3265, 3270, 3271, 3272, 3276, 3281, 3284, 3285, 3287, 3289,
3293, 3294, 3297, 3298, 3300
Dionne T. Bailey
Kee-Wai Lau 3208, 3222, 3223, 3236, 3237, 3241, 3247, 3249, 3248, 3250,
3253, 3260, 3271, 3285, 3286, 3292, 3297
Tom Leong 3251, 3257

Kathleen E. Lewis 3257
Carl Libis 3276
Sotiris Louridas 3298
Tai hi Maekawa 3250
Thanos Magkos 3221, 3260, 3286, 3287, 3297
Salem Maliki 3202, 3203, 3207, 3208, 3211, 3216, 3221, 3222, 3223, 3230,
3233, 3235, 3236, 3237, 3240, 3241, 3243, 3244, 3246, 3247, 3248, 3250,
3251, 3253, 3256, 3257, 3259, 3260, 3266, 3270, 3271, 3272, 3276, 3281,
3222, 3223, 3251, 3253, 3263, 3276, 3285, 3286, 3287,
3284, 3287, 3293, 3297
3230, 3233, 3234, 3235, 3236, 3237, 3238, 3242, 3237, 3238, 3239, 3243,
Dung Nguyen Manh 3281, 3287

Pavlos Maragoudakis 3221, 3251
Petrus Martins 3253
G. Milanova 3221, 3222, 3223
D.M. Milosevi 3222
Dragoljub Milosevi 3221, 3222, 3223
Todor Mitev 3236
Andrea Munaro 3235, 3236, 3251, 3256,
3244, 3246, 3247, 3248, 3250, 3251, 3252, 3253, 3255, 3262, 3263, 3265,
3294
3293
Roy Barbara 3201, 3202, 3203, 3213, 3221, 3222, 3223, 3233, 3234, 3236,
3237, 3238, 3242, 3241, 3243, 3244, 3247, 3250, 3251, 3252, 3257, 3264,
3265, 3266, 3270, 3271, 3272, 3278, 3285, 3286, 3287, 3290, 3291, 3293,
3294, 3297, 3298, 3300
Ri ardo Barroso Campos 3203, 3250, 3279, 3289,

Mi hel Bataille 3201, 3202, 3206, 3207, 3208, 3211,
3221, 3223, 3228,
3266, 3267, 3270, 3271, 3279, 3280, 3282, 3284, 3287, 3289, 3290, 3292,
3294, 3295, 3296, 3297, 3298
Brian D. Beasley 3294, 3297

Fran is o Bellot Rosado 3202, 3203
Mihaly Ben ze 3204, 3205, 3213, 3214,
3280, 3287, 3289, 3290, 3291

3216, 3223, 3228, 3229, 3239,
3247, 3249, 3269, 3292, 3293, 3294
Manuel Benito 3251, 3255, 3257, 3258

K.S. Bhanu 3284
Paul Bra ken 3222, 3262, 3286
Katrina Bri ker 3221
Whitney Bullo k 3257
Sarah Burnham 3271
Elsie M. Campbell 3212, 3222, 3223, 3251, 3253, 3263, 3276, 3285, 3286,
3287, 3293
Cao Minh Quang 3212, 3214,
3264, 3276, 3278, 3287, 3289,
Vedula N. Murty 3214, 3221, 3222, 3223

N. Nadeau 3262
Virgil Ni ula 3241, 3242, 3260, 3264, 3265, 3270, 3271, 3273, 3278, 3279,
3221, 3222, 3223, 3235, 3236, 3242, 3246,
Jose H. Nieto 3205, 3221, 3237, 3248, 3251, 3252, 3253, 3258
Vi tor Oxman 3299
Fran is o Pala ios Qui~nonero 3212
Mi hael Parmenter 3237, 3238, 3250, 3287, 3294
Paolo Perfetti 3281
Pantelimon George Popes u 3269, 3281
A hilleas Pavlos Porfyriadis 3223
D.E. Prithwijit 3272
Stanley Rabinowitz 3297, 3298
Alex Remorov 3226, 3229, 3230, 3234, 3235, 3236, 3237, 3238, 3241, 3243,
3244, 3246, 3247, 3248, 3250
s ar Ciaurri Ramrez 3251, 3255, 3257, 3258

O
Vasile C^rtoaje 3209, 3210, 3274, 3300
Chip Curtis 3202, 3203, 3206, 3216, 3217, 3219, 3222, 3223, 3227, 3230,
Henry Ri ardo 3204, 3212, 3247, 3293

Juan-Bos o Romero Marquez 3221, 3223, 3297
Luz Ron al 3251, 3255, 3257, 3258
Xavier Ros 3227, 3237, 3247, 3248, 3249, 3251, 3253, 3259, 3260, 3262,
3235, 3237, 3238, 3239, 3241, 3243, 3244, 3245, 3246, 3247, 3248, 3251,
3281, 3286, 3287
3247, 3248, 3249, 3250, 3251, 3260, 3265, 3270, 3276, 3287
3252, 3256, 3257, 3259, 3262, 3264, 3265, 3266, 3269, 3270, 3271, 3275,
3278, 3281, 3283, 3284, 3286, 3287, 3289, 3290, 3291, 3292, 3293, 3296,
3297, 3298, 3299, 3300
Paul Deiermann 3284, 3285

Apostolis K. Demis 3202, 3203, 3205, 3207, 3213, 3214, 3221, 3233, 3234,
3235, 3238, 3251, 3256, 3258, 3264, 3265, 3267, 3270, 3289, 3292, 3297,
Deborah Salas-Smith 3257

Bill Sands 3257
Robert P. Sealy 3286
Joel S hlosberg 3202, 3203,
3204, 3205, 3207, 3211, 3218, 3221, 3227,
3235, 3238, 3244, 3250, 3253, 3255, 3265, 3266, 3270, 3271, 3272, 3276,
3278, 3283, 3285, 3287, 3292, 3293, 3297
Hasan Denker 3222, 3223

Nikolaos Dergiades 3300
M.N. Deshpande 3284
Jose Luis Daz-Barrero 3212, 3247, 3263, 3269, 3281, 3286, 3287
Charles R. Diminnie 3212, 3222, 3223, 3250, 3251, 3253, 3263, 3276, 3285,
Bob Serkey 3287

Andrew Siefker 3253
D.J. Smeenk 3203, 3221, 3223, 3233, 3234, 3244, 3250, 3265, 3278, 3287
Digby Smith 3286
David R. Stone 3212, 3237, 3240, 3241, 3257
Edmund Swylan 3203, 3217, 3221, 3234, 3236, 3237, 3243, 3244, 3250,
3286, 3287, 3293
3257
3298
Marian Din a 3205

Oleh Faynshteyn 3293, 3297, 3298
Emilio Fernandez Moral 3251, 3255, 3257, 3258
Ovidiu Furdui 3218, 3226, 3227, 3261, 3262, 3288
Ian June L. Gar es 3251
Fran is o Javier Gar a Capita n 3202, 3290, 3291
Oliver Geupel 3257, 3264, 3266, 3268, 3276, 3278, 3279, 3281, 3283, 3286,
3287, 3289, 3290, 3293, 3294, 3295, 3296, 3297, 3298, 3299, 3300
Karl Havlak 3253, 3276, 3285, 3286, 3287, 3293

John Hawkins 3212, 3237, 3240, 3241, 3257
G.P. Henderson 3201
Ri hard I. Hess 3203, 3206, 3208, 3219, 3221,
3223, 3236, 3248, 3250, 3287, 3297,
3298, 3300
George Tsapakidis 3222, 3223, 3265, 3270, 3276, 3286, 3287, 3291, 3293,
3297
3222, 3227, 3250, 3251,
3253, 3262, 3265, 3270, 3276, 3287, 3292, 3294, 3296, 3297, 3298
John G. Heuver 3203, 3221, 3256

Joe Howard 3204, 3212, 3221, 3222, 3223, 3227, 3235, 3236, 3237, 3246,
3247, 3248, 3251, 3260, 3271, 3276, 3281
Peter Hurthig 3250

Walther Janous 3239,
Alexandros Sygelakis 3209, 3212

Winfer C. Tabares 3251
Ashley Tangeman 3253
Marian Tetiva 3246
Daniel Tsai 3203, 3286
Panos E. Tsaoussoglou 3203, 3222,
3241, 3246, 3247, 3248, 3249, 3250, 3251, 3253,
G. Tsintsifas 3233, 3234, 3243, 3244, 3254, 3255

Pham Van Thuan 3222
Phi Thai Thuan 3236, 3248
Apostolis Vergos 3212, 3248
Dan Vetter 3219
Vo Quo Ba Can 3214, 3222
Stan Wagon 3208, 3222, 3223, 3292, 3300
Peter Y. Woo 3201, 3202, 3207, 3211, 3212, 3213, 3214, 3220, 3222, 3223,
3225, 3233, 3234, 3235, 3236, 3238, 3241, 3243, 3246, 3247, 3248, 3250,
3257, 3260, 3261, 3262, 3263, 3265, 3266, 3267, 3269, 3270, 3271, 3272,
3251, 3252, 3253, 3254, 3255, 3256, 3258, 3259, 3262, 3264, 3265, 3267,
3273, 3274, 3291, 3292, 3294, 3297, 3300
3270, 3271, 3273, 3276, 3278, 3279, 3280, 3281, 3284, 3285, 3286, 3287,
Neven Juri 3259, 3276, 3286

Natalie Kalmink 3221
Georey A. Kandall 3202, 3235, 3250, 3289
Paula Ko a 3253, 3276, 3285, 3286
Va lav Kone ny 3203, 3221, 3234, 3235, 3243,
3289, 3290, 3291, 3292, 3293, 3294, 3297, 3298, 3299
Bingjie Wu 3257, 3272, 3281, 3287

Konstantine Zelator 3203, 3265, 3266, 3272, 3294, 3297, 3298
Titu Zvonaru 3202, 3203, 3221, 3222, 3223, 3233, 3234, 3235, 3236, 3237,
3250, 3256, 3258, 3291,
3297, 3298
3238, 3244, 3246, 3247, 3249, 3250, 3263, 3264, 3265, 3266, 3267, 3271,
3272, 3281, 3287, 3289, 3290, 3291, 3293, 3294, 3297, 3298, 3300
Other Solvers | Groups

Ateneo Problem Solving Group 3221, 3251
Missouri State Univeristy Problem Solving Group 3253, 3257, 3272
Skidmore College Problem Solving Group 3221, 3237, 3291, 3294, 3297
University of Regina Math Central Consultants 3211

CRUXv 34 N 8

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 34 N 8

Uploaded by

Copyright:

Available Formats

449

as an integral part of Crux Mathemati orum with Mathemati al Mayhem.

Veuillez nous transmettre vos solutions aux problemes

Soit x un nombre reel

M373. Propose par Kunal Singh, etudiant,

M374. Propose par Mihaly Ben ze, Brasov, Roumanie.

Soit p un nombre premier xe,

M369. Proposed by the Mayhem Sta .

M370. Proposed by the Mayhem Sta .

for all angles

M371. Proposed by Panagiote Ligouras, Leonardo da Vin i High S hool,

M372. Proposed by the Mayhem Sta .

Determine integers a, b, and c

M373. Proposed by Kunal Singh, student, Kendriya Vidyalaya S hool,

M374. Proposed by Mihaly Ben ze, Brasov, Romania.

M375. Proposed by Ne ulai Stan iu, Saint Mu eni Sava Te hnologi al

Proposed by Dionne Bailey, Elsie Campbell, and Charles R.

Thus, the smallest possible volume for the ylinder is 3456.

M333. Proposed by the Mayhem Sta .

M334. Proposed by the Mayhem Sta .

(b) Determine all integers y for whi h

I. Solution by Edin Ajanovi , student, First Bosniak High S hool, Sarajevo,

Bosnia and Herzegovina.

Let A be an integer su h that A =

Then 3A is an integer and

 \Abastos", Valen ia, Spain; CAO MINH QUANG,

M335. Proposed by the Mayhem sta .

M336. Proposed by the Mayhem Sta .

(x, y) = (0, 0).

has one integer solution only, namely

Consider next |x| + |y| = k, for an integer k with 1 k n. We an

Case 1. The integers x and y satisfy x + y = k, where x 0 and y 0.

Therefore, the number of latti e points inside the region is

 \Abastos", Valen ia, Spain. There were one in orre t

M337. Proposed by the Mayhem Sta .

determine EF in terms of x and y.

Solution by Kunal Singh, student, Kendriya Vidyalaya S hool, Shillong, India.

In part (a), this yields

Sin e the 0 is in row 2 and olumn 3, then 0 = b + C . Similarly,

THE OLYMPIAD CORNER

47th INTERNATIONAL MATHEMATICAL OLYMPIAD

Problems Proposed But Not Used

Contributing Countries. Argentina, Australia, Brazil, Bulgaria, Canada,

where ai is the greatest integer not ex eeding ai , and {ai }

and de ne the sequen e of real numbers a0 , a1 , a2 , . . .

for n 1. Show that an > 0 for n 1.

A3. Let c0 = 1, c1 = 0 and de ne the sequen e c0 , c1 , c2 , . . . by the

if and only if (x, y) S . (By onvention an empty sum is 0.)

A4. Let a1 , a2 , . . . , an be positive real numbers. Prove that

C3. A ake is shaped as an n n square with n2 unit squares. Strawberries

C4. An (n, k)-tournament is a ompetition with n players held in k rounds

C5. A holey triangle is an upward equilateral triangle of side length n with

upward unit triangular holes ut out. A diamond is a unit rhombus with

G1. Let ABCD be a trapezoid with

AB||CD and AB > CD . Points

G2. Let ABCDE be a onvex pentagon su h that

The diagonals BD and CE interse t at

G3. A point D is hosen on the side

of a triangle ABC with

ACB < BAC < 90

in su h a way that BD = BA. The

G4. In triangle ABC , let J be the

entre of the ex ir le tangent to side

M373. Propose par Kunal Singh, etudiant,

M374. Propose par Mihaly Ben ze, Brasov, Roumanie.

M369. Proposed by the Mayhem Sta.

M370. Proposed by the Mayhem Sta.

M372. Proposed by the Mayhem Sta.

M374. Proposed by Mihaly Ben ze, Brasov, Romania.

M333. Proposed by the Mayhem Sta.

M334. Proposed by the Mayhem Sta.

\Abastos", Valen ia, Spain; CAO MINH QUANG,

M335. Proposed by the Mayhem sta.

M336. Proposed by the Mayhem Sta.

\Abastos", Valen ia, Spain. There were one in orre t

M337. Proposed by the Mayhem Sta.

and dene the sequen e of real numbers a0 , a1 , a2 , . . .

A3. Let c0 = 1, c1 = 0 and dene the sequen e c0 , c1 , c2 , . . . by the

(a) Prove that f (n + 1) > f (n) for innitely many n.

(b) Prove that f (n + 1) < f (n) for innitely many n.

be ause K must be 0. The expression simplies to J A modulo 9, whi h

For d = 7 the word w of the Proposition has length 55. A dierent